JP7183221B2 - Modified antigen-binding polypeptide constructs and uses thereof - Google Patents

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application supersedes U.S. Provisional Application No. 62/003,663 filed May 28, 2014 and U.S. Provisional Application No. 62/154,055 filed April 28, 2015 All rights reserved, and these applications are hereby incorporated by reference in their entireties for all purposes.

This application is based on PCT/CA2013/050914 filed November 28, 2013; U.S. Provisional Application No. 61/730,906 filed November 28, 2012; No. 61/761,641, U.S. Provisional Application No. 61/818,874 filed May 2, 2013 and U.S. Provisional Application No. 61/869,200 filed Aug. 23, 2013 , the entire disclosure of each is incorporated herein by reference in its entirety for all purposes.

SEQUENCE LISTING This application contains a Sequence Listing which is provided electronically in ASCII format and is incorporated herein by reference in its entirety. A copy of said ASCII was made on May 29, 2015, 97993-945204(000110PC)_SL. txt and is 27,012 bytes in size.

Bispecific antibodies can bind to two different epitopes. These epitopes can be on the same antigen or each epitope can be on a different antigen. This bispecific antibody feature makes it an attractive tool for a variety of therapeutics, with the therapeutic benefit of targeting or mobilizing more than one molecule in the treatment of disease. One approach to forming bispecific antibodies involves the co-expression of two unique antibody heavy chains and two unique antibody light chains. Accurately forming bispecific antibodies in a format resembling the wild-type remains a challenge, as antibody heavy chains tend to associate relatively randomly with antibody light chains. As a result of this random pairing, co-expression of two antibody heavy chains and two antibody light chains will of course result in random heavy chain-light chain pairing. This mismatch remains a major challenge for the generation of bispecific therapeutics and homogeneous pairing is a prerequisite for good manufacturability and biological efficacy.

Several techniques have been described for preparing bispecific antibodies in which a particular antibody light chain or fragment is paired with an antibody heavy chain or fragment. A review of various approaches to address this issue can be found in Klein et al. (2012) mAbs 4:6, 1-11. International Patent Application PCT/EP2011/056388 (International Application Publication No. 2011/131746) (Patent Document 1) describes a "Fab arm" between two monospecific IgG4 or IgG4-like antibodies incubated under reducing conditions. Describes an in vitro method to generate heterodimeric proteins that introduces asymmetric mutations in the CH3 regions of two monospecific starting proteins to direct the directional exchange of "" or "half-molecules" .

Schaefer et al. (Roche Diagnostics GmbH) assemble two heavy chains and two light chains derived from two existing antibodies into a human bivalent bispecific IgG antibody without the use of artificial linkers. A method is described (PNAS (2011) 108(27): 11187-11192 (Non-Patent Document 2)). This method involves exchanging the heavy and light chain domains within the antigen-binding fragment (Fab) half of the bispecific antibody.

Strop et al. (Rinat-Pfizer Inc.) describe a method for generating stable bispecific antibodies by separately expressing and purifying two antibodies of interest and then mixing them together under specific redox conditions. (J. Mol. Biol. (2012) 420: 204-19 (Non-Patent Document 3)).

Zhu et al. (Genentech) engineered mutations at the VL/VH interface of a diabody construct consisting of a variable domain antibody fragment, completely lacking the constant domains, resulting in a heterodimeric diabodies. (Protein Science (1997) 6:781-788). Similarly, Igawa et al. (Chugai) also engineered mutations in the VL/VH interface of single-chain diabodies to promote selective expression of the diabodies and inhibit conformational isomerization (Protein Engineering, Design & Selection (2010) 23:667-677 (Non-Patent Document 5)).

US Patent Publication No. 2009/0182127 (Novo Nordisk, Inc.) modifies amino acid residues at the Fc and CH1:CL interfaces of a light-heavy chain pair to The generation of bispecific antibodies is described by reducing the ability of the chains to interact with the other pair of heavy chains.

U.S. Patent Publication No. 2014/0370020 (Chugai) describes modulating the association of the CH1 and CL regions of an antibody by replacing amino acids present at the interface of these regions with charged amino acids. do.

International Patent Application PCT/EP2011/056388 (International Application Publication No. 2011/131746) U.S. Patent Publication No. 2009/0182127 U.S. Patent Publication No. 2014/0370020

Klein et al., (2012) mAbs 4:6, 1-11 PNAS (2011) 108(27): 11187-11192 J. Mol. Biol. (2012) 420:204-19 Protein Science (1997) 6:781-788 Protein Engineering, Design & Selection (2010) 23:667-677

comprising at least a first heterodimer and a second heterodimer, the first heterodimer comprising a first immunoglobulin heavy chain polypeptide sequence (H1) and a first immunoglobulin light chain polypeptide sequence (H1); comprising a peptide sequence (L1), the second heterodimer comprising a second immunoglobulin heavy chain polypeptide sequence (H2) and a second immunoglobulin light chain polypeptide sequence (L2), the first At least one of the H1 or L1 sequences of the heterodimer is different from the corresponding H2 or L2 of the second heterodimer such that H1 and H2 each comprise at least a heavy chain variable domain ( _VH domain) and heavy L1 and L2 each comprise _{at least a light chain variable domain (VL domain) and a light chain constant domain (C L domain )} , and at least of H1, H2, L1 and L2 1 comprises at least one amino acid modification in at least one constant domain and/or at least one variable domain, wherein H1 preferentially pairs with L1 over L2, and H2 preferentially with L2 over L1; Described herein are isolated antigen-binding polypeptide constructs that specifically pair.

In some embodiments, the construct further comprises a heterodimeric Fc, wherein the Fc comprises at least two _CH3 sequences, the Fc with or without one or more linkers, the first The dimerized _CH3 sequence that binds the heterodimer and the second heterodimer has a melting temperature (Tm) of about 68° C. or greater as measured by differential scanning calorimetry (DSC). , the construct is bispecific.

In some embodiments, the at least one amino acid modification is selected from at least one amino acid modification shown in the Tables or Examples.

In some embodiments, H1, H2, L1 and L2 are co-expressed in a cell or mammalian cell, or H1, H2, L1 and L2 are co-expressed in a cell-free expression system, or H1, H2, L1 and When L2 is co-produced, or when H1, H2, L1 and L2 are co-produced via redox production methods, H1 preferentially pairs with L1 over L2 and H2 over L1. also preferentially pairs with L2.

In some embodiments, at least one of H1, H2, L1 and L2 is such that H1 preferentially pairs with L1 over L2 and/or H2 preferentially pairs with L2 over L1 at least one amino acid modification in the V _H and/or V _L domains and at least one amino acid modification in the C _H1 and/or C _L domains, such as

In some embodiments, when H1 comprises at least one amino acid modification in the C _H1 domain, at least one of L1 and L2 comprises at least one amino acid modification in the C _L domain and/or H1 contains at least one amino acid modification in the _VH domain, then at least one of L1 and L2 contains at least one amino acid modification in the _VL domain.

In some embodiments, H1, L1, H2 and/or L2 comprise at least 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10 amino acid modifications. In some embodiments, at least one of H1, H2, L1 and L2 is associated with at least 2, 3, 4, 5, 6, 7, 8, 9 at least one constant domain and/or at least one variable domain. or contains 10 amino acid modifications.

In some embodiments, when both L1 and L2 are co-expressed with at least one of H1 and H2, the relative pairing of at least one of the H1-L1 and H2-L2 heterodimer pairs is 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65 for the corresponding H1-L2 or H2-L1 heterodimer pair, respectively , 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90 , 91, 92, 93, 94, 95, 96, 97, 98 or 99% of the relative pairings of the modified H1-L1 or H2-L2 heterodimer pairs have at least one amino acid Greater than the respective relative pairings observed in the corresponding H1-L1 or H2-L2 heterodimer pairs with no modification.

In some embodiments, the thermal stability, as measured by the melting temperature (Tm) as measured by DSF, of at least one of the first and second heterodimers is equal to that of the counterpart without at least one amino acid modification. within about 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10° C. of the Tm of the heterodimer. In some embodiments, the thermal stability, as measured by the melting temperature (Tm) as measured by DSF, of each heterodimer comprising at least one amino acid modification is comparable to that without the at least one amino acid modification. Within about 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10°C of the Tm of the heterodimer. In some embodiments, the thermal stability, as measured by the melting temperature (Tm) as measured by DSF, of each heterodimer comprising at least one amino acid modification is compared to that of the counterpart without at least one amino acid modification. within about 0, 1, 2, or 3°C of the Tm of the heterodimer.

In some embodiments, the affinity of each heterodimer for the antigen it binds is that of the corresponding unmodified heterodimer for the same antigen as measured by surface plasmon resonance (SPR) or FACS. Within about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 times.

In some embodiments, at least one of H1 and L1 has at least one domain comprising at least one amino acid modification that provides greater amino acid steric complementarity when H1 pairs with L1 over L2. including. In some embodiments, at least one of H2 and L2 has at least one domain comprising at least one amino acid modification that provides greater amino acid steric complementarity when H2 pairs with L2 than with L1. including. In some embodiments, at least one of H1 and L1 comprises at least one amino acid modification that results in greater electrostatic complementarity between charged amino acids when H1 pairs with L1 over L2. Contains one domain. In some embodiments, at least one of H2 and L2 comprises at least one amino acid modification that results in greater electrostatic complementarity between charged amino acids when H2 pairs with L2 over L1. Contains one domain.

In some embodiments, the at least one amino acid modification is a set of mutations set forth in at least one of the Tables or Examples.

In some embodiments, the construct further comprises an Fc comprising at least two _CH3 sequences, wherein the Fc comprises the first heterodimer and the second, with or without one or more linkers. binds to the heterodimer of

In some embodiments, the Fc is human Fc, human IgG1 Fc, human IgA Fc, human IgG Fc, human IgD Fc, human IgE Fc, human IgM Fc, human IgG2 Fc, human IgG3 Fc, or human IgG4 Fc. In some embodiments, the Fc is a heterodimeric Fc. In some embodiments, the Fc comprises one or more modifications in at least one of the _CH3 sequences. In some embodiments, the dimerization C _H3 sequence is about 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 77.5, 78, 79, 80 as measured by DSC. , 81, 82, 83, 84 or 85° C. or higher. In some embodiments, the Fc, when produced, is about , is a heterodimer formed with a purity greater than 93, 94, 95, 96, 97, 98 or 99%, or the Fc is about 75 when expressed or expressed via a single cell , 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99% It is a heterodimer formed with greater purity. In some embodiments, the Fc has one or more modifications in at least one of the _CH3 sequences that promote formation of a heterodimeric Fc with stability comparable to a wild-type homodimeric Fc. include. In some embodiments, the Fc further comprises at least one _CH2 sequence. In some embodiments, the Fc _CH2 sequence comprises one or more modifications. In some embodiments, the Fc comprises one or more modifications that facilitate selective binding of Fc gamma receptors.

In some embodiments, Fc is
i) a heterodimeric IgGl Fc having the modification L351Y_F405A_Y407V in the first Fc polypeptide and the modification T366L_K392M_T394W in the second Fc polypeptide;
ii) a heterodimeric IgG1 Fc having the modification L351Y_F405A_Y407V in the first Fc polypeptide and the modification T366L_K392L_T394W in the second Fc polypeptide;
iii) a heterodimeric IgG1 Fc having the modification T350V_L351Y_F405A_Y407V in the first Fc polypeptide and the modification T350V_T366L_K392L_T394W in the second Fc polypeptide;
iv) a heterodimeric IgG1 Fc having the modification T350V_L351Y_F405A_Y407V in the first Fc polypeptide and the modification T350V_T366L_K392M_T394W in the second Fc polypeptide; or v) having the modification T350V_L351Y_S400E_V in the first Fc polypeptide; A heterodimeric IgG1 Fc having a modification T350V_T366L_N390R_K392M_T394W in the second Fc polypeptide.

In some embodiments, Fc is attached to the heterodimer by one or more linkers, or Fc is attached to H1 and H2 by one or more linkers. In some embodiments, one or more linkers are one or more polypeptide linkers. In some embodiments, one or more linkers comprise one or more antibody hinge regions. In some embodiments, one or more linkers comprise one or more IgG1 hinge regions. In some embodiments, one or more linkers contain one or more modifications. In some embodiments, one or more modifications to one or more linkers promote selective binding of Fc gamma receptors.

In some embodiments, at least one amino acid modification is at least one amino acid mutation, or at least one amino acid modification is at least one amino acid substitution.

In some embodiments, the sequences of each of H1, H2, L1 and L2 are derived from human sequences.

In some embodiments, the construct is multispecific or bispecific. In some embodiments, constructs are multivalent or bivalent.

In some embodiments, the heterodimers described herein preferentially pair to form bispecific antibodies. For example, in some embodiments, heavy chain polypeptide sequences H1 and H2 comprise a full-length heavy chain sequence comprising the heavy chain constant domain (C _H1 domain), C _H2 domain and C _H3 domain. In some embodiments, the ratio of correctly paired heavy and light chains (eg, H1-L1:H2-L2) in the bispecific antibody is 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, greater than 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99%.

Also described herein is an isolated polynucleotide or set of isolated polynucleotides comprising at least one sequence encoding a construct or a heavy or light chain described herein. In some embodiments, the polynucleotide or set of polynucleotides is cDNA.

Also described herein are vectors or sets of vectors comprising one or more of the polynucleotides or sets of polynucleotides described herein. In some embodiments, the vector or set of vectors is selected from the group consisting of plasmids, multicistronic vectors, viral vectors, non-episomal mammalian vectors, expression vectors and recombinant expression vectors.

Also described herein are isolated cells containing a polynucleotide or set of polynucleotides described herein, or a vector or set of vectors described herein. In some embodiments, the cells are hybridomas, Chinese Hamster Ovary (CHO) cells or HEK293 cells.

Also described herein are pharmaceutical compositions comprising the constructs described herein and a pharmaceutically acceptable carrier. In some embodiments, the composition comprises one or Further comprising a plurality of substances.

Also described herein is the use of a construct described herein or a pharmaceutical composition described herein in the treatment of a disease or disorder, or cancer or vascular disease in a subject, or in the manufacture of a medicament. be done.

Also described herein are methods of treating a subject having a disease or disorder, or cancer or vascular disease, comprising administering a construct described herein or a composition described herein to the subject. be.

A method of obtaining a construct described herein from a host cell culture comprising: (a) obtaining a host cell culture comprising at least one host cell comprising one or more nucleic acid sequences encoding the construct. and (b) recovering the construct from the host cell culture.

(a) obtaining H1, L1, H2 and L2; and (b) preferentially pairing H1 with L1 over L2 and H2 with L2 over L1. , (c) obtaining the construct are also described herein.

obtaining a polynucleotide or set of polynucleotides encoding at least one construct and determining an optimal ratio of each of the polynucleotides or set of polynucleotides for introduction into at least one host cell, comprising: Optimal ratios of expression of H1, L1, H2 and L2 compared to the mismatched H1-L2 and H2-L1 heterodimer pairs formed upon expression of H1, L1, H2 and L2 determining by evaluating the amount of H1-L1 and H2-L2 heterodimer pairs formed during the process and selecting a preferred optimal ratio, wherein the preferred optimal ratio of polynucleotides or polynucleotides resulting in expression of the construct; transfecting the at least one host cell with an optimal ratio of the polynucleotide or set of polynucleotides; Also described herein are methods of preparing the constructs described herein comprising culturing the cells to express the constructs.

In some embodiments, selecting the optimal ratio is assessed by transfection in a transient transfection system. In some embodiments, transfection of at least one host cell with a preferred optimal ratio of polynucleotides or sets of polynucleotides results in optimal expression of the construct. In some embodiments, the construct comprises an Fc comprising at least two _CH3 sequences, wherein the Fc is a first heterodimer and a second Binds heterodimers. In some embodiments, the Fc is a heterodimer, optionally comprising one or more amino acid modifications.

a first heterodimer comprising a first immunoglobulin heavy chain polypeptide sequence (H1) and a first immunoglobulin light chain polypeptide sequence (L1) and a second immunoglobulin heavy chain polypeptide sequence (H2) ) and a second immunoglobulin light chain polypeptide sequence (L2), wherein H1 and H2 each comprise at least the heavy chain variable domain ( _VH domain) and a heavy chain constant domain (C _H1 domain), L1 and L2 each comprising at least a light chain variable domain ( _VL domain) and a light chain constant domain (C _L domain), and complementary mutation data A data set, wherein the set comprises data representing the mutations or subsets of these mutations presented in the Tables or Examples, and H1 preferentially pairs with L1 over L2 and/or H2 , computer-executable code for determining a likelihood of preferentially pairing with L2 over L1 is also described herein.

a first heterodimer comprising a first immunoglobulin heavy chain polypeptide sequence (H1) and a first immunoglobulin light chain polypeptide sequence (L1) and a second immunoglobulin heavy chain polypeptide sequence (H2) ) and a second immunoglobulin light chain polypeptide sequence (L2), wherein H1 and H2 each comprise at least the heavy chain variable domain ( _VH domain) and a heavy chain constant domain (C _H1 domain), wherein L1 and L2 each comprise at least a light chain variable domain ( _VL domain) and a light chain constant domain (C _L domain), wherein complementary obtaining a data set, wherein the data set of targeted mutations comprises data representing the mutations presented in the Tables or Examples or a subset of these mutations; and H1 preferentially with L1 over L2. Also disclosed herein is a computer-implemented method of determining preferential pairing comprising: determining by a computer processor the likelihood that H2 will preferentially pair with L2 over L1. listed in In some embodiments, the method further comprises producing the constructs described herein.

A method of producing a bispecific antigen binding polypeptide construct, wherein the bispecific construct comprises a first immunoglobulin heavy chain polypeptide sequence (H1) of a first monospecific antigen binding polypeptide and a first immunoglobulin light chain polypeptide sequence (L1) and a second immunoglobulin heavy chain polypeptide sequence (H2) of a second monospecific antigen binding polypeptide and a second immunoglobulin light chain polypeptide sequence (L2), wherein H1 and H2 each comprise at least a heavy chain variable domain ( _VH domain) and a heavy chain constant domain (C _H1 domain), and L1 and L2 each comprise at least a light chain variable domain ( _VL domain) and a light chain constant domain ( _CL domain), and one or more of the datasets described herein. introducing complementary mutations into the first heterodimer and/or the second heterodimer; and introducing the first heterodimer and the second heterodimer into at least one host cell. and co-expressing to produce an expression product comprising the bispecific construct is also described herein.

In some embodiments, the method further comprises determining the amount of bispecific construct in the expression product relative to other polypeptide products to select a favorable subset of complementary mutations. In some embodiments, bispecific constructs are more than 70% (e.g., 75%, 80%, 85%, 90%, 91%, 92%, 93%) compared to other polypeptide products. , 94%, 95%, 96%, 97%, 98% or greater than 99%) purity. In some embodiments, the dataset is a dataset described herein. In some embodiments, the methods add additional amino acid modifications to at least one of H1, H2, L1 or L2 to increase the purity of the bispecific construct compared to other polypeptide products. further comprising the step of allowing In some embodiments, the construct comprises an Fc comprising at least two _CH3 sequences, wherein the Fc is a first heterodimer and a second Binds heterodimers. In some embodiments, the Fc is a heterodimer, optionally comprising one or more amino acid modifications. In some embodiments, the antigen binding polypeptide is an antibody, Fab or scFv.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at L124, K145, D146, Q179 and S186, and L1 and/or L2 are Q124, S131 , V133, Q160, S176, T178 and T180. For example, in some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from L124R, L124E, K145M, K145T, D146N, Q179E, Q179K, S186R and S186K. , L1 and/or L2 comprise at least one amino acid modification or set of amino acid modifications selected from Q124E, S131R, S131K, V133G, Q160E, S176R, S176D, T178D, T178E and T180E. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L124E, K145M, K145T and Q179E, or combinations thereof, and L1 comprises the group consisting of S131R, S131K, V133G and S176R, or combinations thereof. wherein H2 comprises an amino acid modification selected from the group consisting of L124R, D146N, Q179K, S186R and S186K or combinations thereof, L2 comprises Q124E, V133G, Q160E, S176D, T178D, Amino acid modifications selected from the group consisting of T178E and T180E or combinations thereof. In some embodiments, H1 comprises amino acid modifications L124E, K145T and Q179E, L1 comprises amino acid modifications S131K, V133G and S176R, H2 comprises amino acid modifications L124R and S186R, and L2 comprises amino acid modification V133G. , S176D and T178D.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at L124, L143, K145, D146, Q179 and S186, and L1 and/or L2 comprises Q124. , V133, Q160, S176, T178 and T180. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from L124E, L124R, L143E, L143D, K145T, K145M, D146N, Q179K, S186R and S186K. , L1 and/or L2 comprise at least one amino acid modification or set of amino acid modifications selected from Q124K, Q124E, V133G, Q160K, S176R, S176D, T178E, T178K, T178R, T178D and T180E. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L124E, L143E, L143D, K145T and K145M or combinations thereof, wherein L1 is Q124K, V133G, Q160K, S176R, T178K and T178R or H2 comprises an amino acid modification selected from the group consisting of combinations of these, wherein H2 comprises an amino acid modification selected from the group consisting of L124R, D146N, Q179K, S186R and S186K or combinations thereof, L2 comprises Q124E, V133G, including amino acid modifications selected from the group consisting of S176D, T178E, T178D and T180E or combinations thereof. In some embodiments, H1 comprises amino acid modifications L124E, L143E and K145T, L1 comprises amino acid modifications Q124K, V133G and S176R, H2 comprises amino acid modifications L124R and Q179K, L2 comprises amino acid modification V133G , S176D and T178E. In some embodiments, H1 comprises amino acid modifications L124E, L143E and K145T, L1 comprises amino acid modifications Q124K, V133G and S176R, H2 comprises amino acid modifications L124R and S186R, L2 comprises amino acid modification V133G , S176D and T178D.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications in Q39, L45, L124, L143, F122 and H172, and L1 and/or L2 comprise Q38. , P44, Q124, S131, V133, N137, S174, S176 and T178. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from Q39E, Q39R, L45P, F122C, L124E, L124R, L143F, H172T and H172R; and/or L2 comprises at least one amino acid modification or set of amino acid modifications selected from Q38R, Q38E, P44F, Q124C, S131T, S131E, V133G, N137K, S174R, S176R, S176K, S176D, T178Y and T178D . In some embodiments, H1 comprises an amino acid modification selected from the group consisting of Q39E, L45P, F122C, L124E, L143F, H172T and H172R, or combinations thereof, wherein L1 is Q38R, P44F, Q124C, S131T, V133G, N137K, S174R, S176R, S176K and T178Y or combinations thereof, and H2 comprises an amino acid modification selected from the group consisting of Q39R, L124R and H172R or combinations thereof. , L2 comprise amino acid modifications selected from the group consisting of Q38E, S131E, V133G, S176D and T178D or combinations thereof. In some embodiments, H1 comprises amino acid modifications Q39E and L124E, L1 comprises amino acid modifications Q38R, V133G and S176R, H2 comprises amino acid modifications Q39R and L124R, L2 comprises amino acid modifications Q38E, V133G and S176D. In some embodiments, H1 comprises amino acid modifications L45P and L124E, L1 comprises amino acid modifications P44F, V133G and S176R, H2 comprises amino acid modifications L124R, and L2 comprises amino acid modifications V133G, S176D and T178D. including. In some embodiments, H1 comprises amino acid modifications L124E and L143F, L1 comprises amino acid modifications V133G and S176R, H2 comprises amino acid modifications L124R, and L2 comprises amino acid modifications V133G, S176D and T178D. . In some embodiments, H1 comprises amino acid modifications F122C and L124E, L1 comprises amino acid modifications Q124C, V133G and S176R, H2 comprises amino acid modifications L124R, and L2 comprises amino acid modifications V133G and S176D. . In some embodiments, H1 comprises amino acid modifications L124E and H172T, L1 comprises amino acid modifications V133G, N137K, S174R and S176R, H2 comprises amino acid modifications L124R and H172R, and L2 comprises amino acid modification V133G. , S176D and T178D.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at L124, A125, H172 and K228, and L1 and/or L2 are S121, V133, N137. , S174, S176 and T178. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from L124E, L124R, A125S, A125R, H172R, H172T and K228D; (ii) L1 and /or L2 comprises at least one amino acid modification or set of amino acid modifications selected from S121K, V133G, N137K, S174R, S176K, S176R, S176D and T178D. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L124E, A125S, H172R and K228D, or combinations thereof, and L1 is selected from the group consisting of S121K, V133G and S176R, or combinations thereof. wherein H2 comprises amino acid modifications selected from the group consisting of L124R, A125R and H172T or combinations thereof, and L2 comprises the group consisting of V133G, N137K, S174R, S176D and T178D or combinations thereof Amino acid modifications selected from In some embodiments, H1 comprises amino acid modifications L124E and K228D, L1 comprises amino acid modifications S121K, V133G and S176R, H2 comprises amino acid modifications L124R and A125R, and L2 comprises amino acid modifications V133G and S176D. including. In some embodiments, H1 comprises amino acid modifications L124E and H172R, L1 comprises amino acid modifications V133G and S176R, H2 comprises amino acid modifications L124R and H172T, and L2 comprises amino acid modifications V133G, S174R and S176D. including.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications in L124, A139 and V190, and L1 and/or L2 in F116, V133, L135 and S176. It contains at least one amino acid modification or set of amino acid modifications. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from L124E, L124R, A139W, A139G and V190A, and L1 and/or L2 are F116A, It comprises at least one amino acid modification or set of amino acid modifications selected from V133G, L135V, L135W, S176R and S176D. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L124E and A139W or combinations thereof, and L1 is selected from the group consisting of F116A, V133G, L135V and S176R or combinations thereof wherein H2 comprises an amino acid modification selected from the group consisting of L124R, A139G and V190A or combinations thereof, and L2 comprises an amino acid modification selected from the group consisting of V133G, L135W and S176D or combinations thereof including. In some embodiments, H1 comprises amino acid modifications L124E and A139W, L1 comprises amino acid modifications F116A, V133G, L135V and S176R, H2 comprises amino acid modifications L124R, A139G and V190A, L2 comprises amino acid Includes modifications V133G, L135W and S176D.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications in Q39, L45, K145, H172, Q179 and S186, and L1 and/or L2 comprise Q38. , P44, Q124, S131, Q160, T180 and C214. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from Q39E, Q39R, L45P, K145T, H172R, Q179E and S186R, and L1 and/or L2 contains at least one amino acid modification or set of amino acid modifications selected from Q38R, Q38E, P44F, Q124E, S131K, Q160E, T180E and C214S. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of Q39E, L45P, K145T, H172R and Q179E, or combinations thereof, and L1 comprises the group consisting of Q38R, P44F and S131K, or combinations thereof. wherein H2 comprises an amino acid modification selected from the group consisting of Q39R, H172R and S186R or combinations thereof, and L2 comprises Q38E, Q124E, Q160E, T180E and C214S or combinations thereof Amino acid modifications selected from the group consisting of In some embodiments, H1 comprises amino acid modifications Q39E, K145T and Q179E, L1 comprises amino acid modifications Q38R and S131K, H2 comprises amino acid modifications Q39R and S186R, L2 comprises amino acid modifications Q38E, Q124E. , Q160E and T180E. In some embodiments, H1 comprises amino acid modifications L45P, K145T, H172R and Q179E, L1 comprises amino acid modifications P44F and S131K, H2 comprises amino acid modifications H172R and S186R, and L2 comprises amino acid modifications Q124E. , Q160E and T180E.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at A139, L143, K145, Q179 and V190, and L1 and/or L2 are F116, Q124. , L135, Q160, T178 and T180. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from A139W, A139G, L143E, K145T, Q179E, Q179K and V190A, and L1 and/or L2 contains at least one amino acid modification or set of amino acid modifications selected from F116A, Q124R, Q124E, L135V, L135W, Q160E, T178R and T180E. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of A139W, L143E, K145T and Q179E, or combinations thereof, and L1 comprises the group consisting of F116A, Q124R, L135V and T178R, or combinations thereof. wherein H2 comprises an amino acid modification selected from the group consisting of A139G, Q179K and V190A or combinations thereof, and L2 comprises the group consisting of Q124E, L135W, Q160E and T180E or combinations thereof Amino acid modifications selected from In some embodiments, H1 comprises amino acid modifications A139W, L143E, K145T and Q179E, L1 comprises amino acid modifications F116A, Q124R, L135V and T178R, H2 comprises amino acid modification Q179K, L2 comprises amino acid Includes modifications Q124E, L135W, Q160E and T180E.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications in Q39, L143, K145, D146, H172 and Q179, and L1 and/or L2 comprise Q38. , Q124, Q160, T178 and T180. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from Q39E, Q39R, L143E, K145T, D146G, H172R, Q179E and Q179K, and L1 and/or Or L2 comprises at least one amino acid modification or set of amino acid modifications selected from Q38R, Q38E, Q124R, Q124E, Q160K, Q160E, T178R and T180E. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of Q39E, L143E, K145T, H172R and Q179E, or combinations thereof, and L1 is from Q38R, Q124R, Q160K and T178R, or combinations thereof. wherein H2 comprises an amino acid modification selected from the group consisting of Q39R, H172R and Q179K or combinations thereof, L2 comprises Q38E, Q124E, D146G, Q160E and T180E or these It includes amino acid modifications selected from the group consisting of combinations. In some embodiments, H1 comprises amino acid modifications Q39E, L143E, K145T and Q179E, L1 comprises amino acid modifications Q38R, Q124R, Q160K and T178R, H2 comprises amino acid modifications Q39R, H172R and Q179K, L2 contains amino acid modifications Q38E, Q124E, Q160E and T180E.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications in L45, L143, K145, D146, H172 and Q179, and L1 and/or L2 comprise Q38. , P44, Q124, N137, Q160, S174, T178, T180 and C214. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from L45P, L143E, K145T, D146G, H172R, H172T, Q179E and Q179K, (ii) L1 and/or L2 comprise at least one amino acid modification or set of amino acid modifications selected from Q38E, P44F, Q124R, Q124E, N137K, Q160K, Q160E, S174R, T178R, T180E and C214S. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L45P, L143E, K145T, H172R and Q179E, or combinations thereof, and L1 is from P44F, Q124R, Q160K and T178R, or combinations thereof. wherein H2 comprises an amino acid modification selected from the group consisting of D146G, H172R, H172T and Q179K or combinations thereof, L2 comprises Q38E, Q124E, N137K, Q160E, S174R, Amino acid modifications selected from the group consisting of T180E and C214S or combinations thereof. In some embodiments, H1 comprises amino acid modifications L45P, L143E and K145T, L1 comprises amino acid modifications P44F, Q124R, Q160K and T178R, H2 comprises amino acid modifications D146G and Q179K, L2 comprises amino acid modifications Includes modifications Q38E, Q124E, Q160E and T180E. In some embodiments, H1 comprises amino acid modifications L143E, K145T and H172R, L1 comprises amino acid modifications Q124R, Q160K and T178R, H2 comprises amino acid modifications H172T and Q179K, L2 comprises amino acid modifications Q124E , Q160E, N137K, S174R and T180E.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at L124, L143, K145 and Q179, and L1 and/or L2 are at Q124, S131, V133. , S176, T178 and T180. In some embodiments, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications selected from L124W, L124A, L143E, L143F, K145T, Q179E and Q179K, and L1 and/or L2 contains at least one amino acid modification or set of amino acid modifications selected from Q124R, Q124K, Q124E, S131K, V133A, V133W, S176T, T178R, T178L, T178E and T180E. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L124W, L143E, K145T and Q179E or combinations thereof, and L1 is Q124R, Q124K, S131K, V133A, S176T, T178R and T178L or comprising amino acid modifications selected from the group consisting of combinations thereof, H2 comprising amino acid modifications selected from the group consisting of L124A, L143F and Q179K or combinations thereof, L2 comprising Q124E, V133W, S176T, T178L, Amino acid modifications selected from the group consisting of T178E and T180E or combinations thereof. In some embodiments, H1 comprises amino acid modifications L124W, L143E, K145T and Q179E, L1 comprises amino acid modifications Q124R, V133A, S176T and T178R, H2 comprises amino acid modifications L124A, L143F and Q179K, L2 contains amino acid modifications Q124E, V133W, S176T, T178L and T180E.

