Crystallization of antibody or antigen-binding fragments

Abstract

This article provides methods and compositions for improving the crystallization of antibodies, antigen-binding fragments (e.g., Fab or Fab') or fusion proteins and Fab / Fab' / mAb:antigen complexes.

Description

Invention Field

[0001] This disclosure relates to methods and compositions for improving the crystallization of monoclonal antibodies (mAbs) or antigen-binding fragments (e.g., Fab or Fab') or fusion proteins and Fab / Fab' / mAb:antigen complexes.

[0002] background

[0003] Antibody therapies are one of the most important drug categories. By the end of 2019, 90 monoclonal antibody drugs for treating immune diseases, infectious diseases, cardiovascular diseases, cancer, etc. had been approved in the United States and Europe, with sales of $115 billion in 2018 (Kaplon et al., Mabs, 2020. 12(1):1703531; Lu et al. J Biomed Sci. 2020; 27:1). Although rodent antibodies were once developed for human use, followed by a long period of humanization, in the last two decades they have transitioned to fully human discovery platforms, such as phage and yeast display (Parmley and Smith, Gene, 1988. 73(2):305-18; Boder and Wittrup, Nat. Biotechnol., 1997. 15(6):553-7) or by immunizing rodents with human germline banks (Lonberg, Handb. Exp. Pharmacol., 2008.(181):69-97). In these platforms, engineering is not essential for humanization but continues to be used to address other issues: affinity, cross-reactivity, post-translational modifications, hydrophobicity, electrostatics, viscosity, and immunogenicity. Furthermore, antibody characterization continues to become more complex, especially with the development of new antibody-derived forms, such as antibody-drug conjugates and bispecific antibodies (Carter and Lazar, Nat. Rev. Drug Disc., 2018. 17(3)197-223).

[0004] Modeling antibody structures has become an integral part of predicting the behavior of potential therapeutic agents, particularly for properties such as hydrophobicity, stability, charge / dipole moment, and deamidation tendency (Xu Y et al., Mabs, 2019. 11(2)239-264). This modeling is typically based on publicly available crystal structures. Due to the difficulty in modeling CDR structures, especially heavy chain CDR3, modeling crystal structures with highly similar (or identical) Fab crystal structures should improve the accuracy of predicted antibody properties. Available high-resolution Fab structures will provide the best basis for calculations.

[0005] The structure of Fab:antigen complexes has even greater value. They can provide information on the crystal structure of Fab and epitope:complementary site information. Obtaining the structure of a Fab:antigen complex is the only method to directly determine the relative 3-dimensional positions of the antigen and Fab to a single atom precision. Epitopes can be seen, rather than inferred. Amino acid side chains can be examined, and hypotheses can be formed about their roles in affinity and cross-reactivity, and calculations can be performed to predict and engineer affinity (up or down) and cross-reactivity. Furthermore, the structure of a Fab:antigen complex can serve as a reference when considering other mutations and can be an important cross-check in determining the effectiveness of different assay forms.

[0006] However, crystal structure determination is challenging and costly (Slabinski et al., Protein Sci, 2007.16(11):2472-82). The most difficult step is often generating ordered crystals from purified proteins. Many factors can hinder protein crystallization: purity, stability, disorder (between domains, rings, or terminals), surface charge, and hydrophobicity, among others (McPherson, Crystallization of biological macromolecules, 1999. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY). Obtaining good diffraction crystals can take days or years, or may be abandoned simply after tremendous effort.

[0007] Although crystallization and structural determination of Fabs are relatively easy compared to many other protein classes, particularly membrane proteins, extensive screening and optimization of crystallization methods remain essential. Like other proteins, some Fabs require significant crystallization optimization, and instances of completely intractable Fabs exist. Fab:antigen complexes are often easier to crystallize than individual antigens (hence the use of Fabs as “crystallization partners”) (Griffin et al., Clin. Exp. Immunol., 2011. 165(3):285-91), but can still be difficult and require extensive screening and optimization. The individual attention required for crystallization and structural refinement, ranging from days to months, makes these two steps the most costly in the sequence-to-final structure process. Difficult cases are particularly and negatively impact the overall average. The effort required for Fab crystallization may explain why so few structures have been used for engineering or computation.

[0008] Edmundson and Borrebaeck examined the crystal stacking interactions and β-sheet formation between the light chain constant domain (CL) and the heavy chain first constant domain (CH1) of human Fab and observed the importance of the “stacked triplet” (three alternating residues that tend to form β-sheets). (Edmundson and Borrebaeck, Immunotechnology, 1998.3(4):309-17).

[0009] There is a need to improve the crystallization of Fab (e.g., human Fab) and / or human Fab:antigen complexes containing the human CL domain of the κ chain.

[0010] Detailed Explanation

[0011] This document provides methods and compositions for significantly improving the crystallization of human antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins containing the human antibody light chain κ constant domain (Cκ), and Fab / Fab' / mAb:antigen complexes. The methods and compositions described herein are generally applicable to most human Fabs, some human mAbs, and fusion proteins containing human Fab or mAbs that include human Cκ, and significantly improve their crystallization, including making them more likely to crystallize faster, at higher resolution, and at lower concentrations and / or from heterogeneous mixtures.

[0012] In one aspect, this document provides antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising variant Cκ, wherein the variant Cκ comprises amino acid QGTTS at positions 199 to 203 of human Cκ (positions numbered according to Kabat notation corresponding to positions 92 to 96 of SEQ ID NO:1), and amino acids at positions 198 and 204 of human Cκ (positions numbered according to Kabat notation corresponding to positions 91 and 97 of SEQ ID NO:1) are missing in the variant Cκ. In some embodiments, the variant Cκ further comprises alanine at position 126 of human Cκ (position numbered according to Kabat notation corresponding to position 19 of SEQ ID NO:1). In some embodiments, the variant Cκ further comprises proline at position 214 of human Cκ (position numbered according to Kabat notation corresponding to position 107 of SEQ ID NO:1).

[0013] Unless otherwise specified, the amino acid residues in the antibodies or antigen-binding fragments described herein are numbered according to the Kabat numbering system (Kabat et al., Sequences of Proteins of Immunological Interest, 5th ed., Bethesda, MD: US Dept. of Health and Human Services, Public Health Service, National Institutes of Health, 1991).

[0014] In some embodiments, this document provides antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising a variant Cκ domain, wherein the variant Cκ domain comprises SEQ ID NO:3. In some embodiments, this document provides antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising a variant Cκ domain, wherein the variant Cκ domain comprises SEQ ID NO:4. In some embodiments, this document provides antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising a variant Cκ domain, wherein the variant Cκ domain comprises SEQ ID NO:5. In some embodiments, this document provides antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising a variant Cκ domain, wherein the variant Cκ domain comprises SEQ ID NO:6.

[0015] In some embodiments, the antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein further comprises a human light chain variable domain (VL) and a human heavy chain variable domain (VH). In some embodiments, the antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein further comprises a human IgG CH1 domain, such as a human IgG1 or IgG4 CH1 domain. In some embodiments, the antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein further comprises a portion of a human IgG hinge region (e.g., a human IgG1 or IgG4 hinge region).

[0016] In another aspect, this article provides libraries of antibodies, antigen-binding fragments (e.g., human Fab or Fab') or fusion proteins containing variant Cκ, wherein the variant Cκ contains amino acid QGTTS at positions 199 to 203 of human Cκ, and amino acids at positions 198 and 204 of human Cκ are missing in the variant Cκ (all positions are numbered according to the Kabat numbering system).

