WO2022095977A1 - Method for preparing bispecific antibody - Google Patents

Abstract

Provided in the present invention is a method for improving the correct pairing rate of heavy chains and light chains during a preparation process of a bispecific antibody. The method comprises the steps of removing natural interchain disulfide bonds by means of amino acid substitution and simultaneously forming engineered interchain disulfide bonds by means of amino acid substitution on the CH1-CL domain of one Fab arm. The method of the present invention further comprises the step of reversing the charge of a pair of amino acids on the CH1-CL domain of the Fab arm by means of amino acid substitution. The method of the present invention can significantly improve the correct pairing rate of heavy chains and light chains in the production of a bispecific antibody, and is suitable for various antibody types.

Description

Methods of making bispecific antibodies technical field

The present invention relates to the field of biotechnology; in particular, the present invention relates to a novel method for preparing bispecific antibodies and applications thereof.

Background technique

Recombinant monoclonal antibodies (mAbs) can bind antigens with high efficiency and specificity, presenting exciting opportunities in the fields of biomedicine and biotechnology. Many therapeutic monoclonal antibodies have been successfully used to treat a variety of diseases, including cancer, immune diseases and infections. In addition, bispecific antibodies (bsAbs) targeting more than one antigen have shown great potential to maximize the benefits of antibody therapy.

Antibodies are proteins that recognize and specifically bind to antigens (usually called immunoglobulins, Immunoglobulin, abbreviated as Ig). In most mammals, including humans and mice, immunoglobulins consist of two identical heavy chains and two identical light chains. Each heavy and light chain can be divided into two parts: a constant region (C) and a variable region (V). Based on the difference in the structure of the constant region of the heavy chain of antibodies, immunoglobulins can be divided into five classes: IgA, IgD, IgE, IgG and IgM. Each class can have kappa (κ) or lambda (λ) light chains. Based on heavy chain structure, human IgG can be divided into four subclasses: IgG1, IgG2, IgG3, and IgG4. Full-length IgG contains two identical heavy chains and two identical light chains. The amino-terminal (N-terminal) region of the heavy chain is the variable region (VH) and the remainder is the heavy chain constant region (CH). The heavy chain constant region of human IgG consists of three domains, CH1, CH2 and CH3. The region between the CH1 and CH2 domains is called the hinge region. The amino-terminal half of the light chain is a variable region (VL) domain, while the carboxy-terminal (C-terminal) half of the light chain is a constant region (CL) domain. The variable and constant regions of both heavy and light chains are structurally folded into functional units called domains. Each heavy chain associates with the light chain through disulfide bonds and non-covalent interactions to form a heterodimer. Two light chain-heavy chain heterodimers are linked by disulfide bonds in the hinge regions of the two heavy chains, forming a complex Y-shaped antibody (Figure 1). Papain digestion of IgG (molecular weight 150 kDa) yields two identical fragments that retain antigen-binding activity, termed the Fab fragment (molecular weight 45 KDa), and a crystallizable fragment (Fc fragment, molecular weight 50 KDa). Pepsin digestion of IgG produces a fragment called F(ab') ₂ (100 KDa molecular weight), which consists of two Fab fragments (Figure 2). The two arms of a Y-shaped antibody are also called Fab arms. The Fab arm consists of the heavy chain variable region and the CH1 domain and its paired light chain.

IgG1, IgG2, IgG3 and IgG4 showed the greatest sequence diversity within the hinge region (Figure 3), and their interchain disulfide bond structures showed many similarities and differences. The two heavy chains are linked at the hinge region by varying numbers of interchain disulfide bonds: 2 for IgG1 and IgG4, 4 for IgG2, and 11 for IgG3. The light chain of IgG1 passes through kappa and the cysteine at the C-terminus of the CL domain of the lambda chain (cysteine at position 214 in the kappa light chain [EU numbering], cysteine at position 214 in the lambda light chain [Kabat numbering] ]) and a cysteine at the C-terminus of the heavy chain CH1 domain (cysteine at position 220 in the heavy chain [EU numbering]) is attached to the heavy chain by an interchain disulfide bond. Unlike IgG1, the light chain of IgG2, IgG3 or IgG4 is linked by the cysteine at the C-terminus of the CL domain in the kappa and lambda chains to the cysteine at the N-terminus of the CH1 domain in the heavy chain (cysteine at position 131 in the heavy chain). Interchain disulfide bonds between acids [EU numbering]) are attached to the heavy chain (Figure 1, A and B). Although the positions of cysteines in the amino acid sequence differ between IgG1 and other subtypes (IgG2, IgG3, IgG4), their spatial positions are similar to form interchain disulfide bonds. The disulfide bond between the light and heavy chains is important for the stability of the Fab. A review of the structure and function of immunoglobulins can be found in Immunology (Janis Kuby, W.H. Freeman and Company, New York, 1997).

While antibodies are symmetric structural immunoglobulins with two identical heavy and light chains, bispecific antibodies can be asymmetric in structure, with two different heavy chains, each bound to a different light chain . Therefore, antibodies are bivalent and monospecific and can bind specifically to two identical antigens at the same time, but bispecific antibodies are monovalent and bispecific and can bind two different antigens or epitopes .

To efficiently generate bispecific antibodies in intact IgG format, two major difficulties remain to be resolved. One is how to ensure that the heavy chain A for epitope A is only paired with the heavy chain B for epitope B (heterodimerization between heavy chains A and B), and that the antibody is between two identical heavy chains (A with A, B with B) homodimerization is prevented. Another difficulty is how to ensure that the heavy chain A only pairs with its own light chain A, and not with the light chain B of other epitopes; at the same time, the heavy chain B only pairs with its own light chain B, and does not pair with the light chain A. Some methods of producing bispecific antibodies can be found in the review article, Brinkmann U. & Kontermann R.E., The making of bispecific antibodies, mAbs, 9(2017), 182-212).

The so-called "knobs-into-holes" (KIH) method is commonly used to generate antibody heavy chain heterodimerization. In this approach, there are T366W "knobs" mutations in the CH3 domain of heavy chain A and T366S/L368A/Y407V "holes" mutations in the CH3 domain of heavy chain B [EU numbering] . These mutations generate asymmetric spatially complementary structures in the CH3 domains of the A and B heavy chains, thereby promoting heavy chain heterodimerization through hydrophobic interactions.

Another approach exploits electrostatic interactions, promoting heavy chain heterodimerization through electrostatic attraction and avoiding homodimerization between identical heavy chains through electrostatic repulsion. There is a charge interaction between the CH3-CH3 domains of the two heavy chains of the antibody, K409 and D399. In this method, there are K409D and K392D mutations in the CH3 domain of heavy chain A, and D399K and E356K mutations in the CH3 domain of heavy chain B [EU numbering]. These amino acid mutations can promote the heterodimerization of antibody heavy chains A and B through electrostatic interactions, and inhibit the homodimerization of the same heavy chains.

Although amino acid mutations in the CH3 domain can promote antibody heavy chain heterodimerization and inhibit homodimerization between identical heavy chains, these approaches cannot avoid the problem of light chain mismatches. When two antibody heavy chains form a heterodimer, the two different light chains can create four different combinations with the heavy chains, of which only one is a correctly paired bispecific antibody and the other three are mismatched Heavy and light chains.

A straightforward way to avoid the light chain mismatch problem is to use a common light chain. This requires extensive screening to find a common light chain that can pair with two different heavy chains and retain its high affinity for the two different antigens. However, this approach may not work for all antibodies.

Based on rational design, other methods may be developed to achieve the correct pairing of heavy and light chains. The main principle is to alter the heavy and light chain domains in one Fab arm while leaving the other Fab arm wild-type unchanged, which can facilitate pairing of the light chain with its own heavy chain.

Correct pairing of heavy and light chains can be facilitated by swapping the corresponding domains between the antibody heavy and light chains (CrossMab technology), which facilitates pairing of the wild-type light chain with the wild-type heavy chain, Swapped light chains are paired with corresponding swapped heavy chains. However, domain swapping of heavy and light chains may be detrimental to the structural stability of bispecific antibodies.

Another approach to solving the light chain mismatch problem is to alter the domains of the light and heavy chains based on structural design. By altering the interaction between the variable domain (VH-VL) and constant domain (CH1-CL) of one Fab arm, the light chain is facilitated to bind to its own heavy chain. However, changing the variable domain of Fab may have a greater impact on its binding to antigen.

Pairing of the light chain with its own heavy chain can be facilitated by replacing the natural interchain disulfide bonds in the CH1-CL domain with engineered disulfide bonds. Structural modeling predicts three sets of positions in the CH1-CL domain, where new interchain disulfide bonds may be formed by introducing a pair of cysteines. Comparing these three sets of cysteine mutations showed that one of them could efficiently facilitate pairing of the light chain with its own heavy chain with a correct pairing rate of 98%. In this set of mutations, phenylalanine at position 126 of the heavy chain (H) and serine [EU numbering] at position 121 of the light chain (L) were replaced by cysteine (H: F126C/L: S121C; DuetMab technology) . In addition to this group of cysteine mutations, it is not known whether other cysteine mutations can form new interchain disulfide bonds that facilitate pairing of the light chain with its own heavy chain.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a new method for preparing a bispecific antibody, and the correct pairing rate of the heavy chain and the light chain of the bispecific antibody obtained by the method of the present invention will be significantly improved.

In a first aspect, the present invention provides a method for improving the correct pairing rate of heavy and light chains during the preparation of bispecific antibodies, the method comprising the following steps:

In the CH1-CL domain of one Fab arm, the natural interchain disulfide bond was eliminated by amino acid substitution, while an engineered interchain disulfide bond was formed by amino acid substitution.

In a preferred embodiment, the bispecific antibody is derived from any one of IgG, IgA, IgD, IgE and IgM; preferably, from an IgG molecule; more preferably, from a human or non-human, e.g. Primate or rodent IgG molecules.

In a preferred embodiment, the bispecific antibody is derived from human IgGl, IgG2, IgG3, IgG4.

In a preferred embodiment, cysteine is replaced with an amino acid other than cysteine (eg cysteine is mutated to valine or serine, IgG1 heavy chain C220V, IgG2, IgG3 or IgG4 heavy chain C131S, kappa Light chain C214V [EU numbering], lambda light chain C214V [Kabat numbering]) to eliminate natural interchain disulfide bonds, replacing a pair of non-cysteine amino acids with a cysteine pair to form an engineered interchain disulfide bonds.

In a preferred embodiment, the amino acids that form the engineered interchain disulfide bonds are primarily located in the following three regions of the IgG1 heavy chain domain: F126-T135, G166-T187, K218-S219; the following four regions of the kappa CL domain Areas: S114-S121, N158-T164, T172-T180, F209-E213.

In specific embodiments, the amino acids that form the engineered interchain disulfide bonds in the CH1-CL domain of IgG1 are selected from the following table:

In a specific embodiment, the amino acids in the CH1-CL domain of IgG1 that form the engineered interchain disulfide bonds are selected from the group consisting of each pair of cysteine mutations listed as follows: selected in the heavy chain The amino acid [EU numbering] is mutated to cysteine/selected amino acid [EU numbering] in the kappa light chain is mutated to cysteine:

F126C/F118C、L128C/P120C、A129C/P120C、P130C/F118C、S132C/F118C、K133C/S121C、S134C/F118C、G166C/T172C、G166C/S174C、G166C/T180C、H168C/T180C、F170C/Q160C、F170C/ T164C、F170C/T172C、F170C/S174C、F170C/S176C、F170C/T180C、P171C/T172C、P171C/S174C、P171C/T180C、V173C/S174C、Q175C/S174C、Q175C/S176C、Q175C/T180C、S176C/S162C、 S181C/T172C、S181C/S176C、S181C/T180C、S183C/S114C、S183C/F118C、S183C/Q160C、S183C/S174C、S183C/T178C、S183C/T180C、V185C/T172C、V185C/S174C、V185C/T178C、V185C/ T180C, T187C/S114C, T187C/T172C, T187C/S174C, T187C/T178C, K218C/F118C, S219C/F116C, S219C/F118C, S219C/P120C.

In a preferred embodiment, the amino acids that form the engineered interchain disulfide bonds are mainly located in the following three regions of the IgG1 heavy chain domain: F126-S136, G166-T187, V215-S219; the following four regions of the lambda CL domain Regions: S114-S121, G158-Q167, K172-S180, V209-S215.

In a specific embodiment, the amino acid residues that form the engineered interchain disulfide bonds in the CH1 domain of IgG1 and the corresponding CL domain are selected from the following table:

In a specific embodiment, the amino acid residues forming the engineered interchain disulfide bonds in the CH1 domain of IgG1 and the corresponding CL domain are selected from the group consisting of each pair of cysteine mutations listed as follows : Mutation of selected amino acid [EU numbering] to cysteine in heavy chain / Mutation of selected amino acid [Kabat numbering] in lambda light chain to cysteine; S132C/S121C, K133C/T116C, K133C/P211C, S136C/S121C, F170C/G158C, P171C/T162C, P171C/P164C, S176C/T162C, L179C/G158C, S181C/P164C, V215C/T116C, E216C/F118C.

In a preferred embodiment, the amino acids that form the engineered interchain disulfide bonds are mainly located in the following two regions of the IgG4 heavy chain domain: F126-E137, G166-P189; the following four regions of the kappa CL domain: S114 -S121, N158-E165, T172-S182, F209-E213.

In specific embodiments, the amino acids that form the engineered interchain disulfide bonds in the CH1-CL domain of IgG4 are selected from the following table:

In a specific embodiment, the amino acids in the CH1-CL domain of IgG4 that form the engineered interchain disulfide bonds are selected from the group consisting of each pair of cysteine mutations listed as follows: selected in the heavy chain The amino acid [EU numbering] is mutated to cysteine / the selected amino acid [EU numbering] in the kappa light chain is mutated to cysteine;

A129C/F209C、A129C/N210C、P130C/F116C、P130C/F118C、P130C/P119C、P130C/N210C、P130C/R211C、S132C/S114C、S132C/F116C、S132C/F118C、S132C/P120C、S132C/R211C、S132C/ E213C、R133C/P119C、R133C/R211C、R133C/E213C、T135C/F116C、T135C/P120C、G166C/T178C、H168C/N158C、F170C/F182C、V173C/Q160C、V173C/S162C、Q175C/S162C、Q175C/T180C、 S181C/T172C, S181C/S176C, S183C/N158C, S183C/S176C, V185C/E165C, V185C/T178C.

In a preferred embodiment, the amino acids that form the engineered interchain disulfide bonds are mainly located in the following two regions of the IgG4 heavy chain domain: F126-E137, H168-T187; the following four regions of the lambda CL domain: S114 -S121, G158-Q167, K172-S180, V209-S215.

In specific embodiments, the amino acids that form the engineered interchain disulfide bonds in the CH1-CL domain of IgG4 are selected from the following table:

In a specific embodiment, the method further comprises the step of reversing the charge of a pair of amino acids in the CH1-CL domain of the Fab arm by amino acid substitution.

In specific embodiments, the wild-type positively charged lysine at position 213 of the heavy chain is substituted with a negatively charged amino acid (eg, K213E, K213D); the wild-type negatively charged glutamic acid at position 123 of the light chain Substituted with positively charged amino acids (eg, E123K, E123R).

In a preferred embodiment, in the production of an IgG1 kappa bispecific antibody, a charge reversal (eg, K213D/E123K) and different cysteine pairs are combined to form the CH1-CL domain of the Fab arm Engineering the disulfide bonds in a way to achieve the correct pairing of the heavy and light chains:

S132C/F116C、V173C/N158C、S131C/P120C、S132C/S114C、S132C/P120C、K133C/F116C、K133C/P120C、S134C/P120C、T135C/S114C、G166C/S176C、G166C/T178C、H168C/T172C、H168C/ S174C、H168C/T178C、V173C/T180C、Q175C/N158C、Q175C/T172C、S176C/N158C、S176C/Q160C、G178C/N158C、G178C/S162C、G178C/T164C、S181C/S174C、S183C/P120C、S183C/S162C、 V185C/S114C, V185C/P120C, V185C/S176C, T187C/P120C, T187C/S176C, K218C/S114C, K218C/P120C,

(Each pair of cysteines is listed as follows: wild-type amino acid [EU numbering] in the heavy chain is mutated to cysteine / wild-type amino acid [EU numbering] in the kappa chain is mutated to cysteine).

In a preferred embodiment, in the production of an IgG1 lambda bispecific antibody, the CH1-CL domain of the Fab arm is engineered by combining charge inversion (eg K213D/E123K) with different cysteine pairs disulfide bonds to achieve the correct pairing of heavy and light chains, including:

L128C/T116C, A129C/P211C, A129C/T212C, P130C/A210C, P171C/T163C, E216C/T116C, P217C/S215C, K218C/F118C, K218C/P119C

(Each pair of cysteine mutations is listed as follows: wild-type amino acid [EU numbering] in the heavy chain to cysteine / wild-type amino acid [Kabat numbering] in the lambda chain to cysteine.

In specific embodiments, it can also be achieved by combining charge inversion (eg K213D/E123K) with other cysteine pairs listed above to form an engineered disulfide bond in the CH1-CL domain of the Fab arm. Correct pairing of heavy and light chains is achieved.

In a specific embodiment, the formation of an engineered interchain disulfide bond by amino acid substitution and the charge reversal of a pair of amino acid residues of the CH1-CL domain of the Fab arm by amino acid substitution can occur on the same Fab arm, Can also occur on different Fab arms.

In a preferred embodiment, the method further comprises having a T366W "knob" mutation in the CH3 domain of one heavy chain and a T366S/L368A/Y407V "hole" mutation in the CH3 domain of the other heavy chain; And/or, the CH3 domain of heavy chain A has K409D and K392D mutations (EU numbering), and the CH3 domain of heavy chain B has D399K and E356K mutations.

In a second aspect, the present invention provides a method for improving the correct pairing rate of heavy and light chains during the preparation of bispecific antibodies, the method comprising the steps of: replacing one of the CH1-CL domains of the Fab arm by amino acid substitution The amino acid charge is reversed, where the wild-type positively charged lysine at position 213 of the heavy chain is replaced by a negatively charged amino acid (eg, K213E, K213D); the wild-type negatively charged glutamic acid at position 123 of the light chain is replaced by Positively charged amino acid substitutions (eg, E123K, E123R).

In a third aspect, the present invention provides a method for preparing a bispecific antibody, the method comprising adopting the method described in the first aspect or the second aspect in the preparation process of the bispecific antibody so as to increase the concentration of the bispecific antibody Steps for correct pairing rates of heavy and light chains.

In a fourth aspect, the present invention provides a bispecific antibody, the bispecific antibody is prepared by using the method described in the third aspect, or using the method described in the first aspect or the second aspect to improve the Correct pairing rates of heavy and light chains in bispecific antibodies.

In a preferred embodiment, the antibody is a bispecific antibody specifically obtained in the Examples.

It should be understood that within the scope of the present invention, the above-mentioned technical features of the present invention and the technical features specifically described in the following (eg, the embodiments) can be combined with each other to form new or preferred technical solutions. Due to space limitations, it is not repeated here.

Description of drawings

These figures illustrate various aspects of the technology and are not intended to be limiting. For clarity and ease of illustration, the figures have not been drawn to scale, and in some instances various aspects may be exaggerated or exaggerated to facilitate understanding of certain aspects.

Figure 1. Schematic representation of the domains of IgGl and IgG4 antibodies. IgG antibodies are Y-type tetramers with two heavy chains (longer) and two light chains (shorter). The light chain is linked to the heavy chain by interchain disulfide bonds (-S-S-) on the CL and CH1 domains. The two heavy chains are linked together by an interchain disulfide bond (-S-S-) in the hinge region. VL: light chain variable domain, CL: light chain constant domain, VH: heavy chain variable domain, CH1: heavy chain constant domain 1, CH2: heavy chain constant domain 2, CH3: heavy chain constant structure Domain 3.

Figure 2. Schematic representation of the structure of IgG fragments. Papain digestion of IgG yields two identical Fab fragments (retaining antigen binding activity) and one Fc fragment. Pepsin digestion of IgG yields the F(ab') ₂ fragment, which consists of two Fab-like fragments.

Figure 3. Amino acid sequences of the hinge regions of human IgGl, IgG2, IgG3 and IgG4. "_" denotes a cysteine that forms a disulfide bond between the two heavy chains. Sequences were obtained from the IMGT database (www.imgt.org).