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at A139, L143, K145, Q179 and S186, and L1 and/or L2 are F116, Q124. , V133, Q160, T178 and T180. In some embodiments, H1 and/or H2 is at least one amino acid modification or amino acid modification selected from A139C, L143E, L143D, L143R, L143K, K145T, Q179E, Q179D, Q179R, Q179K, S186K and S186R wherein L1 and/or L2 are at least one amino acid modification or set of amino acid modifications selected from F116C, Q124R, Q124K, Q124E, V133E, V133D, Q160K, Q160E, T178R, T178K, T178E and T180E including. In some embodiments, H1 comprises an amino acid modification selected from the group consisting of A139C, L143E, L143D, K145T, Q179E and Q179D or combinations thereof, and L1 comprises F116C, Q124R, Q124K, Q160K, T178R and T178K or a combination thereof, H2 comprises an amino acid modification selected from the group consisting of L143R, L143K, Q179R, Q179K, S186K and S186R, or a combination thereof; L2 comprises: containing amino acid modifications selected from the group consisting of Q124E, V133E, V133D, Q160E, T178E and T180E or combinations thereof. In some embodiments, H1 comprises amino acid modifications A139C, L143E, K145T and Q179E, L1 comprises amino acid modifications F116C, Q124R and T178R, H2 comprises amino acid modifications Q179K, and L2 comprises amino acid modifications Q124E. , Q160E and T180E. In some embodiments, H1 comprises amino acid modifications L143E, K145T and Q179E, L1 comprises amino acid modifications Q124R and T178R, H2 comprises amino acid modifications S186K, and L2 comprises amino acid modifications Q124E, Q160E and T178E. including. In some embodiments, H1 comprises amino acid modifications L143E, K145T and Q179E, L1 comprises amino acid modifications Q124R and T178R, H2 comprises amino acid modifications L143R, and L2 comprises amino acid modifications Q124E and V133E. .

In some embodiments of the constructs, H1 and/or H2 comprise at least one amino acid modification or set of amino acid modifications at L124, L143, K145, D146, Q179, S186 and S188, and L1 and/or L2 are , Q124, S131, V133, Q160, S176, T178 and T180. In some embodiments, H1 and/or H2 have at least one amino acid modification selected from L124A, L143A, L143R, L143E, L143K, K145T, D146G, Q179R, Q179E, Q179K, S186R, S186K and S188L or A set of amino acid modifications, wherein L1 and/or L2 are selected from Q124R, Q124E, S131E, S131T, V133Y, V133W, V133E, V133D, Q160E, Q160K, Q160M, S176L, T178R, T178E, T178F, T178Y and T180E contains at least one amino acid modification or set of amino acid modifications that In some embodiments, H1 comprises an amino acid modification selected from the group consisting of L143E, K145T, Q179E and S188L or combinations thereof, and L1 is selected from the group consisting of Q124R, Q160K and T178R or combinations thereof H2 comprises an amino acid modification selected from the group consisting of L124A, L143A, L143R, L143K, D146G, Q179R, Q179K, S186R and S186K or combinations thereof, L2 comprises Q124E, S131E, Amino acid modifications selected from the group consisting of S131T, V133Y, V133W, V133E, V133D, Q160E, Q160M, S176L, T178E, T178F, T178Y and T180E, or combinations thereof. In some embodiments, H1 comprises amino acid modifications L143E, K145T, Q179E and S188L, L1 comprises amino acid modifications Q124R and T178R, H2 comprises amino acid modifications S186K, and L2 comprises amino acid modifications Q124E, S176L. and T180E. In some embodiments, H1 comprises amino acid modifications L143E, K145T, Q179E and S188L, L1 comprises amino acid modifications Q124R and T178R, H2 comprises amino acid modifications S186K, and L2 comprises amino acid modifications Q124E, S131T. , T178Y and T180E. In some embodiments, H1 comprises amino acid modifications L143E and K145T, L1 comprises amino acid modifications Q124R, Q160K and T178R, H2 comprises amino acid modification S186K, and L2 comprises amino acid modification S131E. In some embodiments, H1 comprises the amino acid modifications L143E and K145T, L1 comprises the amino acid modifications Q124R, H2 comprises the amino acid modifications L143R, and L2 comprises the amino acid modifications Q124E and V133E.

In some embodiments of the constructs, H1 comprises at least one amino acid modification or set of amino acid modifications at F122 and C233 and L1 comprises at least one amino acid modification or set of amino acid modifications at Q124 and C214. . In some embodiments, H1 comprises at least one amino acid modification or set of amino acid modifications selected from F122C and C233S and L1 comprises at least one amino acid modification or amino acid modification selected from Q124C and C214S contains a set of In some embodiments, H1 comprises an amino acid modification selected from the group consisting of F122C and C233S, or combinations thereof, and L1 comprises an amino acid modification selected from the group consisting of Q124C and C214S, or combinations thereof. , H2 contains the wild-type or unmodified amino acid sequence and L2 contains the wild-type or unmodified amino acid sequence. In some embodiments, H1 comprises amino acid modifications F122C and C233S, L1 comprises amino acid modifications Q124C and C214S, H2 comprises a wild-type or unmodified amino acid sequence, and L2 is wild-type or unmodified Contains amino acid sequences.

In some embodiments, the construct is a SMCA design 9561-9095_1, 9561-9095_2, 9121-9373_1, 9121-9373_2, 9116-9349_1, 9116-9349_2, 9134-9521_1, 9134-9521_2, ９２８６－９４０２＿１、９２８６－９４０２＿２、９６６７－９８３０＿１、９６６７－９８３０＿２、９６９６－９８４８＿１、９６９６－９８４８＿２、９０６０－９７５６＿１、９０６０－９７５６＿２、９６８２－９７４０＿１、９６８２－９７４０＿２、９０４９－９７５９＿１、９０４９－９７５９＿２、９８２０－ containing amino acid modifications selected from 9823_1 and 9820-9823_2. In some embodiments, the construct comprises an amino acid Including modification.

In some embodiments, H1 and/or H2 do not contain an amino acid modification at position Q179. In some embodiments, H1 does not contain the amino acid modification Q179E and/or H2 does not contain the amino acid modification Q179K. In some embodiments, L1 does not contain an amino acid modification at position S131. In one embodiment, L1 does not contain the amino acid modification S131K. In some embodiments, L2 does not contain an amino acid modification at position T180. In one embodiment, L2 does not contain the amino acid modification T180E. In some embodiments, the construct does not comprise a combination of amino acid modifications where H1 comprises Q179E, L1 comprises S131K, H2 comprises Q179K and L2 comprises T180E.

In some embodiments, H1 does not contain amino acid modifications at positions Q39 and/or Q179. In some embodiments, H1 does not include amino acid modifications Q39E and/or Q179E. In some embodiments, L1 does not contain an amino acid modification at position Q160. In one embodiment, L1 does not contain the amino acid modification Q160K. In some embodiments, H2 does not contain an amino acid modification at position Q179. In one embodiment, H2 does not contain the amino acid modification Q179K. In some embodiments, L2 does not contain amino acid modifications at positions Q38, Q160 and/or T180. In one embodiment, L2 does not include amino acid modifications Q38E, Q160E and/or T180E. In some embodiments, the construct comprises a combination of amino acid modifications, wherein H1 comprises Q39E and/or Q179E, L1 comprises Q160K, H2 comprises Q179K, and L2 comprises Q38E, Q160E and/or T180E. do not have. For example, in some embodiments, the construct comprises (i) H1 comprises Q179E, L1 comprises Q160K, H2 comprises Q179K, L2 comprises Q160E and T180E, (ii) H1 comprises Q39E and Q179E. L1 comprises Q160K, H2 comprises Q179K, L2 comprises Q38E, Q160E and T180E, or (iii) H1 comprises Q39E, L1 comprises Q160K, H2 comprises Q179K, L2 comprises Q38E, It does not contain combinations of amino acid modifications, including Q160E and T180E.

In some embodiments, H1 does not contain an amino acid modification at position Q179. In some embodiments, H1 does not contain amino acid modifications Q179K or Q179E. In some embodiments, L1 does not contain amino acid modifications at positions Q160 and/or T180. In one embodiment, L1 does not include amino acid modifications Q160E, Q160K and/or T180E. In some embodiments, H2 does not contain an amino acid modification at position Q179. In one embodiment, H2 does not contain amino acid modifications Q179K or Q179E. In some embodiments, L2 does not contain amino acid modifications at positions Q160 and/or T180. In one embodiment, L2 does not include amino acid modifications Q160K, Q160E and/or T180E. In some embodiments, the construct is wherein H1 comprises Q179K or Q179E, L1 comprises Q160E, Q160K and/or T180E, H2 comprises Q179K or Q179E, and L2 comprises Q160K, Q160E and/or T180E. Does not contain combinations of amino acid modifications.

In some embodiments, H1 and/or H2 do not contain an amino acid modification at position Q179. In some embodiments, H1 does not contain the amino acid modification Q179K and/or H2 does not contain the amino acid modification Q179E. In some embodiments, L1 does not contain an amino acid modification at position T180. In one embodiment, L1 does not contain the amino acid modification T180E. In some embodiments, L2 does not contain an amino acid modification at position S131. In one embodiment, L2 does not contain the amino acid modification S131K. In some embodiments, the construct does not comprise a combination of amino acid modifications where H1 comprises Q179K, L1 comprises T180E, H2 comprises Q179E and L2 comprises S131K.

In some embodiments, H1 does not contain an amino acid modification at position Q179. In some embodiments, H1 does not contain the amino acid modification Q179E. In some embodiments, L1 does not contain an amino acid modification at position Q160. In one embodiment, L1 does not contain the amino acid modification Q160K. In some embodiments, H2 does not contain an amino acid modification at position Q179. In one embodiment, H2 does not contain the amino acid modification Q179K. In some embodiments, L2 does not contain an amino acid modification at position T180. In one embodiment, L2 does not contain the amino acid modification T180E. In some embodiments, the construct does not comprise a combination of amino acid modifications where H1 comprises Q179E, L1 comprises Q160K, H2 comprises Q179K and L2 comprises T180E.

In some embodiments, H1 does not contain an amino acid modification at position A139. In some embodiments, H1 does not contain the amino acid modification A139C. In some embodiments, L1 does not contain an amino acid modification at position F116. In one embodiment, L1 does not contain the amino acid modification F116C. In some embodiments, the construct does not comprise a combination of amino acid modifications where H1 comprises A139C and L1 comprises F116C.

In some embodiments, the construct does not contain native disulfide linkages between the heavy and light chains. For example, in some embodiments, the cysteine at position 214 of L1 and/or L2 is modified to another amino acid. In some embodiments, L1 and/or L2 comprise the amino acid modification C214S. In some embodiments, the cysteine at position 233 of H1 and/or H2 is modified to another amino acid. In one embodiment, H1 and/or H2 comprise the amino acid modification C233S.

The embodiments described herein are applicable to constructs in Fab and full antibody formats.

The D3H44 heavy and light chain amino acid sequences aligned to the underlying human germline sequences for the variable, constant and J region segments are shown (figure indication: ^* sequence identity). FIG. 1A shows the human VH germline subgroups (one representative sequence is shown in each family). Sequence identity based on alignment of D3H44 to VH3 and IGHJ3 ^* 02. The human kappa VL germline subgroup is shown (one representative sequence is shown in each family). Sequence identity based on alignment of D3H44 to VKI and IGKJ1 ^* 01. The human lambda VL germline subgroup is shown (one representative sequence is shown in each family). Sequence identity based on alignment of D3H44 to VL1 and IGLJ1 ^* 01. Human CH1 allele sequence is shown. Human kappa and lambda allele sequences are shown. A flowchart for computational modeling to identify key interface residues and to design preferential heavy-light chain pairings is shown. An exemplary set of H1, L1, H2, L2 chains designed such that H1 preferentially pairs with L1 over L2 and H2 preferentially pairs with L2 over L1. A diagram representing the three-dimensional crystal structure of the heavy and light chain interface of the variable region is presented. Mutations introduced at the interface achieve electrostatic and steric complementarity to preferentially form absolute pairs H1-L1 and H2-L2, respectively. On the other hand, there are unfavorable steric and electrostatic mismatches in the imprecise pair that result in a reduced pair shaping tendency as well as reduced stability to the mismatched pair. A high-level schematic of the genetic engineering engineering requirements to form a bispecific Mab (monoclonal antibody) and the assay requirements necessary to quantify heavy and light chain pairs are described. The design goal of engineering a highly purified bispecific Mab (i.e., little or no mispairing HL associations) consists of two unique heavy chains and a unique cognate light chain. Preferential pairing with the chain can be achieved by rational engineering (via the introduction of specific amino acid mutations). This process is shown schematically, where H1 is engineered to preferentially pair with L1 over L2. Similarly, H2 is engineered to preferentially pair with L2 rather than L1. Experimental selection of bispecific Mab designs requires assays that can simultaneously quantify H1-L1:H1-L2 and H2-L2:H2-L1. These assay requirements can be simplified by assuming that each bispecific Fab arm can be engineered independently. In this case, the assay need only quantify H1-L1:H1-L2 or H2-L2:H2-L1, not both simultaneously. A schematic depiction of how heavy and light chains are tagged and how preferential pairing is determined is provided. In this schematic, circles represent cells transfected with the three constructs. The expression product is secreted from the cells and the supernatant (SPNT) flows to the detection device, in this case the SPR chip. Estimating a quantitative estimate of preferential pairing of a heavy chain to two light chains based on the detection levels of two different tags fused to two light chains competing for heavy chain pairing can be done. Boxplots of mean LCCA performance values with at least 86:14 pairings:mispairings of Fab heterodimers in each cluster are shown. Representative UPLC-SEC profiles of A) WT Fab heterodimers and B) representative design Fab heterodimers (H1L1 Fab components of LCCA designs 9735, 9737 and 9740) are shown. Potential heavy chain association products that can be expected when two different light chains are co-expressed with two different heavy chains in a cell are shown. Preferential pairing is assessed by use of SMCA (monoclonal antibody competition assay). Figure 3 shows the polarization/strand utilization preference in the D3H44/trastuzumab bispecific system. Strand utilization was assessed in different species observed by LC-MS. The X-axis represents the DNA ratio of H1:H2:L1:L2 and the Y-axis shows the corresponding ratio of each strand in different transfection experiments. In the equilibrium system all heavy and light chains show 25%. A bias toward utilization of one light chain is observed across all bispecific systems. Figure 3 shows the bias/strand utilization preference in the D3H44/cetuximab bispecific system. Strand utilization was assessed in different species observed by LC-MS. The X-axis represents the DNA ratio of H1:H2:L1:L2 and the Y-axis shows the corresponding ratio of each strand in different transfection experiments. In the equilibrium system all heavy and light chains show 25%. A bias toward utilization of one light chain is observed across all bispecific systems. Figure 3 shows the bias/strand utilization preference in the trastuzumab/cetuximab bispecific system. Strand utilization was assessed in different species observed by LC-MS. The X-axis represents the DNA ratio of H1:H2:L1:L2 and the Y-axis shows the corresponding ratio of each strand in different transfection experiments. In the equilibrium system all H and L chains show 25%. A bias toward utilization of one light chain is observed across all bispecific systems. FIG. 10 shows representative UPLC-SEC profiles of WT heterodimer and modified heterodimer antibodies, respectively. Figure 10a refers to D3H44/trastuzumab WT. 9060-9756_1. Refers to D3H44/Cetuximab WT. Refers to 9820-9823_1. Refers to Trastuzumab/Cetuximab WT. 9696-9848_1. Percent change of correctly matched Fab components to all mismatched Fab components utilizing the same heavy chain in engineered bispecific antibody samples per cluster (for D3H44/trastuzumab and D3H44/cetuximab). H1:L1 changes for all H1 species relative to wild-type, H2:L2 changes for all H2 species relative to trastuzumab/cetuximab wild-type), as well as change boxes for percentage of desired bispecific antibodies relative to wild-type. A whisker diagram is shown. The % change of correctly matched Fab components to all mismatched Fab components utilizing the same heavy chain per cluster is shown in the D3H44/trastuzumab system. The change in the ratio of desired bispecific antibodies to wild-type by cluster is shown in the D3H44/trastuzumab system. Percent change of correctly matched Fab components to all mismatched Fab components utilizing the same heavy chain in engineered bispecific antibody samples per cluster (for D3H44/trastuzumab and D3H44/cetuximab). H1:L1 changes for all H1 species relative to wild-type, H2:L2 changes for all H2 species relative to trastuzumab/cetuximab wild-type), as well as change boxes for percentage of desired bispecific antibodies relative to wild-type. A whisker diagram is shown. The % change of correctly matched Fab components to all mismatched Fab components utilizing the same heavy chain per cluster is shown in the D3H44/cetuximab system. The change in the ratio of desired bispecific antibodies to wild type by cluster is shown in the D3H44/cetuximab system. Percent change of correctly matched Fab components to all mismatched Fab components utilizing the same heavy chain in engineered bispecific antibody samples per cluster (for D3H44/trastuzumab and D3H44/cetuximab). H1:L1 changes for all H1 species relative to wild-type, H2:L2 changes for all H2 species relative to trastuzumab/cetuximab wild-type), as well as change boxes for percentage of desired bispecific antibodies relative to wild-type. A whisker diagram is shown. The percent change of correctly matched Fab components to all mismatched Fab components utilizing the same heavy chain per cluster is shown in the Trastuzumab/Cetuximab system. The change in the ratio of desired bispecific antibodies to wild-type by cluster is shown in the trastuzumab/cetuximab system. Percentage change of correctly matched Fab members to all mismatched Fab members utilizing the same heavy chain is shown per cluster across all bispecific systems. The change in the ratio of desired bispecific antibodies to wild-type by cluster is shown. Note that the values reported in Figure 11 also include estimated changes in engineered bispecific antibody samples whose corresponding wild-type constructs were not evaluated by SMCA. Methods for preparing bispecific antibodies using libraries of obligatory mutation pairs provided herein are shown.

An antigen-binding polypeptide construct (heterodimer) that can comprise a first heterodimer and a second heterodimer, each heterodimer comprising an immunoglobulin heavy chain or fragment thereof and an immunoglobulin light chain (heterodimer) (also called mer pairs) are provided herein. Both heterodimers have one or more amino acid modifications in the immunoglobulin heavy chain constant domain 1 (CH1) and one or more amino acid modifications in the immunoglobulin light chain constant domain (CL), immunoglobulin heavy chain variable one or more amino acid modifications in the domain (VH) and one or more amino acid modifications in the immunoglobulin light chain variable domain (VL), or in both the heavy and light chain constant and variable domains Can include combinations. The modified amino acid is typically part of the light and heavy chain interface, with the heavy chain of the first heterodimer preferentially pairing with one light chain but not the other. Non-matching modifications create preferential pairings between each heavy chain and the desired light chain. Similarly, the second heterodimeric heavy chain may preferentially pair with the second light chain rather than the first.

As indicated above, certain combinations of amino acid modifications described herein promote preferential pairing of heavy chains with certain light chains, thereby negligibly bispecific monoclonals, such that the need to purify the desired heterodimer from undesired or mismatched products is minimized Antibody (Mab) expression can be generated. The heterodimers can exhibit thermal stability comparable to heterodimers without amino acid modifications and demonstrate binding affinities for antigens comparable to heterodimers without amino acid modifications. can also

dual, targeting two different therapeutic targets or targeting two distinct epitopes (overlapping or non-overlapping) within the same antigen using first and second heterodimer designs Specific antibodies can be made.

Methods of preparing heterodimer pairs are also provided herein.

DEFINITIONS Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the claimed subject matter belongs. Where a term in this specification has multiple definitions, the one in this chapter takes precedence. When a reference is made to a URL or other such identifier or address, such identifiers may change and certain information on the Internet may appear and disappear, but the equivalent information , can be found by searching the Internet. Reference thereto is evidence that such information is available and generally disseminated.

It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and do not limit any claimed subject matter. In this application, the use of the singular includes the plural unless specifically stated otherwise.

In this description, any concentration range, percentage range, ratio range, or integer range, unless otherwise indicated, refers to any integer within the recited range, or, where appropriate, fractions thereof (e.g., tenths of an integer). It should be understood to include values of 1 and 1/100) of . As used herein, "about" means ±10% of the indicated range, value, sequence or structure, unless otherwise indicated. The terms "a" and "an", as used herein, refer to "one" of the listed components, unless otherwise indicated or determined from context. It should be understood to refer to "or a plurality". It should be understood that the use of alternatives (eg, “or”) means either one, both, or any combination of the alternatives. As used herein, the terms "include" and "comprise" are used synonymously. In addition, individual single-chain polypeptides or immunoglobulin constructs derived from the various combinations of structures and substituents described herein, each single-chain polypeptide or heterodimer being described separately. It should be understood that there is disclosure in this application to the same extent. Thus, the selection of specific components that form individual single-chain polypeptides or heterodimers is within the scope of this disclosure.

The section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described. All documents or documents cited in this application, including, but not limited to, patents, patent applications, articles, books, manuals and treaty documents, in part, are hereby incorporated by reference in their entirety for any purpose. expressly incorporated into the specification.

It should be understood that the methods and compositions described herein are not limited to the particular methodology, protocols, cell lines, constructs and reagents described herein, as such may vary. be. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the methods and compositions described herein; It should also be understood that you are limited only by the scope of the appended claims.

All publications and patents mentioned herein disclose constructs and methodologies described in the publications that can be used, for example, in connection with the methods, compositions and compounds described herein. For purposes of description and disclosure, the entirety of which is incorporated herein by reference. The publications discussed herein are provided solely for their disclosure prior to the filing date of the present application. Nothing herein is to be construed as an admission that the inventions described herein are not entitled to antedate such disclosure by virtue of prior invention or for any other reason. Nor.

In this application, amino acid names and atom names (e.g., N, O, C, etc.) are used as defined in the Protein DataBank (PDB) (www.pdb.org), which is IUPAC nomenclature. (IUPAC Nomenclature and Symbolism for Amino Acids and Peptides), Eur. J. Biochem., 152, 1 (1985), Eur. J. Biochem., 138, 9 with modifications. -37 (1984).The term "amino acid residue" refers to the twenty naturally occurring amino acids: alanine (Ala or A), cysteine (Cys or C), aspartic acid (Asp or D). , glutamic acid (Glu or E), phenylalanine (Phe or F), glycine (Gly or G), histidine (His or H), isoleucine (Ile or I), lysine (Lys or K), leucine (Leu or L), Methionine (Met or M), Asparagine (Asn or N), Proline (Pro or P), Glutamine (Gln or Q), Arginine (Arg or R), Serine (Ser or S), Threonine (Thr or T), Valine (Val or V), tryptophan (Trp or W) and tyrosine (Tyr or Y) residues are primarily intended to indicate amino acid residues contained therein.

The terms "polypeptide," "peptide" and "protein" are used interchangeably herein to refer to a polymer of amino acid residues. That is, a description directed to a polypeptide applies equally to a description directed to a peptide and a description directed to a protein, and vice versa. The term applies to naturally occurring amino acid polymers as well as amino acid polymers in which one or more amino acid residues are non-naturally encoded amino acids. As used herein, the term encompasses amino acid chains of any length, including full-length proteins in which the amino acid residues are linked by covalent peptide bonds.

The terms "nucleotide sequence" or "nucleic acid sequence" are intended to indicate a continuous stretch of two or more nucleotide molecules. Nucleotide sequences may be of genomic, cDNA, RNA, semi-synthetic or synthetic origin, or any combination thereof.

"Cell," "host cell," "cell line," and "cell culture" are used interchangeably herein and all such terms refer to passages resulting from propagation or culture of cells. should be understood to include "Transformation" and "transfection" are used interchangeably to refer to the process of introducing a nucleic acid sequence into a cell.

The term "amino acid" refers to naturally occurring and non-naturally occurring amino acids, as well as amino acid analogs and mimetics that function similarly to the naturally occurring amino acids. The naturally encoded amino acids are the 20 amino acids (alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine and valine), and pyrrolelysine and selenocysteine. Amino acid analogs have the same basic chemical structure as naturally occurring amino acids, i.e., carbon, carboxyl, amino and R groups bonded to hydrogen, such as homoserine, norleucine, methionine sulfoxide, methionine methylsulfonium. Refers to a compound. Such analogs have modified R groups (eg, norleucine) or modified peptide backbones, but retain the same basic chemical structure as a naturally occurring amino acid. References to amino acids include, for example, naturally occurring proteinogenic L-amino acids, D-amino acids, chemically modified amino acids such as amino acid variants and derivatives, naturally occurring non-proteinogenic amino acids such as alanine, ornithine, etc., and Included are chemically synthesized compounds having properties known in the art to be characteristic of amino acids. Examples of non-naturally occurring amino acids include □-methyl amino acids (eg, methylalanine), D-amino acids, histidine-like amino acids (eg, 2-amino-histidine, hydroxy-histidine, homohistidine), additional methylene They include, but are not limited to, amino acids that have a chain (“homo” amino acids), and amino acids in which the side chain carboxylic acid function is replaced with a sulfonic acid group (eg, cysteic acid). The incorporation of synthetic non-native amino acids, substituted amino acids or non-natural amino acids, including one or more D-amino acids, into the proteins of the invention can be beneficial in a number of different ways. D-amino acid-containing peptides, etc., exhibit increased stability in vitro or in vivo compared to their L-amino acid-containing counterparts. Thus, constructing peptides and the like that incorporate D-amino acids can be particularly useful when greater intracellular stability is desired or sought. More particularly, D-peptides and the like are resistant to endogenous peptidases and proteases, thereby improving the bioavailability of the molecule and extending its life in vivo when such properties are desired. bring. In addition, D-peptides, etc., are not efficiently processed by the restricted presentation of the major histocompatibility complex class II to T helper cells, thus eliciting a humoral immune response throughout the organism. Not likely.

Amino acids are referred to herein by their commonly known three-letter code or by the one-letter code recommended by the IUPAC-IUB Biochemical Nomenclature Commission. Nucleotides may likewise be referred to by their generally accepted single letter codes.

“Conservatively modified variants” applies to both amino acid and nucleic acid sequences. With respect to a particular nucleic acid sequence, "conservatively modified variant" refers to a nucleic acid that encodes an identical or substantially identical amino acid sequence, or a substantially identical sequence if the nucleic acid does not encode an amino acid sequence. Due to the degeneracy of the genetic code, multiple functionally identical nucleic acids encode any given protein. For example, the codons GCA, GCC, GCG and GCU all code for the amino acid alanine. Thus, at all positions where alanine is specified by a codon, the codon can be changed to any of the corresponding codons described without altering the encoded polypeptide. Such nucleic acid variations are "silent variations," which are one species of conservatively modified variations. Every nucleic acid sequence herein that encodes a polypeptide also describes every possible silent variation of the nucleic acid. One skilled in the art will appreciate that each codon in a nucleic acid (except AUG, which is usually the only codon for methionine, and TGG, which is usually the only codon for tryptophan) can be modified to produce a functionally identical molecule. to recognize Accordingly, each silent variation of a nucleic acid which encodes a polypeptide is implicit in each described sequence.

For amino acid sequences, one skilled in the art can make individual substitutions, deletions or modifications to nucleic acid, peptide, polypeptide or protein sequences that alter, add or delete single amino acids or small percentages of amino acids in the encoded sequence. We recognize that additions are "conservative modifications" in which the alteration results in the deletion of an amino acid, the addition of an amino acid, or the substitution of an amino acid with a chemically similar amino acid. Conservative substitution tables providing functionally similar amino acids are known to those of skill in the art. Such conservatively modified variants are in addition to, but not exclusive of, the polymorphisms, variants, interspecies homologues and alleles of the invention.

Conservative substitution tables providing functionally similar amino acids are known to those of skill in the art. The following eight groups each contain amino acids with conservative substitutions for each other.
alanine (A), glycine (G);
aspartic acid (D), glutamic acid (E);
asparagine (N), glutamine (Q);
Arginine (R), Lysine (K);
Isoleucine (I), Leucine (L), Methionine (M), Valine (V);
Phenylalanine (F), Tyrosine (Y), Tryptophan (W);
Serine (S), Threonine (T); and Cysteine (C), Methionine (M).
(See, e.g., Creighton, Proteins: Structures and Molecular Properties (WH Freeman & Co., 2nd ed. (December 1993).)

The terms "identical" or percent "identity", in the context of two or more nucleic acid or polypeptide sequences, refer to two or more sequences or subsequences that are the same. Sequences were determined when compared and aligned for maximum correspondence over the comparison range, or over the designated region using one of the following sequence comparison algorithms (or other algorithms available to those skilled in the art): or by manual alignment and visual inspection, the percentage of amino acid residues or nucleotides that are the same (i.e., at least about 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98% or 99% identity) is "substantially identical" be. This definition is also called the complement of the test sequence. Identity may exist over a region of at least about 50 amino acids or nucleotides in length, or over a region of 75-100 amino acids or nucleotides in length, or, if not specified, over the entire polynucleotide or polypeptide sequence. can be done. Polynucleotides encoding the polypeptides of the invention, including homologues from species other than human, can be prepared by exposing the library under stringent hybridization conditions with labeled probes having the polynucleotide sequences of the invention or fragments thereof. It can be obtained by a process comprising the steps of selecting and isolating full-length cDNA and genomic clones containing said polynucleotide sequence. Such hybridization techniques are well known to those skilled in the art.

A derivative or variant of a polypeptide shares "homology" or "homology" with a peptide if the amino acid sequence of the derivative or variant shares at least 50% identity with the sequence of 100 amino acids of the original peptide. It is said that there is In certain embodiments, a derivative or variant is at least 75% identical to a peptide or fragment of a peptide having the same number of amino acid residues as the derivative. In certain embodiments, a derivative or variant is at least 85% identical to a peptide or fragment of a peptide having the same number of amino acid residues as the derivative. In certain embodiments, the amino acid sequence of the derivative is at least 90% identical to a peptide or fragment of a peptide having the same number of amino acid residues as the derivative. In some embodiments, the amino acid sequence of the derivative is at least 95% identical to a peptide or fragment of a peptide having the same number of amino acid residues as the derivative. In certain embodiments, a derivative or variant is at least 99% identical to a peptide or fragment of a peptide having the same number of amino acid residues as the derivative.

As used herein, an "isolated" polypeptide or construct means a construct or polypeptide that has been identified and separated and/or recovered from a component of its natural cell culture environment. Contaminant components of the natural environment are materials that typically preclude diagnostic or therapeutic use of heteromultimers, and can include enzymes, hormones and other proteinaceous or non-proteinaceous solutes.

In certain embodiments, as used herein, an "isolated" antigen-binding polypeptide construct described herein is identified and separated from components of the natural cell culture environment, and/or heterodimer pairs or “isolated” heterodimer pairs, including recovered heterodimers or heterodimer pairs. Contaminant components of the natural environment are materials that would interfere with diagnostic or therapeutic uses for a heterodimer or antigen-binding polypeptide construct, and may include enzymes, hormones and other proteinaceous or non-proteinaceous solutes.

Heterodimers and antigen-binding polypeptide constructs, and heterodimer pairs will generally be purified to substantial homogeneity. The phrases "substantially homogeneous," "substantially homogeneous form," and "substantially homogeneous" mean that the product is substantially devoid of by-products resulting from undesirable combinations of polypeptides (e.g., homodimers). used to indicate that In this context, the species of interest is a heterodimer comprising H1 and L1 (H1-L1) or H2 and L2 (H2-L2). Contaminants include heterodimers containing H1 and L2 (H1-L2) or H2 and L1 (H2-L1), or homodimers containing H1 and L1 or H2 and L2 (where the Fab portion is included, whether matched or mismatched). Expressed in terms of purity, substantial homogeneity means that the amount of byproducts does not exceed 10% of the total LC-MS intensity of all species present in the mixture, e.g., less than 5%, less than 1% or 0.5%. Means less than and percentages reflect the results of mass spectrometry.

The phrase "selectively (or specifically) hybridizes" means that only certain nucleotide sequences are present in complex mixtures (including, but not limited to, whole cell or library DNA or RNA). When used it is meant to allow molecules to bind, replicate or hybridize under stringent hybridization conditions.

Terms understood by those skilled in the art of antibody technology are each given their art-acquired meaning, unless specifically defined differently herein. Antibodies are known to have variable regions, hinge regions and constant domains. The structure and function of immunoglobulins are reviewed, for example, in Harlow et al., eds., Antibodies: A Laboratory Manual, Chapter 14 (Cold Spring Harbor Laboratory, Cold Spring Harbor, 1988).

As used herein, the terms "antibody" and "immunoglobulin" or "antigen-binding polypeptide construct" are used interchangeably. An "antigen-binding polypeptide construct" refers to a polypeptide substantially encoded by an immunoglobulin gene or one or more fragments thereof that specifically binds to an analyte (antigen). The recognized immunoglobulin genes include the kappa, lambda, alpha, gamma, delta, epsilon and mu constant domain genes, as well as the myriad immunoglobulin variable region genes. Light chains are classified as either kappa or lambda. Heavy chains are classified as gamma, mu, alpha, delta, or epsilon, which in turn define the immunoglobulin isotypes IgG, IgM, IgA, IgD and IgE, respectively. Furthermore, an antibody can belong to one of a number of subtypes, for example an IgG can belong to the IgG1, IgG2, IgG3 or IgG4 subclass.