[0017] In another aspect, this document provides methods for generating crystal structures of antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising variant Cκ as described herein. Such methods may include crystallizing antibodies, antigen-binding fragments, or fusion proteins comprising variant Cκ, wherein the variant Cκ comprises amino acid QGTTS at positions 199 to 203 of human Cκ, and amino acids at positions 198 and 204 of human Cκ are missing in the variant Cκ (all positions are numbered according to the Kabat numbering system). In some embodiments, the methods further comprise constructing a library of antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins comprising variant Cκ as described herein. For example, the variant Cκ may comprise amino acid QGTTS at positions 199 to 203 of human Cκ, and amino acids at positions 198 and 204 of human Cκ are missing in the variant Cκ (all positions are numbered according to the Kabat numbering system). In some embodiments, the variant Cκ comprises SEQ ID NO:3. In some embodiments, the variant Cκ comprises SEQ ID NO:4. In some embodiments, the variant Cκ includes SEQ ID NO:5. In some embodiments, the variant Cκ includes SEQ ID NO:6.

[0018] This document also provides methods for generating crystal structures of antibodies, antigen-binding fragments (e.g., human Fab or Fab'), or fusion proteins containing a human Cκ domain. Such methods may include: generating an antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein containing a variant Cκ, wherein the variant Cκ contains amino acids QGTTS at positions 199 to 203 of human Cκ, and amino acids at positions 198 and 204 of human Cκ are missing in the variant Cκ (all positions are numbered according to Kabat numbering); and crystallizing the antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein containing the variant Cκ. In some embodiments, the variant Cκ further contains alanine at position 126 (according to Kabat numbering) of human Cκ. In some embodiments, the variant Cκ further contains proline at position 214 (according to Kabat numbering) of human Cκ. In some embodiments, the variant Cκ contains SEQ ID NO:3. In some embodiments, the variant Cκ contains SEQ ID NO:4. In some embodiments, the variant Cκ includes SEQ ID NO:5. In some embodiments, the variant Cκ includes SEQ ID NO:6.

[0019] In another aspect, this document provides a method for generating crystal structures of complexes of antigens and antibodies, antigen-binding fragments, or fusion proteins (e.g., human Fab or Fab') binding to said antigens, wherein said antibodies, antigen-binding fragments (e.g., human Fab or Fab') or fusion proteins comprise human Cκ, the method comprising: generating an antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein comprising a variant Cκ, wherein said variant Cκ comprises amino acids QGTTS at positions 199 to 203 of human Cκ, and amino acids at positions 198 and 204 of human Cκ are missing in the variant Cκ (all positions are numbered according to the Kabat numbering system); and co-crystallizing the antigen and the antibody, antigen-binding fragment (e.g., human Fab or Fab'), or fusion protein comprising the variant Cκ. In some embodiments, said variant Cκ further comprises alanine at position 126 (according to the Kabat numbering system) of human Cκ. In some embodiments, said variant Cκ further comprises proline at position 214 (according to the Kabat numbering system) of human Cκ. In some embodiments, variant Cκ includes SEQ ID NO:3. In some embodiments, variant Cκ includes SEQ ID NO:4. In some embodiments, variant Cκ includes SEQ ID NO:5. In some embodiments, variant Cκ includes SEQ ID NO:6.

[0020] As used herein, the terms “a,” “an,” “the,” and similar terms used in the context of this disclosure (particularly in the context of the claims) should be interpreted to cover both the singular and the plural, unless otherwise indicated herein or clearly contradicted by the context.

[0021] As used herein, the term "antibody" refers to an immunoglobulin molecule that binds to an antigen. Embodiments of antibodies include monoclonal antibodies, polyclonal antibodies, human antibodies, humanized antibodies, or chimeric antibodies. The antibodies can be of any class (e.g., IgG, IgE, IgM, IgD, IgA) and any subclass (e.g., IgG1, IgG2, IgG3, IgG4).

[0022] An exemplary antibody is an immunoglobulin G (IgG) antibody composed of four polypeptide chains cross-linked via interchain disulfide bonds: two heavy chains (HC) and two light chains (LC). The amino-terminal portion of each of the four polypeptide chains includes a variable region of approximately 100-125 or more amino acids primarily responsible for antigen recognition. The carboxyl-terminal portion of each of the four polypeptide chains contains a constant region primarily responsible for effector functions. Each heavy chain consists of a heavy chain variable region (VH) and a heavy chain constant region. Each light chain consists of a light chain variable region (VL) and a light chain constant region. IgG isotypes can be further subdivided into subclasses (e.g., IgG1, IgG2, IgG3, and IgG4).

[0023] The VH and VL regions can be further subdivided into hypervariable regions called complementarity-determining regions (CDRs), which are scattered across more conserved regions called framework regions (FRs). CDRs are exposed on the protein surface and are crucial regions for antibody-antigen binding specificity. Each VH and VL consists of three CDRs and four FRs, arranged in the following order from the amino terminus to the carboxyl terminus: FR1, CDR1, FR2, CDR2, FR3, CDR3, FR4. In this paper, the three CDRs of the heavy chain are referred to as "HCDR1, HCDR2, and HCDR3," and the three CDRs of the light chain are referred to as "LCDR1, LCDR2, and LCDR3." These CDRs contain the majority of residues that specifically interact with the antigen. You can use well-known protocols, including Kabat (Kabat et al., “Sequences of Proteins of Immunological Interest,” National Institutes of Health, Bethesda, Md. (1991)), Chothia (Chothia et al., “Canonical structures for the hypervariable regions of immunoglobulins”, Journal of Molecular Biology, 196, 901-917 (1987); Al-Lazikani et al., “Standard conformations for the canonical structures of immunoglobulins”, Journal of Molecular Biology, 273, 927-948 (1997)), North (North et al., “A New Clustering of Antibody CDR Loop Conformations”, Journal of Molecular Biology, 406, 228-256 (2011)) or IMGT (the international ImMunoGeneTics database is available at www.imgt.org; see Lefranc et al., Nucleic Those described in AcidsRes. 1999; 27:209-212) assign amino acid residues to CDRs.

[0024] The term "antigen-binding fragment" refers to an antibody portion that retains the ability to specifically interact with an epitope of an antigen. Examples of antigen-binding fragments include, but are not limited to, Fab or Fab'. A "Fab" fragment consists of a complete antibody light chain containing a light chain variable region (VL) and a light chain constant region (CL), as well as a heavy chain variable region (VH) and a heavy chain first constant region (CH1). Each Fab fragment is monovalent with respect to antigen binding, i.e., it has a single antigen-binding site. A Fab' fragment differs from a Fab fragment in that it has several additional residues at the carboxyl terminus of the CH1 domain, including one or more residues from the antibody hinge region. The Fab or Fab' described herein can be human Fab or Fab' or a chimeric Fab or Fab' containing human CL.

[0025] As used herein, the term "fusion protein" refers to a recombinant protein containing an amino or carboxyl terminus of a human antibody or antibody fragment directly linked to a heteropeptide or polypeptide.

[0026] Brief description of the attached figures

[0027] Figure 1A-1B This illustrates an exemplary G-chain β-fold stacking in a rabbit Fab crystal structure. Figure 1A The image shows HC:HC and LC:LC β stacking in 4ZTO.pdb. Heavy chains are black and light gray. Light chains are dark gray and white. Figure 1B This shows the HC:LC β stack in 4JO1.pdb.