Figure 4. Amino acid sequence comparison and EU numbering of the CH1 domains of human IgG1, IgG2, IgG3 and IgG4. "*" indicates sequence identity. ":" indicates that one of the following groups is fully conserved: STA, NEQK, NHQK, NDEQ, QHRK, MILV, MILF, HY or FYW. "." indicates that one of the following groups is fully conserved: CSA, ATV, SAG, STNK, STPA, SGND, SNDEQK, NDEQHK, NEQHRK, FVLIM or HFY. The CH1 sequences of human IgG1 (P01857), IgG2 (P01859), IgG3 (P01860) and IgG4 (P01861) were obtained from the Uniprot database (www.uniprot.org). The Clustal Omega program in Uniprot was used for amino acid sequence comparison.

Figure 5. Amino acid sequence comparison of the CL domains of human antibody light chain kappa (κ) [EU numbering] and lambda (λ) [Kabat numbering]. "*" indicates sequence identity. ":" indicates that one of the following groups is fully conserved: STA, NEQK, NHQK, NDEQ, QHRK, MILV, MILF, HY or FYW. "." indicates that one of the following groups is fully conserved: CSA, ATV, SAG, STNK, STPA, SGND, SNDEQK, NDEQHK, NEQHRK, FVLIM or HFY. The CL sequences of the human antibody kappa light chain (P01834) and lambda light chain (PODOY3) are available from the Uniprot database (www.uniprot.org). The Clustal Omega program in Uniprot was used for amino acid sequence comparison.

Figure 6. Schematic representation of bispecific IgG structure. (A) shows mutant interchain disulfide bonds in the Fab arm of Antibody 1 and wild-type interchain disulfide bonds in the Fab arm of Antibody 2. The "knob" mutation in the heavy chain of antibody 1 and the mutation of the "hole" in the heavy chain of antibody 2 promote heavy chain heterodimerization. RF mutations in the antibody 2 heavy chain were used to remove antibody homodimers in protein A purification. The 6xHis tag at the C-terminus of the antibody 2 heavy chain was used for western blotting. (B) shows mutant interchain disulfide bonds and charge inversion in the Fab arm of Antibody 1, and wild-type interchain disulfide bonds and residue charges in the Fab arm of Antibody 2.

Figure 7. ELISA identifies a library of bispecific IgG1(kappa) cysteine mutations in which the natural interchain disulfide bond in the CH1-CL domain of one Fab arm is replaced by an engineered disulfide bond (each pair of cysteine Acid mutations are listed as follows: heavy chain wild amino acid [EU numbering] to cysteine / kappa light chain wild amino acid [EU numbering] to cysteine), WT represents two of the two Fab arms Natural interchain disulfide bonds.

Figure 8. ELISA identifies a library of bispecific IgG1 (kappa) cysteine and charge mutations in which the CH1-CL domain K213/E123 in one Fab arm is reversed to opposite charge and native interchain disulfide bonds are engineered The disulfide bond substitutions (charge inversions) are listed as follows: K213 [EU numbering] in the heavy chain is mutated to D or E / E123 [EU numbering] in the light chain is mutated to K or R (K213D/E123K, K213E/E123K , K213D/E123R, K213E/E123R). Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] mutation to cysteine/Kappa light chain wild amino acid [EU numbering] mutation to half Cystine, WT represents the two native interchain disulfide bonds in the two Fab arms.

Figure 9. ELISA identifies a library of bispecific IgG1 (lambda) cysteine mutations in which the natural interchain disulfide bond in the CH1-CL domain of one Fab arm is replaced by an engineered disulfide bond (each pair of cysteine Acid mutations are listed as follows: heavy chain wild-type amino acid [EU numbering] to cysteine / lambda light chain amino acid [Kabat numbering] to cysteine, WT represents the two native chains in the two Fab arms inter-disulfide bond.

Figure 10. ELISA identifies a library of bispecific IgG1 (lambda) cysteine and charge mutations in which the CH1-CL domains K213/E123 in one Fab arm are reversed to opposite charges and the native interchain disulfide bonds are engineered The engineered disulfide bond substitutions (charge inversion) are listed as follows: K213 [EU numbering] in the heavy chain is mutated to D / E123 [Kabat numbering] in the light chain is mutated to K (K213D/E123K). Each pair of cysteine Acid mutations are listed as follows: heavy chain wild amino acid [EU numbering] to cysteine / lambda light chain wild amino acid [Kabat numbering] to cysteine, WT represents the two natural in the two Fab arms Interchain disulfide bonds.

Figure 11. ELISA identifies bispecific IgG1 (kappa/lambda) in which the natural interchain disulfide bond in the CH1-CL domain of one Fab arm is replaced by an engineered interchain disulfide bond (each pair of cysteine mutations Listed as: heavy chain wild amino acid [EU numbering] mutated to cysteine / kappa light chain wild amino acid [EU numbering] mutated to cysteine, WT represents the two native interchains on the two Fab arms disulfide bonds.

Figure 12. ELISA identifies bispecific IgG1 in which the CH1-CL domain K213/E123 in one Fab arm is reversed to opposite charge and the CH1-CL domain native interchain disulfide bond is engineered in the other Fab arm The disulfide bond substitutions (charge inversions) are listed as follows: K213 mutation in heavy chain to D / E123 mutation in light chain to K (K213D/E123K). Each pair of cysteine mutations is listed as follows: heavy Chain wild amino acid [EU numbering] is mutated to cysteine / Light chain wild amino acid [EU numbering] is mutated to cysteine, WT indicates natural interchain disulfide bonds in both Fab arms, and there is no charge inversion Transmutation.

Figure 13. ELISA identifies a library of bispecific IgG4(kappa) cysteine mutations in which the natural interchain disulfide bond in the CH1-CL domain of one Fab arm is replaced by an engineered disulfide bond (each pair of cysteine Acid mutations are listed as follows: heavy chain wild amino acid [EU numbering] to cysteine / kappa light chain wild amino acid [EU numbering] to cysteine. K147D/T129R is represented in CH1 of one Fab arm Lysine 147 was mutated to aspartic acid (K147D), and threonine 129 in CL was mutated to arginine (T129R).

Detailed ways

After extensive and intensive research, the inventors have provided a method and related compositions for the preparation of bispecific antibodies capable of binding two different antigens. In particular, the present invention provides methods for producing bispecific antibodies from two existing antibodies comprising mutating amino acids at the interface of the heavy and light chains. Such mutations include replacement of cysteine with other amino acids to eliminate native interchain disulfide bonds, and replacement of cysteine with other amino acids to form engineered interchain disulfide bonds. On the basis of these mutations, changes in charge interactions through amino acid mutations can also be included. In the presence of these mutations, bispecific antibodies can be generated in which the heavy chain preferentially pairs with its own light chain and which prevents mispairing of heavy and light chains. Thus, the present invention has been completed.

Definition of Terms

Unless otherwise defined, scientific and technical terms used in the present invention shall have the meanings commonly understood by one of ordinary skill in the art.

As used herein, terms in the singular may use the corresponding terms in the plural when appropriate, and vice versa.

As used herein, the term "amino acid" or "amino acid residue" includes both natural and unnatural amino acids.

Amino acids are generally referred to herein by either the well-known three-letter symbols or the one-letter symbols recommended by the IUPAC-IUB Biochemical Nomenclature Committee. Likewise, nucleotides are often represented by recognized one-letter codes.

The term "amino acid mutation" herein includes amino acid substitutions, insertions and/or deletions in a polypeptide sequence. An "amino acid substitution" or "substitution" as used herein refers to the replacement of an amino acid at a particular position in a polypeptide sequence with another amino acid.

As used herein, the term "natural" or "wild-type" amino acid refers to an amino acid residue that occurs naturally at a particular position in a polypeptide and has not been modified by mutation.

The terms "polypeptide," "oligopeptide," "peptide," and "protein" are used interchangeably herein to refer to chains of amino acids of any length. Proteins can be produced by any method known in the art for protein synthesis, in particular by chemical synthesis or by recombinant expression techniques.

As used herein, the terms "vector" and "plasmid" in protein expression are used interchangeably.

As used herein, "cell", "cell line" and "cell culture" in protein expression are used interchangeably.

"Cell transformation" refers to the introduction of exogenous DNA into a cell. It is usually the result of the integration of foreign DNA into the genome or the introduction of a self-replicating plasmid.

Transformation and transfection of host cells can be carried out according to methods well known to those skilled in the art. Suitable transformation methods include viral infection, transfection, conjugation, protoplast fusion, electroporation, particle gun techniques, calcium phosphate precipitation, direct microinjection, and the like. The choice of method generally depends on the type of cells transformed and the environment in which the transformation takes place. A general discussion of these methods can be found in the literature (Ausubel, et al., Short Protocols in Molecular Biology, Wiley & Sons, 1995).

Yeast transformation can be performed using different methods, including spheroplast methods, electroporation, polyethylene glycol methods, alkali metal cation methods, etc. (Gregg JM, Pichia Protocols, Totowa, New Jersey: Humanna Press, 2010).

As used herein, the term "antigen" refers to any substance to which an antibody specifically binds. For example, the antigen can be a protein, polypeptide, carbohydrate, polynucleotide, lipid, or a combination of the foregoing.

As used herein, the term "epitope" refers to a molecular site on an antigen that is recognized and bound by a particular antibody.

As used herein, the terms "antibody" and "immunoglobulin" are used interchangeably in the broadest sense. Antibodies are proteins (immunoglobulins) that recognize and specifically bind to antigens. Antibodies include monoclonal antibodies, polyclonal antibodies, multispecific antibodies comprising at least two distinct epitope binding domains (eg, bispecific antibodies), human antibodies, humanized antibodies, camelid antibodies, chimeric antibodies, antibody fragments, Fusion proteins comprising the antigen-binding portion of an antibody as well as any other modified antibody molecule comprising an antigen-binding site. Antibodies can be derived from any mammal including, but not limited to, humans, monkeys, goats, horses, rabbits, dogs, cats, mice, chickens, camels, sharks, or other animals.

As used herein, the term "recombinant antibody" is intended to include all antibodies produced by host cells, such as yeast or CHO cells, transfected with a recombinant expression vector.

Immunoglobulins are composed of two identical heavy chains and two identical light chains. Based on their heavy chain constant region structure, immunoglobulins can be divided into five classes, namely IgA, IgD, IgE, IgG and IgM. Each class can have kappa (κ) or lambda (λ) light chains. Based on its heavy chain structure, human IgG can be divided into four subclasses, namely IgG1, IgG2, IgG3, IgG4. The "antibodies" of the present invention may be of any class or subclass. Preferably, the antibody of the present invention is human IgG.

Herein, the CH1, CH2, CH3 heavy chain constant regions of human IgG (IgG1, IgG2, IgG3 and IgG4) are numbered using the "EU numbering system" for their amino acid sequences (Edelman GM et al., Proc Natl Acad Sci USA, 63 (1):78-85(1969)). IMGT database (

International ImMunoGeneTics Information System

) fully lists the amino acid sequences of CH1, hinge, CH2 and CH3 constant regions of human IgG1 and their corresponding numbers.

For human IgG kappa light chain constant region, its amino acid sequence numbering also adopts "EU numbering system". The IMGT database completely lists the amino acid sequence of the constant region of the human kappa light chain and the corresponding numbering.

For the human IgG lambda light chain constant region, the amino acid sequence numbering adopts the "Kabat numbering system" (Kabat EA et al, sequences of proteins of immunological interest. 5th Edition-US Department of Health and Human Services, NIH publication, 91-3242 ( 1991)). The IMGT database completely lists the amino acid sequence of the constant region of human lambda light chain and the corresponding number.

Human IgG1 heavy chain constant region and hinge region boundaries are defined in the IMGT database as follows: CH1 constant region is defined as amino acids 118 to 215, hinge region is defined as amino acids 216 to 230, CH2 constant region is defined as amino acids 231 to 340, CH3 constant region is defined as amino acids 231 to 340 Defined as amino acids 341 to 447. The amino acid sequences of the CH1 constant regions of IgG1, IgG2, IgG3 and IgG4 are highly conserved (Fig. 4). However, unlike the hinge region definition of IMGT, based on the crystallographic data, the hinge region on the IgG1 spatial structure is defined as amino acids 221-237. Thus in the present invention, the CH1 constant region of IgG1 is defined as amino acids 118 to 220, the hinge region is defined as amino acids 221-237, and the CH2 constant region is defined as amino acids 238 to 340 [EU numbering]. The human kappa light chain constant region is defined as amino acids 108 to 214 [EU numbering] and the human lambda light chain constant region is defined as amino acids 107A-215 [Kabat numbering] times, as shown in FIG. 5 .

Kabat lists many amino acid sequences for each subtype of antibodies and lists the most common amino acids at each position in each subtype, thereby listing the conserved sequences. Kabat numbers the individual amino acids in the listed sequences, and this numbering has become standard in the art. It will be appreciated that due to allotype and allelic variation in the population, the wild-type amino acid residues at these positions may differ from those listed, and therefore there may be individual differences between the sequences presented and those of the prior art Amino acid differences.

The position of a particular amino acid can vary among the four subtypes of IgG and among IgA, IgD, IgE, and IgM. Therefore, the position of a specific amino acid is not limited to this specific amino acid position in an immunoglobulin, but should include those corresponding amino acid positions in all immunoglobulins.

Those skilled in the art may have different definitions of amino acids corresponding to each domain of an IgG molecule. Thus, the N- or C-terminus of the above domains can be extended or shortened by 1, 2, 3, 4, 5, 6, 7, 8, 9 or even 10 amino acids.

Herein, amino acid mutations are represented by the following methods: wild amino acid, amino acid position, mutated amino acid. Amino acids are represented by one-letter codes, and amino acid positions are using the EU numbering system (IgG heavy chain constant region and kappa light chain constant region), and the Kabat numbering system (Lambda light chain constant region). For example, H(C220V) represents the mutation of wild-type cysteine at position 220 in the antibody heavy chain (H) to valine.

The term "full-length IgG" or "intact IgG" of the present invention refers to a structurally intact IgG, but it may not have all the functions of an IgG. Full-length IgG contains two heavy chains and two light chains. Each heavy chain binds to the light chain through interchain disulfide bonds and non-covalent interactions, forming a heterodimer. The two heavy chains are connected by an interchain disulfide bond in the hinge region.

As used herein, the term "disulfide bond" includes a covalent bond formed between two cysteine residues. Cysteine has a thiol group on it, which can form a disulfide bond with a thiol group on another cysteine. "Intrachain disulfide bond" refers to a disulfide bond formed between two cysteines within the same protein chain. "Interchain disulfide bond" refers to a disulfide bond formed between two cysteines on different chains of the same protein or between two cysteines in different proteins. In the present invention, the terms "cysteine pair" or "cysteine pair" have the same meaning and thus can be used interchangeably, and both refer to two cysteine residues capable of forming a disulfide bond base.

As used herein, the term "protein domain" refers to a portion of a protein that is sterically foldable, has biological functions, and can exist independently of the rest of the protein. Similarly, the term "CH1-CL domain" as used herein refers to the protein structure formed by the interaction of the heavy chain CH1 domain with the light chain CL domain in an antibody.

As used herein, the term "interface" refers to the area where separate protein domains come into contact with each other.

The term "antibody mutant" includes antibodies that do not occur in nature, as well as other non-wild-type antibodies in which at least one amino acid or amino acid side chain structure differs from that of the wild-type antibody.

As used herein, the term "antibody mutant" also includes other forms of antibodies that do not occur naturally, such as bispecific antibodies and antibody fragments (eg, Fab, F(ab')2, etc.).

The terms "Fab portion", "Fab arm", "Fab" or "arm" are used interchangeably herein.

As used herein, "charge inversion" refers to the substitution of an amino acid residue of a certain charge by an amino acid residue of the opposite charge. For example, the wild-type positively charged lysine at position 213 in the CH1 domain is replaced by a negatively charged amino acid (K213E or K213D); the wild-type negatively charged glutamic acid at position 123 in the CL domain is replaced by a positively charged amino acid Amino acid substitution (E123K or E123R).

In some aspects, the antibodies provided herein are bispecific. As used herein, "bispecific antibodies" (bsAbs) are antibodies that have binding specificities for at least two different antigens or at least two different epitopes within the same antigen. Antibodies are symmetric structural immunoglobulins with two identical heavy and light chains, but bispecific antibodies can be asymmetric, with two different heavy chains, each paired with its own light chain. Therefore, an antibody is bivalent and monospecific, which means that it can bind specifically to two of the same antigens or epitopes at the same time, but a bispecific antibody is monovalent and bispecific, with each Fab arm They all specifically bind to different antigens or epitopes, respectively.

Bispecific antibody of the present invention and preparation method thereof

Based on the KIH technique of heavy chain heterodimerization, we constructed a library of IgG mutants with the following substitutions in the CH1 and CL domains of the first Fab arm: (a) mutated wild cysteine to other amino acids to eliminate disulfide bonds between heavy and light chains; (b) predicted wild-type amino acid pairs were mutated to cysteine pairs for the formation of engineered interchain disulfide bonds; (c) in the CH1 domain The naturally positively charged lysine at position 213 is mutated to a negatively charged amino acid (K213E, K213D) substitution; the naturally negatively charged glutamate at position 123 in the CL domain is mutated to a positively charged amino acid (E123K , E123R) substituted. The corresponding amino acids in the CH1 and CL domains of the second Fab arm of IgG were not mutated (wild type, WT). Like mammalian cells, glycoengineered Pichia can correctly carry out cell biological mechanisms such as protein folding, disulfide bond formation, and glycosylation modifications (Ellgaard L. and Helenius A., Quality control in the endoplasmic reticulum, Nat. Rev. Mol. Cell Biol. 4 (2003) 181-191). Thus, a library of IgG mutants can be expressed in Glycoengineered Pichia pastoris, and the expressed IgG mutants can be screened by ELISA to identify those mutated introduced cysteine pairs Engineered to form interchain disulfide bonds to facilitate the correct pairing of heavy and light chains in bispecific antibodies.

The bispecific antibody of the present invention and its preparation method can be applied to various subtypes of human IgG (IgG1, IgG2, IgG3, IgG4) and other classes of Ig. In another aspect, the bispecific antibodies of the present invention and methods of making them can be applied to non-human (eg primate or rodent) different subtypes of IgG (eg murine IgGl, IgG2a, IgG2b or IgG3 antibodies).

In some aspects, the invention provides methods of producing bispecific antibodies. Bispecific antibodies can be prepared as full-length antibodies or antibody fragments, such as F(ab')2. Furthermore, the bispecific antibodies provided herein are easy to express, stable and have low immunogenicity. The bispecific antibody structures described herein provide a good platform for generating bispecific antibodies that can realize the advantages associated with bispecific antibodies while reducing potential therapeutic risks. In certain aspects, bispecific antibodies can specifically bind two different antigens or two different epitopes on the same antigen. The two Fab arms of bispecific antibodies typically contain two distinct variable regions. In some aspects, the binding affinity of the two Fab arms to two independent antigens can be about the same. In some aspects, the binding affinity of the two Fab arms to two separate antigens can be different. In some aspects, the binding affinity of the two Fab arms to two independent epitopes on the same antigen can be about the same. In some aspects, the binding affinity of the two Fab arms to two independent epitopes on the same antigen can be different. In other aspects, the two Fab arms may have the same specificity (eg, bind the same or overlapping epitopes), but may differ in binding affinity. In some aspects, two antibodies with different in vivo potency can be combined into a bispecific antibody, where one Fab arm has high affinity and the other Fab arm has low affinity, which may prevent over- or under-dosing of one of the arms .

In certain aspects, provided herein are "cysteine mutated" polypeptides (eg, heavy and light chains) for use in generating bispecific antibodies and preventing mispairing of heavy and light chains. In some aspects, wild cysteines in the heavy and light chains of antibody alpha against the antigen (or epitope) alpha are mutated to other amino acids to eliminate native interchain disulfide bonds, and the wild amino acids are mutated to cysteine to form Engineered interchain disulfide bonds. Such mutations provided herein are in the CH1 and CL domains. Antigen B has no cysteine mutations in the heavy and light chains of Antigen B. By cysteine mutation of the heavy and light chains of Antibody A, these four polypeptides (heavy and light chains) can be brought together so that the heavy chain of Antibody A pairs correctly with its light chain, while Antibody B The heavy chain of antibody A is correctly paired with its light chain, while preventing the heavy chain of antibody A from mispairing with the light chain of antibody B, and the heavy chain of antibody B and the light chain of antibody A mispairing. As used herein, the term "cysteine unmutated" polypeptides (e.g., heavy and light chains) refers to heavy and light chains that contain wild cysteines to form native interchain disulfide bonds. Such "cysteine-mutated" and "unmutated" heavy and light chains may contain other mutations, eg, in the Fc region described herein and/or known in the art, to promote heavy chain heterodimetry polymerization.