An exemplary immunoglobulin (antibody) structural unit is composed of two pairs of polypeptide chains, each pair comprising one "light" (about 25 kD) and one "heavy" chain (about 50-70 kD). have. The term "light chain" includes full-length light chains and fragments thereof having sufficient variable region sequence to confer binding specificity. A full-length light chain comprises a variable region domain VL and a constant region domain CL. The light chain variable region domain is at the amino terminus of the polypeptide. Light chains include kappa and lambda chains. The term "heavy chain" includes full-length heavy chains and fragments thereof having sufficient variable region sequence to confer binding specificity. A full length heavy chain comprises a variable region domain VH and three constant region domains CH1, CH2 and CH3. The VH domain is at the amino terminus of the polypeptide, the CH domain is at the carboxyl terminus, and CH3 is closest to the carboxyl terminus of the polypeptide. Heavy chains can be of any isotype, including IgG (including IgG1, IgG2, IgG3 and IgG4 subclasses), IgA (including IgA1 and IgA2 subclasses), IgM, and IgE. The term “variable region” or “variable domain” typically comprises approximately 120-130 amino-terminal amino acids in the heavy (VH) chain and about 100-110 amino-terminal amino acids in the light (VL) chain. , generally refers to the portion of the light and/or heavy chain of an antibody that is involved in antigen recognition. A "complementarity determining region" or "CDR" is an amino acid sequence that contributes to antigen-binding properties and affinity. A “framework” region (FR) can help maintain the correct conformation of the CDRs and facilitate binding of the antigen-binding region to the antigen. Structurally, framework regions can be located between the CDRs in an antibody. Variable regions typically exhibit the same general structure, with relatively conserved framework regions (FR) joining the three hypervariable region CDRs. The CDRs of the two chains of each pair are typically aligned by framework regions, which allow binding to a specific epitope. From N-terminus to C-terminus, both light and heavy chain variable regions typically comprise domains FR1, CDR1, FR2, CDR2, FR3, CDR3 and FR4. Amino acid assignments for each domain typically follow the definitions in the Kabat Sequences of Proteins of Immunological Interest (National Institutes of Health, Bethesda, Md. (1987 and 1991)) unless otherwise stated. In certain embodiments, the antigen-binding polypeptide construct comprises at least one immunoglobulin domain of IgG, IgM, IgA, IgD or IgE linked to a therapeutic polypeptide. In some embodiments, the immunoglobulin domains included in the antigen-binding polypeptide constructs provided herein are of immunoglobulin-based constructs, such as diabodies or nanobodies. In certain embodiments, the antigen-binding polypeptide constructs described herein comprise at least one immunoglobulin domain of a heavy chain antibody, such as a camelid antibody. In certain embodiments, the antigen-binding polypeptide constructs provided herein are bovine antibodies, human antibodies, camelid antibodies (single domain and non-single domain), rodent antibodies, humanized antibodies, It contains at least one immunoglobulin domain from a mammalian antibody, such as a non-humanized antibody, a murine antibody or any chimeric antibody. In certain embodiments, the antigen-binding polypeptide constructs provided herein comprise at least one immunoglobulin domain of an antibody generated from a synthetic library.

A "bi-specific," "dual-specific," or "bifunctional" antigen-binding protein or antibody is a hybrid antigen having two different antigen-binding sites. It is a binding protein. Bispecific antigen binding proteins and antibodies are a type of multispecific antigen binding protein antibodies. The two binding sites of bispecific antigen binding proteins and antibodies bind to two different epitopes which may be present on the same or different molecular targets. A "multispecific antigen binding protein" or "multispecific antibody" targets more than one antigen or epitope. A "bivalent antigen binding protein" or "bivalent antibody" contains two antigen binding sites. In some cases, the two binding sites have the same antigen specificity. Bivalent antigen binding proteins and bivalent antibodies can be bispecific. See above. Bivalent antibodies, other than "multispecific" or "multifunctional" antibodies, in certain embodiments, are typically understood to each have identical binding sites.

The term "preferential pairing" is used herein to describe the pattern of pairing between a first polypeptide and a second polypeptide, e.g. Immunoglobulin heavy and light chains in polypeptide constructs and heterodimers. Thus, "preferential pairing" is defined as, when pairing occurs between a first and a second polypeptide, the first polypeptide is in the presence of one or more additional different polypeptides at the same time. Refers to preferential pairing of a peptide with a second polypeptide. Typically, preferential pairing results from modifications (eg, amino acid modifications) of one or both of the first and second polypeptides. Typically, preferential pairing results in the paired first and second polypeptide being the most abundant dimer after pairing occurs. An immunoglobulin heavy chain (H1), when co-expressed with two different immunoglobulin light chains (L1 and L2), pairs statistically equally with both light chains, H1 paired with L1, and L2 paired. It is known in the art to give a mixture of approximately 50:50 for combined H1. In this context, "preferential pairing" means that, for example, when H1 is co-expressed with L1 and L2, the amount of H1-L1 heavy chain-light chain heterodimer is less than that of H1-L2 heterodimer. If it is greater than the amount, it occurs between H1 and L1. Therefore, in this case, H1 preferentially pairs with L1 over L2.

However, in the context of wild-type bispecific antibodies generated from two starting antibody systems, the intrinsic bias in which light chains of one antibody system preferentially pair with heavy chains of both antibody systems is It is also known to exist in some cases. Therefore, when determining the strength of a design in the context of a bispecific antibody binding construct, it may be necessary to assess the degree of pairing of the design relative to that of the wild-type system. Thus, in one embodiment, the design is considered to exhibit preferential pairing when the amount of desired bispecific antibody is greater than the amount of desired bispecific antibody obtained in the wild-type system. be done. In another embodiment, the design is considered to exhibit preferential pairing when the amount of pairing in the weak arm of the antibody is greater than that seen in the wild-type system.

Antibody heavy chains pair with antibody light chains and meet or contact each other at one or more "interfaces." An "interface" includes one or more "contact" amino acids of a first polypeptide that interact with one or more "contact" amino acid residues of a second polypeptide. For example, interfaces exist between the CH3 polypeptide sequences of dimerized CH3 domains, between the CH1 domain of the heavy chain and the CL domain of the light chain, and between the VH domain of the heavy chain and the VL domain of the light chain. The "interface" can be derived from an IgG antibody, eg, a human IgG1 antibody.

The term "amino acid modification" as used herein includes, but is not limited to amino acid mutations, insertions, deletions, substitutions, chemical modifications, physical modifications and rearrangements.

Antigen-Binding Polypeptide Constructs and Heterodimer Pairs Antigen-binding polypeptide constructs described herein can comprise a first heterodimer and a second heterodimer, each heterodimer A dimer results from the pairing of an immunoglobulin heavy chain and an immunoglobulin light chain. The structure and organization of constant and variable domains of immunoglobulin heavy and light chains are well known in the art. Immunoglobulin heavy chains typically comprise one variable (VH) domain and three constant domains, CH1, CH2 and CH3. Immunoglobulin light chains typically comprise one variable (VL) domain and one constant (CL) domain. Various modifications can be made to these exemplary formats.

The antigen-binding polypeptide constructs and heterodimer pairs described herein can comprise a first heterodimer and a second heterodimer, each heterodimer comprising: It includes immunoglobulin/antibody heavy chains or fragments thereof comprising at least VH and CH1 domains, and immunoglobulin/antibody light chains having VL and CL domains. In one embodiment, both the heterodimer pair and the heterodimer of the antigen-binding polypeptide construct comprise a full-length immunoglobulin heavy chain. In another embodiment, the heterodimer pair or both heterodimers of the antigen-binding polypeptide construct comprise a fragment of an immunoglobulin heavy chain comprising at least the VH and CH1 domains. In one embodiment, both heterodimers of the heterodimer pair comprise an amino-terminal fragment of an immunoglobulin heavy chain comprising at least the VH and CH1 domains. In another embodiment, both heterodimers of the heterodimer pair comprise a carboxy-terminal fragment of an immunoglobulin heavy chain comprising at least the VH and CH1 domains.

Each heterodimer of a heterodimer pair is capable of specifically binding an antigen or epitope. In one embodiment, each heterodimeric immunoglobulin heavy chain and immunoglobulin light chain is derived or modified from a known antibody, eg, a therapeutic antibody. A therapeutic antibody is one that is effective in treating a disease or disorder in a mammal having or predisposed to the disease or disorder. Suitable therapeutic antibodies from which the respective heterodimers can be induced include abagovomab, adalimumab, alemtuzumab, aurograb, bapineuzumab, basiliximab, belimumab, bevacizumab, briakinumab, canakinumab, catumaxomab, certolizumab gol, cetuximab, daclizumab, denosumab, efalizumab, galiximab, gemtuzumab ozogamicin, golimumab, ibritumomab tiuxetan, infliximab, ipilimumab, lumiliximab, mepolizumab, motavizumab, muromonab, mycograb, natalizumab, nimotuzumab (nimotuzumab) ), ocrelizumab, ofatumumab, omalizumab, palivizumab, panitumumab, pertuzumab, ranibizumab, reslizumab, rituximab, teplizumab, tocilizumab/atlizumab, tositumomab, trastuzumab, Proxinium™, Rencarex™, ustekinumab , but not limited to.

In one embodiment, each heterodimeric immunoglobulin heavy chain and/or immunoglobulin light chain comprises the proteins listed below, and subunits, domains, motifs and epitopes belonging to the proteins listed below. derived or modified from an antibody that binds to a molecule such as, but not limited to. renin; growth hormones, including human growth hormone and bovine growth hormone; growth hormone-releasing factor; parathyroid hormone; thyroid-stimulating hormone; lipoproteins; Hormones; calcitonin; luteinizing hormone; glucagon; clotting factors such as factor VII, factor VIIIC, factor IX, tissue factor (TF) and von Willebrand factor; : pulmonary surfactant; plasminogen activators such as urokinase or human urinary or tissue-type plasminogen activator (t-PA); bombesin; thrombin; RANTES (regulated on activation normally T-cell expressed and secreted); human macrophage inflammatory protein (MIP-1-alpha); human serum albumin, etc. relaxin A chain; relaxin B chain; prorelaxin; mouse gonadotropin-related peptides; microbial proteins such as beta-lactamase; activin; vascular endothelial growth factor (VEGF); receptors for hormones or growth factors such as EGFR, VEGFR; gamma-IFN); protein A or D; rheumatoid factor; bone-derived neurotrophic factor (BDNF), neurotrophin-3, -4, -5 or -6 (NT-3, NT-4, NT- 5 or NT-6), or nerve growth factors; platelet-derived growth factor (PDGF); fibroblast growth factors such as AFGF and PFGF; epidermal growth factor (EGF); TGF-1, TGF-2 transforming growth factors (TGFs) such as TGF-alpha and TGF-beta, including TGF-3, TGF-4 or TGF-5; insulin-like growth factors I and II (IGF-I and IGF-II); S(1-3)-IGF-I (brain IGF-I), insulin-like growth factor binding protein; CD proteins such as CD34, CD40, CD40L, CD52, CD63, CD64, CD80 and CD147; erythropoietin; osteoinductive factors; immunotoxins; bone morphogenetic proteins (BMPs); colony stimulating factors (CSF) such as GM-CSF and G-CSF; interleukins (ILs) such as IL-1 to IL-13; TNF-alpha, superoxide dismutase; T-cell receptors; surface membrane proteins; facilitating factors; eg parts of the AIDS envelope, viral antigens such as eg gp120; transport proteins; homing receptors; and cell adhesion components such as VCAM; a4/p7 integrins and (Xv/p3 integrins, including a or its subunits; integrin alpha subunits such as CD11a, CD11b, CD51, CD11c, CD41, alpha IIb, alpha IELb; integrin beta subunits such as CD29, CD18, CD61, CD104, beta5, beta6, beta7 and beta8; alphaV Combinations of integrin subunits including but not limited to beta3, alphaVbeta5 and alpha4beta7; members of the apoptotic pathway; IgE; blood group antigens; flk2/flt3 receptors; protein C; Eph receptors such as EphA2, EphA4, EphB2; human leukocyte antigens (HLA) such as HLA-DR; complement proteins such as complement receptors CR1, C1Rq; other complement factors such as C3 and C5; glycoprotein receptors such as GpIbalpha, GPIIb/IIIa and CD200; and fragments of any of the above-listed polypeptides.

In certain embodiments, each heterodimeric immunoglobulin heavy chain and/or immunoglobulin light chain comprises ALK receptor (pleiotrophin receptor), pleiotrophin, KS1/4 pan-cancer antigen (pan- carcinoma antigen); ovarian cancer antigen (CA125); prostatic acid phosphatase; prostate specific antigen (PSA); melanoma-associated antigen p97; melanoma antigen gp75; carcinoembryonic antigen (CEA); polymorphic epithelial mucin antigen; human milk fat globule antigen; colorectal cancer associated antigens such as CEA, TAG-72, CO17-1A, GICA19-9, CTA-1 and LEA; human B lymphoma antigen CD20; CD33; melanoma-specific antigens such as ganglioside GD2, ganglioside GD3, ganglioside GM2 and ganglioside GM3; tumor-specific transplantable cell surface antigen (TSTA); RNA tumors virally induced tumor antigens, including viral T antigens, DNA tumor viruses and envelope antigens; carcinoembryonic antigens alpha-fetoprotein such as CEA of the colon, 514 oncofetal trophoblast glycoprotein and bladder tumor antigen; human lung antigen L6 and human leukemia T-cell antigen Gp37; neoglycoproteins; sphingolipids; breast cancer antigens such as EGFR (epidermal growth factor receptor); HER2 antigen (p185HER2); polymorphic epithelial mucin (PEM); malignant tumor human leukocyte antigen APO-1; differentiation antigens such as I antigen found on fetal erythrocytes; early endoderm I antigen found on adult erythrocytes; I (Ma) found in gastric adenocarcinoma; M18, M39 found in mammary epithelium; SSEA-1 found in myeloid cells; VEP8; TRA-1-85 (blood group H); SCP-1 found in testicular and ovarian cancer; C14 found in rectal adenocarcinoma; F3 found in lung adenocarcinoma; Ley found in cancer cells; TL5 (blood group A); EGF receptor found in A431 cells; E1 series found in pancreatic cancer (blood group B); FC10.2 found in embryonic carcinoma cells; CO-514 (blood group Lea) found in adenocarcinoma; NS-10 found in adenocarcinoma; C O-43 (blood group Leb); G49 found on EGF receptors in A431 cells; MH2 (blood group ALeb/Ley) found in colon adenocarcinoma; 19.9 found in colon cancer; gastric cancer mucin; R24 found in melanoma; 4.2, GD3, D1.1, OFA-1, GM2, OFA-2, GD2 and M1:22:25:8 found in embryonic carcinoma cells and embryos cutaneous T-cell lymphoma antigen; MART-1 antigen; Sialy Tn (STn) antigen; colon cancer antigen NY-CO-45; lung cancer antigen NY-LU- 12 variant A; adenocarcinoma antigen ART1; paraneoplastic associated brain testicular cancer antigen (onconeuronal antigen MA2, paraneoplastic neuronal antigen); neuro-oncological abdominal antigen 2 (Neuro -oncological ventral antigen 2) (NOVA2); hepatocellular carcinoma antigen gene 520; tumor-associated antigen CO-029; tumor-associated antigen MAGE-C1 (cancer/testis antigen CT7), MAGE-B1 (MAGE-XP antigen), MAGE- B2 (DAM6), MAGE-2, MAGE-4-a, MAGE-4-b and MAGE-X2; cancer-testis antigen NY-EOS-1), and fragments of any of the above-listed polypeptides derived or modified from an antibody that specifically binds to a cancer antigen, including but not limited to;

Human antibodies can be divided into isotypes, including IgG, IgA, IgE, IgM and IgD. In one embodiment the Fc is derived from the IgG isotype. In another embodiment, the Fc is derived from the IgA isotype. In another embodiment, the Fc is derived from the IgE isotype. In another embodiment, the Fc is derived from the IgM isotype. In another embodiment, the Fc is derived from the IgD isotype.

Human IgG antibodies can also be divided into subclasses IgG1, IgG2, IgG3 and IgG4. Thus, it is contemplated that in some embodiments the Fc may be derived from the IgG1, IgG2, IgG3 and IgG4 subclasses of antibodies.

Each heterodimer of a heterodimer pair is capable of specifically binding an epitope or antigen. In one embodiment, each heterodimer of the heterodimer pair binds the same epitope. In another embodiment, the first heterodimer of the heterodimer pair specifically binds to an epitope of one antigen and the second heterodimer of the heterodimer pair binds to the same antigen specifically binds to different epitopes of In another embodiment, the first heterodimer of the heterodimer pair specifically binds to an epitope of the first antigen and the second heterodimer of the heterodimer pair binds the second It specifically binds to an epitope on a second antigen that is different from the first antigen. For example, in one embodiment, the first heterodimer specifically binds tissue factor, while the second heterodimer specifically binds the antigen Her2 (ErbB2) and The same is true vice versa. In an alternative embodiment, the first heterodimer specifically binds tissue factor, while the second heterodimer specifically binds EGFR, or vice versa. . In yet another embodiment, the first heterodimer specifically binds EGFR, while the second heterodimer specifically binds antigen Her2, or vice versa. . In another embodiment, the first heterodimer specifically binds to the molecule or cancer antigen described above. In another embodiment, the second heterodimer specifically binds to the molecule or cancer antigen described above.

As indicated above, in some embodiments, each heterodimeric immunoglobulin heavy chain and immunoglobulin light chain is bound from a known therapeutic antibody or to a different target molecule or cancer antigen. It may also be derived from, or genetically engineered from, antibodies that do not exist. The amino acid and nucleotide sequences of many such molecules are readily available (eg GenBank: AJ308087.1 (humanized anti-human tissue factor antibody D3H44 light chain variable region and CL domain); GenBank: AJ308086.1 (humanized anti-human tissue factor antibody D3H44 heavy chain variable region and CH1 domain); GenBank: HC359025.1 (Pertuzumab Fab light chain gene module); GenBank: HC359024.1 (Pertuzumab Fab heavy chain gene module); GenBank: GM685465. 1 (antibody trastuzumab (=Herceptin) - wild type; light chain); GenBank: GM685463.1 (antibody trastuzumab (=Herceptin) - wild type; heavy chain); GenBank: GM685466.1 (antibody trastuzumab (=Herceptin) - GC and GenBank: GM685464.1 (see Antibody Trastuzumab (=Herceptin) - GC Optimized Heavy Chain). The amino acid and nucleotide sequences of cetuximab are also known in the art, for example, Canadian See Drug Bank website, sponsored by the Institutes of Health Research, Alberta Innovates-Health Solutions and The Metabolomics Innovation Center (TMIC), Accession No. DB00002.

In some embodiments, the isolated antigen-binding polypeptide construct comprises at least 80, 85, 90, 91, 92 amino acid sequences set forth in the Tables or Accession Numbers disclosed herein or fragments thereof. , 93, 94, 95, 96, 97, 98, 99 or 100% identical amino acid sequences. In some embodiments, the isolated antigen-binding polypeptide construct comprises at least 80, 85, 90, 91, 92 nucleotide sequences set forth in the tables or accession numbers disclosed herein or fragments thereof. , 93, 94, 95, 96, 97, 98, 99 or 100% identical amino acid sequences encoded by polynucleotides.

Amino Acid Modifications to the Immunoglobulin Heavy and Light Chains At least one heterodimer of the heterodimer pair preferably pairs the heavy chain of the first heterodimer with the light chain of one of the heterodimers. It contains one or more amino acid modifications in the immunoglobulin heavy chain and/or the immunoglobulin light chain so that they do not pair with one another. Similarly, the second heterodimeric heavy chain may preferentially pair with the second light chain rather than the first. Preferential pairing of one heavy chain with one of the two light chains is based on a design set comprising one immunoglobulin heavy chain and two immunoglobulin light chains (referred to as the LCCA design set) and an immunoglobulin heavy chain preferentially pairs with one of the two immunoglobulin light chains over the other when the immunoglobulin heavy chain is co-expressed with both immunoglobulin light chains. Thus, an LCCA design set can include one immunoglobulin heavy chain, a first immunoglobulin light chain and a second immunoglobulin light chain.

In one embodiment, the one or more amino acid modifications comprise one or more amino acid substitutions.

In one embodiment, the preferential pairings demonstrated by the LCCA design set are established by modifying one or more amino acids that are part of the interface between the light and heavy chains. In one embodiment, the preferential pairing demonstrated by the LCCA design set is between the CH1 domain of the immunoglobulin heavy chain, the CL domain of the first immunoglobulin light chain and the CL domain of the second immunoglobulin light chain. Established by modifying one or more amino acids in at least one.

In one embodiment, the one or more amino acid modifications are restricted to conserved framework residues of the variable (VH, VL) and constant (CH1, CL) domains as indicated by the Kabat numbering of the residues. It is For example, Almagro [Frontiers In Bioscience (2008) 13:1619-1633] provides definitions of framework residues based on Kabat, Chotia and IMGT numbering.

In one embodiment, at least one of the heterodimers comprises one or more mutations introduced into immunoglobulin heavy and light chains that are complementary to each other. Complementarity at the heavy and light chain interface can be achieved based on steric and hydrophobic contacts, electrostatic/charge interactions or a combination of various interactions. Complementarity between protein surfaces has been documented in the literature for lock and key fit, knob into hole, protrusion and cavity, donor and acceptor, etc. It has been extensively described, all involving a match of structural and chemical properties between two interacting surfaces. In one embodiment, at least one of the heterodimers is one wherein the mutations introduced into the immunoglobulin heavy chain and the immunoglobulin light chain introduce new hydrogen bonds to the heavy and light chains at the interface. or contain multiple mutations. In one embodiment, at least one of the heterodimers is one wherein the mutations introduced into the immunoglobulin heavy chain and the immunoglobulin light chain introduce new salt bridges to the heavy and light chains at the interface. or contain multiple mutations.

Non-limiting examples of suitable LCCA design sets are provided in the Examples, Tables and Figures. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with at least one amino acid modification in the CH1 domain, a first immunoglobulin light chain with at least one amino acid modification in the CL domain, and an amino acid in the CL domain. It includes a second immunoglobulin light chain that has no modifications. In another embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least one amino acid modification in the CH1 domain, a first immunoglobulin light chain having at least one amino acid modification in the CL domain and at least A second immunoglobulin light chain having a single amino acid modification is included. In another embodiment, the LCCA design set comprises an immunoglobulin heavy chain with at least one amino acid modification in the CH1 domain, a first immunoglobulin light chain with at least two amino acid modifications in the CL domain, and A second immunoglobulin light chain having two amino acid modifications is included. In another embodiment, the LCCA design set comprises an immunoglobulin heavy chain with at least one amino acid modification in the CH1 domain, a first immunoglobulin light chain with at least two amino acid modifications in the CL domain, and A second immunoglobulin light chain having a single amino acid modification is included.

In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with no amino acid modifications in the CH1 domain, a first immunoglobulin light chain with no amino acid modifications in the CL domain, and at least one amino acid in the CL domain. A second immunoglobulin light chain having a modification is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with no amino acid modifications in the CH1 domain, a first immunoglobulin light chain with no amino acid modifications in the CL domain and at least 2 amino acids in the CL domain. A second immunoglobulin light chain having a modification is included.

In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with at least two amino acid modifications in the CH1 domain, a first immunoglobulin light chain with no amino acid modifications in the CL domain, and at least one in the CL domain. A second immunoglobulin light chain having an amino acid modification of In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least two amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least one amino acid modification in the CL domain and at least A second immunoglobulin light chain having a single amino acid modification is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least two amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least one amino acid modification in the CL domain and at least A second immunoglobulin light chain having two amino acid modifications is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least two amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least two amino acid modifications in the CL domain and at least A second immunoglobulin light chain having two amino acid modifications is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least 2 amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least 3 amino acid modifications in the CL domain, and at least A second immunoglobulin light chain having two amino acid modifications is included.

In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with at least 3 amino acid modifications in the CH1 domain, a first immunoglobulin light chain with no amino acid modifications in the CL domain and at least 1 in the CL domain. A second immunoglobulin light chain having an amino acid modification of In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least three amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least one amino acid modification in the CL domain, and at least A second immunoglobulin light chain having a single amino acid modification is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least 3 amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least 3 amino acid modifications in the CL domain and at least A second immunoglobulin light chain having two amino acid modifications is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least 3 amino acid modifications in the CH1 domain, a first immunoglobulin light chain having at least 4 amino acid modifications in the CL domain, and at least A second immunoglobulin light chain with three amino acid modifications is included. In one embodiment, the preferential pairing demonstrated by the LCCA design set is between the VH domain of the immunoglobulin heavy chain, the VL domain of the first immunoglobulin light chain and the VL domain of the second immunoglobulin light chain. Established by modifying one or more amino acids in at least one. Non-limiting examples of suitable LCCA design sets are provided in the table below and in the examples.

In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with no amino acid modifications in the VH domain, a first immunoglobulin light chain with no amino acid modifications in the VL domain, and at least one amino acid in the VL domain. A second immunoglobulin light chain having a modification is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with no amino acid modifications in the VH domain, a first immunoglobulin light chain with no amino acid modifications in the VL domain and at least 2 amino acids in the VL domain. A second immunoglobulin light chain having a modification is included.

In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain with at least one amino acid modification in the VH domain, a first immunoglobulin light chain with no amino acid modification in the VL domain and at least one in the VL domain. A second immunoglobulin light chain having an amino acid modification of In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least one amino acid modification in the VH domain, a first immunoglobulin light chain having at least one amino acid modification in the VL domain and at least A second immunoglobulin light chain having a single amino acid modification is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least one amino acid modification in the VH domain, a first immunoglobulin light chain having at least two amino acid modifications in the VL domain and at least A second immunoglobulin light chain having two amino acid modifications is included.

In one embodiment, the LCCA design set is an immunoglobulin heavy chain with at least two amino acid modifications in the VH domain, a first immunoglobulin light chain with no amino acid modifications in the VL domain, and at least one in the VL domain. A second immunoglobulin light chain having an amino acid modification of In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least two amino acid modifications in the VH domain, a first immunoglobulin light chain having at least two amino acid modifications in the VL domain, and at least A second immunoglobulin light chain having a single amino acid modification is included. In one embodiment, the LCCA design set comprises an immunoglobulin heavy chain having at least two amino acid modifications in the VH domain, a first immunoglobulin light chain having at least one amino acid modification in the VL domain, and at least A second immunoglobulin light chain having a single amino acid modification is included.

In one embodiment, the LCCA design sets are combined to provide a combination comprising two separate immunoglobulin heavy chains (H1 and H2) and two separate immunoglobulin light chains (L1 and L2), wherein When H1, H2, L1 and L2 are co-expressed, H1 preferentially pairs with L1 and H2 preferentially binds with L2.

In some embodiments, the amino acid modifications described herein are in the context of bispecific antibody constructs. For example, the design sets described herein can be incorporated into full-length immunoglobulin heavy chains such that the full-length heavy chains preferentially pair with immunoglobulin light chains. In some embodiments, the full-length immunoglobulin heavy chain contains amino acid modifications in the Fc region that promote dimerization, as described in the Examples.

Transferability of certain amino acid modifications identified herein to other antibodies:
Certain amino acid modifications to the immunoglobulin heavy and light chains identified above have been described for the D3H44 anti-tissue factor extracellular domain antibody, trastuzumab and cetuximab immunoglobulin heavy and light chains, but these amino acids Modifications can be transferred to other immunoglobulin heavy and light chains, resulting in similar patterns of preferential pairing of one immunoglobulin heavy chain with one of the two immunoglobulin light chains, the following aspects are considered and demonstrated herein (see, eg, the Examples, Figures and Tables).

The interface residues of VH:VL and CH1:CL at the interface of immunoglobulin heavy and light chains are relatively well conserved (Padlan et al., 1986, Mol. Immunol. 23(9):951-960). . This sequence conservation is the result of evolutionary restraint and increases the likelihood that a functionally active antigen binding domain will be formed upon combinatorial pairing of light and heavy chains. As a result of this sequence conservation, the sequence modifications in the specific example described above for D3H44, which lead to preferential pairing, can be transferred to other heavy and light chain paired heterodimers to provide nearly equivalent preferential pairing. , as this region exhibits high sequence conservation across antibodies. Furthermore, when sequence differences occur, they are usually distal from the CH1:CL interface. This is especially true in the CH1 and CL domains. However, there is some sequence variability in the antigen binding site with respect to CDR (complementarity determining region) loop residues (and length), especially CDR-H3. Thus, in one embodiment, the heterodimer pairs described herein are such that when the amino acid sequences of the antigen binding sites differ significantly from the D3H44 antibody, at least one heterodimer is one or more amino acid modifications in the VH and/or VL domains distal to the CDR loops. In another embodiment, the heterodimer pairs described herein are such that when the amino acid sequence of the antigen binding site is substantially the same as the D3H44 antibody, at least one heterodimer is one or Heterodimers containing multiple amino acid modifications in the VH and/or VL domains proximal or distal to the CDR loops are included.

In one embodiment, the amino acid modifications described herein can be transferred to the immunoglobulin heavy and light chains of human or humanized IgG1/κ-based antibodies. Non-limiting examples of such IgG1/kappa chains include ofatumumab (human) or trastuzumab, pertuzumab or bevacizumab (humanized).

In another embodiment, the amino acid modifications described herein can be transferred to the immunoglobulin heavy and light chains of antibodies utilizing commonly used VH and VL subgroups. Non-limiting examples of such antibodies include pertuzumab.

In one embodiment, the amino acid modifications described herein can be transferred to the immunoglobulin heavy and light chains of antibodies with near-germline frameworks. Examples of such antibodies include obinutuzumab.

In one embodiment, the amino acid modifications described herein are modified immunoglobulin heavy and light chains of antibodies with VH:VL interdomain angles that approximate those observed on average for heavy and light chain pairs. It is possible to migrate to a chain. Examples of this type of antibody include, but are not limited to, pertuzumab. In another embodiment, the amino acid modifications described herein can be transferred to the immunoglobulin heavy and light chains of antibodies with underlying CL and CH1 domains. Suitable examples of such antibodies include, but are not limited to trastuzumab.

In some embodiments, a particular subset of the amino acid modifications described herein are utilized in the variable domains of the antigen binding constructs provided above.

The examples, figures and tables demonstrate that amino acid modifications (e.g., in one or more Fab fragments containing variable and constant regions) can be transferred to other immunoglobulin heavy and light chains and have been demonstrated to result in similar patterns of preferential pairing of heavy chains with one of the two immunoglobulin light chains.

Preferential Pairing As described above, at least one heterodimer of the antigen-binding construct/heterodimer pair described herein comprises a heavy chain of one heterodimer, For example, H1 preferentially pairs with one of the light chains, e.g., L1, and not with the other light chain, L2, and the other heterodimeric heavy chain, H2, preferentially pairs with light chain, L2. The immunoglobulin heavy chain and/or the immunoglobulin light chain may contain one or more amino acid modifications so that it pairs positively and does not pair with light chain L1. In other words, the desired preferential pairings are considered between H1 and L1 and between H2 and L2. For example, the preferential pairing of H1 and L1 is such that when H1 is combined with L1 and L2, the yield of the H1-L1 heterodimer is reduced relative to the H2/L2 pair without one or more amino acid modifications. , for each pairing of the corresponding H1/L1 pair, is considered to occur when the yield of the mismatched H1-L2 heterodimer is greater than. Similarly, the preferential pairing of H2 and L2 is such that when H2 is combined with L1 and L2, the yield of the H2-L2 heterodimer is reduced to that of the H2/L2 pair without one or more amino acid modifications. is considered to occur when the yield of the mismatched H2-L1 heterodimer is greater than for each pairing of the corresponding H1/L1 pair to . In this context, heterodimers comprising H1 and L1 (H1-L1) or H2 and L2 (H2-L2) are preferentially paired, exactly paired, absolute pairs or desired are referred to herein as heterodimers of, whereas heterodimers comprising H1 and L2 (H1-L2) or H2 and L1 (H2-L1) are referred to herein as mismatched heterodimers. called in the book. The set of mutations corresponding to the two heavy and two light chains intended to achieve selective H1-L1 and H2-L2 pairing is called the design set.