[0028] Figure 2A-2C This demonstrates the incompatibility between human Cκ FG rings and rabbit-like LC:LC stacking. Figure 2A The structure comparison shown is between human Fab (dark gray) from 4NZU.pdb and rabbit Fab crystal stacks (medium and light gray) from 4ZTO. It shows that the rabbit CκFG ring (medium gray) is more compact than the longer and convex human Cκ FG ring and is compatible with β-fold stacking, which is not the case for the longer and convex human Cκ FG ring. Figure 2B This shows sequence alignments of FG loops from the Cκ and Cλ domains in humans, mice, and rabbits. The rabbit FG loop is two residues shorter and resembles the FG loop from the Cλ domain. Figure 2C The second potential interfering site is the human K126 side chain.

[0029] Figures 3A-3C This shows the crystal packing in the Cκ construct. Figure 3A Crystal stacking is visible in the parental Fab of dupilumab. G-chain β-stacking is absent. Figure 3BThis shows a crystal stacking plane in the Crystal Kappa version. Each κ constant region (light gray) forms a β-sheet with the nearby CH1 domain (black). Figure 3C Showing the β-sheet formed between the G chain (white) in the Cκ domain and the G chain (dark gray) in the CH1 domain. The fold extends from Cκ T205 to S211 and from CH1N216 to R222 (residues numbered sequentially in the crystal structure), is pseudo-symmetric, and is centered between Cκ V208 (position 205 according to Kabat numbering) and CH1 V219 (Kabat numbering).

[0030] Figures 4A-4B This shows the crystal structure designed by Crystal Kappa. Figure 4A Crystal Kappa version of dupilumab Fab, which is complexed with the extracellular domain of human IL4R, is shown (top right). Figure 4B Crystal Kappa version of human IL17 dimer complexed with secukinumab Fab is shown (top and middle).

[0031] Figures 5A-5C The column fractions were visualized for crystallization. Column fractions from size exclusion chromatography were directly used in vapor diffusion crystallization experiments under four conditions. Figure 5A Display the chromatogram. Figure 5B Images of each crystallization droplet shown on day 9. Figure 5C Displays the score assigned to each condition, where any crystallization receives a score of 90 or higher.

[0032] Figures 6A-6B This shows the crystallization and structural determination of a full-length mAb designed using Crystal Kappa. Figure 6A It is an image showing a full-length mAb crystal. Figure 6B The image shows a band diagram of the obtained structure, in which the antibody heavy chain is black (including the Fc glycosylation in the bar) and the antibody light chain is light gray.

[0033] Figures 7A-7B This shows the crystallization of a peptide-Fab complex designed using Crystal Kappa. Figure 7A This is an image showing a crystal of the L14H18 light chain of Kappa fused with Tau peptide after 4 hours. Figure 7B This is a banding diagram of the resulting 2.3 Å structure of the Tau peptide (white) bound to L14H18 Fab, where most of the ordered Gly-Ser linkers are attached to the N-terminus (gray) of the Fab light chain. The Fab heavy chain is shown in black. Example

[0034] Example 1: Crystallization of Fab and Fab:antigen complex

[0035] Rabbit Fab crystal packing analysis and design of human crystallizable κ

[0036] Visual examination of crystal stacking interactions in 36 publicly available and proprietary rabbit Fab crystal structures revealed that constant domain β-chain to β-chain crystal stacking is common (Table I). In 68% of the 19 stored rabbit Fab structures (including Fab complexes), LC to LC β interactions occurred, forming continuous β-sheets between the two Fab molecules. Figure 1A More than a third of these structures also exhibit HC:HC β-stacking interactions. Overall, 84% of these structures form some form of β-fold stacking interaction within the G-chain of the constant structural domain. The internal experience is similar, but there are more instances of Fab with HC:HC and LC:LC crystal stacking interactions, and fewer instances with HC:LC stacking (Table I and...). Figure 1B Crystals of Fab that form HC:HC and LC:LC or form HC:LC stacked interactions have a continuous, constant array of structural domains, each of which forms a typical FabHC:LC interaction as well as a β-fold with another structural domain.

[0037] Table I. Stacking of F-chain β-sheet crystals of rabbit Fab and Fab complex structures

[0038] PDB* resolution HC:HC LC:LC HC:LC 4HBC 1.54 x 4JO1 2.03 x 4JO2 2.50 x 4JO3 2.60 x x 4JO4 2.27 x x 4ZTO 2.30 x x 4ZTP 1.63 x 5DRN 1.99 x 5DS8 1.95 x 5DSC 2.40 x 5DTF 1.90 x 5DUB 2.00 5M63 2.74 x x 5V6L 2.55 x x 6CEZ 2.40 6CJK 1.80 x 6I9I 1.98 x 6PEH 2.30 x x 6T3F 3.20 x Total PDB 19 37% 68% 16% HC:HC & LC:LC 6 32% any 17 84% Lilly inside 17 29% 59% 6% HC:HC & LC:LC 10 59% any 14 82%

[0039] * The percentage of storage PDB structures formed by HC:HC, LC:LC, or HC:LC G-chain β-folds. Also indicates the percentage of structures with HC:HC and LC:LC stacks and the percentage of structures with any G-chain β-folds. Lilly internal statistics are listed below the PDB statistics.

[0040] An investigation of dozens of human Fab structures revealed the absence of such interactions. Alignment of human Fab structures with rabbit Fab structures showed that the longer FG loops present in the human Cκ domain form curved and convex conformations that interfere with β-sheet stacking interactions. Figure 2A This longer FG loop is shared by the mouse Cκ domain, but not by the rabbit Cκ domain (nor by the λ-constant Cλ domain). Figure 2B Visual examination of the human Fab aligned with the rabbit stacking interactions also indicated that human Cκ lysine 126 (K126, Kabat number) may be a barrier to the stacking of β-sheets on the contralateral side of this domain. Figure 2C ).

[0041] Several variants of the Fab fragment labeled with six histidine (H6) were generated by mutating the FG loop or K126 of Cκ, and their effects on crystallization were tested. In the case of the FG loop, this involved replacing the heptameric human sequence HQGLSSP (according to Kabat numbers, positions 198 to 204) between the structurally homologous T and V (according to Kabat numbers, positions 197 and 205, respectively) with the pentamer rabbit sequence QGTTS (according to Kabat numbers, positions 199 to 203). The resulting mutants were two residues shorter, with histidine at position 198 and proline at position 205 missing, and are referred to herein as ΔQGTTSΔ (according to Kabat numbers, positions 198 to 204, see SEQ ID NO:3) or “Crystal Kappa” design. In some variants, K126 was mutated to alanine (K126A, Kabat number, see SEQ ID NO:2, 4, 6).

[0042] Independent of crystal packing analysis, few ordered C-terminal interchain disulfide bonds were observed in the Fab structure. This could be due to conformational heterogeneity or heterogeneous oxidation. To address this, two variants were designed. One variant, called “GEP*”, was generated by removing disulfide bonds via mutating the C-terminal κ-chain cysteine ​​to proline (according to Kabat number C214P, see SEQ ID NO:5 or 6) and mutating cysteine ​​127 of the IgG4 heavy chain to alanine (according to Kabat number C127A). The second variant, called “ESKCGGH6”, was generated by mutating κ-cysteine ​​to proline (according to Kabat number C214P) and creating a new disulfide bond partner at the C-terminus of the heavy chain by mutating Tyr 229 to cysteine ​​(according to Kabat number Y229C).