In some aspects, the wild-type cysteine in the heavy chain of IgGl, IgG2, IgG3, or IgG4 that forms the disulfide bond between the CHl and CL chains is mutated to other amino acids. The wild cysteine at position 220 [EU numbering] of the IgG1 heavy chain was mutated to other amino acids. The wild cysteine at position 131 [EU numbering] of IgG2, IgG3 and IgG4 heavy chains was mutated to other amino acids. In some aspects, the wild cysteines that form the CH1 and CL interchain disulfide bonds in IgG kappa and lambda light chains are mutated to other amino acids. The wild cysteine at position 214 [EU numbering] of the IgG kappa light chain was mutated to other amino acids. The wild cysteine at position 214 [Kabat numbering] of the IgG lambda light chain was mutated to other amino acids. In some aspects, such other amino acids include naturally occurring and/or non-classical amino acids. Other naturally occurring amino acids include glycine, alanine, valine, leucine, isoleucine, proline, serine, threonine, methionine, histidine, lysine, arginine, glutamic acid amino acid, aspartic acid, glutamine, asparagine, phenylalanine, tyrosine and tryptophan. Non-classical amino acids include, but are not limited to, ornithine, diaminobutyric acid, norleucine, pyranalanine, thienylalanine, naphthalanine, and phenylglycine. Preferred other amino acids are valine, serine or alanine.

In some aspects, the wild-type amino acids in the CH1 and CL domains of IgG1 (kappa) are mutated to different cysteine pairs. Table 1 summarizes the mutation of wild-type amino acids in IgG1(kappa) to different cysteine pairs that can form engineered interchain disulfide bonds to facilitate the correct pairing of heavy and kappa light chains in bispecific antibody production.

Listing 1: Mutation of IgG1 (kappa) cysteine pairs to form engineered interchain disulfide bonds to facilitate proper pairing of heavy and kappa light chains.

F126C/F118C、L128C/P120C、A129C/P120C、P130C/F118C、S132C/F118C、K133C/S121C、S134C/F118C、G166C/T172C、G166C/S174C、G166C/T180C、H168C/T180C、F170C/Q160C、F170C/ T164C、F170C/T172C、F170C/S174C、F170C/S176C、F170C/T180C、P171C/T172C、P171C/S174C、P171C/T180C、V173C/S174C、Q175C/S174C、Q175C/S176C、Q175C/T180C、S176C/S162C、 S181C/T172C、S181C/S176C、S181C/T180C、S183C/S114C、S183C/F118C、S183C/Q160C、S183C/S174C、S183C/T178C、S183C/T180C、V185C/T172C、V185C/S174C、V185C/T178C、V185C/ T180C, T187C/S114C, T187C/T172C, T187C/S174C, T187C/T178C, K218C/F118C, S219C/F116C, S219C/F118C, S219C/P120C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine / kappa chain wild amino acid [EU numbering] to cysteine, for example: F126C/F118C Indicates the mutation of F (phenylalanine) at position 126 of the heavy chain to C (cysteine) / the mutation of F (phenylalanine) at position 118 of the kappa light chain to (cysteine).

In some aspects, the wild-type amino acids in the CH1 and CL domains of IgG1 (lambda) are mutated to different cysteine pairs. Table 2 summarizes the mutation of IgG1 (lambda) wild-type amino acids to different cysteine pairs to enable the formation of engineered interchain disulfide bonds to facilitate the correct pairing of heavy and lambda light chains in bispecific antibody production.

Listing 2: Mutation of IgG1 (lambda) cysteine pairs to form engineered interchain disulfide bonds to facilitate correct pairing of heavy and lambda light chains.

S132C/S121C, K133C/T116C, K133C/P211C, S136C/S121C, F170C/G158C, P171C/T162C, P171C/P164C, S176C/T162C, L179C/G158C, S181C/P164C, V618C/T118C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine / lambda chain wild amino acid [Kabat numbering] to cysteine. For example: S132C/S121C means S (serine) at position 132 of the heavy chain was mutated to C (cysteine)/S (serine) at position 121 of the lambda light chain was mutated to C (cysteine).

Provided herein are bispecific antibodies having cysteine mutations in the CH1 and CL domains, these mutated antibodies may further comprise one or more mutations in the Fc region described below. An Fc region comprising one or more mutations is referred to herein as a "mutated Fc region." The interface between a pair of antibody Fcs can be mutated to promote heavy chain heterodimerization, including but not limited to "KIH" and electrostatic interaction mutations.

In some aspects, provided herein are mutations with engineered interchain disulfide bonds in the CH1 and CL domains of one Fab arm and native interchain disulfide bonds in the CH1 and CL domains of the other Fab arm Antibody. Typically, the CH1 and CL domains of one Fab arm are mutated as follows: wild cysteine is mutated to other amino acids, and wild amino acid is mutated to cysteine, thereby forming new disulfide bonds in the CH1 and CL domains to replace the natural disulfide Sulfur bond. For example, Tables 1 and 2 summarize the mutation of IgG1 wild-type amino acids to a pair of cysteines to form engineered interchain disulfide bonds in the CH1 and CL domains.

In certain aspects, provided herein are "charge mutant" polypeptides (eg, heavy and light chains) for use in generating bispecific antibodies and preventing mispairing of heavy and light chains. In some aspects, the positively charged wild lysine at position 213 [EU numbering] of the antibody alpha heavy chain against antigen alpha is mutated to a negatively charged amino acid, such as an aspartic acid and glutamic acid (K213E, K213D) substitution. The negatively charged wild-type glutamic acid at position 123 of the antibody alpha light chain [EU numbering in kappa, Kabat numbering in lambda] was mutated to positively charged amino acids such as lysine and arginine (E123K, E123R). Such mutations provided herein that alter charge polarity are located in the CH1 and CL domains. Antibody B heavy and light chains against Antigen B do not have these "charge mutations". By "charge-mutating" the heavy and light chains of Antibody A, these four polypeptides (heavy and light) can be brought together so that Antibody A's heavy chain pairs correctly with its light chain, while Antibody B's heavy chain pairs correctly with its light chain. The heavy chain is correctly paired with its light chain while preventing the mispairing of the heavy chain of Antibody A with the light chain of Antibody B, which in turn prevents the mispairing of the heavy chain of Antibody B with the light chain of Antibody A. As used herein, the term "unmutated" means that the heavy and light chains do not contain mutations that alter the charge polarity of wild-type amino acids in CH1 and CL as described herein. Such "charge-mutated" and "unmutated" heavy and light chains may contain other mutations, eg, mutations in the Fc region described herein and/or known in the art, to facilitate heavy chain heterodimerization change.

Provided herein are bispecific antibodies with charge mutations in the CH1 and CL domains, these mutated antibodies may further comprise one or more mutations in the Fc region described below (mutated Fc region) to promote heavy chain heterodimerization transformations, including but not limited to "KIH" and electrostatic interaction mutations.

In certain aspects, provided herein are "cysteine and charge mutant" polypeptides (eg, heavy and light chains) for use in generating bispecific antibodies and preventing mispairing of heavy and light chains. In some aspects, the antibody alpha heavy chain CH1 domain against antigen alpha comprises the following mutations: (a) wild cysteine is mutated to other amino acids (eg, IgG1C220V; IgG2, IgG3, and IgG4C131S) to eliminate native interchain disulfide bonds, Wild amino acids were mutated to cysteine (as listed in Tables 1, 2, 3, 4, 5) to form engineered interchain disulfide bonds; (b) the positively charged wild lysine at position 213 was mutated to Negatively charged amino acids (eg K213D, K213E). The CL domain of the alpha light chain of an antibody against antigen alpha contains the following mutations: (a) the wild cysteine is mutated to other amino acids (eg kappa and lambda light chain C214V) to eliminate the native interchain disulfide bond, the wild amino acid is mutated to Cysteines (as listed in Tables 1, 2, 3, 4, 5) to form new interchain disulfide bonds; (b) negatively charged wild glutamic acid at position 123 was mutated to a positively charged amino acid (eg E123K, E123R). Antibody B heavy and light chains against Antigen B do not have these "cysteine and charge mutations". By "cysteine and charge mutagenesis" of the heavy and light chains of Antibody A, these four polypeptides (heavy and light chains) can be brought together so that the heavy chain of Antibody A is paired correctly with its light chain , while the heavy chain of antibody B correctly pairs with its light chain, while preventing the mispairing of the heavy chain of antibody A with the light chain of antibody B, and the mispairing of the heavy chain of antibody B with the light chain of antibody A. As used herein, the term "unmutated" refers to heavy and light chains that do not contain cysteine and charge mutations in the CH1 and CL domains described herein. Such "cysteine and charge mutated" and "unmutated" heavy and light chains may contain other mutations, such as those in the Fc region described herein and/or known in the art, to promote heterogeneity of the heavy chain. dimerization. Table 3 summarizes the mutation of IgG1(kappa) wild-type amino acids to different pairs of cysteines that synergize with "charge mutations" to form engineered interchain disulfide bonds that promote heavy chain and Correct pairing of kappa light chains.

Listing 3: Mutations of IgG1 (kappa) wild-type amino acids to different cysteine pairs, under the synergistic effect of "charge mutations", can form engineered interchain disulfide bonds to facilitate the correct pairing of heavy and light chains.

S132C/F116C、V173C/N158C、S131C/P120C、S132C/S114C、S132C/P120C、K133C/F116C、K133C/P120C、S134C/P120C、T135C/S114C、G166C/S176C、G166C/T178C、H168C/T172C、H168C/ S174C、H168C/T178C、V173C/T180C、Q175C/N158C、Q175C/T172C、S176C/N158C、S176C/Q160C、G178C/N158C、G178C/S162C、G178C/T164C、S181C/S174C、S183C/P120C、S183C/S162C、 V185C/S114C, V185C/P120C, V185C/S176C, T187C/P120C, T187C/S176C, K218C/S114C, K218C/P120C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine/kappa chain wild amino acid [EU numbering] to cysteine).

In some aspects, "cysteine and charge mutations" can be applied to the CH1 and CL domains of IgG1 (lambda). Table 4 summarizes the mutation of IgG1 (lambda) wild-type amino acids to different cysteine pairs that, under the synergistic effect of "charge mutations", form engineered interchain disulfide bonds that promote heavy chain and Correct pairing of lambda light chains.

Listing 4: Mutations of IgG1 (lambda) wild-type amino acids to different cysteine pairs, under the synergistic effect of "charge mutations" (eg, K213D/E123K), can form engineered interchain disulfide bonds, promoting heavy chain and Correct pairing of light chains.

L128C/T116C, A129C/P211C, A129C/T212C, P130C/A210C, P171C/T163C, E216C/T116C, P217C/S215C, K218C/F118C, K218C/P119C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine/lambda chain wild amino acid [Kabat numbering] to cysteine).

In some aspects, wild-type amino acids in the CH1 and CL domains of IgG4 (kappa) are mutated to a pair of cysteines. Table 5 summarizes the mutation of wild-type amino acids in IgG4(kappa) to different cysteine pairs that can form engineered interchain disulfide bonds to facilitate the correct pairing of IgG4 heavy and kappa light chains.

Listing 5: IgG4 (kappa) cysteine pair mutations to form engineered interchain disulfide bonds in the CH1 and CL domains to facilitate correct pairing of heavy and kappa light chains in bispecific antibody production.

A129C/F209C、A129C/N210C、P130C/F116C、P130C/F118C、P130C/P119C、P130C/N210C、P130C/R211C、S132C/S114C、S132C/F116C、S132C/F118C、S132C/P120C、 S132C/R211C、S132C/ E213C、R133C/P119C、R133C/R211C、R133C/E213C、T135C/F116C、T135C/P120C、G166C/T178C、H168C/N158C、F170C/F182C、V173C/Q160C、V173C/S162C、Q175C/S162C、Q175C/T180C、 S181C/T172C, S181C/S176C, S183C/N158C, S183C/S176C, V185C/E165C, V185C/T178C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine/kappa light chain wild amino acid [EU numbering] to cysteine).

In some aspects, cysteine pair mutations such as those listed in Tables 1, 2, and 5 can also form engineered interchain disulfide bonds in synergy with "charge mutations" (eg, K213D/E123K), Facilitates the correct pairing of heavy and light chains in bispecific antibody production.

In certain aspects, among the bispecific antibodies provided herein, the bispecific antibodies have "cysteine and charge mutations" in the CH1 and CL domains of antibody A, comprising: (a) a wild-type cysteine mutation to Other amino acids to eliminate native interchain disulfide bonds, wild amino acids were mutated to different cysteine pairs (as listed in Tables 1, 2, 3, 4 and 5) to form engineered interchain disulfide bonds; (b ) The positively charged wild lysine at position 213 in CH1 is mutated to a negatively charged amino acid (eg K213D, K213E); the negatively charged wild glutamic acid at position 123 in CL is mutated to a positively charged amino acid (eg E123K, E123R). These "cysteine and charge mutations" are absent in the CH1 and CL domains of Antibody B. Antibody A and Antibody B may further comprise one or more mutations in the Fc region described below (mutated Fc region) to promote heavy chain heterodimerization, including but not limited to "KIH" and electrostatic interaction mutations.

In some aspects, the bispecific antibodies provided herein are bispecific antibodies having cysteine mutations in the CH1 and CL domains of Antibody A and charge mutations in the CH1 and CL domains of Antibody B. The CH1 and CL domains of antibody A contain the following mutations: wild cysteine to other amino acids to eliminate native interchain disulfide bonds, wild amino acids to cysteine pairs (as in Tables 1, 2, 3, 4 and 5) to form new interchain disulfide bonds. The CH1 and CL domains of antibody B contain the following mutations: positively charged wild lysine at position 213 in CH1 is mutated to a negatively charged amino acid (eg K213D, K213E); negatively charged wild glutamic acid at position 123 in CL Mutations to positively charged amino acids (eg E123K, E123R). Antibody A and Antibody B may further comprise one or more mutations in the Fc region described below (mutated Fc region) to promote heavy chain heterodimerization, including but not limited to "KIH" and electrostatic interaction mutations.

In an advantageous aspect, the bispecific antibodies of the invention are derived from human IgG molecules. Bispecific antibodies of the invention can generally be generated from any suitable type of immunoglobulin. In another aspect, the bispecific antibodies of the invention are derived from non-human (eg primate or rodent) IgG molecules (eg murine IgGl, IgG2a, IgG2b or IgG3 antibodies).

Host cell lines for antibody expression are preferably mammalian cells; they have the cellular biological machinery for correct steric folding, disulfide bond formation and glycosylation modifications. Such mammalian host cells include, but are not limited to: CHO (Chinese hamster ovary), 293 (human kidney), CVI (monkey kidney cell line), COS (CVI with SV40T antigen), R1610 (Chinese hamster fibroblast) , BALBC/3T3 (mouse fibroblasts), HAK (hamster kidney cell line), SP2/0 (mouse myeloma) and RAJI (human lymphocytes). CHO cells are a particularly preferred expression system. Methods for producing recombinant antibodies in these cells are described in review articles, eg: Makrides, S.C, Components of vectors for gene transfer and expression in mammalian Cells. Protein Expr. Purif. 17 (1999) 183-202.

The host cell line used for antibody expression may also preferably be yeast. Pichia pastoris (Pichia pastoris) and glycoengineered Pichia are more preferred host cells. Yeasts include but are not limited to: Pichia pastoris, Saccharomyces cerevisiae, Saccharomyces cerevisiae, Hans. polymorpha, Kluyveromyces, Kluyveromyces lactis, Candida albicans, Aspergillus nigricans, Aspergillus niger, Aspergillus oryzae , Trichoderma reesei, Chrysosporium lucknowense, Fusarium graminearum. Like mammalian cells, yeast cells have cellular biological mechanisms for correct steric folding, disulfide bond formation, and glycosylation modifications. However, the N-glycosylation mechanisms of yeast and mammalian cells are not identical. While mammalian cells and yeast cells synthesize new peptide chains in the endoplasmic reticulum cavity, the new peptide chains undergo the same N-glycosylation initiation step and modification process. First, the precursor oligosaccharide G1c ₃ Man ₉ GlcNAc ₂ is converted to Connected to the Asn residue in the conserved sequence of the new peptide chain Asn-X-Thr/Ser (X is any amino acid except Pro), and then in the glycoside hydrolase such as glucoside hydrolase I, II and mannoside hydrolase I Under the action of , the sugar chain of the protein is processed to form a Man ₈ GlcNAc ₂ sugar chain structure, and then the protein with this sugar chain is transported to the Golgi apparatus. However, in mammalian cells and in the yeast Golgi, the process of further modification of protein sugar chains is completely different. In the Golgi of mammalian cells, the Man ₈ GlcNAc ₂ sugar chain on the protein first removes three mannoses under the action of mannoside hydrolase I (MnsI) to form the Man ₅ GlcNAc ₂ sugar chain structure; Under the action of glucosamine transferase I (GnTI), one N-acetylglucosamine is added to form GlcNAcMan ₅ GlcNAc ₂ sugar chain structure; then under the action of mannoside hydrolase II (MnsII), two mannoses are removed to form GlcNAcMan ₃ GlcNAc ₂ sugar chain structure; then add another N-acetylglucosamine under the action of N-acetylglucosamine transferase II (GnTII) to form GlcNAc ₂ Man ₃ GlcNAc ₂ sugar chain structure; Under the action of (GalT) and sialyltransferase (ST), complex sugar chain structures of Gal ₂ GlcNAc ₂ Man ₃ GlcNAc ₂ and Sia ₂ Gal ₂ GlcNAc ₂ Man ₃ GlcNAc ₂ are formed. But in the yeast cell Golgi, under the action of α-1,6-mannosyltransferase (Ochlp) encoded by the OCH1 gene, the Man ₈ GlcNAc ₂ sugar chain on the protein first receives an α-1,6-mannose, Form Man ₉ GlcNAc ₂ sugar chain structure, and then continue to add mannose under the action of various other mannose transferases to form a high mannose sugar chain structure (Kornfeld, R. & Kornfeld, S. Assembly of asparagine-linked oligosaccharides . Annu. Rev. Biochem. 54, 631–664, 1985). In order to express antibodies with human N-glycosylated structures in yeast, glycosylation engineering of yeast is required. In addition to deletion of endogenous genes such as OCH1 and BMT, stable expression of many exogenous genes is required, including mannoside hydrolase I (MnsI), N-acetylglucosaminyltransferase (GnTI), mannosidase hydrolase II (MnsII) , N-acetylglucosaminyltransferase (GnTII), galactosyltransferase (GalT), sialyltransferase (ST), and six biosynthetic genes to supply galactose and sialic acid. Glycosylation-engineered Pichia pastoris has been successfully used to generate antibodies (Li H, Sethuraman N, Stadheim TA, Zha D, Prinz B et al. (2006) Optimization of humanized IgGs in glycoengineered Pichia pastoris. Nat Biotechnol 24:210 -215).

Advantages of the present invention:

1. The method of the present invention can significantly improve the correct pairing rate of heavy chain and light chain in bispecific antibody production;

2. The method of the present invention is applicable to various antibody types, including but not limited to IgG, IgA, IgD, IgE and IgM; more preferably IgG1, IgG2, IgG3, IgG4;

3. The method of the present invention can directly combine two antibodies into a bispecific antibody.

4. The purification method of the bispecific antibody of the present invention is similar to that of the monoclonal antibody.

5. The bispecific antibody of the present invention has a complete immunoglobulin structure, good stability and low immunogenicity.

The technical solutions of the present invention are further described below in conjunction with specific implementation cases, but the following examples do not constitute limitations to the present invention, and all the various application methods adopted according to the principles and technical means of the present invention belong to the scope of the present invention. In the following examples, the experimental methods without specific conditions are usually in accordance with conventional conditions, or in accordance with the conditions suggested by the manufacturer. Percentages and parts are by weight unless otherwise indicated.

Material

The chemicals, enzymes, media, and solutions used to create, validate, and apply libraries are commonly used and well known to those skilled in the art of molecular and cellular biology; they are available from many companies, including Thermo Fisher Scientific, Invitrogen, Sigma, New England BioLabs, Takara Biotechnology, Toyobo, TransGen Biotech, Vazyme Biotech and Generay Biotechnology, among others. Many of these are available in kit form. DNA can be chemically synthesized by some of these companies. Antibody sequence data were mainly obtained from the Uniprot database (www.uniprot.org) and the IMGT database (www.imgt.org).

method

Unless otherwise stated, methods used in the present invention, including polymerase chain reaction (PCR), restriction enzyme cloning, DNA purification, bacterial, yeast and eukaryotic cell culture, transformation, transfection and western blotting, are performed in molecular and This is carried out by standard methods well known to those skilled in the art of cell biology. Amino acid mutations (eg, substitutions, deletions, and insertions) described herein can be made using any method known in the art. These methods include, but are not limited to, PCR-extension overlap mutagenesis, site-directed mutagenesis, or cassette mutagenesis (see generally Sambrook J et al. (Molecular Cloning A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y., 2001; Ausubel F M et al. Current Protocols in Molecular Biology, Wiley InterScience, 2010; and Gregg JM (Pichia Protocols, Totowa, New Jersey: Humanna Press, 2010).