Thus, in one embodiment, when a heterodimer of one immunoglobulin heavy chain is co-expressed with two immunoglobulin light chains, the relative yield of the desired heterodimer is greater than 55%. In another embodiment, when one immunoglobulin heavy chain in the heterodimer is co-expressed with two immunoglobulin light chains, the relative yield of the desired heterodimer is greater than 60%. In another embodiment, when one immunoglobulin heavy chain in the heterodimer is co-expressed with two immunoglobulin light chains, the relative yield of the desired heterodimer is greater than 70%. In another embodiment, when one immunoglobulin heavy chain in the heterodimer is co-expressed with two immunoglobulin light chains, the relative yield of the desired heterodimer is greater than 80%. In another embodiment, when one immunoglobulin heavy chain in the heterodimer is co-expressed with two immunoglobulin light chains, the relative yield of the desired heterodimer is greater than 90%. In another embodiment, when one immunoglobulin heavy chain in the heterodimer is co-expressed with two immunoglobulin light chains, the relative yield of the desired heterodimer is greater than 95%.

In the example above, the preferential pairing of H1-L1 is such that the desired amount of H1-L1 heterodimer is less than the amount of mismatched H1-L2 heterodimer when H1 is co-expressed with L1 and L2. is considered to occur when greater than the amount. Similarly, the preferential pairing of H2-L2 is such that the amount of the desired H2-L2 heterodimer is higher than that of the mismatched H2-L2 heterodimer when H2 is co-expressed with L1 and L2. is considered to occur when is also large. Thus, in one embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the desired ratio of heterodimers to mismatched heterodimers is: greater than 1.25:1. In one embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the desired ratio of heterodimers to mismatched heterodimers is 1. Greater than 5:1. In another embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the ratio of desired heterodimers to mismatched heterodimers is 2: greater than 1. In another embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the ratio of desired heterodimers to mismatched heterodimers is 3: greater than 1. In another embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the ratio of desired heterodimers to mismatched heterodimers is 5: greater than 1. In another embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the ratio of desired heterodimers to mismatched heterodimers is 10: greater than 1. In another embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the ratio of desired heterodimers to mismatched heterodimers is 25: greater than 1. In another embodiment, when one immunoglobulin heavy chain in a heterodimer is co-expressed with two immunoglobulin light chains, the desired heterodimer to mismatched heterodimer ratio is 50:1. exceed.

In some embodiments, the heterodimers described herein preferentially pair to form bispecific antibodies. In some embodiments, the construct comprises heterodimers that preferentially pair to form a bispecific antibody selected from D3H44/trastuzumab, D3H44/cetuximab and trastuzumab/cetuximab. In some embodiments, bispecific antibodies comprise amino acid modifications set forth in Tables 28a-28c.

In some embodiments, two full-length heavy chain constructs are co-expressed with two unique light chain constructs to generate ten possible antibody species: H1-L1:H1-L1, H1-L2:H1 -L2, H1-L1: H1-L2, H2-L1: H2-L1, H2-L2: H2-L2, H2-L1: H2-L2, H1-L1: H2-L1, H1-L2: H2-L2 , H1-L2:H2-L1 and H1-L1:H2-L2. The H1-L1:H2-L2 species are considered to be precisely matched bispecific antibody species. In some embodiments, the DNA ratio is selected to yield the maximum amount of correctly matched bispecific antibody species. For example, in some embodiments the ratio of H1:H2:L1:L2 is 15:15:53:17. In some embodiments, the ratio of H1:H2:L1:L2 is 15:15:17:53.

In some embodiments, the percentage of correctly matched bispecific species compared to all species is at least 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75% %, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100% (e.g., Tables 29a-29c and 30a-30c). In some embodiments, the rate of correctly matched bispecific species co-expresses the corresponding wild-type H1, H2, L1 and L2 without the amino acid modifications listed in Tables 28a-28c. greater than the rate of correctly matched bispecific antibody species obtained by In some embodiments, the rate of correctly matched bispecific species co-expresses the corresponding wild-type H1, H2, L1 and L2 without the amino acid modifications listed in Tables 28a-28c. at least 5%, 10%, 15%, 20%, 30%, 40%, 50%, 60%, 70% compared to the percentage of correctly matched bispecific antibody species obtained by or increased by 75%.

Thermostability of Heterodimers In addition to promoting preferential pairing, amino acid substitutions were chosen such that the mutations did not destabilize the Fab heterodimer. Thus, in most cases, Fab heterodimer stability measurements were very close to wild-type Fab (eg, within 3°C of wild-type Fab).

Thus, in some embodiments, each heterodimer of a heterodimer pair described herein comprises the same immunoglobulin heavy and light chains but is described herein It has thermal stability comparable to heterodimers with no amino acid modifications in the CH1, CL, VH or VL domains. In one embodiment, thermal stability is determined by measuring the melting temperature or Tm. Thus, in one embodiment, the thermostability of the heterodimers described herein comprises the same immunoglobulin heavy and light chains but CH1, CL, VH or Within about 10°C of the heterodimer with no amino acid modifications in the VL domain. Thus, in one embodiment, the thermostability of the heterodimers described herein comprise the same immunoglobulin heavy and light chains but CH1, CL, VH described herein. or within about 5°C of a heterodimer with no amino acid modification in the VL domain. In another embodiment, the thermostable heterodimers described herein are CH1, CL, VH or VL comprising the same immunoglobulin heavy and light chains but described herein. Within about 3°C of the heterodimer with no amino acid modifications in the domains. In another embodiment, the thermostable heterodimers described herein are CH1, CL, VH or VL comprising the same immunoglobulin heavy and light chains but described herein. Within about 2°C of the heterodimer with no amino acid modifications in the domains. In another embodiment, the thermostable heterodimers described herein are CH1, CL, VH or VL comprising the same immunoglobulin heavy and light chains but described herein. Within about 1.5°C of the heterodimer with no amino acid modifications in the domains. In another embodiment, the thermostable heterodimers described herein are CH1, CL, VH or VL comprising the same immunoglobulin heavy and light chains but described herein. Within about 1°C of the heterodimer with no amino acid modifications in the domains. In another embodiment, the thermostable heterodimers described herein are CH1, CL, VH or VL comprising the same immunoglobulin heavy and light chains but described herein. Within about 0.5°C of the heterodimer with no amino acid modifications in the domains. In another embodiment, the thermostable heterodimers described herein are CH1, CL, VH or VL comprising the same immunoglobulin heavy and light chains but described herein. Within about 0.25° C. of the heterodimer with no amino acid modifications in the domains.

Further, in some embodiments, the thermostability of the heterodimers described herein comprise CH1, CL, CH1, CL, as described herein, but comprising the same immunoglobulin heavy and light chains. Surprisingly improved (ie increased) compared to heterodimers with no amino acid modifications in the VH or VL domains. Thus, in one embodiment, the thermostability of the heterodimers described herein comprise the same immunoglobulin heavy and light chains but CH1, CL, VH or about 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8 compared to heterodimers without amino acid modifications in the VL domains , 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2.0, 2 .1, 2.2, 2.3, 2.4, 2.5, 3.0, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7 , 3.8, 3.9, 4.0, 4.5, 5.0° C., or more.

Affinity of Heterodimers for Antigens In one embodiment, each heterodimer of the heterodimer pair comprises the same immunoglobulin heavy and light chains but is described herein as CH1 , CL, VH or VL domains with the same or comparable affinities to their respective antigens as heterodimers without amino acid modifications. In one embodiment, the heterodimers of the heterodimer pair comprise the same immunoglobulin heavy and light chains but have amino acid modifications in the CH1, CL, VH or VL domains described herein. Each antigen has an affinity within about 50-fold of that of a non-heterodimer. In one embodiment, the heterodimers of the heterodimer pair comprise the same immunoglobulin heavy and light chains but have amino acid modifications in the CH1, CL, VH or VL domains described herein. Each antigen has an affinity within about 25-fold of that of the heterodimer. In one embodiment, the heterodimers of the heterodimer pair comprise the same immunoglobulin heavy and light chains but have amino acid modifications in the CH1, CL, VH or VL domains described herein. have affinities for their respective antigens that are within about 10-fold of the heterodimers that do not. In another embodiment, the heterodimer of the heterodimer pair comprises the same immunoglobulin heavy and light chains but has amino acid modifications in the CH1, CL, VH or VL domains described herein. have affinities for their respective antigens that are within about 5-fold of the heterodimers that do not. In another embodiment, the heterodimer of the heterodimer pair comprises the same immunoglobulin heavy and light chains but has amino acid modifications in the CH1, CL, VH or VL domains described herein. have affinities for each antigen within about 2.5-fold of the heterodimer that does not exist. In another embodiment, the heterodimer of the heterodimer pair comprises the same immunoglobulin heavy and light chains but has amino acid modifications in the CH1, CL, VH or VL domains described herein. have affinities for their respective antigens that are within about 2-fold of the heterodimers that do not. In another embodiment, the heterodimer of the heterodimer pair comprises the same immunoglobulin heavy and light chains but has amino acid modifications in the CH1, CL, VH or VL domains described herein. have affinities for each antigen within about 1.5-fold of the heterodimer that does not exist. In another embodiment, the heterodimer of the heterodimer pair comprises the same immunoglobulin heavy and light chains but has amino acid modifications in the CH1, CL, VH or VL domains described herein. have approximately the same affinities for their respective antigens as heterodimers that do not.

Additional Selective Modifications In one embodiment, the heterodimeric pairs of immunoglobulin heavy and light chains described herein are combined so that the covalent bond prevents preferential pairing of the heavy and light chains. or further modified (i.e., by covalent attachment of different types of molecules) so as not to affect the ability of the heterodimer to bind antigen or affect the stability of the heterodimer can do. Such modifications include, for example, by glycosylation, acetylation, pegylation, phosphorylation, amidation, derivatization with known protecting/blocking groups, proteolytic cleavage, linkage to cellular ligands or other proteins, and the like. including, but not limited to, Any of a number of chemical modifications can be performed by known techniques including, but not limited to, specific chemical cleavage, acetylation, formylation, metabolic synthesis of tunicamycin, and the like.

In another embodiment, the heterodimeric paired immunoglobulin heavy and light chains described herein are incorporated into therapeutic agents or drug moieties that modify a given biological response (either directly or indirectly). can be conjugated). The therapeutic agent or drug moiety should not be construed as limited to classical chemotherapeutic agents. For example, the drug moiety can be a protein or polypeptide that possesses a desired biological activity. Such proteins include, for example, abrin, ricin A, Onconase (or another cytotoxic RNase), toxins such as pseudomonas exotoxin, cholera toxin or diphtheria toxin; tumor necrosis factor, alpha interferon, beta interferon, nerve growth factor, platelet-derived growth factor, tissue plasminogen activator, apoptotic agents such as TNF-alpha, TNF-beta, AIM I (see International Publication No. WO97/33899), AIM II (see International Publication No. WO97/34911), Fas ligand (Takahashi et al., 1994, J. Immunol., 6:1567) and VEGI (see International Publication No. WO99/23105), thrombogenic agents or proteins such as anti-angiogenic agents, e.g., angiostatin or endostatin; biological response modifiers such as leukin-6 (“IL-6”), granulocyte-macrophage colony-stimulating factor (“GM-CSF”) and granulocyte-colony stimulating factor (“G-CSF”)), or growth factors (eg, growth hormone (“GH”)).

Furthermore, in an alternative embodiment, the antibody can be conjugated to therapeutic moieties such as radioactive materials or macrocyclic chelators useful for conjugating radioactive metal ions (for examples of radioactive materials see , see above). In certain embodiments, the macrocyclic chelator is 1,4,7,10-tetraazacyclododecane-N,N′,N″,N″-tetraacetic acid ( DOTA). Such linker molecules are generally known in the art and are described in Denardo et al., 1998, Clin Cancer Res. 4:2483; Peterson et al., 1999, Bioconjug. Chem. 10:553; and Zimmerman et al., 1999, Nucl. Med. Biol. 26:943.

In some embodiments, heterodimeric immunoglobulin heavy and light chains are expressed as fusion proteins that include tags to facilitate purification and/or testing and the like. As referred to herein, a "tag" is any added series of amino acids provided to a protein, either at the C-terminus, N-terminus or internally, that contribute to the identification or purification of the protein. Suitable tags include albumin binding domain (ABD), His-tag, FLAG-tag, glutathione-s-transferase, hemagglutinin (HA) and maltose-binding protein, among others known to those skilled in the art to be useful for purification and/or testing. This includes, but is not limited to, tags that are Such tagged proteins can also be modified to contain cleavage sites such as thrombin, enterokinase or factor X cleavage sites for easy removal of the tag before, during or after purification.

In some embodiments, one of the lower cysteine residues of the Fab domain in the light chain (position 214, Kabat numbering) and heavy chain (position 233, Kabat numbering) that forms an interchain disulfide bond, or Multiple modifications can be serine, or alanine, or non-cysteine, or another amino acid.

It is contemplated that additional amino acid modifications can be made to the immunoglobulin heavy chain to increase the level of preferential pairing and/or thermal stability of heterodimer pairs. For example, additional amino acid modifications can be made to the immunoglobulin heavy chain Fc domain to induce preferential pairing of heterodimeric pairs over homodimeric pairs. Such amino acid modifications are known in the art and include, for example, those described in US Patent Publication No. 2012/0149876. Alternatively, preferential pairs for heterodimer pairs over homodimer pairs, such as, for example, "knob-into-hole", charged residues with ionic interactions and strand exchange engineered domain (SEED) technology Alternative strategies that lead to convergence can also be used. The latter strategy is described in the art and reviewed in Klein et al., supra. Further discussion of Fc domains follows below.

Fc Domain In embodiments in which the antigen-binding polypeptide construct comprises a full-length immunoglobulin heavy chain, the construct comprises an Fc. In some embodiments, the Fc comprises at least one or two CH3 domain sequences. In some embodiments, when the antigen-binding polypeptide construct comprises a heterodimer comprising only the Fab region of the heavy chain, the Fc, with or without one or more linkers, is bound to the first Binds the heterodimer and/or the second heterodimer. In some embodiments, the Fc is human Fc. In some embodiments, Fc is IgG or IgG1 Fc. In some embodiments, the Fc is a heterodimeric Fc. In some embodiments, the Fc comprises at least one or two CH2 domain sequences.

In some embodiments, the Fc comprises one or more modifications in at least one of the _CH3 domain sequences. In some embodiments, the Fc comprises one or more modifications in at least one of the _CH2 domain sequences. In some embodiments, Fc is a single polypeptide. In some embodiments, Fc is multiple polypeptides, eg, two polypeptides.

In some embodiments, the Fc comprises one or more modifications in at least one of the _CH3 sequences. In some embodiments, the Fc comprises one or more modifications in at least one of the _CH2 sequences. In some embodiments, Fc is a single polypeptide. In some embodiments, Fc is multiple polypeptides, eg, two polypeptides.

In some embodiments, the Fc is an Fc as described in patent applications PCT/CA2011/001238 filed November 4, 2011 and PCT/CA2012/050780 filed November 2, 2012, wherein The entire disclosure of each is incorporated herein by reference in its entirety for all purposes.

In some aspects, the constructs described herein comprise a heterodimeric Fc comprising an asymmetrically modified CH3 domain. A heterodimeric Fc can comprise two heavy chain constant domain polypeptides: a first heavy chain polypeptide and a second heavy chain polypeptide, which can be used interchangeably, provided that The proviso is that Fc comprises one first heavy chain polypeptide and one second heavy chain polypeptide. Generally, the first heavy chain polypeptide comprises a first CH3 sequence and the second heavy chain polypeptide comprises a second CH3 sequence.

Two CH3 sequences containing one or more amino acid modifications introduced asymmetrically generally result in a heterodimeric rather than homodimeric Fc when the two CH3 sequences dimerize. As used herein, an "asymmetric amino acid modification" means that an amino acid at a particular position in a first CH3 sequence differs from an amino acid in a second CH3 sequence at the same position, such that the amino acid in the first and second CH3 sequences refers to any modification that preferentially pairs to form a heterodimer rather than a homodimer. This heterodimerization is a modification of only one of the two amino acids at the same corresponding amino acid position in each sequence, or each sequence at the same corresponding position in each of the first and second CH3 sequences. can be the result of modification of both amino acids. The first and second CH3 sequences of the heterodimeric Fc can contain one or more than one asymmetric amino acid modification.

Table X provides the amino acid sequence of the human IgG1 Fc sequence corresponding to amino acids 231-447 of the full-length human IgG1 heavy chain. The CH3 sequence includes amino acids 341-447 of a full-length human IgG1 heavy chain.

Typically, an Fc can comprise two contiguous heavy chain sequences (A and B) that are capable of dimerizing. In some embodiments, one or both sequences of the Fc are at one or both of the following positions, using EU numbering: L351, F405, Y407, T366, K392, T394, T350, S400 and/or N390. Contains multiple mutations or modifications. In some embodiments, the Fc comprises a mutated sequence shown in Table X. In some embodiments, the Fc comprises variants 1A-B mutations. In some embodiments, the Fc comprises variants 2A-B mutations. In some embodiments, the Fc comprises variants 3A-B mutations. In some embodiments, the Fc comprises variants 4A-B mutations. In some embodiments, the Fc comprises variants 5A-B mutations.

(Table X)

The first and second CH3 sequences can contain the amino acid mutations described herein with reference to amino acids 231-447 of a full-length human IgG1 heavy chain. In one embodiment, the heterodimeric Fc comprises a modified CH3 domain, wherein the first CH3 sequence has amino acid modifications at positions F405 and Y407 and the second CH3 sequence has amino acid modifications at position T394. In one embodiment, the heterodimeric Fc has one or more amino acid modifications selected from L351Y, F405A and Y407V in the first CH3 sequence and T366L, T366I, K392L in the second CH3 sequence , K392M and T394W.

In one embodiment, the heterodimeric Fc is a first CH3 sequence having amino acid modifications at positions L351, F405 and Y407 and a second CH3 sequence having amino acid modifications at positions T366, K392 and T394. , one of the first or second CH3 sequences further comprises an amino acid modification at position Q347 and the other CH3 sequence further comprises an amino acid modification at position K360. In another embodiment, the heterodimeric Fc is wherein a first CH3 sequence has amino acid modifications at positions L351, F405 and Y407 and a second CH3 sequence has amino acid modifications at positions T366, K392 and T394. , one of the first or second CH3 sequences further comprises an amino acid modification at position Q347 and the other CH3 sequence further comprises an amino acid modification at position K360, and one or both of said CH3 sequences further comprise the amino acid modification T350V Further comprising modified CH3 domains.

In one embodiment, the heterodimeric Fc has amino acid modifications at positions L351, F405 and Y407 in the first CH3 sequence and amino acid modifications at positions T366, K392 and T394 in the second CH3 sequence. , one of the first or second CH3 sequences further comprises a D399R or D399K amino acid modification, and the other CH3 sequence comprises one or more of T411E, T411D, K409E, K409D, K392E and K392D. Contains the CH3 domain. In another embodiment, the heterodimeric Fc is wherein a first CH3 sequence has amino acid modifications at positions L351, F405 and Y407 and a second CH3 sequence has amino acid modifications at positions T366, K392 and T394. , one of the first or second CH3 sequences further comprises a D399R or D399K amino acid modification, and the other CH3 sequence comprises one or more of T411E, T411D, K409E, K409D, K392E and K392D; One or both of the CH3 sequences comprise modified CH3 domains further comprising the amino acid modification T350V.

In one embodiment, the heterodimeric Fc is a first CH3 sequence having amino acid modifications at positions L351, F405 and Y407 and a second CH3 sequence having amino acid modifications at positions T366, K392 and T394. , wherein one or both of said CH3 sequences further comprises a T350V amino acid modification.

In one embodiment, the heterodimeric Fc comprises a modified CH3 domain comprising the following amino acid modifications, wherein "A" comprises an amino acid modification to the first CH3 sequence and "B" is A second amino acid modification to the CH3 sequence is included. Ａ：Ｌ３５１Ｙ＿Ｆ４０５Ａ＿Ｙ４０７Ｖ、Ｂ：Ｔ３６６Ｌ＿Ｋ３９２Ｍ＿Ｔ３９４Ｗ、Ａ：Ｌ３５１Ｙ＿Ｆ４０５Ａ＿Ｙ４０７Ｖ、Ｂ：Ｔ３６６Ｌ＿Ｋ３９２Ｌ＿Ｔ３９４Ｗ、Ａ：Ｔ３５０Ｖ＿Ｌ３５１Ｙ＿Ｆ４０５Ａ＿Ｙ４０７Ｖ、Ｂ：Ｔ３５０Ｖ＿Ｔ３６６Ｌ＿Ｋ３９２Ｌ＿Ｔ３９４Ｗ、Ａ：Ｔ３５０Ｖ＿Ｌ３５１Ｙ＿Ｆ４０５Ａ＿Ｙ４０７Ｖ、Ｂ：Ｔ３５０Ｖ＿Ｔ３６６Ｌ＿Ｋ３９２Ｍ＿Ｔ３９４Ｗ、Ａ：Ｔ３５０Ｖ＿Ｌ３５１Ｙ＿Ｓ４００Ｅ＿Ｆ４０５Ａ＿Ｙ４０７Ｖ及び／またはＢ：Ｔ３５０Ｖ＿Ｔ３６６Ｌ＿Ｎ３９０Ｒ＿Ｋ３９２Ｍ＿Ｔ３９４Ｗ。

One or more asymmetric amino acid modifications can facilitate the formation of heterodimeric Fc, in which the heterodimeric CH3 domain has a stability comparable to that of the wild-type homodimeric CH3 domain. In certain embodiments, one or more asymmetric amino acid modifications promote formation of heterodimeric Fc domains, wherein the heterodimeric Fc domains have a stability comparable to wild-type homodimeric Fc domains. In certain embodiments, the one or more asymmetric amino acid modifications are such that the heterodimeric Fc domain has stability as observed by melting temperature (Tm) in differential scanning calorimetry studies, and the melting temperature corresponds to Facilitating the formation of heterodimeric Fc domains that are within 4°C of the melting temperature observed in the symmetric wild-type homodimeric Fc domain. In some embodiments, the Fc comprises one or more modifications in at least one of the _CH3 sequences that promote formation of a heterodimeric Fc with stability comparable to a wild-type homodimeric Fc. .

In one embodiment, CH3 domain stability can be assessed by measuring the melting temperature of the CH3 domain, eg, by differential scanning calorimetry (DSC). Accordingly, in further embodiments, the CH3 domain has a melting temperature of about 68° C. or higher. In another embodiment, the CH3 domain has a melting temperature of greater than or equal to about 70°C. In another embodiment, the CH3 domain has a melting temperature of greater than or equal to about 72°C. In another embodiment, the CH3 domain has a melting temperature of greater than or equal to about 73°C. In another embodiment, the CH3 domain has a melting temperature of greater than or equal to about 75°C. In another embodiment, the CH3 domain has a melting temperature of greater than or equal to about 78°C. In some embodiments, the dimerization C _H3 sequence is about It has a melting temperature (Tm) of 83, 84 or 85°C or higher.

In some embodiments, heterodimeric Fc comprising modified CH3 sequences can be formed with at least about 75% purity compared to homodimeric Fc in the expression product. In another embodiment, the heterodimeric Fc is formed with a purity greater than about 80%. In another embodiment, the heterodimeric Fc is formed with a purity greater than about 85%. In another embodiment, the heterodimeric Fc is formed with a purity greater than about 90%. In another embodiment, the heterodimeric Fc is formed with a purity greater than about 95%. In another embodiment, the heterodimeric Fc is formed with a purity greater than about 97%. In some embodiments, the Fc, when expressed, is about , heterodimers formed with purities greater than 93, 94, 95, 96, 97, 98 or 99%. In some embodiments, the Fc is about 75,76,77,78,79,80,81,82,83,84,85,86,87,88,89 when expressed through a single cell , 90, 91, 92, 93, 94, 95, 96, 97, 98 or greater than 99% purity.

Additional methods for modifying monomeric Fc polypeptides to promote formation of heterodimeric Fc are described in International Patent Publication No. WO 96/027011 (Nobu into Hall), Gunasekaran et al. (Gunasekaran K. et al. (2010) Ｊ Ｂｉｏｌ Ｃｈｅｍ．２８５，１９６３７－４６，ｅｌｅｃｔｒｏｓｔａｔｉｃ ｄｅｓｉｇｎ ｔｏ ａｃｈｉｅｖｅ ｓｅｌｅｃｔｉｖｅ ｈｅｔｅｒｏｄｉｍｅｒｉｚａｔｉｏｎ）、Ｄａｖｉｓら（Ｄａｖｉｓ，ＪＨ．ら（２０１０）Ｐｒｏｔ Ｅｎｇ Ｄｅｓ Ｓｅｌ；２３（４）：１９５－２０２，ｓｔｒａｎｄ ｅｘｃｈａｎｇｅ ｅｎｇｉｎｅｅｒｅｄ ｄｏｍａｉｎ（ＳＥＥＤ） technology) and Labrijn et al. [Efficient generation of stable bispecific IgG1 by controlled Fab-arm exchange. Labrijn AF, Meesters JI, de Goeij BE, van den Bremer ET, Neijssen J, van Kampen MD, Strumane K, Verploegen S, Kundu A, Gramer MJ, van Berkel PH, van de Winkel P.W. Proc Natl Acad Sci USA. 2013 Mar 26;110(13):5145-50. In some embodiments, the isolated constructs described herein are stable and manufacturable compared to antigen-binding constructs that bind antigen, as well as antigen-binding constructs that do not contain the same Fc polypeptide. It includes dimeric Fc polypeptide constructs with excellent biophysical characteristics such as ease. Numerous mutations in the Fc heavy chain sequence are known in the art to selectively alter the affinity of an antibody Fc for different Fc gamma receptors. In some embodiments, the Fc comprises one or more modifications that facilitate selective binding of Fc gamma receptors.

The CH2 domain is amino acids 231-340 of the sequence shown in Table X. Exemplary mutations are listed below.

S298A/E333A/K334A, S298A/E333A/K334A/K326A (Lu Y, Vernes JM, Chiang N, et al. J Immunol Methods. 2011 Feb 28; 365(1-2):132-41); F243L/R292P/Y300L V305I/P396L, F243L/R292P/Y300L/L235V/P396L (Stavenhagen JB, Gorlatov S, Tuaillon N, et al. Cancer Res. 2007 Sep 15; 67(18):8882-90; Nordstrom, Sorlatov JL, Gorlatov JL, et al. Breast Cancer Res. 2011 Nov 30; 13(6):R123); F243L (Stewart R, Thom G, Levens M, et al. Protein Eng Des Sel. 2011 Sep; 24(9):671-8.), S298A/E333A /K334A (Shields RL, Namenuk AK, Hong K, et al. J Biol Chem. 2001 Mar 2; 276(9):6591-604); Proc Natl Acad Sci USA.2006 Mar 14;103(11):4005-10); 3926-33); /I332E, S239E/S267E/H268D, L234F/S267E/N325L, G237F/V266L/S267D, and WO2011/120134 and WO2011/120135, which are incorporated herein by reference. Other mutations that have been Therapeutic Antibody Engineering (William R. Strohl and Lila M. Strohl, Woodhead Publishing series in Biomedicine No 11, ISBN 1 907568 37 enumeration 9, Oct 2012) lists the mutation on page 283.

In some embodiments, the CH2 domain comprises one or more asymmetric amino acid modifications. In some embodiments, the CH2 domain comprises one or more asymmetric amino acid modifications that promote selective binding of FcγRs. In some embodiments, the CH2 domain allows isolation and purification of the isolated constructs described herein.

FcRn Binding and PK Parameters As is known in the art, binding to FcRn recycles endocytosed antibodies from endosomes into the blood stream (Raghavan et al., 1996, Annu Rev Cell Dev Biol 12:181). -220; Ghetie et al., 2000, Annu Rev Immunol 18:739-766). This process, coupled with interference with renal filtration due to the large size of the full-length molecule, results in favorable antibody serum half-lives ranging from 1-3 weeks. Fc binding to FcRn also plays a major role in antibody transport. Thus, in one embodiment, constructs of the invention are capable of binding to FcRn.

Additional Modifications to Improve Effector Function In some embodiments, the constructs described herein can be modified to improve effector function. Such modifications are known in the art and include afucosylation, or antibody directed to activated receptors, primarily FCGR3a for ADCC and C1q for CDC. includes engineering the affinity of the Fc portion of . Table Y below summarizes the various designs reported in the literature for effector function manipulation.

(Table Y)

Thus, in one embodiment, the constructs described herein comprise a dimeric Fc comprising one or more amino acid modifications shown in the table above that confer improved effector function. can be done. In another embodiment, the construct can be defucosylated to improve effector function.

Linkers Constructs described herein can comprise one or more heterodimers described herein operably linked to an Fc described herein . In some embodiments, the Fc is attached to one or more heterodimers with and without one or more linkers. In some embodiments, the Fc directly binds one or more heterodimers. In some embodiments, the Fc is attached to one or more heterodimers by one or more linkers. In some embodiments, the Fc is attached to each heterodimeric heavy chain by a linker.

In some embodiments, one or more linkers are one or more polypeptide linkers. In some embodiments, one or more linkers comprise one or more antibody hinge regions. In some embodiments, one or more linkers comprise one or more IgG1 hinge regions.

Methods of Preparing Dimer Pairs As described above, the heterodimer pairs described herein can comprise a first heterodimer and a second heterodimer. , each heterodimer comprising an immunoglobulin heavy chain or fragment thereof comprising at least the VH and CH1 domains, and an immunoglobulin light chain comprising the VL and CL domains. Heterodimeric immunoglobulin heavy and light chains can be readily prepared using recombinant DNA techniques known in the art.例えば、Ｓａｍｂｒｏｏｋ及びＲｕｓｓｅｌｌ、Ｍｏｌｅｃｕｌａｒ Ｃｌｏｎｉｎｇ：Ａ Ｌａｂｏｒａｔｏｒｙ Ｍａｎｕａｌ（Ｃｏｌｄ Ｓｐｒｉｎｇ Ｈａｒｂｏｒ Ｌａｂｏｒａｔｏｒｙ Ｐｒｅｓｓ、Ｃｏｌｄ Ｓｐｒｉｎｇ Ｈａｒｂｏｒ，Ｎ．Ｙ．、第３版、２００１）；Ｓａｍｂｒｏｏｋら、Ｍｏｌｅｃｕｌａｒ Ｃｌｏｎｉｎｇ：Ａ Ｌａｂｏｒａｔｏｒｙ Ｍａｎｕａｌ（Ｃｏｌｄ Ｓｐｒｉｎｇ Ｈａｒｂｏｒ Ｌａｂｏｒａｔｏｒｙ Ｐｒｅｓｓ , Cold Spring Harbor, NY, 2nd ed. 1989); Short Protocols in Molecular Biology (Ausubel et al., John Wiley and Sons, New York, 4th ed. 1999); and Applications of Recombinant DNA (ASM Press, Washington, D.C., 2nd ed. 1998), recombinant nucleic acid methods, nucleic acid synthesis, cell culture, transgene integration. and for recombinant protein expression. Alternatively, the heterodimers and heterodimer pairs described herein can be chemically synthesized.

The immunoglobulin heavy and light chain nucleic acid and amino acid sequences of antibodies from which heterodimers are induced are known in the art or can be readily determined through the use of nucleic acid and/or protein sequencing methods. can. Methods for genetically fusing the tags described herein to immunoglobulin heavy and/or light chains are known in the art and some are described below and in the Examples.

For example, methods of expressing and co-expressing immunoglobulin heavy and light chains in a host are well known in the art. Additionally, methods of tagging heavy and/or light chains using recombinant DNA technology are well known in the art. Suitable expression vectors and hosts for expression of heavy and light chains are also well known in the art and are described below.

Bispecific antibody production methods that do not rely solely on the use of single clones or transient cell lines expressing all four chains are known in the art (Gramer, et al. (2013) mAbs 5 , 962; Strop et al. (2012) J Mol Biol 420, 204). These methods rely on post-production arm exchange under redox conditions of two pairs of light and heavy chains involved in the formation of bispecific antibodies (redox production). In this approach, H1:L1 and H2:L2 pairs can be expressed in two different cell lines to produce two Fab arms independently. The two Fab arms are then mixed under selective redox conditions to achieve reassociation of the two unique heavy chains H1 and H2, resulting in dual specificity involving the L1:H1:H2:L2 chains. form antibodies; One may consider using the designed libraries/datasets described herein in the production of double antibodies using redox production methods or modifications thereof.

In certain embodiments, cell-free protein expression systems are utilized to co-express polypeptides (eg, heavy and light chains of polypeptides) without the use of living cells. Rather, all the components necessary to transcribe DNA into RNA and to transcribe RNA into proteins (e.g., ribosomes, tRNAs, enzymes, cofactors, amino acids) are in solution used in vitro. provided. In certain embodiments, in vitro expression requires a reaction solution containing (1) genetic templates (mRNA or DNA) encoding heavy and light chain polypeptides, and (2) the necessary transcriptional and translational molecular machinery. and In certain embodiments, the cell extract contains components of the reaction solution, e.g., RNA polymerase for mRNA transcription, ribosomes for polypeptide translation, tRNA, amino acids, enzymatic cofactors, energy sources and precise Substantially supplies the essential cellular components for protein folding. Cell-free protein expression systems can be prepared using lysates obtained from bacterial cells, yeast cells, insect cells, plant cells, mammalian cells, human cells or combinations thereof. Such cell lysates can provide the precise composition and proportions of enzymes and components required for translation. In some embodiments, the cell membrane is removed, leaving only the cytoplasmic and organelle components of the cell.