[0043] Fab crystallization

[0044] Crystallization results for the two Fabs (G6 and dupilumab) before and after the incorporation of these mutations, individually or in combination, are shown in Table II. One Fab was derived from a publicly available dupilumab (Dupixent™) sequence (available at https: / / www.kegg.jp / entry / D10354). The second Fab was part of an internal discovery effort targeting the cell surface receptor. Two 96-well crystallization screenings were performed using Fabs with identical purification, set to approximately 10 mg / ml, streaked with unrelated Fab crystals (to eliminate random differences due to nucleation), and analyzed at the same time point (9 days). Both parental Fabs produced crystal hits under some conditions, and the dupilumab Fab crystals even produced a 2.0 Å dataset and structure. The K126A mutation (Table II) and any disulfide bond variants (not shown) did not produce significantly more crystallization conditions. The most significant difference was with the incorporation of the ΔQGTTSΔFG loop (i.e., Crystal Kappa design). The Fab with this design yielded crystals under approximately 90–114 of the 192 conditions for the G6 Fab; and under approximately 113–136 of the 192 conditions for the dupilumab Fab. Crystals harvested directly from these screenings (i.e., not optimized in subsequent screenings) produced a high-resolution dataset (Table II). The best dupilumab Fab diffracted from ΔQGTTSΔ alone to 1.4 Å (CK1.0); and the best G6 Fab diffracted from the combination of ΔQGTTSΔ mutation with K126A and intrachain disulfide bonds to 1.15 Å (CK1.5).

[0045] Table II. Crystallization of two Fab species and their variants

[0046] HC LC G6 Xtal** G6 Res (Å) Dup Xtal** Dup Res (Å) Parent* <![CDATA[ESKYGH6]]> Wild type κ 1 4 2.0 CK0.1 <![CDATA[ESKYGH6]]> K126A 2 3.4 NA CK1.0 <![CDATA[ESKYGH6]]> ∆QGTTS∆ 94 1.5 113 1.4 CK1.1 <![CDATA[ESKYGH6]]> K126A ΔQGTTSΔ 114 1.3 136 2.1 CK1.2 <![CDATA[ESKYGH6 C127A]]> ∆QGTTS∆ GEP* 93 1.7 123 1.9 CK1.3 <![CDATA[ESKCGGH6]]> ∆QGTTS∆ GEP* 90 1.2 125 1.7 CK1.4 <![CDATA[ESKYGH6 C127A]]> K126A ∆QGTTS∆ GEP* 104 1.4 NA CK1.5 <![CDATA[ESKCGGH6]]> K126A ∆QGTTS∆ GEP* 110 1.15 121 2.2

[0047] *The parental light chain is wild-type κ. The parental heavy chain terminates at the sequence ESKYG and includes an H6 tag for purification purposes.

[0048] The "Xtal" column indicates how many crystals of any type were produced from the conditions of the two 96-well screens around day 9. Several types of crystals were sent from each construct with harvestable crystals, and the optimal resolution dataset was indicated.

[0049] Crystal structure of the modified Fab

[0050] Fifty-nine datasets were collected for the G6 variant, covering 11 crystal forms. Structures were resolved for seven crystal forms and refined for five: P212121 with a 43x75x165 Å unit cell (5 refined structures), P43212 77x77x330 (3), P212121 66x74x91 (1), P1 53x65x67 85x71x84 (1), and C2 206x103x70 β=92.7º (1). Forty-four datasets were collected for the dupilumab Fab variant, covering 11 crystal forms. Parse the structure for 4 datasets and refine 3 datasets: P21 53x66x135 β=91.6º (1), P212121 59x73x105 (1), P43212 74x74x185 (1).

[0051] Fab variants without ΔQGTTSΔ (including the parental variant) do not have stacking with extended β-sheet interactions. For example, the parental dupilumab Fab stacks with various types of interactions, but none form a continuous β-sheet. Figure 3A On the other hand, all structures derived from Fab with ΔQGTTSΔFG rings stack up, forming β-sheets between the G chains of the Cκ domain and the G chains of the CH1 domain. Figure 3B and 3C This interaction is similar to, but not the same as, seen in rabbit Fab crystal stacking. It involves... Figure 1B The same chain seen in [the text], but more extensive, involving 7 residues on both sides, such as [the chain]. Figure 1A The H:H or L:L interaction observed in the β-fold. The pseudo-symmetry center of this β-fold lies between Cκ V208 (position 205 according to Kabat number) and CH1 V219 (Kabat number).

[0052] The K126A mutation does not appear to affect crystal stacking or diffraction quality. The highest resolution structure of the G6 Fab incorporates this mutation, but its effect is not systematic. For example, the highest resolution structure of the dupilumab Fab only incorporates the FG ring mutation ΔQGTTSΔ. Disulfide removal (C127A+GEP*) or intrachain disulfide bonding (ESKC+GEP*) also does not appear to affect crystal stacking or diffraction. The structure is obtained with ordered intrachain disulfide bonds. The temperature factor in this region is higher than the average seen in other structures with ordered interchain disulfide bonds.

[0053] Fab: Crystallization of antigen complexes

[0054] Crystallization mutations were applied to the Fab:antigen complex. In the case of G6 Fab, the CK1.5 variant of Fab was utilized because it diffracted best as a standalone Fab. For dupilumab Fab and the other four complexes, CK1.0 (i.e., ΔQGTTSΔ only) was utilized. All CH1 domains were IgG4 and had the same C-terminal hexahistine (H6) tag (SEQ ID NO: 7). Within a limited screening and adoption timeframe, none of the parental complexes (i.e., antigen:Fab complexes without any crystallization engineering applied to Fab) produced any crystals (Table III). All engineered complexes produced crystals, from 4 conditions (out of 192 conditions) for the G6-receptor complex to 87 conditions for the H4-receptor complex (Figure 4C and Table III). Four of the six complexes produced structures, mostly at low resolution. The GITR complex crystals were essentially standalone Fab crystals after structure resolution, and the diffraction of the TIGIT complex crystals was not good enough to produce a dataset. The crystals of both dupilumab and secukinumab require optimization to achieve their respective 3 and 3.2 Å resolutions (…). Figures 4A-4B In the former case, the crystal from the initial screening diffracted up to 5 Å, while in the latter case it did not diffract at all.

[0055] Table III. Crystallization of Fab: Antigen Complex

[0056] Fab antigen Parental crystals CK crystal Resolution (Å) Notes refer to G6 receptor 0 4 3.6 CK1.5 Undisclosed molecules Dupilumab IL4Ra 0 34 3.0 Optimized CAS: 1190264-60-8; Sequence obtained from https: / / www.kegg.jp / entry / D10354 Secukinumab IL17a 0 54 3.2 Optimized CAS: 1229022-83-6; Sequence obtained from https: / / www.kegg.jp / entry / D09967 h2155 GITR 0 54 2.6 Fab Only US2013108641 h22G2 TIGIT 0 30 9 diffraction only US2016176963 H4 receptor 0 87 2.6 Undisclosed molecules

[0057] The complexes were screened and scored in two crystallization screenings after 9 days. The number of conditions for any type of crystal from the parental complex (without crystallization mutations) is indicated in the "Parental Crystals" column. The conditions for crystals from the CK1.0 Fab antigen complex are indicated in the column labeled "CK Crystals". The best diffraction or dataset from these screenings or subsequent optimizations is indicated in the "Resolution" column.