DH5α competent cells were used for plasmid construction and amplification according to the manufacturer's (Vazyme Biotech) instructions.

Strains were cultured in Luria-Bertani (LB) medium (10g/L trypsin, 5g/L yeast extract and 5g/L sodium chloride) or LB plates (10g/L trypsin, 5g/L yeast extract) containing appropriate antibiotics and 5g/L sodium chloride, 20g/L agar). Antibiotics were added at the following concentrations: 100 mg/L ampicillin, 50 mg/L kanamycin, 25 mg/L Zeocin, and 100 mg/L blasticidin.

Transformation of the Pichia strain was performed by electroporation with a MicroPulserTM electroporation device following the manufacturer's (BioRad) instructions.

Pichia pastoris strains were cultured in YPD medium (10g/L yeast extract, 20g/L peptone, 20g/L glucose) and YPD plate (10g/L yeast extract, 20g/L peptone, 20g/L glucose, 20g/L agar) for growth and selection. Antibiotics were added at the following concentrations: 250 mg/L sulfate G-418, 100 mg/L Zeocin, and 300 mg/L blasticidin.

Culture and select on YNB medium without amino acids (6.7g/L yeast nitrogen base, 20g/L glucose) and YNB plates without amino acids (6.7g/L yeast nitrogen base, 20g/L glucose, 20g/L agar) Pichia pastoris vegetative strain, supplemented with amino acids as needed.

Pichia strains were grown in BMGY medium (10g/L yeast extract, 20g/L peptone, 100mM potassium phosphate, pH 6.0, 13.4g/L yeast nitrogen base, 0.4mg/L biotin and 10ml/L glycerol) , and then induced antibody expression in BMMY medium (10g/L yeast extract, 20g/L peptone, 100mM potassium phosphate, pH 6.0, 13.4g/L yeast nitrogen base, 0.4mg/L biotin and 10ml/L methanol) .

Example 1. Design of IgG1 (Kappa) CH1 and CL Domain "Cysteine Mutation" Libraries

The CH1 domain sequences of IgG1, IgG2, IgG3 and IgG4 are highly conserved. The light chain is linked to the heavy chain by interchain disulfide bonds in the CH1-CL domain. In the present invention, we wish to mutate the CH1-CL domain of IgG so that its natural interchain disulfide bonds are replaced by engineered interchain disulfide bonds. Using the Fab crystal structure of human IgG1 (kappa) from the Protein Data Bank (PDB code: 1VGE) as a representative structure, cysteine mutations in the CH1 and CL domains that may interact to form interchain disulfide bonds were analyzed and designed site. where the cysteine pairs that form natural interchain disulfide bonds are mutated to valine, and new cysteine pairs are introduced at different positions in the CH1-CL domain to form engineered interchain disulfide bonds, which are represented by Therefore, a library of IgG1(kappa) mutants was designed. Table 6 lists the introduction of new cysteine pairs at different amino acid positions in the CH1-CL domain of IgG1 (Kappa) that may form engineered interchain disulfide bonds.

Table 6. Library of cysteine mutations in the CH1-CL domain of IgG1 (kappa). Regions where cysteine mutations cluster together are called a group. The CH1 domain and the CL domain may form an interchain disulfide bond. Each group is listed separately.

Example 2. Construction and expression of IgG1 (Kappa) "cysteine mutation" library

Trastuzumab (ERBB2Ab) heavy chain (HoleRF-His) containing T366S/L368A/Y407V "hole" mutations, H435R, Y436F (RF) mutations [EU numbering] and a C-terminal 6xHis tag for use as IgG1 (Kappa) Representative of the heavy chain (SEQ ID NO: 1). Trastuzumab heavy chain (HoleRF-His) codon-optimized DNA was synthesized and used as template for PCR amplification (SEQ ID NO: 2).

PCR 1, TraHF (SEQ ID NO: 3, this primer has Xho I restriction enzyme site) and TraH R (SEQ ID NO: 4, this primer has Not I restriction enzyme site) primer pair for trastuzumab PCR amplification of anti-heavy chain (HoleRF-His) using synthetic DNA as template. The PCR product was digested with Xho I and Not I and inserted into the same digestion site in pPIC9 (Invitrogen) to construct an expression vector for the trastuzumab heavy chain (HoleRF-His), which was named pPIC9-TraH ( HoleRF-His).

Wild-type trastuzumab kappa light chain was used as a representative of IgGl (kappa) light chain (SEQ ID NO: 5). Trastuzumab kappa light chain codon-optimized DNA was synthesized and used as template for PCR amplification (SEQ ID NO:6).

PCR 2, TraκF (SEQ ID NO:7, this primer has Xho I restriction enzyme site) and TraκR (SEQ ID NO:8, this primer has Not I restriction enzyme site) primer pair for Trastuzumab kappa PCR amplification of the light chain using synthetic trastuzumab Kappa light chain as template. The PCR product was digested with Xho I and Not I, and inserted into the same digestion site of pPICZαA (Invitrogen) to construct an expression vector for trastuzumab kappa light chain, named pPICZα-Traκ.

ARG2-ADH1TT fragment (SEQ ID NO:9) of Pichia ARG2 3' and 5' homologous sequences (with Afe I restriction enzyme site and ADH1 terminator sequence) was synthesized and used as PCR amplification template.

PCR 3, ARG2F (SEQ ID NO: 10, the primer has a BamHI restriction enzyme site) and ADH1TT R (SEQ ID NO: 11, the primer has a BamHI restriction enzyme site) primer pair for PCR amplification, synthesis The ARG2-ADH1TT fragment was used as template. The PCR product was digested with BamHI restriction enzyme and inserted into the same site in pPICZα-Traκ to obtain the pPICZα-Traκ-ARG2 expression vector for integration at the Pichia ARG2 locus.

PCR 4, PICZ F (SEQ ID NO: 12) and PICZ R (SEQ ID NO: 13) primer pairs were used for PCR amplification of Zeocin-free pPICZα linear fragments, pPICZα vector was used as template.

PCR 5, Kan F (SEQ ID NO: 14) and Kan R (SEQ ID NO: 15) primer pairs were used to PCR amplify the kanamycin fragment and the pPIC9K vector (Invitrogen) was used as template.

The kanamycin fragment was inserted into the pPICZα linear fragment using the ClonExpress II one-step cloning kit (Vazyme) to generate the pEG vector, in which kanamycin was used instead of Zeocin in the pPICZα vector.

PCR 6, EGF (SEQ ID NO: 16) and EGR (SEQ ID NO: 17) primer pairs were used for PCR amplification of pEG linear fragments, pEG vector was used as template.

The Clazakizumab (IL6Ab) heavy chain (Knob) containing the T366W "knob" mutation and the C220V mutation [EU numbering] was used as a representative of the IgG1 (Kappa) heavy chain (SEQ ID NO: 18). Codon-optimized DNA of clazakizumab heavy chain (Knob) was synthesized and used as template for PCR amplification (SEQ ID NO: 19).

PCR 7, heavy chain N-terminal primer ClaH-Nt (SEQ ID NO: 20) and C-terminal primer ClaH-Ct (SEQ ID NO: 21) respectively and the corresponding reverse primer (R) in Table 7, and the corresponding forward To primers (F), different primer pairs were composed for PCR amplification of N-terminal and C-terminal heavy chains, using synthetic clazakizumab heavy chain (Knob) as template.

PCR 8. Using the ClaH-Nt and ClaH-Ct primer pairs, the N-terminal and C-terminal heavy chain PCR products were ligated by overlap extension PCR. In this way, clazakizumab heavy chain mutants were generated, containing the wild-type amino acid mutation to cysteine, the T366W "knob" mutation and the C220V mutation.

Table 7. PCR primers for mutation of wild-type amino acid to cysteine in the heavy chain of Clazakizumab [EU numbering]

mutation	Reverse primer (R)	Forward primer (F)
ClaH-F126C	ClaH-F126CR (SEQ ID NO:22)	ClaH-F126CF (SEQ ID NO:23)
ClaH-L128C	ClaH-F126CR (SEQ ID NO:22)	ClaH-L128CF (SEQ ID NO:24)
ClaH-A129C	ClaH-F126CR (SEQ ID NO:22)	ClaH-A129CF (SEQ ID NO:25)
ClaH-P130C	ClaH-F126CR (SEQ ID NO:22)	ClaH-P130CF (SEQ ID NO: 26)
ClaH-S131C	ClaH-F126CR (SEQ ID NO:22)	ClaH-S131CF (SEQ ID NO:27)
ClaH-S132C	ClaH-F126CR (SEQ ID NO:22)	ClaH-S132CF (SEQ ID NO:28)
ClaH-K133C	ClaH-F126CR (SEQ ID NO:22)	ClaH-K133CF (SEQ ID NO:29)
ClaH-S134C	ClaH-F126CR (SEQ ID NO:22)	ClaH-S134CF (SEQ ID NO:30)
ClaH-T135C	ClaH-F126CR (SEQ ID NO:22)	ClaH-T135CF (SEQ ID NO:31)
ClaH-G166C	ClaH-G166CR (SEQ ID NO:32)	ClaH-G166CF (SEQ ID NO:33)
ClaH-H168C	ClaH-G166CR (SEQ ID NO:32)	ClaH-H168CF (SEQ ID NO:34)
ClaH-F170C	ClaH-G166CR (SEQ ID NO:32)	ClaH-F170CF (SEQ ID NO:35)
ClaH-P171C	ClaH-G166CR (SEQ ID NO:32)	ClaH-P171CF (SEQ ID NO:36)
ClaH-V173C	ClaH-G166CR (SEQ ID NO:32)	ClaH-V173CF (SEQ ID NO:37)
ClaH-Q175C	ClaH-G166CR (SEQ ID NO:32)	ClaH-Q175CF (SEQ ID NO:38)
ClaH-S176C	ClaH-S176CR (SEQ ID NO:39)	ClaH-S176CF (SEQ ID NO:40)
ClaH-S177C	ClaH-S176CR (SEQ ID NO:39)	ClaH-S177CF (SEQ ID NO:41)
ClaH-G178C	ClaH-S176CR (SEQ ID NO:39)	ClaH-G178CF (SEQ ID NO:42)
ClaH-L179C	ClaH-S176CR (SEQ ID NO:39)	ClaH-L179CF (SEQ ID NO:43)
ClaH-S181C	ClaH-S181CR (SEQ ID NO:44)	ClaH-S181CF (SEQ ID NO:45)
ClaH-S183C	ClaH-S181CR (SEQ ID NO:44)	ClaH-S183CF (SEQ ID NO:46)
ClaH-V185C	ClaH-S181CR (SEQ ID NO:44)	ClaH-V185CF (SEQ ID NO:47)
ClaH-T187C	ClaH-S181CR (SEQ ID NO:44)	ClaH-T187CF (SEQ ID NO:48)
ClaH-K218C	ClaH-K218CR (SEQ ID NO:49)	ClaH-K218CF (SEQ ID NO:50)
ClaH-S219C	ClaH-K218CR (SEQ ID NO:49)	ClaH-S219CF (SEQ ID NO:51)
ClaH-V220C	ClaH-K218CR (SEQ ID NO:49)	ClaH-V220CF (SEQ ID NO:52)

The clazakizumab heavy chain mutant was inserted into the pEG linear fragment using the ClonExpress II one-step cloning kit (Vazyme) to generate an expression vector library named pEG-ClaH(X#C) (ClaH: clazakizumab heavy chain, (X#C): The wild-type amino acid number # is mutated to cysteine). For example, pEG-ClaH(L128C) represents a mutation of leucine at position 128 of the heavy chain to cysteine. Each vector expresses the clazakizumab heavy chain, which contains the T366W "knob" mutation, the C220V mutation and the wild-type amino acid mutation listed in Table 6 to cysteine. A vector, pEG-ClaH, expresses the clazakizumab heavy chain, which contains the T366W "knob" mutation and the wild-type cysteine at position 220, which forms native interchain disulfide bonds.

An ARG4-ADH1TT fragment (SEQ ID NO: 53) with Pichia ARG4 3' and 5' homologous sequences and Sma I restriction enzyme site and ADH1 terminator sequence was synthesized and used as a template for PCR amplification.

PCR 9, ARG4F (SEQ ID NO:54, the primer has a BamHI restriction site) and ADH1TT R (SEQ ID NO:55, the primer has a BamHI restriction site) primer pair was used for PCR The ARG4-ADH1TT fragment was amplified, using the synthetic ARG4-ADH1TT fragment as a template. The PCR product was digested with BamHI restriction enzyme and inserted into the same site of pPIC6α (Invitrogen) to generate the pPIC6α-ARG4-ADH1TT expression vector, which can integrate at the Pichia ARG4 locus.

The Clazakizumab kappa light chain containing the C214V mutation was used as a representative of the IgG1 kappa (κ) light chain (SEQ ID NO: 56). Codon-optimized DNA of the light chain of clazakizumab kappa with the C214V mutation was synthesized and used as a template for PCR amplification (SEQ ID NO: 57).

PCR 10, light chain N-terminal primer Claκ-Nt (SEQ ID NO:58, this primer has Xho I restriction site) and light chain C-terminal primer Claκ-Ct (SEQ ID NO:59, this primer has Not I restriction enzyme site) and the corresponding reverse primer (R) and forward primer (F) in Table 8 to form different primer pairs for PCR amplification of N-terminal and C-terminal kappa light chain, synthetic clazakizumab kappa chain as a template.

PCR 11, PCR products of N-terminal and C-terminal kappa light chains were ligated by overlap extension PCR using the Claκ-Nt and Claκ-Ct primer pairs. It generated a clazakizumab kappa light chain mutant containing a mutation of the wild-type amino acid to cysteine and a C214V mutation. The C-terminal wild amino acid can be mutated to cysteine by direct PCR amplification of the clazakizumab kappa light chain using Claκ-Nt and the corresponding reverse primer (R) in Table 8. The clazakizumab kappa chain with wild cysteine at position 214 can be directly PCR amplified using the Claκ-Nt and Claκ-V214C reverse primers (R).

Table 8. PCR primers for mutation of wild amino acid to cysteine in the light chain of Clazakizumab kappa (κ) [EU numbering]

mutation	Reverse primer (R)	Forward primer (F)
Claκ-S114C	Claκ-S114CR (SEQ ID NO:60)	Claκ-S114CF (SEQ ID NO:61)
Claκ-F116C	Claκ-S114CR (SEQ ID NO:60)	Claκ-F116CF (SEQ ID NO:62)
Claκ-F118C	Claκ-S114CR (SEQ ID NO:60)	Claκ-F118CF (SEQ ID NO:63)
Claκ-P119C	Claκ-S114CR (SEQ ID NO:60)	Claκ-P119CF (SEQ ID NO:64)
Claκ-P120C	Claκ-S114CR (SEQ ID NO:60)	Claκ-P120CF (SEQ ID NO:65)
Claκ-S121C	Claκ-S114CR (SEQ ID NO:60)	Claκ-S121CF (SEQ ID NO:66)
Claκ-N158C	Claκ-N158CR (SEQ ID NO:67)	Claκ-N158CF (SEQ ID NO:68)
Claκ-Q160C	Claκ-N158CR (SEQ ID NO:67)	Claκ-Q160CF (SEQ ID NO:69)
Claκ-S162C	Claκ-N158CR (SEQ ID NO:67)	Claκ-S162CF (SEQ ID NO:70)
Claκ-T164C	Claκ-N158CR (SEQ ID NO:67)	Claκ-T164CF (SEQ ID NO:71)
Claκ-T172C	Claκ-T172CR (SEQ ID NO:72)	Claκ-T172CF (SEQ ID NO:73)
Claκ-S174C	Claκ-T172CR (SEQ ID NO:72)	Claκ-S174CF (SEQ ID NO:74)
Claκ-S176C	Claκ-T172CR (SEQ ID NO:72)	Claκ-S176CF (SEQ ID NO:75)
Claκ-T178C	Claκ-T172CR (SEQ ID NO:72)	Claκ-T178CF (SEQ ID NO:76)
Claκ-T180C	Claκ-T172CR (SEQ ID NO:72)	Claκ-T180CF (SEQ ID NO:77)
Claκ-F209C	Claκ-F209CR (SEQ ID NO:78)
Claκ-N210C	Claκ-N210CR (SEQ ID NO:79)
Claκ-R211C	Claκ-R211CR (SEQ ID NO:80)
Claκ-G212C	Claκ-G212CR (SEQ ID NO:81)
Claκ-E213C	Claκ-E213CR (SEQ ID NO:82)
Claκ-V214C	Claκ-V214CR (SEQ ID NO:83)

Clazakizumab kappa light chain mutants were digested with Xho I and Not I, respectively, and inserted into the same digestion site of pPIC6α-ARG4-ADH1TT to generate a kappa light chain mutant expression vector library named pPIC6-ARG4-Claκ(X#C ) (Claκ: clazakizumab kappa light chain, (X#C): wild amino acid number # mutated to cysteine). For example, pPIC6-ARG4-Claκ(S114C) represents a mutation of the wild-type serine at position 114 of the kappa light chain to a cysteine. As shown in Table 8, each vector can express the clazakizumab kappa light chain containing the C214V mutation and the wild amino acid mutation to cysteine, and one vector can express the clazakizumab kappa chain with wild cysteine at position 214.

The glycoengineered Pichia strain GS2-1 (his4, PpBMT2-SfMNS1::och1, ScMNN10-AtMNS1::pno1) (PpBMT2-SfMNS1 is Fusion protein, including Pichia pastoris BMT2 N-terminal 1-74 amino acids and Spodoptera frugiperda MnsI catalytic domain 164-670 amino acids; ScMNN10-AtMNS1 is a fusion protein, including Saccharomyces cerevisiae MNN10 N-terminal 1-116 amino acids and Arabidopsis thaliana MnsI catalytic domain domain 78-560 amino acids). The structure of the protein N-glycan expressed with the GS2-1 strain mainly contained Man5GlcNAc2.

The expression vector of pPICZα-Traκ-ARG2 was linearized with the restriction enzyme Afe I within the ARG2 3' and 5' homologous sequences and electroporated into GS2-1. The linear expression vector was integrated into the ARG2 locus by recombination of the ARG2 5' and 3' homologous sequences. Transformed cells were grown on YPD plates supplemented with 100 mg/L Zeocin. This resulted in a new expression strain GS2-Traκ, which can express the trastuzumab wild-type kappa light chain.

The pPIC9-TraH (HoleRF-His) expression vector was linearized with restriction enzyme Sal I, electroporated into the GS2-Traκ strain, and integrated into the his4 locus. Transformed cells were selected on YNB plates. This resulted in a new expression strain GS2-TraHκ to express the trastuzumab wild-type light chain and the Fc-mutated heavy chain containing the T366S/L368A/Y407V "hole" mutations, H435R, Y436F(RF) mutations and C 6xHis tag at the end.

Each expression vector of the clazakizumab heavy chain mutant library pEG-ClaH(X#C) was linearized with restriction enzyme Pme I, electroporated into the GS2-TraHκ strain, and integrated at the AOX1 locus. Transformed cells were selected on YPD plates supplemented with 250 mg/L G-418 sulfate. This generated a library of expression strains GS2-TraHκ-ClaH (X#C). These strains can express trastuzumab with wild-type light chain and Fc mutant heavy chain, as well as clazakizumab heavy chain with cysteine mutation in CH1 and "knob" mutation in Fc.

Each expression vector of the clazakizumab kappa light chain mutant library pPIC6-ARG4-Claκ(X#C) was linearized in the ARG4 3' and 5' homologous sequences with restriction enzyme Sma I and then electroporated into the expression strain library GS2- TraHκ-ClaH (X#C). The linear expression vector was integrated into the ARG4 locus by recombination of the ARG4 3' and 5' homologous sequences. Transformed cells were grown on YPD plates supplemented with 300 mg/L blasticidin. This created the expression strain library GS2-TraHκ-ClaH(X#C)κ(X#C). These strains can express asymmetric bispecific antibodies. In asymmetric bispecifics, half of the bispecific are clazakizumab heavy and light chains with engineered interchain disulfide bonds in the CH1 and CL domains of the Fab arms and a "knob" in the Fc mutation. The other half are trastuzumab heavy and light chains with native interchain disulfide bonds in the CH1 and CL domains of the Fab arm, a "hole" and RF mutation in the Fc, and C-terminus of the heavy chain with 6xHis tags,

The expression strain of GS2-TraHκ-ClaH(X#C)κ(X#C) was cultured in BMGY medium for 72 hours at 24°C and 240rpm with shaking. Cells were then pelleted by centrifugation at 3000g for 5 minutes, resuspended in BMMY medium, cultured with shaking at 24°C and 240rpm, and methanol (1% concentration) was added to the medium every day for continuous expression induction for 72 hours. Subsequently, the medium supernatant was harvested by centrifugation (3000 g, 10 min), and the supernatant was frozen at -20°C until used in the next step.