Several cell-free protein expression systems are described by Carlson et al. (2012) Biotechnol. Adv. 30:1185-1194, and are known in the art. For example, cell-free protein expression systems are available that are based on prokaryotic or eukaryotic cells. Examples of prokaryotic cell-free expression systems include those in E. coli. Examples of eukaryotic cell-free expression systems are available, eg, those based on rabbit reticulocytes, wheat germ and insect cell extracts. Such prokaryotic or eukaryotic protein expression systems are commercially available from companies such as Roche, Invitrogen, Qiagen and Novagen. Those skilled in the art can readily select suitable cell-free protein expression systems to produce polypeptides (eg, heavy and light chain polypeptides) capable of pairing with each other. Additionally, cell-free protein expression systems can be supplemented with chaperones (eg, BiP) and isomerases (eg, disulfide isomerase) to improve the efficiency of IgG folding.

In some embodiments, cell-free expression systems are utilized to co-express heavy and light chain polypeptides from DNA templates (transcription and translation) or mRNA templates (translation only).

Vectors and Host Cells Recombinant expression of heavy and light chains requires the construction of expression vectors containing polynucleotides encoding heavy or light chains (eg, antibodies or fusion proteins). Once a polynucleotide encoding a heavy or light chain is obtained, vectors for producing the heavy or light chain can be produced by recombinant DNA technology using techniques well known in the art. Thus, methods of expressing a polynucleotide containing a heavy or light chain encoding nucleotide sequence to prepare a protein are described herein. Methods which are well known to those skilled in the art can be used to construct expression vectors containing heavy or light chain coding sequences and appropriate transcriptional and translational control signals. These methods include, for example, in vitro recombinant DNA techniques, synthetic techniques and in vivo genetic recombination. Accordingly, the invention provides replicable vectors comprising a nucleotide sequence encoding a heavy or light chain operably linked to a promoter.

The expression vector is transferred into host cells by conventional techniques, and the transfected cells are then cultured by conventional techniques to produce the modified heavy or light chains used in the methods of the invention. In certain embodiments, the heavy and light chains used in the methods are co-expressed in the host cell for expression of the whole immunoglobulin molecule, detailed below.

A variety of host-expression vector systems may be utilized to express the modified heavy and light chains. Such host-expression systems represent vehicles by which the coding sequences of interest can be produced and subsequently purified, and, when transformed or transfected with appropriate nucleotide coding sequences, produce modified heavy and light chains thereof. Cells capable of expression in situ are also represented. These include microorganisms such as bacteria (e.g. E. coli and B. sutilis) transformed with recombinant bacteriophage DNA, plasmid DNA or cosmid DNA expression vectors containing modified heavy and light chain coding sequences; Yeast transformed with recombinant yeast expression vectors containing modified heavy and light chain coding sequences (e.g., Saccharomyces, Pichia); recombinant containing modified heavy and light chain coding sequences; Insect cell lines infected with viral expression vectors (e.g., baculovirus); infected with recombinant viral expression vectors (e.g., cauliflower mosaic virus, CaMV; tobacco saic virus, TMV) or modified heavy and light chain encoding. plant cell lines transformed with recombinant plasmid expression vectors (e.g. Ti plasmids) containing the sequences; or promoters derived from mammalian cell genomes (e.g. metallothionein promoters) or mammalian virus-derived promoters (e.g. For example, mammalian cell lines (e.g., COS, CHO, BHK, HEK-293, NSO and 3T3 cells) with recombinant expression constructs containing the adenovirus late promoter; vaccinia virus 7.5K promoter); is not limited to In certain embodiments, bacterial cells such as Escherichia coli or eukaryotic cells are used to express modified heavy and light chains that are recombinant antibody or fusion protein molecules. For example, mammalian cells such as Chinese hamster ovary cells (CHO), together with vectors such as the human cytomegalovirus major intermediate-early gene promoter element, are efficient expression systems for antibodies (Foecking et al., 1986, Gene 45:101; and Cockett et al., 1990, Bio/Technology 8:2). In certain embodiments, the expression of the nucleotide sequences encoding the respective heterodimeric immunoglobulin heavy and light chains is regulated by constitutive, inducible or tissue-specific promoters.

In mammalian host cells, a number of viral-based expression systems are available. When adenovirus is used as the expression vector, the modified heavy and light chain coding sequences of interest can be ligated into an adenovirus transcription/translation control complex, such as the late promoter and trimolecular leader sequences. This chimeric gene can then be inserted into the adenoviral genome by in vitro or in vivo recombination. Insertions into non-essential regions of the viral genome (e.g., region E1 or E3) result in recombinant viruses in which expression of modified heavy and light chains in infected hosts is practical and possible (Logan & Shenk, 1984, Proc. Natl. Acad. Sci. USA 81:355-359). Specific initiation signals may also be required for efficient translation of inserted antibody coding sequences. These signals include the ATG initiation codon and adjacent sequences. Furthermore, the initiation codon must be in phase with the reading frame of the desired coding sequence to ensure translation of the entire insert. These exogenous translational control signals and initiation codons can be of a variety of origins, both natural and synthetic. The efficiency of expression may be enhanced by inclusion of appropriate transcription enhancer elements, transcription terminators, etc. (see, eg, Bittner et al., 1987, Methods in Enzymol. 153:516-544).

Expression of the heterodimeric immunoglobulin heavy and light chains can be controlled by any promoter or enhancer element known in the art. Promoters that can be used to control the expression of genes encoding modified heavy and light chains (eg, antibodies or fusion proteins) include the SV40 early promoter region (Bernoist and Chambon, 1981, Nature 290:304-310), Routine The promoter contained in the 3' long terminal repeat of sarcoma virus (Yamamoto et al., 1980, Cell 22:787-797), the herpes thymidine kinase promoter (Wagner et al., 1981, Proc. Natl. Acad. Sci. US. A. 78.1441-1445), the regulatory sequences of the metallothionein gene (Brinster et al., 1982, Nature 296:39-42), the tetracycline (Tet) promoter (Gossen et al., 1995, Proc. Nat. Acad. Sci. USA 89:5547). -5551); the β-lactamase promoter (Villa-Kamaroff et al., 1978, Proc. Natl. Acad. Sci. USA 75:3727-3731) or the tac promoter (DeBoer et al., 1983, Proc. Natl. Acad. Sci USA 80:21-25; see also "Useful proteins from recombinant bacteria" in Scientific American, 1980, 242:74-94); (Herrera-Estrella et al., Nature 303:209-213) or the cauliflower mosaic virus 35S RNA promoter (Gardner et al., 1981, Nucl. Acids Res. 9:2871) and the promoter of the photosynthetic enzyme ribulose bisphosphate carboxylase (Herrera-Estrella et al., 1984, Nature 310:115-120); yeast or other fungal promoter elements such as the Gal4 promoter, ADC (alcohol dehydrogenase) promoter, PGK (phosphoglycerol kinase) promoter, alkaline phosphatase promoter, and The following animal transcriptional regulatory regions that exhibit tissue specificity and have been utilized in transgenic animals: the elastase I gene regulatory region (Swi ft et al., 1984, Cell 38:639-646; Ornitz et al., 1986, Cold Spring Harbor Symp. Quant. Biol. 50:399-409; MacDonald, 1987, Hepatology 7:425-515), insulin gene control region active in pancreatic beta cells (Hanahan, 1985, Nature 315:115-122), immunoglobulin gene control active in lymphoid cells. Regions (Grosschedl et al., 1984, Cell 38:647-658; Adames et al., 1985, Nature 318:533-538; Alexander et al., 1987, Mol. Cell. Biol. 7:1436-1444), ovary, breast, lymph and mouse mammary tumor virus regulatory region active in mast cells (Leder et al., 1986, Cell 45:485-495), albumin gene regulatory region active in liver (Pinkert et al., 1987, Genes and Devel. 1:268-276), liver alpha-fetoprotein gene regulatory region active in the liver (Krumlauf et al., 1985, Mol. Cell. Biol. 5:1639-1648; Hammer et al., 1987, Science 235:53-58), the alpha 1-antitrypsin gene active in the liver control regions (Kelsey et al., 1987, Genes and Devel. 1:161-171), beta-globin gene control regions active in myeloid cells (Mogram et al., 1985, Nature 315:338-340; Kollias et al., 1986, Cell 46 :89-94), myelin basic protein gene control region active in brain oligodendrocyte cells (Readhead et al., 1987, Cell 48:703-712), myosin light chain-2 gene control region active in skeletal muscle ( Sani, 1985, Nature 314:283-286), neuron-specific enolase (NSE) active in neurons (Morelli et al., 1999, Gen. Virol. 80:571-83), brain-derived neurons active in neurons trophic factor (BDNF) gene control region (Tabuchi et al., 1998, Biochem. Biophysic. Res. Com. 253:818-823), glial fibrillary acidic protein (GFAP) promoter active in astrocytes (Gomes et al., 1999) , Braz J Med Biol Res 32(5):619-631; Morelli et al., 1999, Gen. Virol. 80:571-83) and the gonadotropin-releasing hormone gene control region active in the hypothalamus (Mason et al., 1986, Science 234:1372-1378).

In addition, a host cell strain may be chosen which modulates the expression of the inserted sequences, or modifies and processes the gene product in the specific desired fashion. Expression by certain promoters can be elevated in the presence of certain inducers, thereby controlling expression of the genetically modified fusion protein. Furthermore, different host cells have characteristic and specific mechanisms for translational and post-translational processing and modification (eg, protein glycosylation, phosphorylation). Appropriate cell lines or host systems are chosen to ensure the desired modification and processing of the foreign protein expressed. For example, expression in bacterial systems will produce a non-glycosylated product and expression in yeast will produce a glycosylated product. Eukaryotic host cells that have the cellular machinery (eg, glycosylation and phosphorylation) to correctly process the primary transcript of the gene product can be used. Such mammalian host cells include CHO, VERY, BHK, Hela, COS, MDCK, HEK-293, 3T3, WI38, NSO, especially for example SK-N-AS, SK-N-FI, SK-N. - DZ human neuroblastoma (Sugimoto et al., 1984, J. Natl. Cancer Inst. 73:51-57), SK-N-SH human neuroblastoma (Biochim. Biophys. Acta, 1982, 704:450 - 460), Daoy human cerebellar medulloblastoma (He et al., 1992, Cancer Res. 52:1144-1148), DBTRG-05MG glioblastoma cells (Kruse et al., 1992, In Vitro Cell. Dev. Biol. 28A: 609-614), IMR-32 human neuroblastoma (Cancer Res., 1970, 30:2110-2118), 1321 N1 human astrocytoma (Proc. Natl. Acad. Sci. USA, 1977, 74:4816). ), MOG-G-CCM human astrocytoma (Br. J. Cancer, 1984, 49:269), U87MG human glioblastoma-astrocytoma (Acta Pathol. Microbiol. Scand., 1968, 74:465). -486), A172 human glioblastoma (Olopade et al., 1992, Cancer Res. 52:2523-2529), C6 rat glioma cells (Benda et al., 1968, Science 161:370-371), Neuro-2a mouse nerve blastoma (Proc. Natl. Acad. Sci. USA, 1970, 65:129-136), NB41A3 mouse neuroblastoma (Proc. Natl. Acad. Sci. USA, 1962, 48:1184-1190), SCP Ovine choroid plexus (Bolin et al., 1994, J. Virol. Methods 48:211-221), G355-5, PG-4 feline normal astrocytes (Haapala et al., 1985, J. Virol. 53:827-833) , Mpf ferret brain (Trowbridge et al., 1982, In Vitro 18:952-960) and, for example, CTX TNA2 rat normal cerebral cortex, such as CRL7030 and Hs578Bst (Radany et al., 1992, Proc. Natl. Acad.Sci.USA 8 9:6467-6471), including but not limited to normal cell lines. Furthermore, different vector/host expression systems can affect processing reactions to different extents.

For long-term, high-yield production of recombinant proteins, stable expression is often preferred. For example, cell lines that stably express the modified heavy and light chains (eg, antibodies or fusion proteins) of the invention can be engineered. As an alternative to using expression vectors containing viral origins of replication, host cells can be transfected with DNA and selectable markers controlled by appropriate expression control elements (e.g., promoters, enhancers, sequences, transcription terminators, polyadenylation sites, etc.). can be transformed using After introduction of exogenous DNA, modified cells were grown in rich medium for 1-2 days and then switched to selective medium. A selectable marker in the recombinant plasmid confers resistance to selection and allows cells to stably integrate the plasmid into their chromosomes, grow to form foci, and then clone into cell lines. can be propagated.

A number of selection systems are available, including herpes simplex virus thymidine kinase (Wigler et al., 1977, Cell 11:223), hypoxanthine guanine phosphoribosyltransferase (Szybalska & Szybalski, 1962, Proc. Natl. Acad. Sci. USA). 48:2026) and adenine phosphoribosyltransferase (Lowy et al., 1980, Cell 22:817) genes can be used in tk-, hgprt- or aprt- cells, respectively. It also confers resistance to methotrexate as a selection criterion for the dhfr gene (Wigler et al., 1980, Natl. Acad. Sci. USA 77:3567; O'Hare et al., 1981, Proc. Natl. Acad. Sci. 78:1527), confers resistance to mycophenolic acid as a selection criterion for the gpt gene (Mulligan & Berg, 1981, Proc. Natl. Acad. Sci. USA 78:2072), aminoglycoside G- as a selection criterion for the neo gene. 418 (Colberre-Garapin et al., 1981, J. Mol. Biol. 150:1) and hygromycin as a selection criterion for the hygro gene (Santerre et al., 1984, Gene 30:147). ), antimetabolic resistance can be used.

Co-Expression of Heavy and Light Chains The heterodimeric immunoglobulin heavy and light chains described herein can be co-expressed in mammalian cells, as indicated above. In one embodiment, one heavy chain is co-expressed with two different light chains of the LCCA design set described above, and the heavy chain preferentially pairs with one of the two light chains. In another embodiment, two unique heavy chains are co-expressed with two unique light chains, each heavy chain preferentially pairing with one of the light chains.

Testing of Heterodimer Pairs As described above, the heterodimer of at least one of the heterodimer pairs described herein is a heterodimer with two unique weights of the heterodimer pair. When the chain and two unique light chains are co-expressed in a mammalian cell, the heavy chain of the first heterodimer preferentially pairs with one light chain but not the other. One or more amino acid modifications can be included in the immunoglobulin heavy chain and/or the immunoglobulin light chain. Similarly, the heavy chain of the second heterodimer preferentially pairs with the second light chain rather than the first. The degree of preferential pairing can be assessed, for example, using the methods described below. The affinity of each heterodimer of the heterodimer pair for the corresponding antigen can be tested as described below. The thermal stability of each heterodimer of a heterodimer pair can also be tested as described below.

Method for measuring preferential matching LCCA
In one embodiment, preferential pairing of immunoglobulin heavy and light chains is determined by performing a light chain competition assay (LCCA). Commonly-owned patent application PCT/US2013/063306, filed October 3, 2013, describes various embodiments of LCCAs and is hereby incorporated by reference in its entirety for all purposes. The method allows quantitative analysis of the pairing of heavy chains and specific light chains within a mixture of co-expressed proteins, wherein when the heavy and light chains are co-expressed, one specific immunoglobulin heavy chain is associated with two immunoglobulin heavy chains. It can be used to determine preferential association with either immunoglobulin light chain. The method is briefly described below. At least one heavy chain and two different light chains were co-expressed in the cells in a ratio such that the heavy chains were pair-restricted pairwise, optionally secreted proteins were separated from the cells and bound to the heavy chains. separating the immunoglobulin light chain polypeptides from the rest of the secreted protein to produce an isolated heavy chain paired fraction, detecting the amount of each distinct light chain in the isolated heavy chain fraction; The relative abundance of each different light chain in the isolated heavy chain fraction is analyzed to determine the ability of at least one heavy chain to selectively pair with one light chain.

The method provides reasonable throughput, is robust (ie, insensitive to small variations in practice, such as user or flow rate), and accurate. The method provides a sensitive assay that can measure the effects of small variations in protein sequence. Disordered protein-protein, domain-domain, chain-chain interactions over large areas usually require multiple mutations (replacements) to introduce selectivity. There is no need to isolate and purify the protein product, which allows for more efficient sorting. Further details regarding embodiments of this method are provided in the Examples.

Alternative Methods for Determining Preferential Pairing An alternative method for detecting preferential pairing is to use differences in molecular weight to identify each distinct species and identify relative heterodimers each containing a light chain. Including using LC-MS (liquid chromatography-mass spectroscopy) to quantify the population. Antigenic activity assays can also be used to quantify the relative heterodimer populations each containing a light chain, and the extent of binding measured (compared to control) can be used to Estimate the heterodimer population that

Additional methods such as SMCA are described in the Examples, Figures and Tables.

Thermal Stability Thermal stability of a heterodimer can be determined according to methods known in the art. The melting temperature of each heterodimer indicates thermal stability. The melting point of the heterodimer can be determined using techniques such as differential scanning calorimetry (Chen et al. (2003) Pharm Res 20:1952-60; Ghirlando et al. (1999) Immunol Lett 68:47-52 ). Alternatively, the thermal stability of heterodimers can be measured using circular dichroism (Murray et al. (2002) J. Chromatogr Sci 40:343-9).

Affinity to Antigen The binding affinity of the heterodimer to the corresponding antigen and the off-rate of the interaction can be determined by competitive binding assays according to methods well known in the art. One example of a competitive binding assay involves incubating a labeled antigen (e.g. 3H or 125I) and a molecule of interest (e.g. a heterodimer of the invention) in the presence of increasing amounts of unlabeled antigen and A radioimmunoassay comprising detecting molecules bound to a labeled ligand. The affinities and off-rates of the heterodimers of the invention for antigens can be determined from saturation data by Scatchard analysis.

The kinetic parameters of the heterodimers described herein can also be determined using surface plasmon resonance (SPR)-based assays known in the art (e.g., BIAcore kinetic analysis). can. Reviews of SPR-based techniques include Mullet et al., 2000, Methods 22: 77-91; Dong et al., 2002, Review in Mol. Biotech. , 82:303-23; Fivash et al., 1998, Current Opinion in Biotechnology 9:97-101; Rich et al., 2000, Current Opinion in Biotechnology 11:54-61. In addition, U.S. Pat. Nos. 6,373,577, 6,289,286, 5,322,798, 5,341,215, 6,268,125 Both SPR instruments and SPR-based methods that measure protein-protein interactions as described are contemplated in the methods of the present invention. FACS can also be used to measure affinity and is known in the art.

Using a Library of Bispecific Antibody Mutagenesis Design Sets to Generate Predefined Mab1 and Mab2 of Bispecific Antibodies In one embodiment, two basic and Described herein is a bispecific antibody mutation design set aimed at selectively forming bispecific antibodies, starting from two antibodies Mab1 and Mab2. The design set consists of cognate mutations corresponding to Fab1, Fab2 and Fc, respectively. In one embodiment, the design set library is represented by the design sets contained in any of Tables 5, 12 or 15-17. Mutations are introduced at the interface of the light and heavy chains of Fab1 to achieve selective pairing of the two chains in the presence of competing light and heavy chains of Fab2. Selective pairing directs favorable complementary mutations to two obligate light and heavy chains based on steric, hydrophobic or electrostatic complementarity between specific hotspot framework residues at the interface. while making these mutated residues participate in unfavorable interfacial interactions into non-absolute chain pairs. In each design set, selective pairwise mutations were introduced at the interface of the light and heavy chains of Fab2 to provide absolute selection of the two chains in the presence of competing light and heavy chains of Fab1. Target match can also be achieved. The mutations are aimed at reducing mismatches between the light chain of Fab1 and the heavy chain of Fab2 and vice versa. Mutations are introduced at the Fc interface to achieve selective pairing of heavy chains to form an asymmetric antibody molecule comprising two different heavy chains. Genetic engineering of specific interface residue positions in the light and heavy chains of antibodies often has deleterious effects such as loss of antigen-binding affinity, stability, solubility, aggregation properties, etc. of the antibody. can result in A number of relevant properties can be affected, such as kon and koff rates, melting temperature (Tm), stability to stress conditions such as acid, base, oxidation, freeze/thaw, agitation, pressure, and the like. This is often influenced by the complementarity determining regions (CDRs) of the antibody of interest. If the CDRs of different antibodies are generally not identical, the effect of the mutation design set on the properties described above may not be the same across all antibodies. Provided herein is a method for producing Mab1 and Mab2 of any two available antibodies of a bispecific antibody with outstanding purity compared to other contaminant-containing, incorrectly paired antibody-like structures. Presented in The light and heavy chains of Mab1 and Mab2 were co-expressed after introduction of the cognate mutations of each mutation design set, and the expressed antibody products were analytically sorted into other Mab-like species expressed into protein products. To estimate the purity of the preferred bispecific antibody. In some embodiments, analytical sorting procedures may be based on LC-MS techniques. In some embodiments, analytical sorting procedures may be based on charge-based separations, such as capillary isoelectric focusing (cIEF) techniques or chromatographic techniques. An example of a screening technique is presented in Example 9 based on the SMCA procedure. In some embodiments, outstanding purity of a bispecific antibody is defined as greater than 70% of all Mab-like species obtained in the expressed protein product. In some embodiments, outstanding purity of a bispecific antibody is defined as greater than 90% of all Mab-like species obtained in the expressed protein product. The procedure for the preparation and selection of defined Mab1 and Mab2 of the bispecific Mab design set is shown schematically in FIG.

Pharmaceutical Compositions The present invention also provides pharmaceutical compositions comprising the heterodimers or heterodimer pairs described herein. Such compositions comprise a therapeutically effective amount of the heterodimer or heterodimer pair and a pharmaceutically acceptable carrier. In certain embodiments, the term "pharmaceutically acceptable" means having been approved by a federal or state governmental regulatory agency, or a United States Pharmacopoeia or other generally accepted drug for use in animals, particularly humans. It means that it has been presented in the pharmacopoeia. The term "carrier" refers to a diluent, adjuvant, excipient, or vehicle with which the therapeutic is administered. Such pharmaceutical carriers can be water and oils, including those of petroleum, animal, vegetable or synthetic origin, such as sterile liquid peanut oil, soybean oil, mineral oil, sesame oil, and the like. Water is a preferred carrier when the pharmaceutical composition is administered intravenously. Saline solutions and aqueous dextrose and glycerol solutions can also be employed as liquid carriers, particularly for injectable solutions. Suitable pharmaceutical excipients include starch, glucose, lactose, sucrose, gelatin, taproot, rice, wheat, chalk, silica gel, sodium stearate, glycerol monostearate, talc, sodium chloride, skimmed milk powder, glycerol, propylene. , glycol, water, ethanol, and the like. The composition, if desired, can also contain minor amounts of wetting or emulsifying agents, or pH buffering agents. These compositions can take the form of solutions, suspensions, emulsions, tablets, pills, capsules, powders, sustained release formulations, and the like. The composition can be formulated as a suppository, with traditional binders and carriers such as triglycerides. Oral formulations can include standard carriers such as pharmaceutical grades of mannitol, lactose, starch, magnesium stearate, sodium saccharine, cellulose, magnesium carbonate, and the like. Examples of suitable pharmaceutical carriers include E. W. Martin, "Remington's Pharmaceutical Sciences". Such compositions will contain a therapeutically effective amount of the compound, preferably in purified form, together with a suitable amount of carrier so as to provide the form for proper administration to the patient. The formulation should suit the mode of administration.

In certain embodiments, a composition comprising a heterodimer or heterodimer pair is formulated according to routine procedures in pharmaceutical compositions adapted for intravenous administration to humans. Typically, compositions for intravenous administration are solutions in sterile isotonic aqueous buffer. If desired, the composition can also contain a solubilizing agent and a local anesthetic such as lignocaine to relieve pain at the injection site. Generally, the ingredients are provided either separately or mixed together in unit dosage form, for example, as a lyophilized powder or anhydrous concentrate contained in a hermetically sealed container such as an ampoule or sachet indicating the quantity of active agent. . Compositions to be administered by infusion can be dispensed with an infusion bottle containing sterile pharmaceutical grade water or saline. Where the composition is administered by injection, an ampoule of sterile water for injection or saline can be provided so that the ingredients can be mixed prior to administration.

In certain embodiments, the compositions described herein are formulated as neutral or salt forms. Pharmaceutically acceptable salts include those with anions such as those derived from hydrochloric, phosphoric, acetic, oxalic, tartaric, etc., and with sodium, potassium, ammonium, calcium, ferric hydroxide, isopropylamine, Included are those formed by cations such as those derived from triethylamine, 2-ethylaminoethanol, histidine, procaine, and the like.

The amount of a composition described herein that is effective in treating, inhibiting and preventing a disease or disorder associated with aberrant expression and/or activity of a therapeutic protein can be determined by standard clinical techniques. can be done. In addition, in vitro assays may optionally be employed to help identify optimal dosage ranges. The precise dose to be prescribed will also depend on the route of administration, and the seriousness of the disease or disorder, and should be decided according to the judgment of the physician and each patient's circumstances. Effective doses are extrapolated from dose-response curves derived from in vitro or animal model test systems.

Uses of Heterodimer Pairs As described above, the heterodimer pairs described herein can comprise a first heterodimer and a second heterodimer. , each heterodimeric immunoglobulin heavy chain and/or immunoglobulin light chain comprises one or more modifications from a known therapeutic antibody or from a known antibody binding molecule. It is therefore contemplated that heterodimers containing modifications to these antibodies can be used to treat or prevent known therapeutic antibodies or the same diseases, disorders or infections for which known antibodies can be used.

In another embodiment, the heterodimer pairs described herein are combined with other therapeutic agents known in the art for the treatment or prevention of cancer, autoimmune diseases, inflammatory disorders or infectious diseases. can be used to your advantage. In certain embodiments, heterodimeric pairs described herein are combined with monoclonal or chimeric antibodies, lymphokines, or hematopoietic growth factors such as IL-2, IL-3 and IL-7. can be used as a molecule, which serves, for example, to increase the number or activity of effectors that interact with the molecule and to increase the immune response. The heterodimer pairs described herein are advantageously combined with one or more drugs used to treat diseases, disorders or infections, such as, for example, anti-cancer, anti-inflammatory or anti-viral agents. can also be used for

Kits The invention additionally provides kits comprising one or more heterodimer pairs. The individual components of the kit may be packaged in separate containers and a label prescribed by a regulatory agency regulating the manufacture, use, or sale of pharmaceuticals or biological products may be associated with such containers. , the label reflects authorization by an authority to make, use or sell. Kits can optionally contain instructions or instructions outlining methods of use or dosing regimens for heterodimer pairs.

When one or more components of the kit are provided as a solution, such as an aqueous or sterile aqueous solution, the container means is itself an inhaler, syringe, pipette, eye dropper or other such device. can be used, from which solutions can be administered to a subject, or can be applied or mixed with other components of the kit.

Components of the kit may also be provided in dried or lyophilized form, and the kit may additionally contain solvents suitable for reconstitution of the lyophilized components. Regardless of the number or type of containers, the kits of the invention can also include devices to assist in administering the compositions to the patient. Such a device can be an inhaler, nasal sprayer, syringe, pipette, forceps, measuring spoon, eye dropper or similar medically approved delivery vehicle.

Computer Implementation In one embodiment, a computer includes at least one processor coupled to a chipset. Memory, storage, keyboards, devices, graphics adapters, pointing devices and network adapters are also coupled to the chipset. A display is coupled to the graphics adapter. In one embodiment, chipset functionality is provided by a memory controller hub and an I/O controller hub. In another embodiment, the memory is directly coupled to the processor rather than the chipset.

A storage device is any device capable of holding data, such as a hard drive, compact disc read-only memory (CD-ROM), DVD, or solid state memory device. The memory holds instructions and data used by the processor. The pointing device, which can be a mouse, trackball or other type of pointing device, is used in conjunction with the keyboard to enter data into the computer system. A graphics adapter displays images and other information on a display. A network adapter couples a computer system to a local or wide area network.

A computer may have components different from those described above and/or other components, as is known in the art. Additionally, the computer may lack certain components. Further, storage devices can be local and/or remote to the computer (eg, incorporated within a storage area network (SAN)).

Computers, as known in the art, are adapted to execute computer program modules to provide the functionality described herein. As used herein, the term "module" refers to computer program logic utilized to provide specified functionality. Accordingly, modules may be implemented in hardware, firmware and/or software. In one embodiment, program modules are stored in a storage device, loaded into memory and executed by a processor.

The examples and embodiments described herein are for illustrative purposes only and in that light various modifications or alterations will be suggested to those skilled in the art and are within the spirit and scope of this application. as well as within the scope of the appended claims.

Below are examples of specific embodiments for carrying out the present invention. Examples are provided for illustrative purposes only and are not intended to limit the scope of the invention in any way. Efforts have been made to ensure accuracy with respect to numbers used (eg amounts, temperature, etc.) but some experimental errors and deviations should, of course, be allowed for.

The practice of the present invention employs, unless otherwise indicated, conventional methods of protein chemistry, biochemistry, recombinant DNA technology and pharmacology within the skill of the art. Such techniques are explained fully in the literature. For example, T. E. Creighton, Proteins: Structures and Molecular Properties (WH Freeman and Company, 1993); L. Lehninger, Biochemistry (Worth Publishers, Inc., current edition); Sambrook et al., Molecular Cloning: A Laboratory Manual (2nd edition, 1989); See Remington's Pharmaceutical Sciences, 18th ed. (Easton, Pennsylvania: Mack Publishing Company, 1990); Carey and Sundberg Advanced Organic Chemistry 3rd ed.

Example 1: Molecular Modeling and Computer-Guided Engineering of the Fab Interface Heavy and light chain sequences that can be screened in the context of other antibodies (Abs) or fragments thereof using structural and computer molecular modeling-guided techniques. A library of mutation designs was generated to identify mutations that exhibited the desired specificity in the antibody of interest. The design strategy to engineer preferential heavy (H)-light (L) chain pairing involved first identifying a representative Fab (ie, D3H44).

As shown in Table 1, the main criteria for this Fab were to be human/humanized, to have commonly used VH and VL subgroups and minimal framework region mutations. was to contain Additionally, a structural consideration was that the VH:VL interdomain angle should be close to the average observed in antibodies. After selecting Fab D3H44, computational analysis of the Fab interface was performed to identify and understand residues important for heavy and light chain interactions by using a bifurcated approach.

The first approach, involving a global analysis of sequence conservation across Fab variable and constant interfaces, was through alignment of known antibody sequences and structures. An alignment of the constant and variable domain sequences of various antibody subgroups is shown in FIG. FIG. 1A shows an alignment of representative human VH germline subgroups. FIG. 1B shows a representative human kappa VL germline subgroup alignment. FIG. 1C shows an alignment of representative human lambda VL germline subgroups. FIG. 1D shows an alignment of human CH1 allele sequences. FIG. 1E shows an alignment of human kappa and lambda allele sequences. The second approach involved analysis of the D3H44 crystal structure interface using a number of molecular modeling tools (eg, ResidueContacts™), as shown in FIG. These analyzes led to the identification of a list of hotspot locations for engineering preferential HL pairings. Hotspot locations determined in these analyzes are listed in Table 2. These positions and amino acids are primarily framework residues (with the exception of some located in the CDR3 loops) and are mostly conserved in lambda light chains. Amino acids of the parental D3H44 sequences with Kabat numbering are presented in Tables 3a-3b.

Potential mutations at hotspot positions and positions flanking hotspots of interest in the three-dimensional crystal structure were then simulated and identified via computational mutagenesis and packing/modeling with Zymepack™. . Zymepack™, given an input structure and a set of mutations, modifies the residue types of the input structure according to the supplied mutations, generating a new structure that approximates the physical structure of the mutein. It is a software suite that In addition, Zymepack characterizes muteins by calculating various quantitative metrics. These metrics include measurements of steric and electrostatic complementarity, which can be correlated with mutein stability, binding affinity or heterodimer specificity.