[0058] Because the secukinumab:IL17 complex produces low-resolution but diffracted crystals, it was chosen for further comparison of the CH1 domain isotype and C-terminus. Four new constructs were created: a short 5-residue IgG4 version ending with the sequence DKRVESK (untagged, SEQ ID NO:11), an IgG4 version ending with DKRVH6 (tagged, SEQ ID NO:12), and an IgG1 version ending with the sequence KSC with or without the H6 tag (SEQ ID NO:10) or without the H6 tag (SEQ ID NO:9). These were purified and screened at 10 mg / ml, but also at 5 mg / ml, as previously described. The shorter IgG4 version produced fewer crystals than the original versions (24 for the tagged version and 18 for the untagged version, compared to 54 for the original tagged version). The IgG1 version gave a similar number of crystals under similar conditions (88 for the tagged version and 43 for the untagged version, compared to 54 for the original tagged version). The 10 mg / ml IgG4-labeled version diffracted to 4.2 Å, just as the parental version, which initially did not diffract but was optimized to produce a 3.2 Å dataset. On the other hand, the 10 mg / ml IgG1-labeled version directly produced a 2.7 Å dataset from the initial screening, and the 5 mg / ml unlabeled IgG1 version produced a 2.4 Å dataset from the initial screening.

[0059] Column-fed fractional crystallization (CFC)

[0060] The robust crystallization of the engineered Fab allows for direct column-stage fractionation. When crystallized directly from the column stage, the ΔQGTTSΔ variant of G6Fab produces the same crystal form as purified and concentrated samples at 10 mg / ml. Figures 5A-5C The CFC structure has a fairly good resolution of 2.4 Å (compared to 1.4 Å).

[0061] The design described here is applied to several targets, utilizing comparable isotypes and C-termini, employing the same purification, crystallization, and crystal harvesting procedures (in parallel as much as possible), utilizing crystal seeding to reduce nucleation variability, evaluating the same crystallization experiments over the same time period, and having the same scientists perform purification and crystallization in all experiments.

[0062] The variant Cκ domain (ΔQGTTSΔ) increased the crystallization frequency of human Fab by 50-fold. The G6 parental Fab (fully human) produced crystals under only one condition, but the G6 variant containing ΔQGTTSΔ produced crystals under approximately 90–114 conditions. The dupilumab parental Fab produced crystals under four conditions, but the dupilumab variant containing ΔQGTTSΔ produced crystals under approximately 113–136 conditions. Furthermore, the modified FG ring of the Cκ domain (“crystallizable Kappa” or “CrystalKappa”) enabled the Fab:antigen complex to crystallize, but the fold increase could not be calculated because no crystals were produced without the Crystal Kappa design. Most Crystal Kappa versions produced crystals under 30–90 crystallization conditions for the complex (note: statistics are from two plates at a relatively conservative time point).

[0063] The results from the complexes were less encouraging compared to the Fab results. While all the complexes using Crystal Kappa produced crystals, a significant advantage over any other effort, only four out of six produced complex datasets good enough to resolve and refine, and these tended to be low-resolution. One (h2155 + GITR) produced crystals from Fab only.

[0064] The 3.0 Å structure of dupilumab Fab complexed with human IL4R shows epitopes that largely overlap with those binding to IL4 and IL13, explaining its blocking activity. Figure 4A The central part of the epitope is a CD loop (Ul-Haq2016), which explains why dupilumab has a very different sequence in this region (L). 67 L 68 vs. Q 67 S 68 The cynomolgus monkey IL4R showed no cross-reactivity. The 3.0 Å secukinumab IL17 complex (and its 2.4 Å modified structure) showed that two Fabs bound to an IL17 dimer with discontinuous epitopes, each Fab binding portions of two IL17 chains. Figure 4B The H4 complex produces a crystal stacking arrangement of three Fabs in the asymmetric unit (not shown). One Fab forms a typical two HC:LC β-stacking interaction for the CK design. Another Fab forms one on the LC side but not on the HC side. And the third Fab forms an HC:LC interaction with the second Fab, and its LC forms an LC:LC β-stacking interaction, which is the only such interaction observed so far.

[0065] Further refinement of the secukinumab construct shortened the C-terminus of the IgG4 construct and included the IgG1 version for the first time, comparing the H6-labeled version with the unlabeled version. 10 mg / ml and 5 mg / ml were compared to reduce crystal crowding for harvesting purposes. In this series, the IgG1 version outperformed IgG4, generating a 2.7 Å dataset directly for the labeled (10 mg / ml) version and a 2.4 Å dataset for the unlabeled Fab (5 mg / ml) version, while the 3.0 Å dataset from the CK1.0 construct was obtained after optimization and screening of a large number of crystals. Interestingly, G1, G4 (both versions), labeled, and unlabeled all produced isomorphic crystals.

[0066] Regarding crystallization rate, all described crystals grew within a week, and for those examined more frequently, crystals appeared within hours. Regarding concentration, column-fractionated crystallization experiments showed that crystals could be obtained from samples diluted to 0.1 mg / ml, at least for Fab alone. Besides its effect on the desired concentration, CFC has other potential advantages. For example, the characteristics of proteins at the column leading edge may differ from those at the trailing edge, and one or the other may produce more in the determined structure. The fact that CFC experiments and Fab crystallized under more than half of the high-concentration crystallization conditions suggests that, for Fab, the Crystal Kappa design should allow for a significantly simplified set of screening conditions.

[0067] In summary, this paper presents a crystallizable variant of the human constant κ domain (variant Cκ) that significantly increases the frequency of crystal formation of Fab and Fab:antigen complexes, yields high-resolution structures of Fab (and Fab:peptide complexes), and in most cases, generates at least low-resolution datasets and structures of Fab:protein complexes. The Crystal Kappa design appears to allow overnight crystallization from diluted samples under limited screening conditions. The Crystal Kappa design should make Fab structure determination robust even with smaller screenings and fewer proteins, and accelerate complex structure determination, including Fab chaperone complexes with difficult targets.

[0068] Materials and methods

[0069] Engineering modification and molecular biology

[0070] Rabbit Fab crystal stacking was analyzed in Pymol using structures available in protein databases and Eli Lilly's proprietary structure database. Alignment of LC constant domains from different species was performed using BLAST. Amino acid sequences of the variable domains of dupilumab and secukinumab were obtained from the Kyoto Encyclopedia of Genes and Genomes (www.kegg.jp; Kanehisa2000), entries D10354 and D09967. Sequences of h2155 and h22G2 were obtained from patents US2013108641 and US2016176963. Expression vectors were created by synthesizing the corresponding DNA fragments into gblocks (IDT, CoralvilleIA) and cloning them into mammalian expression vectors using standard techniques.

[0071] Expression and purification

[0072] Fab and the antigen ECD were expressed in mammalian cell culture CHO cells. Cell culture supernatant containing the protein was harvested and the clarified medium was purified by immobilized metal affinity chromatography (IMAC) using His Trap™ Excel (GE Healthcare) with PBS buffer containing 15 mM imidazole at pH 7.5 as the binding buffer. The protein was then eluted on a 10-column volume gradient in PBS containing 0.3 M imidazole at pH 7.4. The eluent was collected and concentrated using a Millipore 10 kDa rotary concentrator. The concentrated IMAC pool was loaded onto a Superdex 75 or Superdex 200 (GE Healthcare) column. For the crystallization tray, the protein was further concentrated to 10 mg / ml.

[0073] Proteins were characterized using analytical size exclusion chromatography (Waters) and SDS-PAGE gel chromatography (data not shown).