Example 3. Screening to identify IgG1 (Kappa) "cysteine mutation" library

The concentration of bispecific antibodies in the culture supernatant was measured by ELISA. Briefly, 100 μL/well of 5 μg/mL AGL protein (Leeanntech), 50 mM sodium carbonate buffer (pH 9.6) was added to 96-well plates (Maxisorp Nunc-Immuno, Thermo Scientific) and coated overnight at 4°C. After washing the plates 3 times with PBS-T (PBS with 0.05% Tween-20), the plates were blocked with 2% nonfat dry milk in PBS-T for 1 hour at 37°C. After washing the plate 3 times with PBS-T, 100 μL/well of human IgG standard (initial concentration 2.5ug/ml) and the culture supernatant at 1:2 serial dilution in PBS-T were added at 37°C Incubate for 1 hour. Plates were washed 3 times with PBS-T and 100 μL/well of AGL-HRP (Leeanntech) diluted 1:5000 in PBS-T and 0.5% nonfat dry milk was added. Plates were incubated at 37°C for 1 hour, washed 3 times with PBS-T, and 100 μL/well of TMB (Leeanntech) was added. An additional 100 μL/well of 2M H ₂ SO ₄ was added to stop the colorimetric reaction, and the absorbance (A450 nm) of each well was measured at 450 nm.

Bispecific antibodies were purified by protein A affinity chromatography using MabSelect SuRe resin (GE Healthcare) according to the manufacturer's instructions. Since the RF mutation in the trastuzumab heavy chain abolishes its binding to protein A, byproducts such as trastuzumab pore-pore heavy chain homodimers and half-antibodies (one pore heavy and light chain) can be found in Protein A is easily removed from bispecific antibodies in purification. Briefly, the collected supernatant was mixed with MabSelect SuRe resin and shaken for 1 hour at room temperature, then MabSelect SuRe resin was washed with 25 mM sodium phosphate buffer pH 7.0, 1 M sodium chloride. Bispecific antibodies were eluted from MabSelect SuRe resin with 50 mM sodium citrate (pH 3.0) and neutralized to pH 6.5 with 1 M disodium hydrogen phosphate (pH 8.9).

In the present invention, we used an ELISA method to compare whether different interchain disulfide bonds in bispecific antibodies can promote the correct pairing of heavy and light chains. Bispecific antibodies with correctly paired heavy and light chains can bind to the antigen and produce a signal in the ELISA, but bispecific antibodies with mismatched heavy and light chains cannot bind to the antigen and produce no signal in the ELISA. At the same concentration of the bispecific antibody, it binds to the antigen and produces a high signal in the ELISA, indicating that the correct pairing rate of heavy and light chains is high, and vice versa, the correct pairing rate of heavy and light chains is low.

The bispecific antibody purified from protein A was diluted 1:2 with PBS and divided into two parts. One part was used to measure the concentration of the antibody by ELISA, and the other part was used to measure the binding of the antibody to the antigen by ELISA. The concentration of purified antibody was determined by ELISA. Briefly, 100 μL/well of 5 μg/mL AGL protein, 50 mM sodium carbonate buffer (pH 9.6) was added to the plate and coated overnight at 4°C. After washing the plates 3 times with PBS-T, the plates were blocked with 2% nonfat dry milk in PBS-T for 1 hour at 37°C. After washing the plate 3 times with PBS-T, 100 μL/well of purified bispecific antibody diluted in PBS was added and incubated at 37°C for 1 hour. The plate was washed 3 times with PBS-T, and 100 μL/well of AGL-HRP diluted 1:5000 in PBS-T was added. Plates were incubated at 37°C for 1 hour, washed 3 times with PBS-T, and 100 μL/well of TMB was added. An additional 100 μL/well of 2M H ₂ SO ₄ was added to stop the colorimetric reaction, and the absorbance (A450 nm) of each well was measured at 450 nm. The concentration of the bispecific antibody can be expressed by the absorbance value measured by the corresponding ELISA reaction.

ELISA measures the binding of antibody to antigen. Briefly, 100 μL/well of 1 μg/mL human HER2/ErbB2 protein (His tag) (Sino Biological), 50 mM sodium carbonate buffer (pH 9.6) was added to the plate and coated overnight at 4°C. After washing the plate 3 times with PBS-T, the plate was blocked with 2% nonfat milk in PBS-T for 1 hour at 37°C. After washing the plate 3 times with PBS-T, 100 μL/well of purified bispecific antibody diluted in PBS was added and incubated at 37°C for 1 hour. The plate was washed 3 times with PBS-T, and 100 μL/well of AGL-HRP diluted 1:5000 in PBS-T was added. Plates were incubated at 37°C for 1 hour, washed 3 times with PBS-T, and 100 μL/well of TMB was added. An additional 100 μL/well of 2M H ₂ SO ₄ was added to stop the colorimetric reaction, and the absorbance (A450 nm) of each well was measured at 450 nm. The binding amount of the bispecific antibody to the antigen can be expressed by the absorbance value (A450nm) measured by the corresponding ELISA reaction.

In an asymmetric bispecific antibody, half of the bispecific antibody has trastuzumab heavy and light chains, its Fab arms CH1 and CL domains contain native interchain disulfide bonds, and its heavy chain Fc contains natural interchain disulfide bonds The "hole" and RF mutations, which contain a 6xHis tag at the C-terminus of its heavy chain, while the other half contains clazakizumab heavy and light chains, contain engineered interchain disulfide bonds in the Fab arm CH1 and CL domains, and its heavy chain Fc contains a "knob" mutation. According to reports, when phenylalanine 126 in clazakizumab heavy chain CH1 is mutated to cysteine (F126C) and serine 121 in light chain CL is mutated to cysteine (S121C), an engineered interchain can be formed. Disulfide bonds, this asymmetric bispecific antibody can have 98% correct pairing of heavy and light chains. The asymmetric bispecific antibody was designated herein as TraHκ-ClaH(F126C)κ(S121C), abbreviated as F126C/S121C, and used as a positive control. However, when both Fab arms have native interchain disulfide bonds, the resulting bispecific antibody has only 25% of the heavy and light chains correctly paired. This bispecific antibody was named TraHκ-ClaHκ, abbreviated as WT, and was used as a negative control (Yariv Mazor, Vaheh Oganesyan, Chunning Yang, Anna Hansen, Jihong Wang, Hongji Liu, Kris Sachsenmeier, Marcia Carlson, Dhanesh V Gadre, Martin Jack Borrok, Xiang-Qing Yu, William Dall'Acqua, Herren Wu, and Partha Sarathi Chowdhury, mAbs 7, 377--389; 2015).

Bispecific antibodies with correctly paired heavy and light chains can bind to the antigen and produce a signal in the ELISA, but bispecific antibodies with mismatched heavy and light chains cannot bind to the antigen and produce no signal in the ELISA. The ELISA absorbance value (A450nm) on the horizontal axis represents the concentration of the bispecific antibody, and the ELISA absorbance value (A450nm) on the vertical axis represents the binding amount of the bispecific antibody to the antigen. The data were processed by Excel to obtain a scatter plot. Within the effective ELISA color value range, the higher the position of the curve, that is, the higher the color value of the ELISA reaction between the bispecific antibody and the same amount of antigen under the same concentration conditions, the higher the value of the heavy chain and light chain of the bispecific antibody antibody. The higher the pairing rate, the lower the mismatching rate. The lower the curve position, the lower the rate of correct pairing of the heavy and light chains of the bispecific antibody, and the higher the rate of mismatches. As shown in Figure 7, under the same antibody concentration conditions (ELISA absorbance value on the horizontal axis), the positive control F126C/S121C of the reported bispecific antibody has 98% correct pairing of heavy and light chains, thus showing a high Absorbance values for the antibody/antigen binding ELISA (vertical axis), but the negative control WT for the bispecific antibody is expected to have only 25% of the heavy and light chains correctly paired, thus showing very low absorbance values for the antibody/antigen binding ELISA. Under the same antibody concentration conditions (ELISA absorbance value on the horizontal axis), many of the bispecific antibodies we constructed had the same or higher antibody/antigen binding absorbance value as the positive control F126C/S121C. This indicates that our bispecific antibodies have 98% or more correct pairing of heavy and light chains. Therefore, in the production of IgG1 kappa bispecific antibodies, by replacing the native interchain disulfide bonds with different cysteine pairs, an engineered disulfide bond is formed in the CH1-CL domain of one Fab arm, which can Achieve proper pairing of heavy and light chains, including:

F126C/F118C、L128C/P120C、A129C/P120C、P130C/F118C、S132C/F118C、K133C/S121C、S134C/F118C、G166C/T172C、G166C/S174C、G166C/T180C、H168C/T180C、F170C/Q160C、F170C/ T164C、F170C/T172C、F170C/S174C、F170C/S176C、F170C/T180C、P171C/T172C、P171C/S174C、P171C/T180C、V173C/S174C、Q175C/S174C、Q175C/S176C、Q175C/T180C、S176C/S162C、 S181C/T172C、S181C/S176C、S181C/T180C、S183C/S114C、S183C/F118C、S183C/Q160C、S183C/S174C、S183C/T178C、S183C/T180C、V185C/T172C、V185C/S174C、V185C/T178C、V185C/ T180C, T187C/S114C, T187C/T172C, T187C/S174C, T187C/T178C, K218C/F118C, S219C/F116C, S219C/F118C, S219C/P120C,

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine / kappa light chain wild amino acid [EU numbering] to cysteine)

Example 4. Construction and expression of IgG1 (Kappa) "cysteine and charge mutation" library

In the IgG1 (Kappa) CH1 and CL domains, a pair of charged amino acids (K213 in the heavy chain and E123 in the kappa light chain, EU numbering) are involved in charge-charge interactions. This pair of amino acids is conserved among IgG1, IgG2, IgG3 and IgG4 and serves the same purpose. At neutral pH (pH=7.0), aspartic acid (D) and glutamic acid (E) are negatively charged, and lysine (K), arginine (R) and histidine (H) are positively charged Electricity. Attractive interactions can occur between amino acids with opposite charges, while repulsion occurs between amino acids with similar charges. As shown in Table 9, mutating this pair of amino acids in the CH1 and CL domains of the antibody alpha heavy and light chains to reverse their charge polarity (eg K213D and E123K) may result in mismatched heavy and light chains Unfavorable mutual repulsion occurs between the chains (for example, the negative charge of the K213D mutation of the antibody A heavy chain and the negative charge of the antibody B light chain E123 repel each other; the positive charge of the antibody A light chain E123K mutation and the positive charge of the antibody B heavy chain K213 charges repel each other). Introducing mutated interchain disulfide bonds or/and charge reversal in the CH and CL domains facilitates the correct pairing of heavy and light chains in bispecific antibodies.

Table 9. Mutations with charge polarity reversal in heavy and light chains.

PCR 1, EGF and EGR primer pairs were used to PCR amplify pEG linear fragments (as described in Example 2) using the pEG vector as a template.

PCR 2, using ClaH-Nt and ClaH-K213DR primer pair (SEQ ID NO:84), ClaH-K213DF (SEQ ID NO:85) and ClaH-Ct primers for N-terminal and C-terminal heavy chain PCR amplification, respectively In addition, each expression vector of the clazakizumab mutant library pEG-ClaH(X#C) was used as a template. The N-terminal and C-terminal heavy chains were ligated by overlap extension PCR using the ClaH-Nt and ClaH-Ct primer pairs. This resulted in a mutated clazakizumab heavy chain containing a T366W "knob" mutation, a C220V mutation, a cysteine mutation and a K213D charge mutation.

PCR 3, using ClaH-Nt and ClaH-K213DR primer pair, ClaH-K213EF (SEQ ID NO: 86) and ClaH-Ct primer pair for N-terminal and C-terminal heavy chain PCR amplification, respectively, using clazakizumab heavy chain Each expression vector of the mutant library pEG-ClaH (X#C) was used as a template. The N-terminal and C-terminal heavy chains were ligated by overlap extension PCR using the ClaH-Nt and ClaH-Ct primer pairs. This resulted in a mutated clazakizumab heavy chain containing a T366W "knob" mutation, a C220V mutation, a cysteine mutation and a K213E charge mutation.

Mutated clazakizumab heavy chains were inserted into pEG linear fragments using the ClonExpress II one-step cloning kit (Vazyme) to construct clazakizumab mutant expression vector libraries, designated pEG-ClaH (X#CD) and pEG-ClaH (X#CE). Each vector can express a mutated clazakizumab heavy chain, including the T366W "knob" mutation, the C220V mutation, the wild-type amino acid mutation to cysteine, and the K213D or K213E mutation.

PCR 4, using the corresponding primer pairs in Table 10-1 to carry out PCR amplification of the N-terminal and C-terminal Kappa light chain respectively, using each expression vector pPIC6α-Claκ (X#C) of the Kappa light chain mutant library as a template. The N-terminal and C-terminal kappa chains were ligated by overlap extension PCR using the Claκ-Nt and Claκ-Ct primer pairs. This created a mutated clazakizumab kappa chain containing the C214V mutation, the cysteine mutation and the E123K mutation.

Table 10-1 Clazakizumab light chain N-terminal and C-terminal PCR amplification primer pairs

PCR 5, using the corresponding primer pairs in Table 10-2 to carry out PCR amplification of the N-terminal and C-terminal Kappa light chain respectively, using each expression vector pPIC6α-Claκ (X#C) of the Kappa light chain mutant library as a template . The N-terminal and C-terminal kappa chains were ligated by overlap extension PCR using the Claκ-Nt and Claκ-Ct primer pairs. This created a mutated clazakizumab kappa chain containing the C214V mutation, the cysteine mutation and the E123R mutation.

Table 10-2 Clazakizumab light chain N-terminal and C-terminal PCR amplification primer pairs

The mutated clazakizumab kappa chain was digested with Xho I and Not I and inserted into the same digestion site of pPIC6α-ARG4 to generate a library of kappa light chain mutant expression vectors designated pPIC6α-ARG4-Claκ(X#CK) and pPIC6α- ARG4-Claκ (X#CR). Each vector can express a mutated clazakizumab kappa light chain that contains a C214V mutation, a cysteine mutation, and an E123K or E123R mutation.

The expression vectors pEG-ClaH(X#CD) and pEG-ClaH(X#CE) of the clazakizumab heavy chain mutant library were linearized with restriction endonuclease Pme I, electroporated into the GS2-TraHκ strain, and integrated in AOX1 on the locus. Transformed cells were selected on YPD plates supplemented with 250 mg/L G-418 sulfate. This resulted in the expression strain libraries GS2-TraHκ-ClaH (X#CD), GS2-TraHκ-ClaH (X#CE).

Each expression vector of the clazakizumab kappa light chain mutation library pPIC6α-ARG4-Claκ (X#CK) and pPIC6α-ARG4-Claκ (X#CR) was linearized with Sma I and electroporated into the expression strain library GS2-TraHκ-ClaH ( X#CD) and GS2-TraHκ-ClaH (X#CE). The linear expression vector was integrated at the ARG4 locus. Transformed cells were grown on YPD plates supplemented with 300 mg/L blasticidin. This generated the expression strain libraries GS2-TraHκ-ClaH(X#CD)κ(X#CK), GS2-TraHκ-ClaH(X#CD)κ(X#CR), GS2-TraHκ-ClaH(X#CE) κ(X#CK), GS2-TraHκ-ClaH(X#CE)κ(X#CR). These strains can express asymmetric bispecific antibodies. In an asymmetric bispecific antibody, half of the bispecific antibody is clazakizumab heavy and light chains with engineered interchain disulfide bonds and/or a pair of charge mutations in the CH1 and CL domains of the Fab arms, in Has a "knob" mutation in the Fc; the other half are Trastuzumab heavy and light chains with native interchain disulfide bonds in the CH1 and CL domains of the Fab arm, and a "hole" and RF in the Fc Mutated, with a 6xHis tag at the C-terminus of the heavy chain.

As described in Example 2, the expression strains were grown in BMGY medium and the antibodies were induced in BMMY medium. The supernatant harvested by centrifugation was frozen at -20°C until used in the next step.

Example 5. Screening to identify IgG1 (Kappa) "cysteine and charge mutation" library

Bispecific antibodies were purified by protein A affinity chromatography as described in Example 3, and the concentration of purified antibody was determined by ELISA.

As described in Example 3, ELISA was used to compare whether different interchain disulfide bonds and charge mutations promote correct pairing of heavy and light chains in bispecific antibodies comprising in the CH1 and CL domains A pair of novel cysteines and charge-reversal mutations (eg K213D and E123K). As shown in Figure 8A, under the same antibody concentration conditions (the horizontal axis ELISA absorbance value), the bispecific antibody positive control F126C/S121C had 98% of the heavy and light chains correctly paired, and had a high antibody/antigen binding ELISA Absorbance values (vertical axis), but the bispecific antibody negative control WT had only 25% of the heavy and light chains correctly paired and had very low absorbance values for antibody/antigen binding (vertical axis). Under the same antibody concentration conditions (horizontal axis ELISA absorbance value), the two bispecific antibodies containing charge inversion (K213D/E123K, K213E/E123K) in the CH1-CL domain of the Fab arm had higher values than the positive control ( Antibody/antigen binding absorbance values of F126C/S121C). However, the antibody/antigen binding absorbance values of the other two bispecifics containing charge inversions (K213D/E123R, K213E/E123R) in the CH1-CL domain of the Fab arm were significantly lower than the positive control F126C/S121C. Thus, in IgG bispecific antibody production, correct pairing of heavy and light chains can be facilitated by using charge-reversal mutations (e.g. K213D/E123K, K213E/E123K) (each pair of charge-reversal mutations is listed as follows: Lysine K at position 213 of the heavy chain is mutated to a negatively charged amino acid D or E / glutamic acid E at position 123 of the light chain is mutated to a positively charged amino acid K or R).

To assess the effect of a new pair of cysteines in the CH1 and CL domains of one Fab arm in combination with charge reversal on the correct pairing of heavy and light chains, a bispecific antibody, TraHκ-ClaH(S132C)κ, was used. (F116C), abbreviated as S132/F116C and the following four bispecific antibodies with electrical mutations as examples:

TraHκ-ClaH(S132C, K213D)κ(F116C, E123K), abbreviated as S132C, K213D/F116C, E123K;

TraHκ-ClaH(S132C, K213E)κ(F116C, E123K), abbreviated as S132C, K213E/F116C, E123K;

TraHκ-ClaH(S132C, K213D)κ(F116C, E123R), abbreviated as S132C, K213D/F116C, E123R,

TraHκ-ClaH(S132C, K213E)κ(F116C, E123R), abbreviated as S132C, K213E/F116C, E123R.

As shown in Figure 8B, under the same antibody concentration conditions (the horizontal axis ELISA absorbance value), the bispecific antibody S132C/F116C containing a pair of new cysteines in the CH1-CL domain of one Fab arm had the same positive Similar antibody/antigen binding absorbance values for control F126C/S121C. 4 bispecific antibodies with novel cysteine pairing and charge reversal Can further increase the absorbance value of antibody/antigen binding.

To assess the effect of a new pair of cysteines in the CH1 and CL domains of one Fab arm in combination with charge reversal on the correct pairing of heavy and light chains, a bispecific antibody (TraHκ-ClaH(V173C) Kappa (N158C), abbreviated as V173C/N158C and the following four bispecific antibodies are another example:

TraHκ-ClaH(V173C, K213D)κ(N158C, E123K), abbreviated as V173C, K213D/N158C, E123K;

TraHκ-ClaH(V173C, K213E)κ(N158C, E123K), abbreviated as V173C, K213E/N158C, E123K;

TraHκ-ClaH(V173C, K213D)κ(N158C, E123R), abbreviated as V173C, K213D/N158C, E123R;

TraHκ-ClaH(V173C, K213E)κ(N158C, E123R), abbreviated as V173C, K213E/N158C, E123R.