Figure 3 depicts a subset of hotspot positions at the heavy and light chain interfaces of variable domains and how mutations are introduced at these interface positions to drive obligate selective chain pairing; It also demonstrates how to avoid the formation of incorrect strand pairings. Computational methods, including Zymepack™, were used to model and calculate steric complementarity based on energetic factors such as van der Waals packing, cavitation effects and intimate contacts with hydrophobic groups. . Similarly, electrostatic interaction energies were modeled and evaluated based on Coulomb interactions of charge, hydrogen bonding and desolvation effects. Simulating both preferred heavy and light chain pairing models such as H1:L1 (or H2:L2) and imprecise pairing models such as H1:L2 (and H2:L1) resulting from the introduction of mutations of interest to calculate relative steric and electrostatic scores. This suggests that a particular set of mutations may have a dominant energy, i.e. greater steric and/or electrostatic complementarity, in favored (absolute) heavy-light chain pairs compared to imprecise (non-absolute) pairs. It made it possible to decide whether to bring about sexuality. The calculated steric and electrostatic energies are the free energy components associated with light and heavy chain pairing. Thus, greater steric and electrostatic complementarity indicates greater free energy associated with absolute pairwise pairing compared to non-absolute pairwise pairing. Greater steric and electrostatic complementarity results in preferential (selective) pairing of obligate heavy and light chains compared to non-absolute pairs.

Example 2 Design Selection and Description Using the techniques described in Example 1, heavy-light chain heterodimer pairs (i.e., H1-L1 and H2- L2) was designed. The heterodimers are designed in pairs, called "designs" or "design sets", containing a set of substitutions in the H1, L1, H2 and L2 chains that promote preferential pairing (Table 5). The design set was first tested as the "LCCA design" (Table 4), in which one heavy chain is co-expressed with two light chains, to assess relative pairing. Amino acid substitutions are identified by reference to Tables 3a, 3b using the Kabat numbering system.

Using the design library listed in Table 30 of International Patent Application PCT/CA2013/050914 as a starting point, the LCCA designs shown in Table 4 and some of the design sets shown in Table 5 were identified. Some of the designs in Tables 4 and 5 are new independent designs. The core design is shown in Table 6 with associated unique identifiers. Most designs cover the constant region only, but some designs incorporate modifications in the variable region as well. It was proposed that these designs would further induce pairing specificity and at the same time could be conveniently transferred to other antibody systems.

In the derived designs, using as a starting point the library of designs listed in Table 30 of International Patent Application PCT/CA2013/050914, the designs were clustered by structural similarity and analyzed by differential scanning calorimetry (DSC). Rankings were based on the measured strength of pairing specificity, efficacy and stability of antigen binding. The designs were then combined (see, eg, Table 7) and/or optimized (see, eg, Tables 8 and 9) to yield derived designs. In combination, at least one design exhibited high pairing specificity and the other designs exhibited a range of favorable pairing specificities. The designs selected for combination and/or optimization all showed minimal/no effect on antigen binding and minimal/no effect on melting temperature (Tm). did not show.

Independent designs alone (classified as Independent in the Design Type column of Table 5) and in combination with derived designs (classified as Independent/Combination in the Design Type column of Table 5). See also Table 10) were tested.

The design was packed into a molecular model of D3H44 and metrics were calculated (as described in Example 1). Top designs were then selected based on risk (possible impact on stability and immunogenicity) and impact (considering proposed strength to induce pairing specificity). These high-level designs are shown in Table 5.

Example 3 Preparation of Fab Constructs Encoding D3H44 IgG Heavy and D3H44 IgG Light Chains Wild-type Fab heavy and light chains of the anti-tissue factor antibody D3H44 were prepared as follows. D3H44 Fab light chain (AJ308087.1) and heavy chain (AJ308068.1) sequences were taken from GenBank (Table 3c), genes were synthesized and codons were optimized for mammalian expression. Light chain vector inserts consisting of 5'-EcoRI cleavage site-HLA-A signal peptide-HA or FLAG tag-light chain Ig clone-'TGA stop'-BamH1 cleavage site-3' were ligated into pTT5 vector (Durocher Y et al., Nucl. Acids Res. 2002;30, No. 2e9). The resulting vector plus insert was sequenced to confirm the correct reading frame and sequence of the coding DNA. Similarly consisting of 5′-EcoR1 cleavage site-HLA-A signal peptide-heavy chain clone (ends at T238, see Table 3a)-ABD ₂ -His ₆ tag-TGA stop-BamH1 cleavage site-3′ The heavy chain vector insert was ligated into the pTT5 vector (ABD, albumin binding domain). The resulting vector plus insert was also sequenced to confirm the correct reading frame and sequence of the coding DNA. Various Fab D3H44 constructs containing amino acid substitutions for the design set were generated by gene synthesis or by site-directed mutagenesis (Braman J, Papworth C & Greener A., Methods Mol. Biol. (1996) 57 :31-44).

To facilitate evaluation of preferential pairing by competition assay SPR sorting, heavy and light chains were tagged at the C- and N-termini, respectively. The ABD ₂ -His ₆ heavy chain tag specifically allowed the HL complex to be captured on the anti-His tag SPR chip surface, while the FLAG and HA light chain tags affected the relative L1 and L2 populations. made it possible to quantify

Example 4 Evaluation of Preferential Pairing of Fab Heterodimers Containing Either Constant Domain Modifications or Combinations of Constant and Variable Domain Modifications in D3H44 IgG Light and/or Heavy Chains Table 12 LCCA Design Set Amino Acids Constructs encoding D3H44 IgG heavy and light chains in Fab format containing modifications were prepared as described in Example 3. The ability of the constructs to preferentially pair to form the desired H1-L1 heterodimer in the context of the LCCA design set (H1, L1, L2) was determined using the light chain competition assay (LCCA).

LCCA quantifies the relative pairing of one heavy chain with at least two unique light chains and can be summarized as follows. One D3H44 heavy chain Fab construct is co-expressed with two unique D3H44 light chain Fab constructs and relative light chain pairing specificities (eg H1-L1:H1-L2) determined by competition assay-SPR selection and this was done twice. The amount of L1 (designed to preferentially pair with the H chain) is reduced relative to L2 (e.g., L1:L2 = 1:3 based on weight), while the heavy chain is used in limited amounts. The LCCA selectivity was biased by holding (ie, H1:L1+L2 at 1:3) to identify strong drivers. The amount of each heterodimer (i.e., H1-L1 and H1-L2) formed was determined by binding of the heavy chain to the SPR chip via His-tag pull-down followed by antibodies specific for these tags. Determined by detection of the amount of each light chain tag (HA or FLAG) used. Selected heterodimer hits were subsequently validated via validation of the light chain competition assay, which showed that L1:L2 DNA ratios were 1:3 and 1:9 upon transfection. variable, while the heavy chain was retained in limited amounts. Also note that the light chain tag (HA or FLAG) does not affect LCCA pairing in the D3H44 system (see Example 10 of International Patent Application PCT/CA2013/050914). A schematic of the assay design is shown in FIG. Figure 5 shows how heavy and light chains are tagged and how preferential pairing is assessed. Experimental details of the LCCA are provided below.

Transfection Method LCCA designs containing one heavy and two light chain constructs, prepared as described in Example 3, were transfected into CHO-3E7 cells as follows. CHO-3E7 cells were grown at 1.7-2×10 ⁶ cells/cell in FreeStyle™ F17 medium (Invitrogen Cat. Cultured at 37° C. with a density of ml. A total of 2 μg of DNA was transfected using PEI-pro (Polyplus transfection number 115-375) at a DNA:PEI ratio of 1:2.5 for a total volume of 2 ml. 24 hours after adding the DNA-PEI mixture, the cells were transferred to 32°C. Supernatants were tested for expression on day 7 by non-reducing SDS-PAGE analysis followed by visualization of bands by Coomassie Blue staining. The H:L ratio is as shown in Table 11.

Competition Assay SPR Method The extent of preferential D3H44 light chain pairing to D3H44 heavy chain in the LCCA design was assessed by using SPR-based readout of unique epitope tags located at the N-terminus of each light chain.

Supply of Surface Plasmon Resonance (SPR). GLC sensor chip, Biorad ProteOn amine coupling kit (1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysulfosuccinimide (sNHS) and ethanolamine), and 10 mM sodium acetate buffer The liquid was obtained from Bio-Rad Laboratories (Canada) Ltd. (Mississauga, ON). 4-(2-Hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer, ethylenediaminetetraacetic acid (EDTA) and NaCl were purchased from Sigma-Aldrich (Oakville, ON). A 10% Tween 20 solution was purchased from Teknova (Hollister, Calif.).

SPR biosensor assay. All surface plasmon resonance assays were performed on a BioRad ProteOn XPR36 instrument (Bio-Rad Laboratories (Canada) Ltd., Mississauga, ON) with PBST running buffer (PBS, Teknova Inc, 0.05% Tween20). ) at a temperature of 25°C. An anti-penta-His capture surface was generated using a GLM sensor chip activated with a 1:5 dilution of standard BioRad sNHS/EDC solution injected at 100 μL/min for 140 seconds in the analyte (horizontal) direction. Immediately after activation, a 25 μg/mL solution of anti-penta-His antibody (Qiagen Inc.) in 10 mM NaOAc, pH 4.5 was applied in the analyte (horizontal) direction at a flow rate of 25 μL/min to approximately 3000 resonance units ( RU) was injected until immobilized. Remaining active groups are terminated by injecting 1 M ethanolamine in the direction of the analyte for 140 seconds at a flow rate of 100 μL/min, which creates a false activated interspot for blank reference. I also made sure.

Selection of heterodimers for binding to anti-FLAG (Sigma Inc.) and anti-HA (Roche Inc.) monoclonal antibodies was performed in two steps. Indirect capture of the heterodimer on the anti-penta-His surface in the ligand direction, followed by injection of anti-FLAG and anti-HA in the analyte direction. Initially, a single injection of PBST at 100 μL/min for 30 seconds in the ligand direction was used for baseline stabilization. For each heterodimer capture, the unpurified heterodimer in cell culture medium was diluted to 4% with PBST. 1-5 heterodimers or controls (i.e., controls containing 100% HA light chain or 100% FLAG light chain) were injected simultaneously into individual ligand channels at a flow rate of 25 μL/min for 240 seconds to allow saturation. Approximately 300-400 RU of heterodimer capture resulted on the anti-penta-His surface. The first ligand channel was left empty and used as a blank control when necessary. This heterodimer capture step was immediately followed by injections of two buffers in the direction of the analyte to stabilize the baseline, followed by 5 nM anti-FLAG and 5 nM anti-HA, respectively. A 120 second injection at 50 μL/min with a second dissociation step was repeated twice, resulting in a set of binding sensorgrams using a buffer basis for each captured heterodimer. Tissue factor (TF) antigen to which the heterodimer binds was also injected into the last remaining analyte channel as an activity control. The heterodimer was regenerated with an 18 second pulse of 0.85% phosphoric acid at 100 μL/min for 18 seconds to prepare the anti-penta-His surface for the next injection cycle. Sensorgrams were aligned and double referenced by using buffer blank injection and interspot, and the resulting sensorgrams were analyzed using ProteOn Manager software v3.0.

Results The LCCA results are shown in Tables 12, 13a and 14a. In Tables 13 and 14, "Unique Identifier" indicates that the unique identifier of the two constituent LCCAs can be oriented in either orientation ((set number of H1L1L2-set number of H2L2L1) or (set number of H2L2L1-set number of H1L1L2) ), so it may not exactly match Table 5. The rating of the preferential match for each LCCA design is shown in the last three columns of Table 12. The same data is also included in the context of design pairs in columns 5, 6 and 8 or 10, 11 and 13 of Tables 13a and 14a. Each unique set of H1, L1 and L2 mutations (LCCA design) was assigned a unique number or "set number" (eg, 9567 or 9087). When the data is represented by the H1L1H2L2 format (Fab versus format or design set), such design set is therefore indicated by a "unique identifier" consisting of the set number of the two constituent LCCAs (eg 9567-9087). be Note that most LCCA experiments were performed on constructs containing interchain Fab disulfide bonds located in constant domains (H/C233-L/C214, Kabat numbering). In Tables 13(a and b) and 14(a and b), two complementary LCCA sets (H1, L1, L2 and H2, L2, L1 ) are presented in Fab vs. format. The presence of tags (L chain: HA and FLAG, and H chain: ABD ₂ -His ₆ ) did not affect the predicted neutral pairings of approximately 50%:50% in D3H44 WT.

In this table, the LCCA data reported are median in ratio format (H1-L1:H1-L2 and H2-L2:H2-L1) normalized to L1:L2 DNA ratio=1:1. In addition, the LCCA data were normalized to 100%, as the sum of L1 and L2 was observed to differ significantly from 100% in some variants. This difference in total light chain fraction is believed to be due in part to non-specific binding fluctuations that occur during the initial heterodimer capture on the SPR chip. Both LCCA normalized ratios are presented in the table because the LCCA experiments were performed at two different L1:L2 DNA ratios (L1:L2 of 1:3 and 1:9, respectively). LCCA data were obtained when the experimental data obtained did not meet the criteria for inclusion (e.g., Fab capture on the SPR chip was less than 100, or total LCCA for L1 and L2 was outside the range of 60-140). ), so we did not report on some LCCA experiments.

Table 12 covers all of the LCCA designs (530) for which data were obtained. Of the 530 LCCA designs, 490 (92.5%) of these LCCAs had at least 60% exact matches ( normalized to L1:L2 DNA ratio=1:1). The remaining LCCA designs were predominantly neutral LCCA designs (32/530 or 6.0%), as well as a small percentage of LCCA designs that produced inconsistent results (8/530 or 1.0%). 5%). The designs shown in Table 12 are predominantly electrostatic (based on specific drivers that utilize hydrogen bonding or charge-charge interactions), some designs are steric complementarity and /or also include interchain covalent disulfide bonds. Some designs also included mutations for the formation of new disulfide bonds (formed by H/C233-L/C214) in the absence of natural interchain disulfide bonds.

Tables 13(a and b) and 14(a and b) cover 447 designs for which LCCA data are present in both heterodimers of the design set. Tables 13a and 14a demonstrate that the computational approach described in Example 1 yielded preferential pairings of H1-L1 over H1-L2 and H2-L2 over H2-L1 across a diverse design set and variants thereof. demonstrated that it resulted in the achievement of preferential pairing.

Table 13(a and b) covers designs with average LCCA performance with a Fab heterodimer pairing:mispairing of at least 86:14 (normalized median L1:L2 ratio of H1 -L1:H1:L2 and H2-L2:H2-L1 are 1:1), while Table 14(a and b) show less than 86:14 pairing:mismatching for Fab heterodimers. covers designs with an average LCCA performance of . The performance of each LCCA was normalized to 100%, similarly the L1:L2 DNA ratio was normalized to 1:1 (described above in this example), and both scalar values (ln(r1/f1) or ln(r2/f2)), where r1 and r2 correspond to the median experimental ratios H1L1:H1L2 and H2L2:H2L1, respectively, and f1 and f2 correspond to the corresponding experimental ratios. ), described by the ratio of matched and mismatched Fab heterodimers. Each design also has an associated average LCCA performance scalar value (0.5(ln(r1/f1)+ln(r2/f2))), which is also normalized to 100% and similarly L1:L2 DNA The ratio is also normalized to 1:1 (described above in this example). In addition, the scalar ranges of the respective LCCAs of the designs (LCCA1 and LCCA2 corresponding to H1L1:H1L2 and H2L2:H2L1) are shown. 354 of the 447 Mab designs (79.2%) show an average LCCA performance that is at least 86:14 (Tables 13a and b). The designs in Table 13(a and b) were further characterized into 13 clusters based on design similarity. The designs within each cluster were ordered by highest to lowest average LCCA performance scalar value.

In addition, the LCCA data in Table 13a are also graphically represented in FIG. FIG. 7 shows boxplots of mean LCCA performance values with at least 86:14 pairings:mispairings of Fab heterodimers in each cluster. The bottom of the boxplot shows the first quartile (Q1), which is the ratio of the minimum and median values such that values below the first quartile represent the lowest 25% of the data. Intermediate average LCCA performance values. Horizontal bars within the boxplot indicate the second quartile, which is the cluster median mean LCCA performance value. The top of the boxplot shows the third quartile (Q3), which is the highest and median values such that values above the third quartile represent the highest 25% of the data. is the median average LCCA performance value of . The interquartile area is the difference between Q3 and Q1. Whiskers extending vertically in both directions indicate data ranges of values both within Q1-(1.5 ^* IQR) or Q3+(1.5 ^* IQR). Horizontal bars that cap the whiskers indicate the maximum and longest values within the range. Data and whiskers lying outside the boxplot plot are identified as outliers, the median outlier is indicated by a dashed line (1.5 ^* IQR to 3 ^* IQR different from Q1 or Q3), extreme outliers are , indicated by the + sign (differs from Q1 or Q3 by more than 3 ^* IQR).

Example 5: Scale-up of biophysical characterization To test for thermostability and antigen binding, precisely matched heterodimers indicated by a unique set of identifiers (Table 5) were tested as follows. (typically 20 ml) and purified. Heterodimeric heavy and light chains were expressed in 20 ml cultures of CHO-3E7 cells. CHO-3E7 cells were grown to 1.7-2×10 ⁶ cells in FreeStyle™ F17 medium (Invitrogen Cat# A-1383501) supplemented with 4 mM glutamine and 0.1% Koliphor 188 (Sigma #K4894). /ml and cultured at 37°C. A total of 20 μg of DNA was transfected at a DNA:PEI ratio of 1:2.5 using PEI-pro (Polyplus Cat#115-375) for a total volume of 20 ml. 24 hours after adding the DNA-PEI mixture, the cells were transferred to 32°C.

Cells were centrifuged 7 days after transfection and heterodimers were purified from the supernatant by high-throughput nickel affinity chromatography purification as follows. The supernatant is diluted to 20-25% cell culture supernatant with equilibration buffer (Calcium, Magnesium and Phenollet (Dulbecco-Phosphate Buffered Saline (DPBS) without HyClone™ No. SH30028.02)) was incubated for 12 hours with mixing with HisPur® Ni-NTA resin (Thermos Scientific No. PI-88222), which had also been previously equilibrated in equilibration buffer.The resin was then collected by centrifugation and Transferred to a well frit plate, washed three times with equilibration buffer, and eluted using HIS-Select® elution buffer (Sigma-Aldrich #H5413).

After purification, heterodimer expression was assessed by a non-reducing High Throughput Protein Express assay using the Caliper LabChip GXII (Perkin Elmer #760499). The procedure was performed according to the HT Protein Express LabChip User Guide version 2 LabChip GXII User Manual with the following modifications. Heterodimer samples were added to separate wells of a 96-well plate (BioRad # HSP9601) in either 2 μl or 5 μl (concentration range 5-2000 ng/μl) in 7 μl of HT Protein Express Sample Buffer (Perkin Elmer #760328). ). The heterodimer samples were then denatured at 70°C for 15 minutes. The LabChip instrument was powered using the HT Protein Express LabChip (Perkin Elmer #760499) and Ab-200 assay setup. After use, the chips were cleaned with MilliQ water and stored at 4°C.

Example 6 Measurement of Thermal Stability of Fab Heterodimers by DSF To assess thermal stability, differential scanning fluorescence (DSF) was used as a high-throughput method to measure wild-type, unmodified heavy chain- All correctly matched heterodimers were selected compared to light chain pairs. Heterodimers were prepared as described in Example 5.

Thermal Stability Measurements Thermal stability of all heterodimer pairs was measured using DSF as follows. Each heterodimer was purified as described in Example 5 and diluted to 0.5 mg/mL with DPBS (HyClone catalog number SH30028.02). For most samples, a working stock of Sypro Orange gel stain (Life Technologies Cat. No. S-6650) was prepared by diluting 4 μL of Sypro Orange gel stain into 2 ml of DPBS. DSF samples were prepared by adding 14 μL of 0.5 mg/mL protein to 60 μL of diluted Sypro Orange gel stain working solution. However, below 0.5 mg/mL protein, each DSF sample was prepared by adding 14 μL of undiluted protein to 60 μl of Sypro Orange dye working solution (diluted 1:1500 in DPBS). DSF analysis was then performed in duplicate on 20 μl aliquots using a Rotor-Gene 6000 qPCR instrument (QiaGen Inc). Each sample was scanned from 30° C. to 94° C. using 1° C. intervals with 10 seconds equilibration at each step and a 30 second wait time at the start. An excitation filter of 470 nM and an emission filter of 610 nM with a gain of 9 were used. Data were analyzed by Rotor-Gene 6000 software using the maximum from the first derivative of the Tm denaturation curve. The remaining DSF samples were similarly prepared and analyzed with the following protocol modifications that did not alter the measured Tm values. 1) Working solution was prepared by diluting 1 μL Sypro Orange gel stain in 2 ml DPBS. 2) A 30 μl aliquot was analyzed. 3) A gain of 10 was used.

The DSF results are shown in Tables 12, 13b and 14b. The thermal stability (DSF values and DSF values compared to wild type) of H1:L1 Fabs in the context of LCCA design are shown in columns 3 and 4 of Table 12. The same DSF values are also included in columns 7 and 8 of Tables 13b and 14b in the context of design pairs. For each Fab heterodimer for which replicates are performed, the reported Tm value is the median. A comparison of the Tm value of the Fab heterodimer to that of the wild-type Fab heterodimer (the wild-type Fab construct containing the HA tag, with a median Tm of 81.0° C.) is shown in the H1L1_dTm_dsf column. report to In some Fab heterodimers lacking the natural interchain disulfide bond (between H chain C233 and L chain C214), the H1L1_dTm_dsf value was higher than the corresponding wild-type Fab lacking the natural interchain disulfide. Note that it was not determined as it was not evaluated. In addition, some Fab heterodimers were associated with poor experimental quality (e.g., low yields, low intensities, partially masked peaks and repeat-to-repeat variability of Fab heterodimers of more than 1 °C). Note that no Tm values were reported (17/230 or 7.4% Feb heterodimers) due to sex). In some of these Fab heterodimers, putative Tm values corresponding to those of similar Fab heterodimers differing only in the presence/absence or nature of the binding L chain tag (HA or FLAG) were substituted. was reported to For estimated Tm values, the corresponding wild-type Tm value (81.2° C.) is the median value obtained from all wild-type Fab heterodimer constructs (i.e., Fab constructs containing HA or FLAG tags). . HA or FLAG tags do not significantly affect the Tm values of wild-type Fab heterodimers. Overall, Fab heterodimers showed similar Tm values compared to WT. Of the Fab heterodimers that contain native interchain disulfides and for which DSF data are also available, 93% (195/209) of the Fab heterodimers lose ≤3°C compared to WT showed that. Furthermore, the most affected Fab heterodimer showed a loss of 6.5°C compared to WT. Table 12 covers the LCCA designs in order of decreasing Tm rank.

Additionally, 13 amino acid substitutions were identified that generally improved the stability of the Fab heterodimer (see Table 34). Stabilizing mutations were identified after comparison of Fab heterodimers containing stabilizing mutations with similar Fab heterodimers differing in the absence of stabilizing mutations. Heavy chain stabilizing mutations include A125R, H172R, L143F, Q179D, Q179E, Q39R, S188L and V190F. Light chain stabilizing mutations include Q124E, Q124R, Q160F, S176L and T180E. Overall, stabilizing mutations increased stability from 0.4°C to 2.1°C. Heavy chain stabilizing mutations A125R, H172R, L143F, Q179D, Q179E, Q39R, S188L and V190F have stabilities of 0.4°C to 0.6°C, 0.4°C to 2.1°C, 0.4°C to 0.6°C, respectively. 4°C, 0.5°C to 0.6°C, 0.5°C to 0.8°C, 1.1°C to 1.6°C, 0.4°C to 1.2°C and 1°C increments. Light chain stabilizing mutations Q124E, Q124R, Q160F, S176L and T180E have stabilities of 0.4°C to 0.5°C, 0.8°C to 0.9°C, 0.6°C and 0.4°C, respectively. ~1.0°C and 0.5°C increments.

Example 7 Determination of Antigen Affinity of Fab Heterodimers Whether amino acid substitutions have any effect on the ability of Fab heterodimers to bind tissue factor to antigen binding of the heterodimers was evaluated to determine The affinity of each Fab heterodimer for tissue factor was determined by SPR as follows.

SPR supply. GLC sensor chips, Biorad ProteOn amine coupling kits (EDC, sNHS and ethanolamine), and 10 mM sodium acetate buffer were obtained from Bio-Rad Laboratories (Canada) Ltd. (Mississauga, ON). PBS running buffer with 0.05% Tween 20 (PBST) was obtained from Taknoca Inc. (Hollister, Calif.).

Batch of Fab heterodimers. Purified Fab heterodimers were tested in three batches, A, B and C. Batches A and B were stored at 4°C for approximately 1 month before performing the SPR assay, while batch C Fab heterodimers were stored at 4°C for approximately 2 months before performing the SPR assay. saved for a month. Fab heterodimers of Batch C are marked with a "+" next to the corresponding KD values in Table 12.

All surface plasmon resonance assays were performed using a BioRad ProteOn XPR36 instrument (Bio-Rad Laboratories (Canada) Ltd. (Mississauga, ON)) with PBST running buffer at a temperature of 25°C. An anti-penta-His capture surface was generated using a GLC sensor chip activated with a 1.5 dilution of standard BioRad sNHS/EDC solution injected at 100 μL/min for 140 seconds in the analyte (horizontal) direction. Immediately after activation, a 25 μg/mL solution of anti-penta-His antibody (Qiagen Inc.) in 10 mM NaOAc, pH 4.5 was applied in the analyte (horizontal) direction at a flow rate of 25 μL/min to approximately 3000 resonance units ( RU) was injected until immobilized. Remaining active groups are terminated by injecting 1 M ethanolamine in the direction of the analyte for 140 seconds at a flow rate of 100 μL/min, which creates a false activated interspot for blank reference. I also made sure.

Selection of Fab heterodimers binding to TF antigen was performed in two steps. Indirect capture of Fab heterodimers on the anti-penta-His antibody surface in the ligand direction, followed by simultaneous 5 purified antigen concentrations and 1 buffer blank for double referencing in the analyte direction. Injection. The baseline was first stabilized with a 30 second injection of 100 uL/min in the ligand direction of one buffer. One to five variants or controls at a concentration of 3.4 μg/ml in PBST were co-injected into individual ligand channels at a flow rate of 25 μL/min for 240 seconds. This resulted in an average entrapment of approximately 1000 RU on the anti-penta-His surface for batches A and B, and an average entrapment of approximately 600 RU on the anti-penta-His surface for batch C. The first ligand channel was left empty and used as a blank control when necessary. This capture step was immediately followed by two buffer injections of 100 μL/min for 30 sec each in the analyte direction to stabilize the baseline, then 60 nM, 20 nM, 6.7 nM, 2.2 nM and Antigen (TF) at 0.74 nM was co-injected at 50 μL/min for 120 seconds with a dissociation step of 600 seconds along with a buffer blank. The capture antibody surface was regenerated for the next injection cycle by regenerating two 18 second pulses of 0.85% phosphoric acid at 100 μL/min for 18 seconds. Sensorgrams were aligned and double referenced by using buffer blank injection and interspot, and the resulting sensorgrams were analyzed using ProteOn Manager software v3.1. Double-referenced sensorgrams were fitted to a 1:1 binding model. Rmax values for each antigen were normalized to the antibody capture level for each variant and compared to the 100% control.

Antigen affinity (KD) values for Fab heterodimer samples are reported in Tables 12, 13b and 14b. The KD values (range of KD and KD values and median KD compared to wild type) of H1:L1 Fabs in the context of LCCA design are shown in columns 5, 6 and 7 of Table 12, respectively. The same KD values are shown in columns 3 (KD of H1-L1 Fab heterodimer), 4 (KD change of H1-L1 Fab heterodimer compared to wild type), 5 (H2-L2) in Tables 13b and 14b. Fab heterodimer KD), as well as in the context of design pairs of 6 (H2-L2 Fab heterodimer KD change compared to wild type). KD values were determined only in Fab heterodimer samples that showed at least 100 RU of Fab heterodimer capture. The reference wild-type KD (0.157 nM) reflects the median wild-type Fab heterodimer whose light chain contains a FLAG tag. Wild-type Fab heterodimers (containing either FLAG or HA tags) showed similar KD values, with a 2.6-fold difference observed between maximum and minimum values. In Tables 12, 13b and 14b, KD differences for wild-type antigen binding affinities are shown using a calculation of - (log (KD) design - log (KD) wt), with positive values indicating A decreased KD value of the Fab heterodimer compared to the wild-type binding affinity of , whereas a negative value indicates an increased KD value. Note that some Fab heterodimers lack measured KD values. In some of these cases, Fab heterodimers were evaluated, but SPR experiments showed low Fab heterodimer capture (i.e., less than 100 RU) and thus accurate determination of KD values was not possible. rice field. In Fab heterodimers showing similarity to other Fab heterodimers (i.e. differing only in the presence/absence or nature of the attached L chain tag (HA or FLAG)), similar Fab heterodimers Estimated KD values corresponding to the KD values of are provided instead (shown in Tables 12, 13b and 14b). The corresponding estimated wild-type KD value (0.15 nM) was the median value obtained from all wild-type Fab heterodimer constructs (ie, Fab constructs containing HA or FLAG tags). Overall, the results show that correctly paired heterodimers (from a design standpoint) exhibit wild-type-like binding affinities for antigen (within about 2.3-fold of the reference wild-type affinity). It is shown.

Example 8. Ultra-Performance Liquid Chromatography Size Exclusion Chromatography (UPLC-SEC) profiles of wild-type tagged D3H44 heterodimer and preferentially paired heterodimer C-terminal ABD2-His ₆ tag on heavy chain and light chain Wild-type D3H44 heterodimers (one heavy chain and one light chain) with N-terminal tags (FLAG on one construct and HA on another construct) were prepared by methods known in the art and as described in Example 5. It was expressed and purified by methods similar to those described above. Preferentially or correctly paired heterodimers were individually scaled up and purified by His-tag affinity purification as described in Example 5.

UPLC-SEC was performed using a Waters BEH200 SEC column (2.5 mL, 4.6×150 mm, stainless steel, 1.7 μm particles) set at 30° C. and attached to a Waters Acquity UPLC system with a PDA detector. Carried out. The run time consisted of 7 minutes, the total volume per injection was 2.8 mL, and the running buffer Hyclone DPBS/Modified-Calcium-Magnesium (Part No. SH30028.02) was used at 0.4 ml/min. Elution was monitored by UV absorption in the range 200-400 nm and chromatograms were extracted at 280 nm. Peak integration was performed using Empower3 software.

Figure 6 shows a representative WT Fab heterodimer pair (containing a FLAg tag on the L chain), as well as representative design Fab heterodimer pairs (LCCA-designed 9735, 9737 and 9740 H1L1 Fab components). UPLC-SEC profile. In general, the design Fab heterodimer pairs showed similar UPLC-SEC profiles compared to WT.

Example 9 Evaluation of Preferential Pairing of Heterodimers in Co-Expressed Sets Containing Modifications of Constant Domains or Constant and Variable Domains in Bispecific Antibody Format were also evaluated to determine if specific matching was possible. In this example, to promote unique heavy chain heterodimerization, the Fc region of each heterodimeric full-length heavy chain was modified one heavy chain with modifications T350V, L351Y, F405A and Y407V. and the other heavy chain was asymmetrically modified to contain modifications T350V, T366L, K392L and T394W (EU numbering).

Construct preparation:
The heterodimer design was tested in the context of the following bispecific antibodies: a) D3H44/trastuzumab, b) D3H44/cetuximab, and c) trastuzumab/cetuximab. Note that D3H44 is a human antibody, Trastuzumab is a humanized antibody, and Cetuximab is a chimeric antibody comprising human IgG1 and mouse Fv regions. Constructs encoding D3H44, Trastuzumab and Cetuximab IgG heavy and light chains containing designed amino acid modifications were prepared as follows. The base DNA sequences of the heavy and light chains of D3H44, Trastuzumab and Cetuximab are shown in Table 3C. D3H44, Trastuzumab and Cetuximab light chain sequences were prepared as described in Example 3, except that some sequences lacked tags while others contained FLAG or HA tags. . The D3H44, Trastuzumab and Cetuximab heavy chain sequences were created by adding an IgG1 ^* 0.1 DNA sequence where the full length heavy chain encodes the hinge CH2-CH3 domains and was modified to add the C of the CH1 domain of the Fab heavy chain. Prepared as described in Example 3, except that dimerization was promoted at the ends. The underlying C-terminal heavy chain lysine residue was removed to prevent LC-MS signal heterogeneity due to C-terminal lysine clipping (Lawrence W. Dick Jr. et al., Biotechnol. Bioeng. (2008) 100:1132-43).