[0074] Crystallization and structural determination

[0075] All samples were concentrated to 5–10 mg / ml and placed in vapor diffusion droplets at a 1:1 ratio using Qiagen Classics II and PEGs crystallization screening at room temperature. Droplets were immediately cross-seeded with relevant Fab seeds for Fab crystallization and with complex seeds for complex crystallization. Images of the crystallization trays were taken on days 1, 4, and 9. Prior to freezing in liquid nitrogen, the crystals were transferred to a cryoprotectant solution consisting of a pore solution supplemented with an additional 10% precipitant for the crystallization pores and 25% glycerol, or a pore solution from the mother liquor if the mother liquor contained a cryoprotectant of sufficient concentration for cryoprotection (such as PEG 400, PEG MME550, PEG MME 2K, etc.).

[0076] The structural determination diffraction datasets were collected from the following sources: Lilly Research Collaborative Access Team (LRL-CAT) Beamline 31-ID at Advanced Proton Source (Argonne, IL); Beamline ALS-502 at Advanced Light Source (Berkley, CA); and Beamline I04-1 at Diamond Light Source (Oxfordshire, UK). The data were integrated and simplified using MOSFLM (Leslie AGW & Powell HR. In: Read RJ, & Sussman JL (eds.), Evolving methods for macromolecular crystallography: the structural path to the understanding of the mechanism of action of CBRN Agents. Dordrecht: Springer Netherlands. 2007) and the CCP4 program suite (Winn, et al., Acta Crystallogr D Biol Crystallogr, 2011. 67(Pt 4):235-42). The initial molecular substitution solution was obtained using Phaser (CCP4 kit) (McCoy et al., J. Appl. Crystallogr., 2007.40(Pt 4):658-674). The model was constructed using COOT (Emsley et al., Acta Crystallogr D Biol Crystallogr, 2010. 66(Pt 4):486-501) and refined using Refmac (Murshudov et al., Acta Crystallogr D Biol Crystallogr, 2011. 67(Pt 4):355-67) or Buster (Bricogne et al., BUSTER version 2.11.5. Cambridge, United Kingdom: Global Phasing Ltd. 2011) and validated using an in-house developed scheme.

[0077] Example 2: Crystallization and X-ray structural determination of full-length monoclonal antibody (mAb)

[0078] In addition to Fab fragments, Crystal Kappa design can also be used to crystallize full-length IgG antibodies, as follows. Using standard molecular biology techniques, the light chain of the antibody is modified with Crystal Kappa design in a standard expression vector. The light chain, along with the corresponding heavy chain, is then expressed in an expression system suitable for antibody secretion (such as HEK293 or Chinese hamster ovary cells). The antibody is then purified from the culture medium and concentrated to, for example, 5 mg / ml using techniques such as column purification with MabSelect columns (GE Healthcare). Vapor diffusion is used, followed by screening conditions to grow protein crystals from the purified antibody. The crystals are then isolated, transferred, and frozen, and X-ray data are collected. Using standard techniques, the structure can then be determined from this data by molecular substitution (e.g., using software Phaser, and then using a refined atomic model).

[0079] One antibody structure was generated by incorporating Crystal Kappa design into the κ light chain of an IgG4-P isotype antibody (IgG4 with S241P (according to Kabat number) or S228P (according to EU index number)). Crystals were produced from 96-well screening at 21°C (ComPAS, Qiagen) under several conditions, including 18% ethanol + 100 mM Tris HCl (pH 8.5). By optimizing these conditions, diffraction crystals and a 4 Å dataset were obtained from 20% ethanol and 21°C. Molecular substitution solutions for this dataset yielded a structure of the full-length antibody (two heavy chains, two light chains, and glycosylation) whose conformation was significantly different from the conformations of two known human IgG4 structures (5DK3.pdb and 6GFE.pdb) and more similar to the known human IgG1 structure (1HZH.pdb). Crystal Kappa design facilitated crystal stacking, as did other contact points in the antibody.

[0080] Figures 6A-6B This shows the crystallization and structural determination of the full-length mAb. Figure 6A It is an image showing a crystal with a full-length mAb. Figure 6B The image shows a band diagram of the obtained structure, in which the antibody heavy chain is black (including the rod-shaped Fc glycosylation) and the antibody light chain is light gray.

[0081] Example 3: Generation of fusion protein crystal Fab:antigen peptide complex

[0082] The determination of the crystal structure of a Fab complex with its antigenic peptide can be achieved in two ways. First, and more typically, the antigenic peptide can be purchased or prepared using various techniques (Chandrudu S, Simerska P, Toth I. Chemical methods for peptide and protein production. Molecules. 2013 Apr 12;18(4):4373-88). The peptide can then be dissolved and added to the Fab at a final concentration equal to or greater than the Fab concentration in moles. The complex can then be crystallized and its structure determined, just as with Crystal Kappa Fab alone. A second technique that does not require purchasing the peptide is to insert the coding sequence of the peptide directly into the open reading frame of the Fd or κ light chain to generate a fusion protein with a suitable linker between the peptide and the Fd / κ light chain during the construction of the Fab expression vector. The peptide is then tethered to the Fab via the linker in a precise one-to-one stoichiometric manner in the fusion protein. This tethering also has the benefit of increasing the effective concentration of the peptide and avoiding the absence of low-affinity peptides in the crystal or the resulting structure. The tethering ensures the presence of antigenic peptides in the crystal.

[0083] The anti-Tau antibody L14H18 (WO 2017 / 005734) and its homologous Tau peptide antigen Fab:antigen peptide complex were obtained as follows. A tau peptide (amino acids 231-250) containing the known epitope TPPKSPSSAKSRLQTAPVPM (SEQ ID NO:18) was fused to the N-terminus of the Crystal Kappa light chain of Fd-His6 or L14H18 Fab using a GS adapter (SEQ ID NO:19) by incorporating a DNA sequence (translated into the following 27 amino acids) between the signal sequence of the open reading frame and the maturation start of the Fab chain. Both the Tau peptide-adaptor-Fd-His / Crystal Kappa light chain and the Fd-His / Tau peptide-adaptor-Crystal Kappa light chain versions were expressed in CHO cells, purified by immobilized metal affinity chromatography and gel filtration, concentrated to 4.5 mg / ml, and crystallized in a commercially available screening process. In Classics and PEG screening (Qiagen), dozens of conditions were applied to produce crystals for both versions. The structures of the two complexes were resolved and refined for the Crystal Kappa fusion: a 1.22 Å structure from 20% PEG 3350 / 200 mM sodium potassium tartrate and a 2.3 Å structure from 100 mM sodium acetate pH 4.6 / 25% PEG 4000 / 200 mM ammonium sulfate. Both structures showed a largely extended peptide, with one helical turn binding to the Fab CDR, and differed only in the degree of order at the linker portion.

[0084] Figures 7A-7B The crystallization of the Tau peptide-Fab complex is shown. Figure 7A This is an image showing a crystal of the L14H18 light chain of Kappa fused with Tau peptide after 4 hours. Figure 7B This is a banding diagram of the resulting 2.3 Å structure of the Tau peptide (white) bound to L14H18 Fab, where most of the ordered GS linkers are attached to the N-terminus of the Fab light chain (gray). The Fab heavy chain is shown in black.