As shown in Figure 8C, the bispecific antibody V173C/N158C containing a new pair of cysteines in the CH1-CL domain of one Fab arm had a ratio positive under the same antibody concentration conditions (the horizontal axis ELISA absorbance value). Control F126C/S121C low antibody/antigen binding absorbance values. Two bispecific antibodies containing a new cysteine pair and charge reversal, V173C, K213D/N158C, E123K; V173C, K213E/N158C, E123K had higher antibody/antigen binding absorbance values than the positive controls F126C/S121C. Two other bispecifics containing novel cysteine pairing and charge reversal (V173C, K213D/N158C, E123R; V173C, K213E/N158C, E123R) have better antibody/antigen binding than bispecific V173C/N158C The absorbance value is high, but lower than the antibody/antigen binding absorbance value of the positive control F126C/S121C. These examples show that charge-reversal mutation of heavy chain K213 and light chain E123 in one Fab arm CH1-CL domain can further improve the correct pairing of heavy and light chains of bispecific antibodies with cysteine pair mutations. It appears that the charge inversion of K213D/E123K and K213E/E123K has a significantly greater improvement effect than that of K213D/E123R, K213E/E123R.

In the present invention, in the CH1-CL domain of one Fab arm, charge reversal of K213D/E123K was applied to improve the correct pairing of heavy and light chains of other bispecific antibodies with different cysteine mutations. As shown in Figure 8D, under the same antibody concentration conditions (the horizontal axis ELISA absorbance value), the bispecific antibody containing cysteine mutation in the CH1-CL domain of one Fab arm has similar to the positive control F126C/S121C or lower absorbance values for antibody/antigen binding. The corresponding bispecific antibodies containing cysteine and K213D/E123K charge mutations had higher antibody/antigen binding absorbance values than the positive controls F126C/S121C. Therefore, to further facilitate the correct pairing of heavy and light chains to form engineered disulfide bonds in the production of IgG kappa bispecific antibodies, charge mutations of K213D/E123K can be combined in the CH1-CL domain of one Fab arm and different cysteine pair mutations, including but not limited to:

S132C/F116C、V173C/N158C、S131C/P120C、S132C/S114C、S132C/P120C、K133C/F116C、K133C/P120C、S134C/P120C、T135C/S114C、G166C/S176C、G166C/T178C、H168C/T172C、H168C/ S174C、H168C/T178C、V173C/T180C、Q175C/N158C、Q175C/T172C、S176C/N158C、S176C/Q160C、G178C/N158C、G178C/S162C、G178C/T164C、S181C/S174C、S183C/P120C、S183C/S162C、 V185C/S114C, V185C/P120C, V185C/S176C, T187C/P120C, T187C/S176C, K218C/S114C, K218C/P120C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine/kappa light chain wild amino acid [EU EU numbering] to cysteine, EU numbering).

If necessary, to further facilitate correct pairing of heavy and light chains to form engineered disulfide bonds, the different cysteine pairs identified in Example 3 can be mutated in the CH1-CL domain of one Fab arm and K213D /E123K charge mutations combined.

Example 6. Design of IgG1 (lambda) CH1 and CL Domain "Cysteine Mutation" Libraries

The IgG lambda light chain is linked to the heavy chain by interchain disulfide bonds of the CH1 and CL domains. In the present invention, we mutated the CH1-CL domain of IgG (lambda), mutated the cysteine pair that forms the natural interchain disulfide bond to valine, and introduced a new cysteine pair to form Engineered interchain disulfide bonds to replace natural interchain disulfide bonds. Using the Fab crystal structure of human IgG1 (lambda) from the Protein Data Bank (PDB code: 2FB4) as a representative structure, we analyzed and designed cysteines in the CH1 and CL domains that may interact to form interchain disulfide bonds Mutation site. These cysteine mutation sites in the CH1 and CL domains are listed in Table 11 and were used to construct an IgG1 (lambda) "cysteine mutation" library.

Table 11. Cysteine mutation library of IgG1 (lambda) CH1-CL domain. Regions where cysteine mutations cluster together are called a group. The CH1 domain and the CL domain may form an interchain disulfide bond. Each group is listed separately.

Example 7. Construction and expression of IgG1 (lambda) cysteine mutation library

Fezakinumab (IL22 antibody) heavy chain (HoleRF-His) containing T366S/L368A/Y407V "hole" mutations, H435R, Y436F (RF) mutations and a 6xHis tag at the C-terminus was used as a representative of the IgG1 heavy chain (SEQ ID NO: 102 ). Fezakinumab heavy chain codon-optimized DNA (HoleRF-His) was synthesized and used as template for PCR amplification (SEQ ID NO: 103).

PCR 1, FezH F (SEQ ID NO: 104, primer with Xho I restriction enzyme site) and FezH R (SEQ ID NO: 105, primer with Not I restriction enzyme site) primer pair for Fezakinumab heavy chain (HoleRF- PCR amplification of His) using synthetic DNA as template. The PCR product was digested with Xho I and Not I and inserted into the same digestion site of pPIC9 (Invitrogen) to construct an expression vector for Fezakinumab heavy chain (HoleRF-His), named pPIC9-FezH (HoleRF-His).

Fezakinumab light chain was used as a representative of IgG1 lambda (λ) light chain (SEQ ID NO: 106). Codon-optimized DNA (SEQ ID NO: 107) of Fezakinumab lambda light chain was synthesized and used as template for PCR amplification.

PCR 2 FezλF (SEQ ID NO: 108, this primer has Xho I restriction enzyme site) and FezλR (SEQ ID NO: 109, this primer has Not I restriction enzyme site) primer pair for PCR amplification of fezakinumab lambda light chain , using a synthetic fezakinumab light chain as a template. The PCR product was digested with Xho I and Not I and inserted into the same digestion sites of pPICZα-Traκ-ARG2 (as described in Example 2) to construct an expression vector for fezakinumab light chain, designated pPICZα-Fezλ-ARG2.

PCR 3, EGF and EGR primer pairs were used to PCR amplify pEG linear fragments, using the pEG vector as a template.

An otelixizumab (CD3 mAb) heavy chain (knob) containing the T366W "knob" mutation and the C220V mutation was used as another representative of the IgGl heavy chain (SEQ ID NO: 110). Codon-optimized DNA of the otelixizumab heavy chain (knob) was synthesized and used as a template for PCR amplification (SEQ ID NO: 111).

PCR 4, heavy chain N-terminal primer OteH-Nt (SEQ ID NO: 112) and C-terminal primer OteH-Ct (SEQ ID NO: 113) correspond to the reverse primer (R) and forward primer (F) in Table 12, respectively ) composed of different primer pairs for PCR amplification of N-terminal and C-terminal heavy chains, using the synthetic otelixizumab heavy chain (knob) as a template.

PCR 5, using OteH-Nt and OteH-Ct primer pairs, PCR products of N-terminal and C-terminal heavy chains were connected by overlap extension PCR. In this way, otelixizumab heavy chain mutants were generated containing the wild-type amino acid mutation to cysteine, the T366W "knob" mutation and the C220V mutation.

The otelixizumab heavy chain mutant was inserted into the pEG linear fragment using the ClonExpress II one-step cloning kit (Vazyme) to generate an expression vector library named pEG-OteH(X#C) (OteH: otelixizumab heavy chain, (X#C): in The wild amino acid at position # is mutated to cysteine).

Table 12. PCR primers for mutation of otelixizumab heavy chain wild amino acid to cysteine (EU numbering).

mutation	Reverse primer (R)	Forward primer (F)
OteH-F126C	OteH-F126CR (SEQ ID NO: 114)	OteH-F126CF (SEQ ID NO: 115)
OteH-L128C	OteH-F126CR (SEQ ID NO: 114)	OteH-L128CF (SEQ ID NO: 116)
OteH-A129C	OteH-F126CR (SEQ ID NO: 114)	OteH-A129CF (SEQ ID NO: 117)
OteH-P130C	OteH-F126CR (SEQ ID NO: 114)	OteH-P130CF (SEQ ID NO: 118)
OteH-S131C	OteH-F126CR (SEQ ID NO: 114)	OteH-S131CF (SEQ ID NO: 119)
OteH-S132C	OteH-F126CR (SEQ ID NO: 114)	OteH-S132CF (SEQ ID NO: 120)
OteH-K133C	OteH-F126CR (SEQ ID NO: 114)	OteH-K133CF (SEQ ID NO: 121)
OteH-S134C	OteH-F126CR (SEQ ID NO: 114)	OteH-S134CF (SEQ ID NO: 122)
OteH-T135C	OteH-F126CR (SEQ ID NO: 114)	OteH-T135CF (SEQ ID NO: 123)
OteH-S136C	OteH-F126CR (SEQ ID NO: 114)	OteH-S136CF (SEQ ID NO: 124)

OteH-G166C	OteH-G166CR (SEQ ID NO: 125)	OteH-G166CF (SEQ ID NO: 126)
OteH-H168C	OteH-G166CR (SEQ ID NO: 125)	OteH-H168CF (SEQ ID NO: 127)
OteH-F170C	OteH-G166CR (SEQ ID NO: 125)	OteH-F170CF (SEQ ID NO: 128)
OteH-P171C	OteH-G166CR (SEQ ID NO: 125)	OteH-P171CF (SEQ ID NO: 129)
OteH-V173C	OteH-G166CR (SEQ ID NO: 125)	OteH-V173CF (SEQ ID NO: 130)
OteH-Q175C	OteH-G166CR (SEQ ID NO: 125)	OteH-Q175CF (SEQ ID NO: 131)
OteH-S176C	OteH-S176CR (SEQ ID NO: 132)	OteH-S176CF (SEQ ID NO: 133)
OteH-S177C	OteH-S176CR (SEQ ID NO: 132)	OteH-S177CF (SEQ ID NO: 134)
OteH-G178C	OteH-S176CR (SEQ ID NO: 132)	OteH-G178CF (SEQ ID NO: 135)
OteH-L179C	OteH-S176CR (SEQ ID NO: 132)	OteH-L179CF (SEQ ID NO: 136)
OteH-S181C	OteH-S181CR (SEQ ID NO: 137)	OteH-S181CF (SEQ ID NO: 138)
OteH-S183C	OteH-S181CR (SEQ ID NO: 137)	OteH-S183CF (SEQ ID NO: 139)
OteH-V185C	OteH-S181CR (SEQ ID NO: 137)	OteH-V185CF (SEQ ID NO: 140)
OteH-T187C	OteH-S181CR (SEQ ID NO: 137)	OteH-T187CF (SEQ ID NO: 141)
OteH-V215C	OteH-V215CR (SEQ ID NO: 142)	OteH-V215CF (SEQ ID NO: 143)
OteH-E216C	OteH-V215CR (SEQ ID NO: 142)	OteH-E216CF (SEQ ID NO: 144)
OteH-P217C	OteH-V215CR (SEQ ID NO: 142)	OteH-P217CF (SEQ ID NO: 145)
OteH-K218C	OteH-V215CR (SEQ ID NO: 142)	OteH-K218CF (SEQ ID NO: 146)
OteH-S219C	OteH-V215CR (SEQ ID NO: 142)	OteH-S219CF (SEQ ID NO: 147)
OteH-V220C	OteH-V215CR (SEQ ID NO: 142)	OteH-V220CF (SEQ ID NO: 148)

The Otelixizumab lambda (λ) light chain containing the C214V mutation was used as another representative of the IgG1 lambda light chain (SEQ ID NO: 149). otelixizumab lambda light chain codon-optimized DNA (SEQ ID NO: 150) was synthesized and used as a template for PCR amplification.

PCR 6, light chain N-terminal primer Oteλ-Nt (SEQ ID NO: 151, the primer has Xho I restriction enzyme site) and light chain C-terminal primer Oteλ-Ct (SEQ ID NO: 152, this primer has Not I restriction Enzyme site) and the corresponding reverse primers (R) and forward primers (F) in Table 13 respectively form different primer pairs for PCR amplification of N-terminal and C-terminal lambda light chains, using synthetic otelixizumab lambda chains as template.

PCR 7, using Oteλ-Nt and Oteλ-Ct primer pairs, PCR products of N-terminal and C-terminal lambda light chains were ligated by overlap extension PCR. In this way, otelixizumab lambda light chain mutants containing a mutation of the wild-type amino acid to cysteine and a C214V mutation were generated. The otelixizumab lambda chain can be directly PCR amplified using Oteλ-Nt and the corresponding reverse primer (R) in Table 13 to mutate the C-terminal wild amino acid to cysteine. Direct PCR amplification using Oteλ-Nt and Oteλ-V214C reverse primers (R) generated Otelixizumab lambda chains with wild-type cysteine (Kabat numbering) at position 214.

Table 13. PCR primers for mutation of wild-type amino acid to cysteine in the light chain of Otelixizumab lambda (λ) [Kabat numbering].

mutation	Reverse primer (R)	Forward primer (F)
Oteλ-S114C	Oteλ-S114CR (SEQ ID NO: 153)	Oteλ-S114CF (SEQ ID NO: 154)
Oteλ-T116C	Oteλ-S114CR (SEQ ID NO: 153)	Oteλ-T116CF (SEQ ID NO: 155)
Oteλ-F118C	Oteλ-S114CR (SEQ ID NO: 153)	Oteλ-F118CF (SEQ ID NO: 156)
Oteλ-P119C	Oteλ-S114CR (SEQ ID NO: 153)	Oteλ-P119CF (SEQ ID NO: 157)
Oteλ-S121C	Oteλ-S114CR (SEQ ID NO: 153)	Oteλ-S121CF (SEQ ID NO: 158)
Oteλ-G158C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-G158CF (SEQ ID NO: 160)
Oteλ-E160C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-E160CF (SEQ ID NO: 161)
Oteλ-T162C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-T162CF (SEQ ID NO: 162)

Oteλ-T163C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-T163CF (SEQ ID NO: 163)
Oteλ-P164C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-P164CF (SEQ ID NO: 164)
Oteλ-S165C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-S165CF (SEQ ID NO: 165)
Oteλ-Q167C	Oteλ-G158CR (SEQ ID NO: 159)	Oteλ-Q167CF (SEQ ID NO: 166)
Oteλ-K172C	Oteλ-K172CR (SEQ ID NO: 167)	Oteλ-K172CF (SEQ ID NO: 168)
Oteλ-A174C	Oteλ-K172CR (SEQ ID NO: 167)	Oteλ-A174CF (SEQ ID NO: 169)
Oteλ-S176C	Oteλ-K172CR (SEQ ID NO: 167)	Oteλ-S176CF (SEQ ID NO: 170)
Oteλ-Y178C	Oteλ-K172CR (SEQ ID NO: 167)	Oteλ-Y178CF (SEQ ID NO: 171)
Oteλ-S180C	Oteλ-K172CR (SEQ ID NO: 167)	Oteλ-S180CF (SEQ ID NO: 172)
Oteλ-V209C	Oteλ-V209CR (SEQ ID NO: 173)
Oteλ-A210C	Oteλ-A210CR (SEQ ID NO: 174)
Oteλ-P211C	Oteλ-P211CR (SEQ ID NO: 175)
Oteλ-T212C	Oteλ-T212CR (SEQ ID NO: 176)
Oteλ-E213C	Oteλ-E213CR (SEQ ID NO: 177)
Oteλ-V214C	Oteλ-V214CR (SEQ ID NO: 178)
Oteλ-S215C	Oteλ-S215CR (SEQ ID NO: 179)

Otelixizumab lambda light chain mutants were digested with Xho I and Not I and inserted into the same digestion site of pPIC6α-ARG4 to generate a library of lambda light chain mutant expression vectors, designated pPIC6-ARG4-Oteλ(X#C)(Oteλ: otelixizumab lambda light chain, (X#C): No. # wild amino acid mutated to cysteine, Kabat numbering). For example, pPIC6-ARG4-Otelλ-S114C represents a mutation of the wild-type serine at position 114 of the otelixizumab lambda light chain to a cysteine. As shown in Table 13, each vector expresses the otelixizumab lambda light chain containing the C214V mutation and the wild amino acid mutation to cysteine, and one vector expresses the otelixizumab lambda chain with the wild cysteine at position 214.

As described in Example 3, the ARG2 homologous sequence was digested with the restriction enzyme Afe I, the pPICZα-Fezλ-ARG2 expression vector was linearized, electroporated into GS2-1, and integrated into GS2-1 by recombination of the ARG2 5' and 3' homologous sequences at the ARG2 locus. Transformed cells were grown on YPD plates supplemented with 100 mg/L Zeocin. A new expression strain GS2-Fezλ was thus generated.

The pPIC9-FezH (HoleRF-His) expression vector was linearized with restriction enzyme Sal I, electroporated into the GS2-Fezλ strain, and integrated into the his4 locus of the Pichia genome. Transformed cells were selected on YNB plates. A new expression strain GS2-FezHλ was thus generated.

Each expression vector of the otelixizumab heavy chain mutant library pEG-OteH(X#C) was linearized with restriction enzyme Pme I, electroporated into the GS2-FezHλ strain, and integrated into the AOX1 locus of the Pichia genome. Transformed cells were selected on YPD plates supplemented with 250 mg/L G-418 sulfate. The expression strain library GS2-FezHκ-OteH (X#C) was thus generated.

Each expression vector of the otelixizumab lambda chain mutant library pPIC6-ARG4-Oteλ(X#C) was linearized within the ARG4 3' and 5' homologous sequences with restriction enzyme Sma I, and electroporated into the expression strain library GS2-FezHλ- OteH(X#C). The linear expression vector was integrated at the ARG4 locus by recombination of ARG4 3' and 5' homologous sequences. Transformed cells were grown on YPD plates supplemented with 300 mg/L blasticidin, thus generating the expression strain library GS2-FezHλ-OteH(X#C)λ(X#C). They can express asymmetric bispecific antibodies, half of which are otelixizumab heavy and light chains with engineered interchain disulfide bridges in the CH1 and CL domains of the Fab arms and " Knob" mutation. The other half are Fezakinumab heavy and light chains with native interchain disulfide bonds in the CH1 and CL domains of the Fab arms, a "hole" and RF mutation in the Fc, and a 6xHis tag at the C-terminus of the heavy chain.

As described in Example 2, the expression strains were grown in BMGY medium and the antibodies were induced in BMMY medium. The supernatant harvested by centrifugation was frozen at -20°C until used in the next step.

Example 8. Screening to identify IgG1 (lambda) "cysteine mutant" library

Bispecific antibodies were purified by protein A affinity chromatography as described in Example 3, and the concentration of purified antibody was determined by ELISA.

As described in Example 3, the protein A purified bispecific antibody was diluted 1:2 with PBS and divided into two parts, one part was determined by ELISA on the plate coated with AGL protein, and the other part was measured with human interleukin 22 ( IL22) (Sino Biological) coated plate was determined by ELISA to determine the binding of antibody to antigen.

According to reports, when phenylalanine at position 126 of Otelixizumab heavy chain CH1 was mutated to cysteine (F126C, EU numbering), and serine 121 in lambda light chain CL was mutated to cysteine (S121C, Kabat numbering), Engineered interchain disulfide bonds can be formed to allow the correct pairing of heavy and light chains. This asymmetric bispecific antibody was named FezHλ-OteH(F126C)λ(S121C), abbreviated as F126C/S121C, and used as a positive control. However, when both Fab arms have native interchain disulfide bonds, the resulting bispecific antibody has only 25% of the heavy and light chains correctly paired. This bispecific antibody is named FezHλ-OteHλ, abbreviated as WT, and used as a negative control (Yariv Mazor, Vaheh Oganesyan, Chunning Yang, Anna Hansen, Jihong Wang, Hongji Liu, Kris Sachsenmeier, Marcia Carlson, Dhanesh V Gadre, Martin Jack Borrok, Xiang-Qing Yu, William Dall'Acqua, Herren Wu, and Partha Sarathi Chowdhury, mAbs 7, 377--389; 2015).

As described in Example 3, an ELISA method was used to identify engineered interchain disulfide bonds in bispecific antibodies that facilitate the correct pairing of heavy and light chains.

As shown in Figure 9, under the same antibody concentration conditions (the horizontal axis ELISA absorbance value), the bispecific antibody positive control F126C/S121C showed high antibody/antigen binding ELISA absorbance value (vertical axis), but the bispecific antibody positive control F126C/S121C showed high antibody/antigen binding ELISA absorbance value (vertical axis) The antibody negative control WT showed very low absorbance values for the antibody/antigen binding ELISA. Under the same antibody concentration conditions (ELISA absorbance value on the horizontal axis), many of the bispecific antibodies we constructed had the same or higher antibody/antigen binding absorbance value as the positive control F126C/S121C. This indicates that the heavy and light chains of our bispecific antibodies are able to pair correctly. Therefore, in IgG1 (lambda) bispecific antibody production, an engineered disulfide bond is formed in the CH1-CL domain of one Fab arm by replacing the native interchain disulfide bond with a different cysteine pairing , which enables the correct pairing of heavy and light chains, including:

S132C/S121C, K133C/T116C, K133C/P211C, S136C/S121C, F170C/G158C, P171C/T162C, P171C/P164C, S176C/T162C, L179C/G158C, S181C/P164C, E21C1FC

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine/lambda light chain wild amino acid [Kabat numbering] to cysteine).