Assay format (SMCA)
The ability of heterodimer co-expression set designs to preferentially pair to form bispecific antibodies was assessed as described below. The assay consists of co-expressing the four chains (H1 and L1 chains from one antibody, H2 and L2 chains from the other antibody) and determining the presence of correctly formed bispecific antibodies by mass spectrometry (LC-MS). ) is based on detection by the use of Figure 8 shows the four starting polypeptide chains and the potential products that result from co-expression of these starting polypeptides in the absence of preferential pairing of the heavy and light chains of the heterodimeric pair. Schematic diagrams showing are provided. Two full-length heavy chain constructs were co-expressed with two unique light chain constructs to give 10 possible antibody species: H1-L1: H1-L1, H1-L2: H1-L2, H1-L1: H1-L2, H2-L1: H2-L1, H2-L2: H2-L2, H2-L1: H2-L2, H1-L1: H2-L1, H1-L2: H2-L2, H1-L2: H2- L1 and H1-L1: gave rise to H2-L2. The H1-L1:H2-L2 species is a precisely matched bispecific antibody (see Figure 8). The relative pairing specificity for the amount of H1-L1:H2-L2 of the preferred species relative to others was determined using LC-MS after pA purification and deglycosylation. Where possible, strands were left untagged, provided that all Mab and half-Ab species differed from each other by at least 50 Da. When mass differences fell outside this possibility, N-terminal tags (HA or FLAG) were added to the light chains to provide sufficient mass differences between species.

This assay with bispecific antibody expression and selection steps is called SMCA.

Mass Spectrometry The extent of preferential D3H44 light chain pairing to D3H44 heavy chain in the co-expression set was assessed using mass spectrometry after Protein A purification and native deglycosylation. Since the D3H44/trastuzumab heterodimer contains only Fc N-linked glycans, this system was treated with only one enzyme, N-glycosidase F (PNGase F). Purified samples were deglycosylated by PNGase F as follows. 0.2 U PNGase F/μg antibody in 50 mM Tris-HCl, pH 7.0 was incubated overnight at 37° C. to a final protein concentration of 0.5 mg/mL. In the D3H44/cetuximab and trastuzumab/cetuximab systems, the systems were treated with N-glycosidase F plus multiple exoglycosidases due to additional N-linked glycans in the Fab region of cetuximab. Typically, four enzyme mixtures were used for this purpose. N-glycosidase F, β-galactosidase (Prozyme), β-N-acetylglucosaminidase (New England Biolabs) and neuraminidase. N-glycosidase F removes Fc N-linked glycans, while exoglycosidase cleaves Fab N-linked glycans into the uniform core structure of M3F (GlcNAc ₂ Man ₃ Fuc ₁ ). Purified samples were deglycosylated with four enzyme mixtures as follows. 0.2 U PNGase F/μg antibody, 0.002 U α-neuraminidase/μg antibody, 0.0001 U β-galactosidase/μg antibody and 0.2 U β-N-acetyl in 50 mM Tris-HCl pH 7.0 Glucosaminidase/μg antibody was incubated overnight at 37° C. to a final protein concentration of 0.5 mg/mL. However, in some cases, three enzymatic treatments (N-glycosidase F, β-galactosidase and neuraminidase) were preferred to avoid mass overlap of sample constituents in LC-MS analysis. In these cases, the Fab glucan _was reduced to the _slightly larger structure G0F ( _{Man3GlcNAc2Fuc1GlcNAc2 )} . Purified samples were deglycosylated with the three enzyme mixtures using the same concentrations and conditions as described for the four sample mixtures. After deglycosylation, samples were stored at 4°C prior to LC-MS analysis.

Deglycosylated protein samples were analyzed by crude LC-MS using an Agilent 1100 HPLC system coupled to an LTQ-Orbitrap XL mass spectrometer (ThermoFisher Scientific) via an Ion Max electrospray ion source (ThermoFisher). Sample (5 μg) was injected onto a 2.1×30 mm Poros R2 reverse phase column (Applied Biosystems) with the following gradient conditions: 20% solvent B for 0-3 minutes, 20-90% solvent B for 3-6 minutes. , 90-20% solvent B for 6-7 minutes and 20% solvent B for 7-9 minutes. Solvent A was degassed 0.1% aqueous formic acid and solvent B was degassed acetonitrile. The flow rate was 3 mL/min. The flow was split after the column to direct 100 μL to the electrospray interface. The column was heated to 82.5° C. and the solvent was pre-column heated to 80° C. to improve protein peak shape. The LTQ-Orbitrap XL was calibrated by using ThermoFisher Scientific's LTQ-Positive Ion ESI calibration solution (caffeine, MRFA and Ultramark 1621) and adjusted by using a solution of 10 mg/mL CsI. The cone voltage (source fragmentation setting) was 40 V, the FT resolution was 7,500, and the scan range was m/z 400-4,000. LTQ-Orbitrap XL was adjusted for optimal detection of target proteins (>50 kDa).

A range containing multiply charged ions from full-sized antibodies (m/z 2000-3800) and half-antibodies (m/z 1400-2000) were analyzed using MassLynx's MaxEnt 1 module (Waters), instrument control and data analysis software. were separately deconvoluted into molecular weight profiles using Briefly, raw protein LC-MS data were first opened by Xcalibur's spectral viewing module, QualBrower (Thermo Scientific), and converted using Databridge, a file conversion program provided by Water, Adapted for MassLynx. Transformed protein spectra were observed by MassLynx's Spectrum module and deconvoluted by using MaxEnt 1. The abundance of different antibody species in each sample was determined directly from the molecular weight profiles obtained.

Representative Designs in SMCA Assay A total of 25 representative designs with high average LCCA performance values were selected from clusters 1-12 for testing in SMCA format. Representative designs were selected based on corresponding design sets occupying similar spaces that use similar drivers but also share similar mutations. At least one representative design was selected from each cluster. Several clusters were represented by one representative design (ie clusters 1, 5, 7, 8, 10). The remaining clusters had more than one representative design, either because the cluster was large (i.e. cluster 2) or composed of small clusters (i.e. clusters 3, 4, 6, 9, 11 and 12). did. The designs within each cluster shared sequence similarity, but smaller clusters within the cluster differed by at least one driver and set of mutations. For clusters made up of smaller clusters, additional representative designs were selected from each smaller cluster.

Amino acid substitutions in each cluster are covered in Tables 15-27 and the corresponding representatives in each cluster/small cluster are shown. In Cluster 1, only one design (9134-9521) was selected to represent the cluster because these designs utilized similar electrostatic drivers occupying similar spaces (Table 15). In all members of this cluster, H1 was designed such that negatively charged substitutions (L124E and Q179E) could form salt bridges with positively charged substitutions of L1 (S176R and either S131K or S131R). H2 was designed so that positively charged substitutions (L124R and either Q179K or S186K) could form salt bridges with negatively charged substitutions in L2 (S176D and either T178D or T178E and/or T180E). Mismatched H1L2 and H2L1 are repelled primarily by electrostatic repulsion.

In Cluster 2, two representative designs (9279-9518 and 9286-9402) were chosen to represent large clusters (see Table 16). The designs within this cluster utilized similar electrostatic drivers occupying similar spaces. In all members of this cluster, H1 is replaced by negatively charged substitutions (L124E and L143E or L143D) positively charged substitutions of L1 (S176R and any combination of (Q124K and/or T178K) or (Q124K and Q160K)) It was designed to form a salt bridge with H2 was designed such that the positively charged substitutions (L124R and Q179K or S186K or S186R) can form salt bridges with the negatively charged substitutions of L2 (S176D and T178D or T178E and/or T180E). Mismatched H1L2 and H2L1 are repelled primarily by electrostatic repulsion.

In Cluster 3, 5 representative designs (9338-9748, 9815-9825, 6054-9327, 9066-9335 and 9121-9373) were selected to represent 5 smaller clusters (see Table 17). . All members of this cluster utilize electrostatic drivers similar to H1 (L124E), L1 (S176R), H2 (L124R) and L2 (S176D), indicating preferentially paired heterodimers It allows the formation of salt bridges in the body, while mismatched pairs are repelled primarily by electrostatic repulsion. The 6054-9327 design was chosen to be representative of these small clusters, as it represents designs that primarily utilize constant region drivers. In addition to these electrostatic interactions, one small cluster also contains variable domain steric drivers (L45P in H1 and P44F in L1), therefore representatives containing variable domain drivers were chosen to represent this small cluster. (9338-9748). Another small cluster also contained variable region electrostatic drivers (Q39E in H1, Q38R in L1, Q39R in H2, Q38E in L2) in both Fab heterodimers, thus representatives containing this variable region driver were (9815-9825) was chosen to represent this small cluster. In addition, one small cluster is composed of one member, thus one representative design (9066-9335) contains a modified disulfide between F122C of H1 and Q124C of L1. The remaining small cluster, represented by 9121-9373, primarily utilizes constant region drivers with the additional substitutions of H172T in H1 and S174R in L1, slightly modifying the H1L1 constant region driver interaction while simultaneously , the effect of H172R on HC2 was also explored.

In Cluster 4, two representative designs (9168-9342 and 9118-6098) were chosen to represent two small clusters each (see Table 18). All members of this cluster utilize electrostatic drivers similar to H1 (L124E), L1 (S176R or S176K), H2 (L124R) and L2 (S176D), indicating preferentially paired heterozygous It allows the formation of salt bridges in the dimer, while mismatched pairs are repelled primarily by electrostatic repulsion. One small cluster, represented by 9118-6098, mainly utilizes shared electrostatic drivers for preferential pairing, while another small cluster, represented by 9168-9342, is responsible for the formation of additional salt bridges. Substitutions of H1 (K228D) and L1 (S121K) were further utilized, which allow.

Cluster 5 is represented by unique identifiers 9116-9349 and consists of only one member (see Table 19). This design includes electrostatic drivers at H1 (L124E), L1 (S176R), H2 (L124R) and L2 (S176D) and 3D drivers were utilized. Consequently, in preferentially paired heterodimers, charge displacement allows the formation of salt bridges. In mismatched Fab heterodimers, formation is repelled by electrostatic repulsion and additional steric effects.

For Cluster 6, two representative designs were chosen to represent two small clusters each (see Table 20). All members of this cluster utilize static region-like electrostatic drivers (Q179E in H1, S131K in L1, S186R in H2, and Q124E, Q160E and T180E in L2), which preferentially Allows salt bridge formation in combined heterodimers, while mismatched pairs are repelled primarily by electrostatic repulsion. In addition, small clusters were also composed of different variable region drivers. One small cluster, represented by unique identifiers 9814-9828, utilized electrostatic drivers (Q39E for H1, Q38R for L1, Q39R for H2 and Q38E for L2) for the variable regions. Another small cluster utilized a variable domain steric driver composed of L45P in H1 and P44F in L1. Consequently, in this small cluster, mismatched pairings are further discouraged by the steric effects introduced. This small cluster is represented by designs derived from unique identifiers 9745-9075 that differ only in the absence of Q38E in L2.

In cluster 7, only one design (9060-9756) was chosen to represent the cluster because these designs utilized similar electrostatic and cubic drivers (see Table 21 thing). Shared electrostatic drivers consisted of L143E and Q179E in H1, Q124R in L1, Q179K in H2, and Q124E, Q160E and T180E in L2. The shared steric driver consisted of A139W in H1, F116A_L135V in L1 and L135W in L2. Consequently, in preferentially paired heterodimers, charge displacements in the Fab regions allow formation of salt bridges. In mismatched heterodimers, formation is repelled by electrostatic repulsion and additional steric effects.

In Cluster 8, only one design (9820-9823) was chosen to represent the cluster, as these designs utilized similar electrostatic drivers (see Table 22). . In this variable region, Q39E was used for H1, Q38R for L1, Q39R for H2 and Q38E for L2. In the constant region, L143E was utilized for H1, Q124R, Q160K and T178R for L1, Q179K for H2, and Q124E, Q160E and T180E for L2. In preferentially paired heterodimers, charge displacements in the Fab regions allow formation of salt bridges, whereas in mismatched heterodimers, formation is largely repelled by electrostatic repulsion.

For cluster 9, two representative designs were chosen to represent two small clusters each (see Table 23). All members of this cluster utilize static region-like electrostatic drivers (L143E in H1, Q124R in L1, Q179K in H2, and Q124E, Q160E and T180E in L2), which preferentially Allows salt bridge formation in combined heterodimers, while mismatched pairs are repelled primarily by electrostatic repulsion. In addition, small clusters differ by the presence/absence of variable region drivers (L45P in H1 and P44F in L1). Consequently, in small clusters containing variable region drivers, mismatched pairings are further discouraged by the steric effects introduced. A representative design of this small cluster containing variable region drivers was derived from unique identifiers 9751-9065, differing only in the absence of Q38E in L2. A representative design for each small cluster of variable region replacements is 9611-9077.

In Cluster 10, only one design (9561-9095) was chosen to represent the cluster because they utilized similar electrostatic and steric drivers occupying similar spaces. (See Table 24). Shared electrostatic drivers consisted of L143E and Q179E in H1, also located at Q124R, Q124K or S131K in L1, Q179K in H2, and Q124E and T180E in L2. The shared steric drivers consisted of L124W in H1, V133A in L1 and V133W in L2. Consequently, in preferentially paired heterodimers, charge displacements in the Fab regions allow formation of salt bridges. In mismatched heterodimers, formation is repelled by electrostatic repulsion and additional steric effects.

For Cluster 11, three representative designs (9049-9759, 9682-9740 and 9667-9830) were selected to represent three smaller clusters each (see Table 25). All members of this cluster utilized electrostatic displacements to direct preferential pairing of heterodimers. Consequently, in preferentially paired heterodimers, charge displacements in the Fab regions allow formation of salt bridges. In mismatched heterodimers, formation is repelled primarily by electrostatic repulsion. In the small cluster represented by unique identifiers 9667-9830, the covalent substitutions are negatively charged substitutions in H1 (L143E or L143D and Q179E or Q179D), positively charged substitutions in L1 (T178R or T178K), and positively charged substitutions in H2 ( S186K or S186R, or Q179K or Q179R), and a negative charge substitution (Q124E) in L2. Another small cluster was represented by a single member of this small cluster (unique identifier 9049-9759) and additionally contained substitutions for the formation of modified disulfide bonds. The remaining cluster, represented by unique identifiers 9682-9740, utilized similar drivers for H1 and L1 as the other two smaller clusters, but different constant region H2 and L2 drivers. H2 utilized L143R or L143K and L2 utilized V133E or V133D in addition to the Q124E substitution (shared with two other smaller clusters).

In cluster 12, four representative designs (9696-9848, 9986-9978, 9692-9846 and 9587-9735) were each chosen to represent four smaller clusters (see Table 26). All members of this cluster utilized electrostatic displacements to direct preferential pairing of heterodimers. Some members additionally utilized stereo drivers. A small cluster, represented by unique identifiers 9696-9848, utilized both electrostatic and steric drivers. The shared electrostatic substitutions in this small cluster consist of L143E in H1, Q124R and T178R in L1, similarly located at S186K or S186R, or Q179K or Q179R in H2, and Q124E and T180E in L2. is doing. The shared steric substitutions in this cluster consisted of S188L in H1 and S176L or V133Y or V133W in L2, and in designs utilizing V133Y or V133W in L2, L143A or L124A were also present in H2, had high mutations. In the small cluster represented by unique identifiers 9692-9846, similar electrostatic drivers were utilized compared to the small cluster represented by unique identifiers 9696-9848, and in some members, similarly located A similar substitution T178E was utilized in L2 in place of T180E. In addition, a subset of this small cluster also utilized similar steric drivers, with similarly located substitutions T178Y or T178F in L2 instead of S176L. A small cluster, represented by unique identifiers 9986-9978, utilized only one electrostatic driver to direct preferential pairing. Similar covalent substitutions were utilized for H1, L1 and H2, but a different L2 substitution (S131E) was utilized. The remaining small cluster, represented by unique identifiers 9587-9735, utilized electrostatic drivers similar to H1 and L1 (except T178R of L1 was not utilized by all members within this small cluster). ), different electrostatic clusters were utilized in H2 (L143R or L143K) and L2 (Q124E and V133E or Q124E and V133D). A few members within this small cluster also utilized similar stereoscopic drivers, composed of S188L in H1 and S176L in L2. Overall, in preferentially paired heterodimers, charge displacements in the Fab region allow formation of salt bridges. In mismatched heterodimers, formation is repelled by electrostatic repulsion. Furthermore, in designs that also utilize cubic drivers, shaping is additionally repelled by cubic effects. Cluster 13 consists of one member 9122-9371 (see Table 27). This design utilized a modified disulfide as the covalent driver of preferential pairing of the heterodimer between F122C of H1 and Q124C of L1. In addition, since the design lacked natural interchain disulfides, disulfide bond formation was confirmed by non-reducing and reducing SDS-PAGE gels. Although this design was not tested in SMCA format, modified disulfides were tested in the presence of natural interchain disulfides in combination with additional constant region drivers (Cluster 3, representative design 9066-9335).

Transfection Method A co-expression set containing two heavy chain and two light chain constructs was transfected into CHO-3E7 cells as follows. CHO-3E7 cells were grown at 1.7-1.7°C in FreeStyle™ F17 medium (Invitrogen Cat. Cultured at 37° C. with a density of 2×10 ⁶ cells/ml. A total of 50 μg of DNA was transfected using PEI-pro (Polyplus Catalog No. 115-010) at a DNA:PEI ratio of 1:2.5 for a total volume of 50 ml. 24 hours after adding the DNA-PEI mixture, the cells were transferred to 32° C. and incubated for 7 days before harvesting. Culture medium was harvested by centrifugation and vacuum filtered using a Steriflip 0.2 μM filter. The filtered culture medium was then purified by use of Protein A MabSelect SuRe resin (GE Healthcare No. 17-5438-02) as follows. The filtered culture medium was applied to a column (Hyclone DPBS/modified, no calcium, no magnesium, number SH-300028.02) previously equilibrated with PBS. The heterodimeric antibody species were then washed with PBS and eluted onto an Amicon ultra 15 centrifugal filter Ultracel 10K (Millipore number SCGP00525) with 100 mM citrate pH 3.6. Buffers were then PBS exchanged and samples evaluated by caliper prior to deglycosylation and LC-MS.

H1:H2:L1:H1:H2:L1: A set of L2 DNA ratios were tested in CHO expression. These ratios attempt to compensate for natural differences and/or inherent pairing biases in the expression levels of the heavy and light chains of the two different antibodies. In all bispecific Ab systems, bias was observed across all ratios tested (Figure 9). In the D3H44/trastuzumab system, a bias is observed towards trastuzumab, ie the D3H44 heavy chain preferentially pairs with the trastuzumab light chain (see Figure 9a). With D3H44/cetuximab, a bias is observed towards cetuximab, ie the D3H44 heavy chain preferentially pairs with the cetuximab light chain (see Figure 9b). In the trastuzumab/cetuximab system, a bias is observed towards trastuzumab, ie the cetuximab heavy chain preferentially pairs with the trastuzumab light chain (see Figure 9c).

In testing each of the 25 representative designs in each bispecific Ab system, the H1:H2:L1:L2 DNA ratio used yielded the highest abundance of the bispecific Ab species, while at the same time lower abundance (see Tables 32a, b and c). For the D3H44/trastuzumab system, the ratios used were 15 (H1), 15 (H2), 53 (L1), 17 (L2), where H1 and L1 refer to D3H44 and H2 and L2 , see Trastuzumab. For the D3H44/cetuximab system, the ratios used were 15 (H1), 15 (H2), 17 (L1), 53 (L2), where H1 and L1 refer to trastuzumab and H2 and L2 , see cetuximab. For the D3H44/cetuximab system, the ratios used were 15 (H1), 15 (H2), 53 (L1), 17 (L2), where H1 and L1 refer to D3H44 and H2 and L2 , see cetuximab.

In addition, substitutions present in H1L1 and H2L2 were tested in the bispecific Ab system Antibody 1 (Ab1) and Antibody 2 (Ab2), respectively, in one orientation, and H1L1 and H2L2 in the other "reversed" orientation. The design tested the D3H44/cetuximab and trastuzumab/cetuximab bispecific systems in both orientations so that substitutions present in H2L2 were tested in Ab2 and Ab1, respectively (see Tables 28a and b). The "_1" attached to the unique identifier indicates that the heavy chain and related light chain substitutions (see Table 13a) that give stronger LCCA preferential pairing results indicate that the heavy chain is compared to the light chain of the other antibody. refers to a design placed in an antibody that weakly competes with related light chains as The "_2" attached to the unique identifier indicates that the heavy chain and related light chain substitutions (see Table 13a) that give stronger LCCA preferential pairing results are the heavy chains compared to the light chains of other antibodies. It refers to the opposite, "flipped" orientation, which is placed in antibodies that compete weakly with their associated light chains. For the D3H44/trastuzumab system, designs were tested in the '_1' orientation only (see Table 28c).

SMCA Results The D3H44/trastuzumab line was treated with only one enzyme (PNGase F) to completely deglycosylate. Multi-enzyme treatments generally cleaved the sugars attached to the Fab region to either core M3F (using four enzyme treatments) or G0F (using three enzyme treatments). Overall, in most cases the deglycosylation treatment provided the ability to identify all possible different species identified by LC-MS. In many cases each species was represented by a single LC-MS peak. Exceptions include side peaks (possibly heterogeneity due to adduct or leader peptide truncation) that may also correspond to the desired bispecific species, but due to the ambiguity of the side peaks. Therefore, these minor peaks were not considered for their contribution to bispecific species. In addition, some designs in the D3H44/cetuximab (3519_1, 3522_1) and trastuzumab/cetuximab (9748-9338_1) systems required multiple peaks to explain species due to the variability of the combined high mannose. did. All of these designs introduced glycosylation sites into the cetuximab light chain. In some cases, it was not possible to distinguish some minority species (composing less than 5% of all species) due to low mass separations between species (i.e. differences of less than 50 Da). rice field. Furthermore, the desired bispecific H1-L1_H2-L2 is generally indistinguishable experimentally from the mismatched H1-L2_H2-L1 based on LC/MS. Thus, when the bispecific content is reported in the tables, it cannot be completely ruled out that it does not contain mismatched species of this kind. However, the very low abundances observed in species such as H1-L2_H1-L2 and H2-L1_H2-L1, and H1-L2 and H2-L1 half-antibodies suggest that very, if any, of the bispecific species This indicates that less contamination occurs.

LC-MS data are presented in Tables 29a, 29b and 29c. For comparison, wild-type data are also presented in Tables 33a, 33b, and 33c, with the "SMCA Unique Identifier" column and the "Cluster" column indicated by "NA". All three bispecific wild-type lines showed a biased bias with one light chain preferentially binding to both heavy chains (see Table 33 and Figure 9). Furthermore, at least in the trastuzumab/cetuximab system, tag placement also appears to have a significant impact on H1L1 and H2L2 pairing. Therefore, to assess the effect of the design on the transferability and rate of the desired bispecific species relative to the wild-type, the same H1:H2:L1:L2 DNA ratios as the corresponding wild-type bispecific constructs were used. comparisons were performed, and "change in percentage of H1L1 pairings to wild type (for all H1 species)", "change in percentage (%) of H2L2 pairings to wild type (for all H2 species) ' and 'H1:H2:L1:L2 ratio (%) change relative to wild type' (considering only full-size antibody species) (see Table 29). In evaluating either the H1L1 percent pairing (for all H1 species) or the H2L2 percent pairing (for all H2 species), all species are evaluated for pairing in the Fab region. If the corresponding wild-type bispecific species antibody was not assessed by SMCA, comparisons were made in similar wild-type constructs. Estimated values are indicated by " ^*** " immediately next to the reported value. Similar wild-type constructs selected for comparison were selected as follows. To assess translocation potential, each wild-type construct was subjected to SMCA experiments (different ratios (implemented in ). To assess the effect of the design on the ratio of the desired bispecific species to the wild-type, each wild-type construct was added to the highest H1:H2:L1:L2 ratio (%) (full-size antibody species only). were represented by SMCA experiments (performed at different ratios) that showed In both cases, of all wild-type constructs in the bispecific species, the median value was chosen as the wild-type value for comparison.

In each design, transferability was determined by the percentage of overall H:L pairings across all species for WT, specifically the percentage of H1L1/all H1 species and/or the percentage of H2L2/all H2 species ( %) by noting the increase. In addition, the effect on the rate of the desired bispecific species is also assessed, as half-antibodies, if present, can be removed/minimized by preparative SEC or by further H1:H2:L1:L2 DNA titrations. , highlighting the comparison of full-size antibody species only. Tables 30a, b and c show that preparative SEC can be effective in removing/minimizing half-Ab species. Tables 32a, b and c show that the rate of half-Ab species can also be manipulated during transfection using various DNA titration ratios.

In the D3H44/cetuximab system (Table 29a), all but one design (9327-6054_2) was transferred, assessed by H1L1 pairing across all species with respect to wild type. Most designs (except 9327-6054_2 and 9134-9521_2) also showed increased rates of desired bispecific antibodies when only full Ab species were considered. Furthermore, except for one design (9327-6054_2) that did not migrate, the designs reduced the major mismatch antibody species (H1H2L2L2) observed in the wild type. In addition, with the exception of 9327-6054_2 and the corresponding 9327-6054_1 in the other orientation, the designs were migrated in both orientations, and most designs had similar valid H:L pairings in both orientations. Indicated.

In the D3H44/trastuzumab system (Table 29b), all designs were transferred, assessed by H1L1 pairing across all species with respect to wild type. In addition, all designs showed increased rates of desired bispecific antibodies (when only full Ab species were considered). Furthermore, most designs significantly reduced the major mismatch antibody species (H1H2L2L2) observed in wild type. However, no data were reported for 9611-9077_1 due to lack of expression (Table 28c).

In the trastuzumab/cetuximab system (Table 29c), at least 35 of the 49 designs evaluated H2L2 pairing across all H2 species to demonstrate transitional potential ("Change in % H2L2 pairing relative to wild type (for all H2 species)" column). Designs that may not have migrated include 9279-9518_2, 3522_2, 9815-9825_2, 9327-6054_2, 9118-6098_2, 9748-9338_2, 9692-9846_2, 9587-9735_2, 9814-9828_2, 3519_2, 9782-9 9168-9342_2 and 9066-9335_1 are included (negative values in column "% change in H2L2 pairing relative to wild type (for all H2 species)"), while the other orientation design is transferable. (Note 9279-9518_1 was not tested due to lack of sample). All designs that showed transferability showed a reduction in the rate of the major mismatched antibody species (H1H2L1L1) observed in wild-type experiments. In addition, of the designs that were transferred, only two designs (9134-9521_1 and 9279-9518_2) showed the desired ratio of bispecific Abs compared to wild type when considering only the full antibody species. showed a decrease.

In general, most designs with increased H:L pairings of weakly competing antibodies have resulted in increased rates of the desired bispecific species (considering full-size antibodies only). In terms of orientation, most designs showed similar or better migration potential compared to the H:L pairing compared to the '_2' pairing in the '_1' orientation, with the exception of trastuzumab/cetuximab was primarily observed in the system). In addition, Table 35a covers designs that migrated in both orientations in all three bispecific systems tested (D3H44/cetuximab, D3H44/trastuzumab and trastuzumab/cetuximab). Table 35b shows designs that migrated in one orientation in all three bispecific systems tested (D3H44/cetuximab, D3H44/trastuzumab and trastuzumab/cetuximab) and the other orientation in only one bispecific system. Covers migrated designs in . In addition, in a specific orientation, the same mutation is present in the heavy chains of the three bispecific systems and the weakly competing cognate light chain, with light chain utilization exceeding at least 10%.

With respect to cluster migration potential and performance, in the D3H44/trastuzumab bispecific system, all members within all clusters showed migration potential (see Figure 11a) and the desired bispecific antibody (considering full-size antibodies only) increased (see Figure 11b). In the D3H44/cetuximab bispecific system, all clusters showed transferability and only one member in cluster 3 showed a decrease in H1L1 pairing across all H1 species compared to wild type ( see Figure 11c). All clusters also contained members that showed an increase in the ratio of desired bispecific antibodies to wild-type (considering only full-length antibodies), but three clusters (clusters 1, 3 and 4) Members that showed a decrease in the ratio of desired bispecific antibodies (considering full-length antibodies only) to wild-type were also included (see Table 11d). For the trastuzumab/cetuximab bispecific system, all clusters contain variants showing desirable translocation potential, but only a few clusters (1, 5, 7, 8, 10, 11) are All contain variants that showed migration potential (see Figure 11e). In addition, all clusters contain members that show an increase in the rate of desired bispecific antibodies (considering full-size antibodies only) relative to wild-type (see Figure 11f). In clusters where all members showed migration potential, all members in clusters 5, 7, 8, 10 and 11 increased the desired bispecific antibody rate (considering full-size antibodies only) relative to wild type. also showed.

Overall, when all three bispecific systems are considered together, all members in clusters 1, 5, 7, 8, 10 and 11 show transferability (see Figure 11g), Clusters 5, 7, 8, 10 and 11 contained members that all showed an increase in the rate of desired bispecific antibodies (considering full-size antibodies only) relative to wild type (see Figure 11h). to do).

Overall, most designs with increased H:L pairings of weakly competing antibodies were accompanied by increased rates of the desired bispecific species (considering full-size antibodies only). In terms of orientation, most designs showed similar or better migration potential compared to the H:L pairing compared to the '_2' pairing in the '_1' orientation, with the exception of trastuzumab/cetuximab was primarily observed in the system).

Example 10: Preparative Size Exclusion Chromatography (SEC) of SMCA Bispecific Heterodimeric Antibodies and Parental Mabs Selected for Biophysical Characterization
A subset of SMCA samples was selected for additional biophysical characterization. The majority of these SMCA samples typically show high pairing (greater than about 80% in the H1L1+H2L2/all species column) and low abundance half-antibody species (approximately less than 30%). Preparative SEC was performed as follows. Heterodimeric antibody samples were separated by use of a Superdex 200 10/300 GL (GE Healthcare) column attached to a Pharmacia (GE Healthcare) AKTA Purifier system. Heterodimeric antibody samples (0.3-0.5 ml) in PBS (Hyclone DPBS/modified, no calcium, no magnesium, Cat# SH-300028.02) were added to a 0.5 ml loop filled with PBS. Manually loaded. The sample was then automatically injected onto the column and resolved at 0.5 ml/min with a 1 CV elution volume. Protein elution was monitored at _OD280 and collected in 0.5 ml fractions. For each SMCA sample, those fractions containing the major peak were pooled and further biophysically characterized.

Example 11: Evaluation of preferential pairing of bispecific heterodimers in antibody format after preparative size exclusion chromatography. Bispecific heterodimeric antibodies were analyzed for preferential binding by using LC-MS methods. These samples all increase the percentage of the desired bispecific antibody species and decrease the percentage of half antibody species (Tables 29 and 30).

Example 12: Thermal Stability of SMCA Bispecific Heterodimeric Antibodies After preparative SEC, the thermal stability of selected SMCA bispecific heterodimeric antibodies was measured and the parental D3H44 and trastuzumab monoclonals antibody, as well as the cetuximab 1-arm antibody. In general, a one-arm antibody is composed of one full-length heavy chain, one truncated heavy chain that lacks the Fab region (and incorporates a C233S substitution), and one light chain and is described in Example 9. Refers to constructs with heavy chain heterodimerization achieved as described.

Measurement of Thermal Stability Thermal stability of selected bispecific heterodimeric antibodies and wild-type controls was measured using differential scanning calorimetry (DSC) as follows. After preparative SEC processing, 400 μL samples were used for DSC analysis by VP-Capillary DSC (GE-Healthcare), mainly at concentrations of 0.2 mg/ml or 0.4 mg/ml in PBS. At the beginning of each DSC run, five buffer blank injections were performed to stabilize the baseline, and buffer injections were performed prior to each sample injection for reference. Each sample was scanned from 20-100°C at a rate of 60°C/hr, low feedback, 8 sec filter, 5 min preTstat and 70 psi nitrogen pressure. The resulting thermograms were referenced and analyzed using Origin 7 software.

The results are shown in Tables 31a, b and c. Fab Tm values reported in the table for wild-type were obtained from homodimeric antibodies of D3H44 (79° C.) and trastuzumab (81° C.) and from the one-arm antibody of cetuximab (72° C.). Only two peaks corresponding to Fab Tm are observed for WT D3H44/Cetuximab and Trastuzumab/Cetuximab heterodimeric antibodies. No distinct peaks are observed in CH2 (due to overlap with cetuximab Fab) or CH3 (due to overlap of D3H44 and trastuzumab Fab Tm values). For the WT D3H44/trastuzumab heterodimeric antibody, the similar Tm values for the two Fabs, D3H44 and Trastuzumab, suggest that the peak at 81°C may correspond to both Fabs, whereas the peak at approximately 72°C The peak at may correspond to CH2.