[0085] sequence list

[0086] Wild-type human κ light chain constant structural domain Amino acid sequence (SEQ ID NO:1)

[0087]

[0088] people Variant amino acid sequence (SEQ ID NO:2)

[0089]

[0090] people Variant amino acid sequence (SEQ ID NO:3)

[0091]

[0092] people Variant amino acid sequence (SEQ ID NO:4)

[0093]

[0094] people Variant amino acid sequence (SEQ ID NO:5)

[0095]

[0096] people Variant amino acid sequence (SEQ ID NO:6)

[0097]

[0098] Amino acid sequence of the ESKYGH6 variant of the human IgG4 CH1 domain (SEQ ID NO:7)

[0099]

[0100] The amino acid sequence of the ESKYGH6 C127A variant of the human IgG4 CH1 domain (SEQ ID NO:8)

[0101]

[0102] Wild-type human IgG1 untagged CH1 domain amino acid sequence (SEQ ID NO:9)

[0103]

[0104] Human IgG1 has a tagged CH1 domain amino acid sequence (SEQ ID NO:10)

[0105]

[0106] Human IgG4 untagged CH1 domain (SEQ ID NO:11)

[0107]

[0108] Human IgG4 has a tagged CH1 domain (SEQ ID NO:12)

[0109]

[0110] Wild-type mice Amino acid sequence (SEQ ID NO:13)

[0111]

[0112] Wild-type rabbit Amino acid sequence (SEQ ID NO:14)

[0113]

[0114] Wild-type human λ light chain constant structural domain Amino acid sequence (SEQ ID NO:15)

[0115]

[0116] Wild-type mice Amino acid sequence (SEQ ID NO:16)

[0117]

[0118] Wild-type rabbit Amino acid sequence (SEQ ID NO:17)

[0119]

[0120] Tau peptide (SEQ ID NO:18)

[0121]

[0122] GS connector (SEQ ID NO:19)

[0123] sequence list <110> Eli Lilly <120> Crystallization of antibody or antigen-binding fragments <130> X22176 <150> 62 / 975,269 <151> 2020-02-12 <160> 19 <170> PatentIn version 3.5 <210> 1 <211> 107 <212> PRT <213> Homo sapiens <400> 1 Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu Lys Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr His Gln Gly Leu Ser Ser 85 90 95 Pro Val Thr Lys Ser Phe Asn Arg Gly Glu Cys 100 105 <210> 2 <211> 107 <212> PRT <213> artificial sequence <220> <223> synthetic construction <400> 2 Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu Ala Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr His Gln Gly Leu Ser Ser 85 90 95 Pro Val Thr Lys Ser Phe Asn Arg Gly Glu Cys 100 105 <210> 3 <211> 105 <212> PRT <213> Artificial sequence <220> <223> Synthetic constructs <400> 3 Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu Lys Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr Gln Gly Thr Thr Ser Val 85 90 95 Thr Lys Ser Phe Asn Arg Gly Glu Cys 100 105 <210> 4 <211> 105 <212> PRT <213> Artificial sequence <220> <223> Synthetic constructs <400> 4 Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu Ala Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr Gln Gly Thr Thr Ser Val 85 90 95 Thr Lys Ser Phe Asn Arg Gly Glu Cys 100 105 <210> 5 <211> 105 <212> PRT <213> Artificial Sequence <220> <223> Synthetic construct <400> 5 Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu Lys Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr Gln Gly Thr Thr Ser Val 85 90 95 Thr Lys Ser Phe Asn Arg Gly Glu Pro 100 105 <210> 6 <211> 105 <212> PRT <213> Artificial Sequence <220> <223> Synthetic Construct <400> 6 Arg Thr Val Ala Ala Pro Ser Val Phe Ile Phe Pro Pro Ser Asp Glu 1 5 10 15 Gln Leu Ala Ser Gly Thr Ala Ser Val Val Cys Leu Leu Asn Asn Phe 20 25 30 Tyr Pro Arg Glu Ala Lys Val Gln Trp Lys Val Asp Asn Ala Leu Gln 35 40 45 Ser Gly Asn Ser Gln Glu Ser Val Thr Glu Gln Asp Ser Lys Asp Ser 50 55 60 Thr Tyr Ser Leu Ser Ser Thr Leu Thr Leu Ser Lys Ala Asp Tyr Glu 65 70 75 80 Lys His Lys Val Tyr Ala Cys Glu Val Thr Gln Gly Thr Thr Ser Val 85 90 95 Thr Lys Ser Phe Asn Arg Gly Glu Pro 100 105 <210> 7 <211> 109 <212> PRT <213> artificial sequence <220> <223> synthetic construction <400> 7 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Cys Ser Arg 1 5 10 15 Ser Thr Ser Glu Ser Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr Lys Thr 65 70 75 80 Tyr Thr Cys Asn Val Asp His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Arg Val Glu Ser Lys Tyr Gly His His His His His His 100 105 <210> 8 <211> 109 <212> PRT <213> Artificial Sequence <220> <223> Synthetic Construct <400> 8 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Ala Ser Arg 1 5 10 15 Ser Thr Ser Glu Ser Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr Lys Thr 65 70 75 80 Tyr Thr Cys Asn Val Asp His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Arg Val Glu Ser Lys Tyr Gly His His His His His His 100 105 <210> 9 <211> 103 <212> PRT <213> Homo sapiens <400> 9 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Ser Ser Lys 1 5 10 15 Ser Thr Ser Gly Gly Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr Gln Thr 65 70 75 80 Tyr Ile Cys Asn Val Asn His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Lys Val Glu Pro Lys Ser Cys 100 <210> 10 <211> 109 <212> PRT <213> artificial sequence <220> <223> synthetic construction <400> 10 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Ser Ser Lys 1 5 10 15 Ser Thr Ser Gly Gly Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr Gln Thr 65 70 75 80 Tyr Ile Cys Asn Val Asn His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Lys Val Glu Pro Lys Ser Cys His His His His His His 100 105 <210> 11 <211> 101 <212> PRT <213> Homo sapiens <400> 11 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Cys Ser Arg 1 5 10 15 Ser Thr Ser Glu Ser Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr Lys Thr 65 70 75 80 Tyr Thr Cys Asn Val Asp His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Arg Val Glu Ser Lys 100 <210> 12 <211> 104 <212> PRT <213> artificial sequence <220> <223> synthetic construction <400> 12 Ala Ser Thr Lys Gly Pro Ser Val Phe Pro Leu Ala Pro Cys Ser Arg 1 5 10 15 Ser Thr Ser Glu Ser Thr Ala Ala Leu Gly Cys Leu Val Lys Asp Tyr 20 25 30 Phe Pro Glu Pro Val Thr Val Ser Trp Asn Ser Gly Ala Leu Thr Ser 35 40 45 Gly Val His Thr Phe Pro Ala Val Leu Gln Ser Ser Gly Leu Tyr Ser 50 55 60 Leu Ser Ser Val Val Thr Val Pro Ser Ser Ser Leu Gly Thr Lys Thr 65 70 75 80 Tyr Thr Cys Asn Val Asp His Lys Pro Ser Asn Thr Lys Val Asp Lys 85 90 95 Arg Val His His His His His His 100 <210> 13 <211> 107 <212> PRT <213> House mouse <400> 13 Arg Ala Asp Ala Ala Pro Thr Val Ser Ile Phe Pro Pro Ser Ser Glu 1 5 10 15 Gln Leu Thr Ser Gly Gly Ala Ser Val Val Cys Phe Leu Asn Asn Phe 20 25 30 Tyr Pro Lys Asp Ile Asn Val Lys Trp Lys Ile Asp Gly Ser Glu Arg 35 40 45 Gln Asn Gly Val Leu Asn Ser Trp Thr Asp Gln Asp Ser Lys Asp Ser 50 55 60 [[ID=3C]]Thr Tyr Ser Met Ser Ser Thr Leu Thr Leu Thr Lys Asp Glu Tyr Glu 65 70 75 80 Arg His Asn Ser Tyr Thr Cys Glu Ala Thr His Lys Thr Ser Thr Ser 85 90 95 Pro Ile Val Lys Ser Phe Asn Arg Asn Glu Cys 100 105 [[ID=QC]]<210> 14 <211> 104 <212> PRT <213> European rabbit <400> 14 Gly Asp Pro Val Ala Pro Thr Val Leu Ile Phe Pro Pro Ala Ala Asp 1 5 10 15 Gln Val Ala Thr Gly Thr Val Thr Ile Val Cys Val Ala Asn Lys Tyr 20 25 30 Phe Pro Asp Val Thr Val Thr Trp Glu Val Asp Gly Thr Thr Gln Thr 35 40 45 Thr Gly Ile Glu Asn Ser Lys Thr Pro Gln Asn Ser Ala Asp Cys Thr 50 55 60 Tyr Asn Leu Ser Ser Thr Leu Thr Leu Thr Ser Thr Gln Tyr Asn Ser 65 70 75 80 His Lys Glu Tyr Thr Cys Lys Val Thr Gln Gly Thr Thr Ser Val Val 85 90 95 Gln Ser Phe Asn Arg Gly Asp Cys 100 <210> 15 <211> 106 <212> PRT <213> Homo sapiens <400> 15 Gly Gln Pro Lys Ala Asn Pro Thr Val Thr Leu Phe Pro Pro Ser Ser 1 5 10 15 Glu Glu Leu Gln Ala Asn Lys Ala Thr Leu Val Cys Leu Ile Ser Asp 20 25 30 Phe Tyr Pro Gly Ala Val Thr Val Ala Trp Lys Ala Asp Gly Ser Pro 35 40 45 Val Lys Ala Gly Val Glu Thr Thr Lys Pro Ser Lys Gln Ser Asn Asn 50 55 60 Lys Tyr Ala Ala Ser Ser Tyr Leu Ser Leu Thr Pro Glu Gln Trp Lys 65 70 75 80 Ser His Arg Ser Tyr Ser Cys Gln Val Thr His Glu Gly Ser Thr Val 85 90 95 Glu Lys Thr Val Ala Pro Thr Glu Cys Ser 100 105 <210> 16 <211> 105 <212> PRT <213> Mus musculus <400> 16 Gln Pro Lys Ser Ser Pro Ser Val Thr Leu Phe Pro Pro Ser Ser Glu 1 5 10 15 Glu Leu Glu Thr Asn Lys Ala Thr Leu Val Cys Thr Ile Thr Asp Phe 20 25 30 Tyr Pro Gly Val Val Thr Val Asp Trp Lys Val Asp Gly Thr Pro Val 35 40 45 Thr Gln Gly Met Glu Thr Thr Gln Pro Ser Lys Gln Ser Asn Asn Lys 50 55 60 Tyr Met Ala Ser Ser Tyr Leu Thr Leu Thr Ala Arg Ala Trp Glu Arg 65 70 75 80 His Ser Ser Tyr Ser Cys Gln Val Thr His Glu Gly His Thr Val Glu 85 90 95 Lys Ser Leu Ser Arg Ala Asp Cys Ser 100 105 <210> 17 <211> 105 <212> PRT <213> Artificial Sequence <220> <223> Synthetic Construct <400> 17 Gln Pro Ala Val Thr Pro Ser Val Ile Leu Phe Pro Pro Ser Ser Glu 1 5 10 15 Glu Leu Lys Asp Asn Lys Ala Thr Leu Val Cys Leu Ile Ser Asp Phe 20 25 30 Tyr Pro Arg Thr Val Lys Val Asn Trp Lys Ala Asp Gly Asn Ser Val 35 40 45 Thr Gln Gly Val Asp Thr Thr Gln Pro Ser Lys Gln Ser Asn Asn Lys 50 55 60 Tyr Ala Ala Ser Ser Phe Leu His Leu Thr Ala Asn Gln Trp Lys Ser 65 70 75 80 Tyr Gln Ser Val Thr Cys Gln Val Thr His Glu Gly His Thr Val Glu 85 90 95 Lys Ser Leu Ala Pro Ala Glu Cys Ser 100 105 <210> 18 <211> 20 <212> PRT <213> Artificial sequence <220> <223> Synthetic constructs <400> 18 Thr Pro Pro Lys Ser Pro Ser Ser Ala Lys Ser Arg Leu Gln Thr Ala 1 5 10 15 Pro Val Pro Met 20 <210> 19 <211> 7 <212> PRT <213> Artificial sequence <220> <223> Synthetic constructs <400> 19 Gly Gly Gly Ser Gly Gly Gly 1 5