Example 9. Construction and expression of an IgG1 (lambda) "cysteine and charge mutation" library

PCR 1, EGF and EGR primer pairs were used to PCR amplify pEG linear fragments using the pEG vector as template (as described in Example 2). "

PCR 2, use the corresponding primer pairs in Table 14-1 to carry out PCR amplification of the N-terminal and C-terminal heavy chains respectively, and use each expression vector of the otelixizumab mutant library pEG-OteH (X#C) as a template. Using the OteH-Nt and OteH-Ct primer pairs, the N- and C-terminal heavy chains were linked by overlap extension PCR, thus generating a mutated otelixizumab heavy chain containing the T366W "knob" mutation, the C220V mutation, and the wild-type amino acid mutation to cysteine amino acid and the K213D mutation.

Table 14-1 Otelixizumab heavy chain N-terminal and C-terminal PCR amplification primer pairs

PCR 3, using the corresponding primer pairs in Table 14-2 to carry out PCR amplification of the N-terminal and C-terminal heavy chains, respectively, using each expression vector of the otelixizumab mutant library pEG-OteH (X#C) as a template. Using the OteH-Nt and OteH-Ct primer pairs, the N- and C-terminal heavy chains were linked by overlap extension PCR, thus generating a mutated otelixizumab heavy chain containing the T366W "knob" mutation, the C220V mutation, and the wild-type amino acid mutation to cysteine amino acid and the K213E mutation.

Table 14-2 Otelixizumab heavy chain N-terminal and C-terminal PCR amplification primer pairs

The mutated otelixizumab heavy chain was inserted into pEG linear fragments using the ClonExpress II one-step cloning kit (Vazyme) to construct otelixizumab mutant expression vector libraries, designated pEG-OteH (X#CD) and pEG-OteH (X#CE).

PCR 4, using the corresponding primer pairs in Table 15 to carry out PCR amplification of the N-terminal and C-terminal lambda light chain respectively, using each expression vector of the otelixizumab mutant library pPIC6α-ARG4-OteHλ(X#C) as a template. Using the Oteλ-Nt and Oteλ-Ct primer pairs, the N- and C-terminal lambda light chains were linked by overlap extension PCR, thus generating a mutated otelixizumab lambda light chain containing the T366W "knob" mutation, the C220V mutation, and the wild-type amino acid mutation as Cysteine and E123K mutations.

Table 15 Otelixizumab light chain N-terminal and C-terminal PCR amplification primer pairs

The mutated otelixizumab lambda light chain was digested with Xho I and Not I and inserted into the same digestion site of pPIC6α-ARG4 to generate a library of lambda light chain mutant expression vectors, designated pPIC6α-ARG4-Oteλ (X#CK). Each vector can express mutated otelixizumab lambda light chain, including C214V mutation, wild amino acid mutation to cysteine, and E123K.

Each expression vector of the otelixizumab heavy chain mutant library pEG-OteH(X#CD) was linearized with restriction enzyme Pme I, electroporated into the GS2-FezHλ strain, and integrated at the AOX1 locus of the Pichia genome. Transformed cells were selected on YPD plates supplemented with 250 mg/L G-418 sulfate, thus generating a library of expression strains (GS2-FezHλ-OteH(X#CD).

Each expression vector of the otelixizumab lambda light chain mutant library pPIC6α-ARG4-Oteλ(X#CK) was linearized in the ARG4 3' and 5' homologous sequences with restriction enzyme Sma I, and electroporated into the expression strain library GS2-FezHλ -OteH (X#CD), and integrated into the ARG4 locus. Transformed cells were grown on YPD plates supplemented with 300 mg/L blasticidin, thus generating a library of expression strains GS2-FezHλ-OteH(X#CD)λ(X#CK) for expressing asymmetric bispecific Sexual antibodies. In the asymmetric bispecific antibody, half of which are otelixizumab heavy and light chains, the CH1-CL domains of the Fab arms have an engineered interchain disulfide bond and a pair of charge-reversal mutations, and have " Knob" mutation; the other half are Fezakinumab heavy and light chains with native interchain disulfide bonds in the CH1-CL domains of the Fab arms, "hole" and RF mutations in the Fc, and 6xHis at the C-terminus of the heavy chain Tag, expressing strains were grown in BMGY medium as described in Example 2, and antibody expression was induced in BMMY medium. The supernatant harvested by centrifugation was frozen at -20°C until used in the next step.

Example 10. Screening to identify an IgG1 (lambda) "cysteine and charge mutation" library

As described in Example 3, the protein A purified bispecific antibody was diluted 1:2 with PBS and divided into two parts, one part was determined by ELISA on the plate coated with AGL protein, and the other part was measured with human interleukin 22 ( IL22) (Sino Biological)-coated plates were assayed for antibody-antigen binding by ELISA.

As shown in Figure 10, under the same antibody concentration conditions (the horizontal axis ELISA absorbance value), the positive control F126C/S121C had high antibody/antigen binding ELISA absorbance value (vertical axis), but the negative control WT had very low Antibody/antigen binding ELISA absorbance values. Bispecific antibodies containing cysteine mutations in the CH1-CL domain of one Fab arm had antibody/antigen binding absorbance values similar to or lower than positive controls F126C/S121C. The corresponding bispecific antibodies containing cysteine and charge mutations of K213D/E123K had higher antibody/antigen binding absorbance values than the positive controls F126C/S121C. Therefore, in the production of IgG lambda bispecific antibodies, to further achieve correct pairing of heavy and light chains to form engineered disulfide bonds, K213D/E123K charge mutations can be combined in the CH1-CL domain of one Fab arm Pairs with different cysteines, including:

L128C/T116C, A129C/P211C, A129C/T212C, P130C/A210C, P171C/T163C, E216C/T116C, P217C/S215C, K218C/F118C, K218C/P119C

(Each pair of cysteine mutations is listed as follows: heavy chain wild amino acid [EU numbering] to cysteine/lambda light chain wild amino acid [Kabat numbering] to cysteine).

Example 11. Expression of bispecific antibody IgG1 (kappa/lambda)

The expression vector pEG-ClaH(X#C) for the clazakizumab heavy chain mutant was linearized with the restriction enzyme Pme I as described in Example 2, electroporated into the GS2-FezHλ strain (as described in Example 7), and The AOX1 locus integrated into the Pichia genome. Transformed cells GS2-FezHλ-ClaH (X#C) were selected on YPD plates supplemented with 500 mg/L G-418 sulfate. Subsequently, the expression vector pPIC6-ARG4-Claκ(X#C) of the clazakizumab kappa light chain mutant was linearized with the restriction enzyme Sma I, electroporated into the GS2-FezHλ-ClaH(X#C) strain, and integrated into the ARG4 locus superior. Transformed cells were grown on YPD plates supplemented with 300 mg/L blasticidin. An expression strain GS2-FezHλ-ClaH(X#C)κ(X#C) capable of expressing asymmetric bispecific antibodies was thus produced. In an asymmetric bispecific antibody, one half is clazakizumab heavy chain and kappa light chain with an engineered interchain disulfide bond in the CH1-CL domain of the Fab arm and a "knob" mutation in the Fc, the other half is are fezakinumab heavy chain and lambda light chain with native interchain disulfide bonds in the CH1-CL domains of the Fab arms, a "hole" and RF mutation in the Fc, and a 6xHis tag at the C-terminus of the heavy chain.

As described in Example 2, the expression strains were grown in BMGY medium and the antibodies were induced in BMMY medium. The supernatant harvested by centrifugation was frozen at -20°C until used in the next step.

As described in Example 3, the protein A purified bispecific antibody was diluted 1:2 with PBS and divided into two parts, one part was determined by ELISA on the plate coated with AGL protein, and the other part was measured with human interleukin 22 ( IL22) (Sino Biological) coated plate was determined by ELISA to determine the binding of antibody to antigen.

Taking two bispecific anti-FezHλ-ClaH(K218C)κ(F118C) and FezHλ-ClaH(S132C)κ(F116C) as representative examples, the Fab arm heavy chain CH1 and kappa light chain CL domains of clazakizumab contain Correct pairing of light and heavy chains in bispecific IgG kappa/lambda with a pair of novel cysteine mutations (K218C/F118C or S132C/F116C). In these two asymmetric bispecifics, half are clazakizumab heavy chain and kappa light chain, which contain a new pair of cysteines (K218C/F118C or S132C/F116C) in the Fab arm and in the Fc There is a "knob" mutation. The other half is the fezakinumab heavy chain and lambda light chain, with native interchain disulfide bonds in their Fab arms, a "hole" and RF mutation in the Fc, and a 6xHis tag at the C-terminus of the heavy chain, as shown in Figure 11 , under the same antibody concentration conditions (the horizontal axis ELISA absorbance value), the two IgG1 (kappa/lambda) bispecific antibodies have similar antibody/antigen binding absorbance values as the positive control F126C/S121C.

These examples demonstrate that in the production of an IgG1 kappa/lambda bispecific antibody, it is possible to mutate a cysteine pair in the CH1-CL domain of one Fab arm to form an engineered disulfide bond, replacing the native interchain disulfide bond , to achieve the correct pairing of heavy and light chains. Correct pairing of heavy and light chains can also be achieved by combining charge mutations of K213D/E123K and different cysteine pairings to form engineered disulfide bonds in the CH1-CL domain of the Fab arm, if necessary.

Example 12. Production of bispecific IgG1 with different Fab arms at cysteine mutation and charge mutation

PCR 1, using synthetic trastuzumab heavy chain (HoleRF-His) DNA as template, N- and C-terminal heavy chain with TraH F and ClaH-K213DR primer pair and ClaH-K213D F and TraHR primer pair PCR amplification. The N-terminal and C-terminal heavy chains were ligated by overlap extension PCR using the TraHF and TraHR primer pairs.

PCR 2, using synthetic trastuzumab heavy chain (HoleRF-His) DNA as template, using TraH F and ClaH-K213ER primer pair and ClaH-K213EF and TraHR primer pair for N-terminal and C-terminal heavy chain PCR amplification. The N-terminal and C-terminal heavy chains were ligated by overlap extension PCR using TraHF and TraHR primers.

PCR 1 and 2 generated mutated trastuzumab heavy chains containing K213D or K213E mutations in the CH1 domain, T366S/L368A/Y407V "hole" mutations, H435R, Y436F(RF) mutations in the Fc domain, end contains the 6xHis tag.

The mutated trastuzumab heavy chain was digested with Xho I and Not I and inserted into the same digestion site of pPIC9 (Invitrogen) to construct an expression vector for the trastuzumab heavy chain, named pPIC9-TraH (D) and pPIC9-TraH(E). Both vectors express a mutated trastuzumab heavy chain containing K213D or K213E mutations in the CH1 domain and T366S/L368A/Y407V "hole" mutations in the Fc domain, H435R, Y436F(RF) Mutation and inclusion of a 6xHis tag at the C-terminus.

PCR 3, using synthetic trastuzumab kappa chain as template, PCR amplification of N-terminal and C-terminal kappa chains with TraκF and Claκ-E123K R primer pair and Claκ-E123K F and TraκR primer pair. The N-terminal and C-terminal kappa chains were ligated by overlap extension PCR using the TraκF and TraκR primer pairs.

PCR 4, using synthetic trastuzumab kappa chain as template, PCR amplification of N-terminal and C-terminal kappa chains with TraκF and Claκ-E123R R primer pair and Claκ-E123RF and TraκR primer pair. The N-terminal and C-terminal kappa chains were ligated by overlap extension PCR using the TraκF and TraκR primer pairs.

PCR 3 and 4 products were digested with Xho I and Not I, and inserted into the same digestion site of pPICZα-Traκ-ARG2 to construct expression vectors for the trastuzumab kappa chain, named pPICZα-Traκ(K)-ARG2 and pPICZα -Traκ(R)-ARG2. Both vectors can express mutated trastuzumab kappa chains that contain either E123K or E123R mutations in the CL domain.

Expression vectors for pPICZα-Traκ(K)-ARG2 and pPICZα-Traκ(R)-ARG2 were linearized with Afe I, electroporated into GS2-1 and integrated at the ARG2 locus as described in Example 2. Transformed cells were grown on YPD plates supplemented with 100 mg/L Zeocin. Thus, new expression strains GS2-Traκ(K) and GS2-Traκ(R) were created, which can express the trastuzumab kappa light chain.

The expression vectors of pPIC9-TraH (D) and pPIC9-TraH (E) were linearized with Sal I, electroporated into the GS2-Traκ (K) and GS2-Traκ (R) strains, and integrated into the his4 locus. Transformed cells were selected on YNB plates. Strains of GS2-TraH(D)κ(K), GS2-TraH(D)κ(R), GS2-TraH(E)κ(K) and GS2-TraH(E)κ(R) were thus generated to express Mutated trastuzumab.

Certain expression vectors of the clazakizumab heavy chain mutant library pEG-ClaH(X#C) were linearized with Pme I and electroporated into GS2-TraH(D)κ(K), GS2-TraH(D)κ(R), GS2-TraH(E)κ(K) and GS2-TraH(E)κ(R) strains and integrated at the AOX1 locus. Transformed cells were selected on YPD plates supplemented with 250 mg/L G-418 sulfate. A library of expression strains was thus generated:

GS2-TraH(D)κ(K)-ClaH(X#C),

GS2-TraH(D)κ(R)-ClaH(X#C),

GS2-TraH(E)κ(K)-ClaH(X#C),

GS2-TraH(E)κ(R)-ClaH(X#C).

The expression vector of the clazakizumab kappa chain mutant library pPIC6-ARG4-Claκ(X#C) was linearized with Sma I, electroporated into the above-described library of expression strains, and integrated into the ARG4 locus. Transformed strains were grown on YPD plates supplemented with 300 mg/L blasticidin. So this creates a library of expression strains:

GS2-TraH(D)κ(K)-ClaH(X#C)κ(X#C),

GS2-TraH(D)κ(R)-ClaH(X#C)κ(X#C),

GS2-TraH(E)κ(K)-ClaH(X#C)κ(X#C),

GS2-TraH(E)κ(R)-ClaH(X#C)κ(X#C).

As described in Example 2, expression strains were grown in BMGY medium, and antibody expression was induced in BMMY medium. The supernatant harvested by centrifugation was frozen at -20°C until used in the next step.

As described in Example 3, the protein A purified bispecific antibody was diluted 1:2 with PBS and then divided into two parts, one part was determined by ELISA using AGL protein-coated plate, and the other part was determined by ELISA with human HER2/ErbB2 The protein-coated plates were assayed for antibody-antigen binding by ELISA.

The following three groups of bispecific antibodies are used as representative examples:

TraH(D)κ(K)-ClaH(S219C)κ(P120C), abbreviated as S219C/P120C, K213D/E123K;

TraH(E)κ(K)-ClaH(S219C)κ(P120C), abbreviated as S219C/P120C, K213E/E123K;

TraH(D)κ(R)-ClaH(S219C)κ(P120C), abbreviated as S219C/P120C, K213D/E123R;

TraH(E)κ(R)-ClaH(S219C)κ(P120C), abbreviated as S219C/P120C, K213E/E123R;

TraH(D)κ(K)-ClaH(K218C)κ(F118C), abbreviated as K218C/F118C, K213D/E123K;

TraH(E)κ(K)-ClaH(K218C)κ(F118C), abbreviated as K218C/F118C, K213E/E123K;

TraH(D)κ(R)-ClaH(K218C)κ(F118C), abbreviated as K218C/F118C, K213D/E123R;

TraH(E)κ(R)-ClaH(K218C)κ(F118C), abbreviated as K218C/F118C, K213E/E123R,

TraH(D)κ(K)-ClaH(V173C)κ(N158C), abbreviated as V173C/N158C, K213D/E123K;

TraH(E)κ(K)-ClaH(V173C)κ(N158C), abbreviated as V173C/N158C, K213E/E123K;

TraH(D)κ(R)-ClaH(V173C)κ(N158C), abbreviated as V173C/N158C, K213D/E123R;

TraH(E)κ(R)-ClaH(V173C)κ(N158C), abbreviated as V173C/N158C, K213E/E123R.

Half of these asymmetric bispecific antibodies are clazakizumab heavy chain and kappa light chain, which contain a new pair of cysteines (S219C/P120C, K218C/F118C or V173C/ N158C), contains the "knob" mutation in the "Fc"; the other half are the trastuzumab heavy and light chains, which retain the native interchain disulfide bonds in their Fab arms, but the Fab arm CH1 contains K213D or K213E Charge mutation, including charge mutation of E123K or E123R in Fab arm CL, "hole" and RF mutation in Fc, 6xHis tag in heavy chain C-terminus,

As shown in Figure 12, under the same antibody concentration conditions (ELISA absorbance value on the horizontal axis), all of these bispecific antibodies had higher antibody/antigen binding absorbance values (ELISA absorbance value on the vertical axis) than the positive control F126C/S121C.

These examples demonstrate that, in the production of IgG bispecific antibodies, through charge mutation of the CH1-CL domain of one Fab arm, mutation of a cysteine pair in the CH1-CL domain of the other Fab arm forms an engineered bispecific Sulfide bonds replace natural interchain disulfide bonds, allowing for correct pairing of heavy and light chains.

Example 13. Design of IgG4 (Kappa) CH1 and CL Domain "Cysteine Mutation" Libraries

Although the amino acid sequence and spatial structure of the CH1 domains of IgG1, IgG2, IgG3 and IgG4 are very similar, the positions of the cysteine pairs that form natural disulfide bonds are different. The light chain of IgG1 passes through kappa and the cysteine at the C-terminus of the CL domain of the lambda chain (cysteine at position 214 of the kappa light chain [EU numbering]; cysteine at position 214 of the lambda light chain [Kabat numbering]) It forms an interchain disulfide bond with the cysteine at the C-terminus of the CH1 domain of the heavy chain (cysteine at position 220 of the heavy chain [EU numbering]). In contrast, the light chain of IgG2, IgG3, or IgG4 interacts with the cysteine at the C-terminus of the CL domain in the kappa and lambda chains to the cysteine at the N-terminus of the CH1 domain in the heavy chain (cysteine at position 131 of the heavy chain). acid [EU numbering]) to form interchain disulfide bonds (Figure 1, A and B). Although the positions of cysteines in the amino acid sequence differ between IgG1 and other subtypes (IgG2, IgG3, IgG4), their spatial positions are similar to form interchain disulfide bonds. To understand where IgG2, IgG3 or IgG4 may form other disulfide bonds, we analyzed and designed the CH1 and CL structures using the Fab crystal structure of human IgG4 (kappa) from the Protein Data Bank (PDB code: 5DK3) as a representative structure Mutation sites of cysteine domains in the domains that may interact to form interchain disulfide bonds, an IgG4 (kappa) mutant library was designed in which pairs of heavy and light chain cysteines that form native interchain disulfide bonds were Mutated to serine and valine, and introduced new cysteine pairs at different positions in the CH1-CL domain to form engineered interchain disulfide bonds. Table 16 lists the introduction of new cysteine pairs at different amino acid positions in the CH1-CL domain of IgG4 (Kappa) that may form engineered interchain disulfide bonds.

Table 16. Library of cysteine mutations in the CH1-CL domain of IgG4 (kappa). Regions where cysteine mutations cluster together are called a group. The CH1 domain and the CL domain may form interchain disulfide bonds. Each group is listed correspondingly.

Example 14. Construction and expression of IgG4 (Kappa) cysteine mutation library

Ixekizumab (IL-17A antibody) heavy chain (HoleRF-His) containing T366S/L368A/Y407V "hole" mutations, H435R, Y436F (RF) mutations (EU numbering) and a 6xHis tag at the C-terminus was used as a representative of the IgG4 heavy chain (SEQ ID NO: 202). Ixekizumab heavy chain codon-optimized DNA (HoleRF-His) was synthesized and used as a template for PCR amplification (SEQ ID NO: 203).

PCR 1, IxeHF (SEQ ID NO: 204, primer with Xho I restriction enzyme site) and IxeHR (SEQ ID NO: 205, primer with Not I restriction enzyme site) primer pair for Ixekizumab heavy chain (HoleRF- PCR amplification of His) using synthetic DNA as template. The PCR product was digested with Xho I and Not I and inserted into the same digestion site of pPIC9 (Invitrogen) to construct an expression vector for the heavy chain of ixekizumab (HoleRF-His), designated pPIC9-IxeH (HoleRF-His).