Claims

1. An antibody, antigen-binding fragment, or fusion protein comprising a variant CK, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO: 3, 4, 5, or 6.

2. The antibody, antigen-binding fragment, or fusion protein of claim 1, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

3.

3. The antibody, antigen-binding fragment, or fusion protein of claim 1, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

4.

4. The antibody, antigen-binding fragment, or fusion protein of claim 1, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

5.

5. The antibody, antigen-binding fragment, or fusion protein of claim 1, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

6.

6. The antibody, antigen-binding fragment, or fusion protein of any one of claims 1-5, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a human light chain variable domain (VL) and a human heavy chain variable domain (VH).

7. The antibody, antigen-binding fragment, or fusion protein of any one of claims 1-5, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a human IgG CH1 domain.

8. The antibody, antigen-binding fragment, or fusion protein of claim 7, wherein the human IgG CH1 domain is a human IgG1 or IgG4 CH1 domain.

9. The antigen-binding fragment of any one of claims 1-5, wherein the antigen-binding fragment is a human Fab or Fab'.

10. A method of generating a crystal structure of an antibody, antigen-binding fragment, or fusion protein comprising human Ckappa, the method comprising: generating an antibody, antigen-binding fragment, or fusion protein comprising a variant CK, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO: 3, 4, 5, or 6; and crystallizing the antibody, antigen-binding fragment, or fusion protein comprising the variant CK.

11. A method of generating a crystal structure of a complex of an antigen and an antibody, antigen-binding fragment, or fusion protein that binds the antigen, wherein the antibody, antigen-binding fragment, or fusion protein comprises a human CK, the method comprising: generating an antibody, antigen-binding fragment, or fusion protein comprising a variant CK, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO: 3, 4, 5, or 6; and co-crystallizing the antigen and the antibody, antigen-binding fragment, or fusion protein comprising the variant CK.

12. The method of claim 10 or 11, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

3.

13. The method of claim 10 or 11, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

4.

14. The method of claim 10 or 11, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

5.

15. The method of claim 10 or 11, wherein the amino acid sequence of the variant CK is set forth in SEQ ID NO:

6.

16. The method of claim 10 or 11, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a human VL and a human VH.

17. The method of claim 10 or 11, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a human IgG CHI domain.

18. The method of claim 17, wherein the human IgG CHI domain is a human IgGl or IgG4 CHI domain.

19. The method of claim 10 or 11, wherein the antigen-binding fragment is a human Fab or Fab'.

20. The method of claim 10 or 11, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a human IgG CH2 domain.

21. The method of claim 20, wherein the human IgG CH2 domain is a human IgGl or IgG4 CH2 domain.

22. The method of claim 10 or 11, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a human IgG CH3 domain.

23. The method of claim 22, wherein the human IgG CH3 domain is a human IgGl or IgG4 CH3 domain.

24. The method of claim 10 or 11, wherein the antibody, antigen-binding fragment, or fusion protein further comprises a

Images