Ixekizumab light chain was used as a representative of IgG4 kappa (κ) light chain (SEQ ID NO: 206). Codon-optimized DNA (SEQ ID NO: 207) of the light chain of Ixekizumab kappa was synthesized and used as a template for PCR amplification.

PCR 2 IxeκF (SEQ ID NO:208, this primer has Xho I restriction enzyme site) and IxeκR (SEQ ID NO:209, this primer has Not I restriction enzyme site) primer pair for PCR amplification of Ixekizumab kappa light chain , using a synthetic Ixekizumab light chain as a template. The PCR product was digested with Xho I and Not I and inserted into the same digestion sites of pPICZα-Traκ-ARG2 (as described in Example 2) to construct an expression vector for the light chain of Ixekizumab, designated pPICZα-Ixeκ-ARG2.

PCR 3, EGF and EGR primer pairs were used to PCR amplify pEG linear fragments, using the pEG vector as a template.

The olokizumab (IL6 mAb) heavy chain (knob) containing the T366W "knob" mutation and the C131S mutation [EU numbering] was used as another representative of the IgG4 heavy chain (SEQ ID NO: 210). Codon-optimized DNA (SEQ ID NO: 211) of the olokizumab heavy chain (knob) was synthesized and used as a template for PCR amplification.

PCR 4, heavy chain N-terminal primer OloH-Nt (SEQ ID NO: 212) and C-terminal primer OloH-Ct (SEQ ID NO: 213) correspond to the reverse primer (R) and forward primer (F) in Table 17, respectively ) composed of different primer pairs for PCR amplification of the N-terminal and C-terminal heavy chains, using the synthetic olokizumab heavy chain (knob) as a template.

Table 17. PCR primers for mutation of wild amino acid to cysteine in clazakizumab heavy chain (EU numbering)

PCR 5, using OloH-Nt and OloH-Ct primer pairs, PCR products of N-terminal and C-terminal heavy chains were ligated by overlap extension PCR. In this way, an olokizumab heavy chain mutant was generated, which contains a mutation of the wild-type amino acid to cysteine, a T366W "knob" mutation and a C131S mutation. One of them contains the T366W "knob" mutation and wild-type cysteine at position 134, which can form native interchain disulfide bonds.

PCR 6, heavy chain N-terminal primer OloH-Nt and OloH-K147DR (SEQ ID NO:246) primer pair and OloH-K147DF (SEQ ID NO:247) and C-terminal primer OloH-Ct primer pair were used for PCR, respectively The N- and C-terminal heavy chains were amplified using the heavy chain containing the T366W "knob" mutation and wild-type cysteine 134 as a template.

PCR 7, using OloH-Nt and OloH-Ct primer pairs, PCR products of N-terminal and C-terminal heavy chains were ligated by overlap extension PCR. In this way, an olokizumab heavy chain mutant was generated, containing the K147D mutation, the T366W "knob" mutation and the wild-type cysteine at position 134, capable of forming native interchain disulfide bonds

The olokizumab heavy chain mutant was inserted into the pEG linear fragment using the ClonExpress II one-step cloning kit (Vazyme) to generate an expression vector library named pEG-OloH(X#C) (OloH: olokizumab heavy chain, (X#C): in The wild amino acid at position # is mutated to cysteine). One of the expression vectors is pEG-OloH(K147D) (OloH: olokizumab heavy chain, K147D: lysine 147 is mutated to aspartic acid).

The Olokizumab kappa (κ) light chain containing the C214V mutation was used as another representative of the IgG4 kappa light chain (SEQ ID NO: 248). Olokizumab kappa light chain codon-optimized DNA (SEQ ID NO: 249) was synthesized and used as a template for PCR amplification.

PCR 8, light chain N-terminal primer Oloκ-Nt (SEQ ID NO:250, the primer has Xho I restriction enzyme site) and light chain C-terminal primer Oloκ-Ct (SEQ ID NO:251, this primer has Not I restriction Enzyme site) and the corresponding reverse primers (R) and forward primers (F) in Table 18 respectively form different primer pairs for PCR amplification of N-terminal and C-terminal kappa light chains, using synthetic Olokizumab kappa chains as template.

Table 18. PCR primers for mutation of wild amino acid to cysteine in Olokizumab kappa (κ) light chain [EU numbering]

mutation	Reverse primer (R)	Forward primer (F)
Oloκ-S114C	Oloκ-S114CR (SEQ ID NO:252)	Oloκ-S114CF (SEQ ID NO: 253)
Oloκ-F116C	Oloκ-S114CR (SEQ ID NO:252)	Oloκ-F116CF (SEQ ID NO:254)
Oloκ-F118C	Oloκ-S114CR (SEQ ID NO:252)	Oloκ-F118CF (SEQ ID NO:255)
Oloκ-P119C	Oloκ-S114CR (SEQ ID NO:252)	Oloκ-P119CF (SEQ ID NO:256)
Oloκ-P120C	Oloκ-S114CR (SEQ ID NO:252)	Oloκ-P120CF (SEQ ID NO: 257)
Oloκ-S121C	Oloκ-S114CR (SEQ ID NO:252)	Oloκ-S121CF (SEQ ID NO: 258)
Oloκ-N158C	Oloκ-N158CR (SEQ ID NO:259)	Oloκ-N158CF (SEQ ID NO: 260)
Oloκ-Q160C	Oloκ-N158CR (SEQ ID NO:259)	Oloκ-Q160CF (SEQ ID NO:261)
Oloκ-S162C	Oloκ-N158CR (SEQ ID NO:259)	Oloκ-S162CF (SEQ ID NO: 262)
Oloκ-V163C	Oloκ-N158CR (SEQ ID NO:259)	Oloκ-V163CF (SEQ ID NO:263)
Oloκ-T164C	Oloκ-N158CR (SEQ ID NO:259)	Oloκ-T164CF (SEQ ID NO:264)
Oloκ-E165C	Oloκ-N158CR (SEQ ID NO:259)	Oloκ-E165CF (SEQ ID NO:265)
Oloκ-T172C	Oloκ-T172CR (SEQ ID NO:266)	Oloκ-T172CF (SEQ ID NO: 267)
Oloκ-S174C	Oloκ-T172CR (SEQ ID NO:266)	Oloκ-S174CF (SEQ ID NO: 268)
Oloκ-S176C	Oloκ-T172CR (SEQ ID NO:266)	Oloκ-S176CF (SEQ ID NO: 269)
Oloκ-T178C	Oloκ-T172CR (SEQ ID NO:266)	Oloκ-T178CF (SEQ ID NO: 270)
Oloκ-T180C	Oloκ-T172CR (SEQ ID NO:266)	Oloκ-T180CF (SEQ ID NO:271)

Oloκ-S182C	Oloκ-T172CR (SEQ ID NO:266)	Oloκ-S182CF (SEQ ID NO: 272)
Oloκ-F209C	Oloκ-F209CR (SEQ ID NO:273)
Oloκ-N210C	Oloκ-N210CR (SEQ ID NO:274)
Oloκ-R211C	Oloκ-R211CR (SEQ ID NO:275)
Oloκ-G212C	Oloκ-G212CR (SEQ ID NO: 276)
Oloκ-E213C	Oloκ-E213CR (SEQ ID NO: 277)
Oloκ-V214C	Oloκ-V214CR (SEQ ID NO: 278)

PCR 9, using Oloκ-Nt and Oloκ-Ct primer pair, PCR products of N-terminal and C-terminal light chains were ligated by overlap extension PCR. In this way, Olokizumab kappa light chain mutants containing the wild-type amino acid mutation to cysteine and the C214V mutation were generated. Olokizumab kappa light chain can be directly PCR amplified using Oloκ-Nt and the corresponding reverse primer (R) in Table 18, mutating the C-terminal wild amino acid to cysteine. Direct PCR amplification using the Oloκ-Nt and Oloκ-V214C reverse primers (R) resulted in an Olokizumab kappa light chain with wild-type cysteine at position 214, capable of forming native interchain disulfide bonds.

PCR 10, light chain N-terminal primers Oloκ-Nt and Oloκ-T129R R (SEQ ID NO:279) primer pair and Oloκ-T129RF (SEQ ID NO:280) and C-terminal primer Oloκ-Ct primer pair were used for PCR, respectively N-terminal and C-terminal light chains were amplified using Olokizumab kappa light chain with wild-type cysteine at position 214 as template.

PCR 11, using Oloκ-Nt and Oloκ-Ct primer pair, PCR products of N-terminal and C-terminal heavy chains were ligated by overlap extension PCR. In this way, an olokizumab light chain mutant was generated, containing the T129R mutation, and the wild-type cysteine at position 214, capable of forming native interchain disulfide bonds

Olokizumab kappa light chain mutants were digested with Xho I and Not I and inserted into the same digestion site of pPIC6α-ARG4 to generate a library of light chain mutant expression vectors, designated pPIC6-ARG4-Oloκ(X#C) (Oloκ: Olokizumab kappa light chain, (X#C): wild-type amino acid at position # is mutated to cysteine).

As described in Example 2, the ARG2 homologous sequence was digested with the restriction enzyme Afe I, the pPICZα-Ixeκ-ARG2 expression vector was linearized, electroporated into GS2-1, and integrated into GS2-1 by recombination of the ARG2 5' and 3' homologous sequences at the ARG2 locus. Transformed cells were grown on YPD plates supplemented with 100 mg/L Zeocin. A new expression strain GS2-Ixeκ was thus generated.

The pPIC9-IxeH (HoleRF-His) expression vector was linearized with the restriction enzyme Sal I, electroporated into the GS2-Ixeκ strain, and integrated into the his4 locus of the Pichia genome. Transformed cells were selected on YNB plates. A new expression strain GS2-IxeHκ was thus generated.

Each expression vector of the Olokizumab heavy chain mutant library pEG-OloH(X#C) was linearized with restriction enzyme Pme I, electroporated into the GS2-IxeHκ strain, and integrated into the AOX1 locus of the Pichia genome. Transformed cells were selected on YPD plates supplemented with 250 mg/L G-418 sulfate. The expression strain library GS2-IxeHκ-OloH (X#C) was thus generated.

Each expression vector of Olokizumab kappa light chain mutant library pPIC6-ARG4-Oloκ(X#C) was linearized within ARG4 3' and 5' homologous sequences with restriction enzyme Sma I, and electroporated into expression strain library GS2-IxeHκ -OloH(X#C). The linear expression vector was integrated at the ARG4 locus by recombination of ARG4 3' and 5' homologous sequences. Transformed cells were grown on YPD plates supplemented with 300 mg/L blasticidin, thus generating the expression strain library GS2-IxeHκ-OloH(X#C)κ(X#C). They can express asymmetric bispecific antibodies, half of which are Olokizumab heavy and light chains with engineered interchain disulfide bonds in the CH1 and CL domains of the Fab arms and " Knob" mutation. The other half are Ixekizumab heavy and light chains with native interchain disulfide bonds in the CH1 and CL domains of the Fab arms, a "hole" and RF mutation in the Fc, and a 6xHis tag at the C-terminus of the heavy chain.

As described in Example 2, the expression strains were grown in BMGY medium and the antibodies were induced in BMMY medium. The supernatant harvested by centrifugation was frozen at -20°C until used in the next step.

Example 15. Screening to identify IgG4 (Kappa) cysteine mutation library

As described in Example 3, the bispecific antibody purified from protein A was diluted 1:2 with PBS and then divided into two parts. A part of the bispecific antibody was directly coated on the plate, and anti-Fc-HRP (Invitrogen) was used to pass ELISA. The antibody concentration was determined, and the other part was assayed for the binding of the antibody to the antigen by ELISA using human interleukin 17A (IL17A) (Sino Biological)-coated plates.

According to reports, when lysine 147 in Olokizumab heavy chain CH1 is mutated to aspartic acid (K147D), and threonine 129 in kappa light chain CL is mutated to arginine (T129R) [EU numbering], it can promote remodeling. Chain and light chain are correctly paired up to 80-99%. This asymmetric bispecific antibody is named IxeHκ-OloH(K147D)κ(T129R), abbreviated as K147D/T129R, and used as a positive control. (Maximilian

Carolin Sellmann,Daniel Maresch,Claudia Halbig,Stefan Becker,Lars Toleikis,

Hock, and Florian Rüker, Protein Engineering, Design & Selection 30, 685–696, 2017). We found that the F126C/S121C cysteine pair mutation reported in the literature failed to efficiently form interchain disulfide bonds in IgG4(kappa) and could not be used as a positive control.

As described in Example 3, an ELISA method was used to identify engineered interchain disulfide bonds in bispecific antibodies for promoting correct pairing of heavy and light chains.

As shown in Figure 13, the bispecific antibody positive control (K147D/T129R) showed high antibody/antigen binding ELISA absorbance values (vertical axis) under the same antibody concentration conditions (ELISA absorbance values on the horizontal axis). Under the same antibody concentration conditions (ELISA absorbance value on the horizontal axis), many of the bispecific antibodies we constructed had the same or higher antibody/antigen binding absorbance value as the positive control K147D/T129R. This indicates that the heavy and light chains of our bispecific antibodies are able to pair correctly. Therefore, in IgG4(kappa) bispecific antibody production, an engineered disulfide bond is formed in the CH1-CL domain of one Fab arm by replacing the native interchain disulfide bond with a different cysteine pairing , which enables the correct pairing of heavy and light chains, including:

A129C/F209C、A129C/N210C、P130C/F116C、P130C/F118C、P130C/P119C、P130C/N210C、P130C/R211C、S132C/S114C、S132C/F116C、S132C/F118C、S132C/P120C、S132C/R211C、S132C/ E213C、R133C/P119C、R133C/R211C、R133C/E213C、T135C/F116C、T135C/P120C、G166C/T178C、H168C/N158C、F170C/F182C、V173C/Q160C、V173C/S162C、Q175C/S162C、Q175C/T180C、 S181C/T172C, S181C/S176C, S183C/N158C, S183C/S176C, V185C/E165C, V185C/T178C

(Each pair of cysteine mutations is listed as follows: IgG4 heavy chain wild amino acid [EU numbering] to cysteine/kappa light chain wild amino acid [EU numbering] to cysteine).

Example 16. Design of IgG4 (lambda) CH1 and CL domain chain "cysteine mutated" libraries

Using the Fab crystal structure of human IgG2 (lambda) from the Protein Data Bank (PDB code: 5GKS) as a representative structure, we analyzed and designed cysteines in the CH1 and CL domains that may interact to form interchain disulfide bonds Mutation sites, an IgG4 (kappa) mutant library was designed in which the heavy and light chain cysteine pairs that form native interchain disulfide bonds were mutated to serine and valine, respectively, and in the CH1-CL domain New cysteine pairs were introduced at different positions to form engineered interchain disulfide bonds. Table 19 lists the introduction of new cysteine pairs at different amino acid positions in the CH1-CL domain of IgG4 (lambda) that may form engineered interchain disulfide bonds.

Table 19. Cysteine mutation library of IgG4(lambda) CH1-CL domain. Regions where cysteine mutations cluster together are called a group. The CH1 domain and the CL domain may form an interchain disulfide bond. Each group is listed separately.

The sequences used in the present invention are as follows:

All documents mentioned herein are incorporated by reference in this application as if each document were individually incorporated by reference. In addition, it should be understood that after reading the above teaching content of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of the present application.

Claims (14)

1. A method for improving the correct pairing rate of heavy chain and light chain in the preparation process of bispecific antibody, the method comprising the following steps:

   In the CH1-CL domain of one Fab arm, the natural interchain disulfide bond was eliminated by amino acid substitution, while an engineered interchain disulfide bond was formed by amino acid substitution.
2. The method of claim 1, wherein the amino acids that form the engineered interchain disulfide bond in the CH1-CL domain of IgG1 are selected from the following table:

   
3. The method of claim 2, wherein the engineered interchain disulfide bond forming amino acids in the CH1-CL domain of IgG1 are selected from the group consisting of each pair of cysteine mutations listed as follows : Mutation of selected amino acid [EU numbering] to cysteine in heavy chain / Mutation of selected amino acid [EU numbering] to cysteine in kappa chain: F126C/F118C, L128C/P120C, A129C/P120C, P130C /F118C、S132C/F118C、K133C/S121C、S134C/F118C、G166C/T172C、G166C/S174C、G166C/T180C、H168C/T180C、F170C/Q160C、F170C/T164C、F170C/T172C、F170C/S174C、F170C/S176C 、F170C/T180C、P171C/T172C、P171C/S174C、P171C/T180C、V173C/S174C、Q175C/S174C、Q175C/S176C、Q175C/T180C、S176C/S162C、S181C/T172C、S181C/S176C、S181C/T180C、S183C /S114C、S183C/F118C、S183C/Q160C、S183C/S174C、S183C/T178C、S183C/T180C、V185C/T172C、V185C/S174C、V185C/T178C、V185C/T180C、T187C/S114C、T187C/T172C、T187C/S174C , T187C/T178C, K218C/F118C, S219C/F116C, S219C/F118C, S219C/P120C.
4. The method of claim 1, wherein the amino acids that form the engineered interchain disulfide bond in the CH1-CL domain of IgG1 are selected from the following table:

   
5. The method of claim 4, wherein the engineered interchain disulfide bond forming amino acids in the CH1-CL domain of IgG1 are selected from the group consisting of each pair of cysteine mutations listed as follows : Mutation of selected amino acids in heavy chain [EU numbering] to cysteine / Mutation of selected amino acids in lambda chain [Kabat numbering] to cysteine S132C/S121C, K133C/T116C, K133C/P211C, S136C/ S121C, F170C/G158C, P171C/T162C, P171C/P164C, S176C/T162C, L179C/G158C, S181C/P164C, V215C/T116C, E216C/F118C.
6. The method of claim 1, wherein the amino acids that form the engineered interchain disulfide bonds in the CH1-CL domain of IgG4 are selected from the following table:

   

   
7. The method of claim 6, wherein the engineered interchain disulfide bond forming amino acids in the CH1-CL domain of IgG4 are selected from the group consisting of each pair of cysteine mutations listed in the following manner : Mutation of selected amino acid [EU numbering] to cysteine in heavy chain / Mutation of selected amino acid [EU numbering] to cysteine in kappa chain; A129C/F209C, A129C/N210C, P130C/F116C, P130C /F118C、P130C/P119C、P130C/N210C、P130C/R211C、S132C/S114C、S132C/F116C、S132C/F118C、S132C/P120C、S132C/R211C、S132C/E213C、R133C/P119C、R133C/R211C、R133C/E213C 、T135C/F116C、T135C/P120C、G166C/T178C、H168C/N158C、F170C/F182C、V173C/Q160C、V173C/S162C、Q175C/S162C、Q175C/T180C、S181C/T172C、S181C/S176C、S183C/N158C、S183C /S176C, V185C/E165C, V185C/T178C.
8. The method of claim 1, wherein the amino acids that form the engineered interchain disulfide bonds in the CH1-CL domain of IgG4 are selected from the following table:

   
9. The method of any one of claims 1-8, wherein the method further comprises the step of inverting the charge of a pair of amino acids in the CH1-CL domain of the Fab arm by amino acid substitution.
10. The method of claim 9, wherein the wild-type positively charged lysine at position 213 of the heavy chain is substituted with a negatively charged amino acid (eg, K213E, K213D); the wild-type band at position 123 of the light chain Negatively charged glutamic acid is substituted with positively charged amino acid (eg, E123K, E123R).
11. The method of claim 10, wherein formation of an engineered interchain disulfide bond by amino acid substitution and charge reversal of a pair of amino acids in the CH1-CL domain of the Fab arm by amino acid substitution can occur in the same Fab arm, can also occur on a different Fab arm.
12. A method for improving the correct pairing rate of heavy and light chains in the preparation of bispecific antibodies, the method comprising the steps of inverting the charge of a pair of amino acids in the CH1-CL domains of a Fab arm by amino acid substitution, wherein the heavy chain The wild-type positively charged lysine at chain position 213 is replaced by a negatively charged amino acid (eg, K213E, K213D); the wild-type negatively charged glutamic acid at position 123 of the light chain is replaced by a positively charged amino acid ( For example, E123K, E123R).
13. A method for preparing a bispecific antibody, the method comprising adopting the method according to any one of claims 1 to 12 in the preparation process of the bispecific antibody so as to improve the correctness of the heavy chain and the light chain in the bispecific antibody Pairing rate steps.
14. A bispecific antibody, the bispecific antibody is prepared by the method of claim 13, or the method of any one of claims 1-12 is used to improve the weight of the bispecific antibody. Chain and light chain correct pairing rates.

Images