In Tables 31a, b and c, only Tm values for peaks corresponding to both Fabs were reported unless otherwise indicated. Note also that in some heterodimer samples, protein concentrations were low (less than 0.4 mg/mL), resulting in increased noise at baseline. As a result, in the D3H44/trastuzumab system, some samples produced DSC curves with low peak intensities such that the CH2 peak was difficult to distinguish and could destabilize the Fab. In this case, Tm values of 70-72° C. were also reported (Table 31a). Overall, most heterodimeric antibodies exhibit thermostabilities (3° C. or less) similar to the corresponding wild-type molecules. Furthermore, most heterodimers show no additional peaks suggesting significant instability of the CH2 or CH3 peaks. One exception includes the trastuzumab/cetuximab heterodimeric engineered heterodimeric antibody 9611-9077_2, which exhibits an additional peak at 60° C. that may be due to CH2 destabilization.

Example 13: Determination of Antigen Affinity of Bispecific Heterodimeric Antibodies The Ability of Bispecific Antibodies to Bind to Related Antigens is Determined if Amino Acid Substitutions Have Any Effects on Antigen Binding evaluated for Antigen binding affinities were determined by SPR as follows.

SPR biosensor assay EDC: 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride, NHS: N-hydroxysuccinimide, SPR: surface plasmon resonance, EDTA: ethylenediaminetetraacetic acid, TF: tissue factor, EGFR ECD : epidermal growth factor receptor extracellular domain, Her2 ECD: human epidermal growth factor receptor 2 extracellular domain.

SPR supply. Series S Sensor Chip CM5, Biacore Amine Coupling Kit (NHS, EDC and 1 M ethanolamine), and 10 mM sodium acetate buffer were purchased from GE Healthcare Life Sciences (Mississauga, ON). Recombinant Her2 extracellular domain (ECD) protein was purchased from eBioscience (San Diego, Calif.). PBS running buffer with 1% Tween 20 (PBST) was obtained from Taknova Inc. Goat polyclonal human Fc antibodies were purchased from Jackson Immuno Research Laboratories Inc. (Hollister, Calif.). (West Grove, PA). EDTA was purchased from Bioshop (Burlington, ON).

All surface plasmon resonance assays were performed on a Biacore T200 Surface Plasmon Resonance instrument (GE Healthcare Life Sciences, GE Healthcare Life Sciences, Inc.) at a temperature of 25° C. with PBST running buffer (0.5 M EDTA stock solution was added to give a final concentration of 3.4 mM). (Mississauga, ON)). Anti-human Fc capture surfaces were generated using a Series S Sensor Chip CM5 using default parameters under the Immobilization Wizard with the Biacore T200 control software set to target 2000 resonance units (RU). Selection of antibody variants binding to Her2 ECD, TF or EGFR ECD antigen targets was performed in two steps. Indirect capture of antibody variants to anti-human Fc antibody flow cell surface followed by injection of 5 concentrations of purified antigen for kinetic analysis using single cycle kinetic methodology. Variants or controls were injected at 1 μg/mL into individual flow cells for 60 seconds at a flow rate of 10 μL/min for capture. Typically, this results in entrapment of approximately 50-100 RU on the anti-human Fc surface. The first flow cell was left empty and used as a blank control. Immediately following this capture step, five concentrations of antigen (5 nM, 2.5 nM, 1.25 nM, 0.63 nM and 0.31 nM for TF or EGFR ECD antigens or 40 nm, 20 nm, 10 nm, 5 nm for Her2 ECD antigens) and 2.5 nm) were injected sequentially over all four flow cells at 100 μL/min for 180 s with dissociation steps of 300 s for EGFR ECD, 1800 s for Her2 ECD and 3600 s for TF. The capture antibody surface was regenerated with 10 mM glycine, pH 1.5 at 30 μL/min for 120 seconds to prepare for the next injection cycle. At least two sham buffer injections were performed for each analyte injection used as a reference. The resulting single-cycle kinetic sensorgrams were analyzed using the Biacore T200 BiaEvaluation software and fitted to a 1:1 binding model.

Antigen affinities of the heterodimeric antibodies were evaluated with reference to corresponding wild-type controls: D3H44, trastuzumab OAA and cetuximab OAA Mabs. Antigen affinities were also obtained from wild-type bispecific antibodies, but since SPR capture of WT bispecifics can be heterogeneous (e.g., with capture of mismatched heterodimers), Prevents determination of KD (see Tables 31a and c). For heterodimeric antibodies with antigen binding measured in the D3H44/cetuximab system, antigen affinities were similar to corresponding WT controls (see Table 31b). For most heterodimeric antibodies with antigen binding measured in both the D3H44/trastuzumab and trastuzumab/cetuximab systems, antigen affinities were similar to corresponding WT controls (see Tables 31a and c). thing). Exceptions include 11 engineered antibodies that showed no Her2 binding. In both the D3H44/trastuzumab and trastuzumab/cetuximab systems, no Her2 binding was observed in the six engineered heterodimeric antibodies 9049-9759_1 and 9682-9740_1 and 3522_1. Furthermore, in the Trastuzumab/Cetuximab system, five additional modified antibodies 9696-9848_1, 9561-9095_2, 9611-9077_2, 9286-9402_2 and 9060-9756_2 also lacked binding to Her2. Ten of these 11 engineered antibodies shared constant region mutations in the heavy (L143E_K145T) and light (Q124R_T178R) chains. The other engineered antibody 9286-9402_2 shared the same constant region mutations in the heavy chain (L143E_K145T) and similar mutations in the light chain (Q124K and S176R).

Example 14: Ultra Performance Liquid Chromatography Size Exclusion Chromatography (UPLC-SEC) Profiles of Engineered Heterodimeric Antibodies, and Wild-Type Heterodimeric and Homodimeric Antibodies Engineered Heterodimeric Antibodies, and Control Wild After preparative SEC of bispecific and homodimeric antibodies, a Waters BEH200 SEC column (2.5 mL, 4.6 x 150 mm) was set at 30°C and mounted on a Waters Acuity UPLC system with a PDA detector. , stainless steel, 1.7 μm particles) were used to perform UPLC-SEC. Run time consisted of 7 minutes, total volume per injection was 2.8 mL, running buffer PBS and 0.02% polysorbate 20 or 20 mM NaPO4, 50 mM KCl, 0.02% polysorbate 20. , 10% acetonitrile, pH 7 was used at 0.4 ml/min. Elution was monitored by UV absorption in the range 210-400 nm and chromatograms were extracted at 280 nm. Peak integration was performed using Empower3 software.

FIG. 10 shows UPLC-SEC profiles of a representative modified heterodimeric antibody and a representative WT heterodimer. In most cases, the engineered heterodimeric antibodies showed similar UPLC-SEC profiles to the corresponding WT heterodimeric antibodies, with mean percent monomers in each of D3H44/trastuzumab, D3H44/cetuximab and trastuzumab/cetuximab. was 99.18%, 98.70% and 98.77% (see Tables 31a, 31b and 31c).

Although the invention has been shown and described with reference to preferred and various alternative embodiments, it is understood that various changes in form and detail can be made without departing from the spirit and scope of the invention. It is understood by those skilled in the art.

All references, issued patents, patent publications and sequence accession numbers disclosed herein are hereby incorporated by reference in their entirety for all purposes.

TABLES KEY TO TABLES Table 1: Fab Model Key Criteria Table 2: Hotspot Amino Acid Positions in the Heavy and Light Chain Interface of D3H44 (a Typical Fab Containing a Kappa Light Chain) Table 3A. Kabat numbering table 3B.D3H44, heavy chain amino acid sequences of trastuzumab and cetuximab. D3H44, Kabat numbering table of trastuzumab and cetuximab light chain amino acid sequences 3C: D3H44, trastuzumab and cetuximab amino acid and DNA sequence table 4. LCCA design table 5.H1 has a modification in one immunoglobulin heavy chain and/or two immunoglobulin light chains that preferentially pairs with L1. Design library table6. Core design table7. Examples of combinatorial design Table 8. Examples of modified/optimized designs Table 9. Examples of combinatorial designs including optimization designs Table 10. Examples of combinatorial designs, including independent designs Table 11: H1:L1:L2 DNA ratios used for light chain competition assay and validation Table 12. Evaluation of LCCA Performance, Stability and Antigen Binding of LCCA Designs Arranged to Reduce DSF Values of H1L1 Fab Heterodimers Table 13a. LCCA performance table for designs meeting the LCCA average performance criteria of 86:14 correctly matched Fab heterodimers:mismatched Fab heterodimers 13b. Accurately matched Fab heterodimer: Mismatched Fab heterodimer 86:14 Stability and Antibody Binding Evaluation of Designs Meeting LCCA Average Performance Criteria Table 14a. LCCA performance table for designs performing below the LCCA average performance criteria of correctly matched Fab heterodimer:mismatched Fab heterodimer 86:14. 14b. Accurately matched Fab heterodimers: Stability and antibody binding evaluation of designs performing below the LCCA average performance criteria of 86:14 mismatched Fab heterodimers. Design table for cluster 1 with representative design 16. Design Table for Cluster 2 with Representative Designs 17. Design Table for Cluster 3 with Representative Designs 18. Design table for cluster 4 with representative design 19. Design table for cluster 5 with representative design 20. Design Table for Cluster 6 with representative designs 21 . Design Table for Cluster 7, including representative designs 22. Design table for cluster 8 with representative designs 23 . Design table for cluster 9 with representative designs 24 . Design table for cluster 10, including representative design 25. Design table for cluster 11 containing representative designs 26 . Design table for cluster 12 containing representative designs 27 . Design table 28a. of cluster 13 containing a representative design. SMCA Unique Identifiers for Trastuzumab/Cetuximab Bispecific Systems Table 28b. SMCA Unique Identifiers for the D3H44/Cetuximab Bispecific System Table 28c. SMCA Unique Identifiers for the D3H44/Trastuzumab Bispecific System Table 29a. LC-MS paired data and post-pA yield (mg/L) of heterodimeric antibody derived from the D3H44 (H1L1)/cetuximab (H2L2) bispecific system.
Table 29b. LC-MS paired data and post-pA yield (mg/L) of heterodimeric antibody derived from D3H44 (H1L1)/trastuzumab (H2L2) bispecific system.
Table 29c. LC-MS paired data and post-pA yields (mg/L) of heterodimeric antibodies derived from the Trastuzumab (H1L1)/Cetuximab (H2L2) bispecific system.
Table 30a. LC-MS pairing table of heterodimeric antibodies derived from the D3H44 (H1L1)/Cetuximab (H2L2) bispecific system after preparative SEC 30b. LC-MS pairing table of heterodimeric antibodies derived from the D3H44 (H1L1)/trastuzumab (H2L2) bispecific system after preparative SEC 30c. LC-MS pairing table of heterodimeric antibodies derived from Trastuzumab (H1L1)/Cetuximab (H2L2) bispecific system after preparative SEC 31a. Biophysical Characterization of Selected Designs Derived from the D3H44/Trastuzumab System (Antigen Binding, Thermal Stability, UPLC-SEC)
Table 31b. Biophysical Characterization of Selected Designs Derived from the D3H44/Cetuximab System (Antigen Binding, Thermal Stability, UPLC-SEC)
Table 31c. Biophysical Characterization of Selected Designs Derived from the Trastuzumab/Cetuximab System (Antigen Binding, Thermal Stability, UPLC-SEC)
Table 32a. Effect of DNA titration ratio of wild-type D3H44/trastuzumab system on percentage of antibody species as assessed by LC-MS. H1 and L1 refer to the heavy and light chains of D3H44, respectively. H2 and L2 refer to the heavy and light chains of trastuzumab, respectively.
Table 32b. Effect of DNA titration ratio of wild-type D3H44/cetuximab system on percentage of antibody species as assessed by LC-MS. H1 and L1 refer to the heavy and light chains of D3H44, respectively. H2 and L2 refer to the heavy and light chains of cetuximab, respectively.
Table 32c. Effect of DNA titration ratio of wild-type trastuzumab/cetuximab system on percentage of antibody species as assessed by LC-MS. H1 and L1 refer to the heavy and light chains of trastuzumab, respectively. H2 and L2 refer to the heavy and light chains of cetuximab, respectively.
Table 33a. LC-MS pairing table of wild-type antibody constructs derived from the D3H44 (H1L1)/cetuximab (H2L2) bispecific system 33b. LC-MS pairing table of wild-type antibody constructs derived from the D3H44 (H1L1)/trastuzumab (H2L2) bispecific system 33c. LC-MS pairing table of wild-type antibody constructs derived from the Trastuzumab (H1L1)/Cetuximab (H2L2) bispecific system.34. Stabilizing mutations in Fab heterodimers Table 35a. Design table showing mobility in all three bispecific systems (D3H44/cetuximab, D3H44/trastuzumab and trastuzumab/cetuximab) in both orientations 35b. The mobility in all three bispecific systems (D3H44/cetuximab, D3H44/trastuzumab and trastuzumab/cetuximab) was shown in one orientation and migrated in the other orientation only in one bispecific system while simultaneously migrating to the other orientation. Designs that also meet at least 10% of the chain utilization criteria

(Table 1) Main criteria for Fab model

*Ｋａｂａｔ番号付け Table 2. Hotspot amino acid positions at the heavy and light chain interface of D3H44 (a typical Fab containing a kappa light chain)

*Kabat numbering

(Table 3A) Kabat numbering of the heavy chain amino acid sequences of D3H44, Trastuzumab and Cetuximab

(Table 3B) Kabat numbering of the light chain amino acid sequences of D3H44, Trastuzumab and Cetuximab

(Table 3C) Amino acid and DNA sequences of D3H44, Trastuzumab and Cetuximab

*Ｋａｂａｔ番号付け；ＷＴは、アミノ酸突然変異を有さない野性型免疫グロブリン鎖を指す。
**Ｈ１、Ｌ１及びＬ２突然変異それぞれの特有のセット（ＬＣＣＡフォーマット）が、セット番号または「特有の識別子」として指定された。 (Table 4) LCCA designs in which H1 has modifications in one immunoglobulin heavy chain and/or two immunoglobulin light chains that preferentially pair with L1

*Kabat numbering; WT refers to wild-type immunoglobulin chains with no amino acid mutations.
**The unique set (LCCA format) of each H1, L1 and L2 mutation was designated as the set number or "unique identifier".

*Ｋａｂａｔ番号付け；ＷＴは、アミノ酸突然変異を有さない野性型免疫グロブリン鎖を指す。
**「特有の識別子」は、２つの構成要素ＬＣＣＡの特有の識別子から構成される、またはＳＭＣＡフォーマットのみにより試験された設計の単一識別子から構成される。 (Table 5) Design library

*Kabat numbering; WT refers to wild-type immunoglobulin chains with no amino acid mutations.
** "Unique Identifier" consists of the unique identifier of a two component LCCA or a single identifier of a design tested by the SMCA format only.

*Ｋａｂａｔ番号付け；ＷＴは、アミノ酸突然変異を有さない野性型免疫グロブリン鎖を指す。
**「特有の識別子のセット」は、２つの構成要素ＬＣＣＡの特有の識別子から構成される。 (Table 6) Core design

*Kabat numbering; WT refers to wild-type immunoglobulin chains with no amino acid mutations.
**The "set of unique identifiers" consists of the unique identifiers of the two component LCCAs.

*Ｋａｂａｔ番号付け (Table 7) Example of combination design

*Kabat numbering

*Ｋａｂａｔ番号付け (Table 8) Examples of modified/optimized designs

*Kabat numbering

(Table 9) Example of combinatorial design including optimization design

*Ｋａｂａｔ番号付け；ＷＴは、アミノ酸突然変異を有さない野性型免疫グロブリン鎖を指す。 (Table 10) Examples of Combinatorial Designs Including Independent Designs

*Kabat numbering; WT refers to wild-type immunoglobulin chains with no amino acid mutations.

^追加のＤＮＡ：ＡＫＴｄｄ ｐＴＴ２２は、構成的に活性なタンパク質キナーゼＢ突然変異体（有意な陽性ＡＫＴ突然変異体）を含有するベクターを指す。ｓｓＤＮＡは、サケ精子ＤＮＡを指す。 Table 11. H1:L1:L2 DNA ratios used for light chain competition assay and validation

^Additional DNA: AKTdd pTT22 refers to a vector containing a constitutively active protein kinase B mutant (significantly positive AKT mutant). ssDNA refers to salmon sperm DNA.

*結合Ｌ鎖タグ（ＨＡまたはＦＬＡＧ）の存在／不在のみが異なる他のＦａｂヘテロ二量体から導かれた推定値を示す。
**３３３（Ｈ１）、２５０（Ｌ１）、７４９（Ｌ２）のＬＣＣＡ実験から導かれた値。
***３３３（Ｈ１）、１００（Ｌ１）、８９９（Ｌ２）のＬＣＣＡ実験から導かれた値。
ＮＤは、利用可能なデータがないことを示す。 Table 12. Evaluation of LCCA performance, stability and antigen binding of LCCA designs set to reduce DSF values of H1L1 Fab heterodimers

*Indicates estimates derived from other Fab heterodimers that differ only in the presence/absence of the associated light chain tag (HA or FLAG).
**Values derived from 333 (H1), 250 (L1), 749 (L2) LCCA experiments.
***Values derived from 333 (H1), 100 (L1), 899 (L2) LCCA experiments.
ND indicates no data available.

*値は、Ｌ１：Ｌ２のＤＮＡ比１：３で実施したＬＣＣＡ実験から得て、Ｌ１：Ｌ２のＤＮＡ比を１：１に正規化した。
**値は、Ｌ１：Ｌ２のＤＮＡ比１：９で実施したＬＣＣＡ実験から得て、Ｌ１：Ｌ２のＤＮＡ比を１：１に正規化した。
***「特有の識別子」は、（Ｈ１Ｌ１Ｌ２のセット番号－Ｈ２Ｌ２Ｌ１のセット番号）または（Ｈ２Ｌ２Ｌ１のセット番号－Ｈ１Ｌ１Ｌ２のセット番号）の配向のいずれかの２つの構成要素ＬＣＣＡの特有の識別子からなる。 (Table 13a) LCCA performance of a design meeting the LCCA mean performance criteria of 86:14 correctly matched Fab heterodimers:mismatched Fab heterodimers

*Values are from LCCA experiments performed with L1:L2 DNA ratios of 1:3, normalized to L1:L2 DNA ratios of 1:1.
** Values were obtained from LCCA experiments performed with L1:L2 DNA ratios of 1:9 and normalized to L1:L2 DNA ratios of 1:1.
*** "Unique Identifier" consists of a unique identifier of two component LCCAs in either orientation (H1L1L2 set number - H2L2L1 set number) or (H2L2L1 set number - H1L1L2 set number) .

*結合Ｌ鎖タグ（ＨＡまたはＦＬＡＧ）の存在／不在のみが異なる他のＦａｂヘテロ二量体から導かれた推定値を示す。
**「特有の識別子」は、（Ｈ１Ｌ１Ｌ２のセット番号－Ｈ２Ｌ２Ｌ１のセット番号）または（Ｈ２Ｌ２Ｌ１のセット番号－Ｈ１Ｌ１Ｌ２のセット番号）の配向のいずれかの２つの構成要素ＬＣＣＡの特有の識別子からなる。 (Table 13b) Accurately matched Fab heterodimer: Stability and antibody binding evaluation of designs meeting LCCA average performance criteria with 86:14 mismatched Fab heterodimers

*Indicates estimates derived from other Fab heterodimers that differ only in the presence/absence of the associated light chain tag (HA or FLAG).
**"Unique Identifier" consists of a unique identifier of two component LCCAs in either (H1L1L2 set number-H2L2L1 set number) or (H2L2L1 set number-H1L1L2 set number) orientations.

*値は、Ｌ１：Ｌ２のＤＮＡ比１：３で実施したＬＣＣＡ実験から得て、Ｌ１：Ｌ２のＤＮＡ比を１：１に正規化した。
**値は、Ｌ１：Ｌ２のＤＮＡ比１：９で実施したＬＣＣＡ実験から得て、Ｌ１：Ｌ２のＤＮＡ比を１：１に正規化した。
***「特有の識別子」は、（Ｈ１Ｌ１Ｌ２のセット番号－Ｈ２Ｌ２Ｌ１のセット番号）または（Ｈ２Ｌ２Ｌ１のセット番号－Ｈ１Ｌ１Ｌ２のセット番号）の配向のいずれかの２つの構成要素ＬＣＣＡの特有の識別子からなる。 (Table 14a) Correctly matched Fab heterodimers: LCCA performance of designs performing below the LCCA average performance criteria of 86:14 mismatched Fab heterodimers

*Values are from LCCA experiments performed with L1:L2 DNA ratios of 1:3, normalized to L1:L2 DNA ratios of 1:1.
** Values were obtained from LCCA experiments performed with L1:L2 DNA ratios of 1:9 and normalized to L1:L2 DNA ratios of 1:1.
*** "Unique Identifier" consists of a unique identifier of two component LCCAs in either orientation (H1L1L2 set number - H2L2L1 set number) or (H2L2L1 set number - H1L1L2 set number) .

*結合Ｌ鎖タグ（ＨＡまたはＦＬＡＧ）の存在／不在のみが異なる他のＦａｂヘテロ二量体から導かれた推定値を示す。
**「特有の識別子」は、（Ｈ１Ｌ１Ｌ２のセット番号－Ｈ２Ｌ２Ｌ１のセット番号）または（Ｈ２Ｌ２Ｌ１のセット番号－Ｈ１Ｌ１Ｌ２のセット番号）の配向のいずれかの２つの構成要素ＬＣＣＡの特有の識別子からなる。 (Table 14b) Accurately matched Fab heterodimers: Evaluation of stability and antibody binding of designs performing below the LCCA average performance criteria of 86:14 mismatched Fab heterodimers

*Indicates estimates derived from other Fab heterodimers that differ only in the presence/absence of the associated light chain tag (HA or FLAG).
**"Unique Identifier" consists of a unique identifier of two component LCCAs in either (H1L1L2 set number-H2L2L1 set number) or (H2L2L1 set number-H1L1L2 set number) orientations.

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 15) Cluster 1 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 16) Cluster 2 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 17) Cluster 3 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 18) Cluster 4 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 19) Cluster 5 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計
***代表的な設計は、Ｌ２がＱ３８Ｅ置換を欠いていることを除いて、９７４５－９０７５に類似している。 (Table 20) Cluster 6 designs, including representative designs

*Kabat numbering
**Representative design
***Representative design is similar to 9745-9075 except L2 lacks the Q38E substitution.

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 21) Cluster 7 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 22) Cluster 8 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計
***代表的な設計は、Ｌ２がＱ３８Ｅ置換を欠いていることを除いて、９７５１－９０６５に類似している。 (Table 23) Cluster 9 designs, including representative designs

*Kabat numbering
**Representative design
***Representative design is similar to 9751-9065 except L2 lacks the Q38E substitution.

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 24) Cluster 10 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 25) Cluster 11 designs, including representative designs

*Kabat numbering
**Representative design

*Ｋａｂａｔ番号付け
**代表的な設計 (Table 26) Cluster 12 designs, including representative designs

*Kabat numbering
**Representative design

(Table 27) Cluster 13 designs, including representative designs

*Ｋａｂａｔ番号付け。ＷＴ残基は、Ｄ３Ｈ４４系を指すことに留意すること。 (Table 28a) SMCA Unique Identifiers for Trastuzumab/Cetuximab Bispecific Systems

*Kabat numbering. Note that WT residues refer to the D3H44 system.

*Ｋａｂａｔ番号付け。ＷＴ残基は、Ｄ３Ｈ４４系を指すことに留意すること。 (Table 28b) SMCA Unique Identifiers for the D3H44/Cetuximab Bispecific System

*Kabat numbering. Note that WT residues refer to the D3H44 system.

*Ｋａｂａｔ番号付け。ＷＴ残基は、Ｄ３Ｈ４４系を指すことに留意すること。 (Table 28c) SMCA Unique Identifiers for the D3H44/Trastuzumab Bispecific System

*Kabat numbering. Note that WT residues refer to the D3H44 system.

*完全なＡｂ種のみを考慮した割合（％）
**全ての種を考慮した割合（％）
***野生型に対するＨ１：Ｈ２：Ｌ１：Ｌ２の割合（％）の推定変化 Table 29a: LC-MS paired data and post-pA yields (mg/L) of heterodimeric antibodies derived from the D3H44 (H1L1)/cetuximab (H2L2) bispecific system

*Proportion (%) considering only complete Ab species
** Percentage considering all species (%)
*** Estimated change in percentage of H1:H2:L1:L2 relative to wild type

*完全なＡｂ種のみを考慮した割合（％）
**全ての種を考慮した割合（％）
***野生型に対するＨ１：Ｈ２：Ｌ１：Ｌ２の割合（％）の推定変化 Table 29b LC-MS paired data and post-pA yield (mg/L) of heterodimeric antibody derived from D3H44 (H1L1)/trastuzumab (H2L2) bispecific system

*Proportion (%) considering only complete Ab species
** Percentage considering all species (%)
***Estimated change in percentage of H1:H2:L1:L2 relative to wild type

*完全なＡｂ種のみを考慮した割合（％）
**全ての種を考慮した割合（％）
***野生型に対する推定変化 Table 29c. LC-MS paired data and post-pA yields (mg/L) of heterodimeric antibodies derived from the Trastuzumab (H1L1)/Cetuximab (H2L2) bispecific system.

*Proportion (%) considering only complete Ab species
** Percentage considering all species (%)
***Puted change relative to wild type

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％）
***野生型に対するＨ１：Ｈ２：Ｌ１：Ｌ２の割合（％）の推定変化 (Table 30a) LC-MS pairing of heterodimeric antibodies derived from the D3H44 (H1L1)/cetuximab (H2L2) bispecific system after preparative SEC.

*Only full Ab species considered (%)
**Considering all species (%)
***Estimated change in percentage of H1:H2:L1:L2 relative to wild type

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％）
***野生型に対するＨ１：Ｈ２：Ｌ１：Ｌ２の割合（％）の推定変化 (Table 30b) LC-MS pairing of heterodimeric antibodies derived from the D3H44 (H1L1)/trastuzumab (H2L2) bispecific system after preparative SEC.

*Only full Ab species considered (%)
**Considering all species (%)
*** Estimated change in percentage of H1:H2:L1:L2 relative to wild type

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％）
***野生型に対する推定変化 (Table 30c) LC-MS pairing of heterodimeric antibodies derived from Trastuzumab (H1L1)/Cetuximab (H2L2) bispecific system after preparative SEC.

*Only full Ab species considered (%)
**Considering all species (%)
***Puted change relative to wild type

*報告されたＫＤ値は、単回の測定または２回の測定の平均である。範囲は括弧内に示されている。 (Table 31a) Biophysical characterization (antigen binding, thermostability. UPLC-SEC) of selected designs derived from the D3H44/trastuzumab system.

*KD values reported are the average of single or duplicate determinations. Ranges are shown in brackets.

*報告されたＫＤ値は、単回の測定または２回の測定の平均である。範囲は括弧内に示されている。 (Table 31b) Biophysical characterization (antigen binding, thermostability. UPLC-SEC) of selected designs derived from the D3H44/cetuximab system.

*KD values reported are the average of single or duplicate determinations. Ranges are shown in brackets.

*報告されたＫＤ値は、単回の測定または２回の測定の平均である。範囲は括弧内に示されている。
**ＮＤ＝決定されず。 (Table 31c) Biophysical characterization of selected designs derived from the Trastuzumab/Cetuximab system (antigen binding, thermostability. UPLC-SEC)

*KD values reported are the average of single or duplicate determinations. Ranges are shown in brackets.
**ND = not determined.

(Table 32a) Effect of DNA titration ratio of wild-type D3H44/trastuzumab system on percentage of antibody species as assessed by LC-MS. H1 and L1 refer to the heavy and light chains of D3H44, respectively. H2 and L2 refer to the heavy and light chains of trastuzumab, respectively.

(Table 32b) Effect of DNA titration ratio of wild-type D3H44/cetuximab system on percentage of antibody species as assessed by LC-MS. H1 and L1 refer to the heavy and light chains of D3H44, respectively. H2 and L2 refer to the heavy and light chains of cetuximab, respectively.

(Table 32c) Effect of DNA titration ratio of wild-type trastuzumab/cetuximab system on percentage of antibody species as assessed by LC-MS. H1 and L1 refer to the heavy and light chains of trastuzumab, respectively. H2 and L2 refer to the heavy and light chains of trastuzumab, respectively.

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％） (Table 33a) LC-MS pairing of wild-type antibody constructs derived from the D3H44 (H1L1)/cetuximab (H2L2) bispecific system.

*Only full Ab species considered (%)
**Considering all species (%)

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％） (Table 33b) LC-MS pairing of wild-type antibody constructs derived from the D3H44 (H1L1)/trastuzumab (H2L2) bispecific system.

*Only full Ab species considered (%)
**Considering all species (%)

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％） (Table 33c) LC-MS pairing of wild-type antibody constructs derived from the Trastuzumab (H1L1)/Cetuximab (H2L2) bispecific system.

*Only full Ab species considered (%)
**Considering all species (%)

*ＷＴは、野生型を指す
**ＮＡは、対応するＬＣＣＡセット番号がないので、該当なしを指す Table 34. Stabilizing mutations in Fab heterodimers

*WT refers to wild type
**NA indicates not applicable as there is no corresponding LCCA set number

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％）
***野生型に対する推定変化 (Table 35a) Designs showing mobility in all three bispecific systems (D3H44/cetuximab, D3H44/trastuzumab, and trastuzumab/cetuximab) in both orientations

*Only full Ab species considered (%)
**Considering all species (%)
***Puted change relative to wild type

*完全なＡｂ種のみを考慮した（％）
**全ての種を考慮した（％）
***野生型に対する推定変化 (Table 35b) Mobility in all three bispecific systems (D3H44/cetuximab, D3H44/trastuzumab, and trastuzumab/cetuximab) is shown in one orientation and in the other orientation only in one bispecific system. Designs that migrate and also meet at least 10% of the light chain utilization criteria

*Only full Ab species considered (%)
**Considering all species (%)
***Puted change relative to wild type

Claims (35)

An isolated antigen-binding polypeptide construct comprising at least a first heterodimer and a second heterodimer,
said first heterodimer comprising a first human or humanized immunoglobulin G (IgG) heavy chain polypeptide sequence (H1) and a first human or humanized immunoglobulin kappa light chain polypeptide sequence (L1) and has a Fab region that binds to a first epitope; and said second heterodimer comprises a second human or humanized immunoglobulin G (IgG) heavy chain polypeptide sequence (H2) and comprising a second human or humanized immunoglobulin kappa light chain polypeptide sequence (L2) and having a Fab region that binds a second epitope; at least one of which is different from the corresponding H2 or L2 sequence of said second heterodimer;
H1 and H2 each comprise at least a heavy chain variable domain ( _VH domain) and a heavy chain constant domain (C _H1 domain);
L1 and L2 each comprise at least a light chain variable domain ( _VL domain) and a light chain constant domain ( _CL domain);
H1, H2, L1 and L2 comprise a set of amino acid modifications, wherein H1 preferentially pairs with L1 over L2, H2 preferentially pairs with L2 over L1;
The thermal stability of the Fab regions of the first and second heterodimers as measured by the melting temperature (Tm) determined by differential scanning calorimetry compared to that of the heterodimer without the set of amino acid modifications. 1 . within 5°C and the set of amino acid modifications are:
a) H1 contains amino acid substitutions 124R and 172T, L1 contains amino acid substitutions 133G, 174R and 176D, H2 contains amino acid substitutions 124E and 172R, and L2 contains amino acid substitutions 133G and 176R;
b) H1 contains amino acid substitutions 139W, 143E, 145T and 179E, L1 contains amino acid substitutions 116A, 124R, 135V and 178R, H2 contains amino acid substitutions 179K and L2 contains amino acid substitutions 124E, 135W, 160E and 180E including;
c) H1 contains amino acid substitutions 124A, 143F and 179K, L1 contains amino acid substitutions 124E, 133W, 176T, 178L and 180E, H2 contains amino acid substitutions 124W, 143E, 145T and 179E, and L2 contains amino acid substitution 124R , 133A, 176T and 178R;
d) H1 contains amino acid substitutions 143E, 145T and 179E, L1 contains amino acid substitutions 124R and 178R, H2 contains amino acid substitutions 143R and L2 contains amino acid substitutions 124E and 133E;
e) H1 contains amino acid substitutions 143E, 145T and 179E, L1 contains amino acid substitutions 124R and 178R, H2 contains amino acid substitutions 179R, and L2 contains amino acid substitutions 124E, 178E and 180E; or f) H1 contains contains amino acid substitutions 143E, 145T and 179E, L1 contains amino acid substitutions 124R and 178R, H2 contains amino acid substitutions 179K, and L2 contains amino acid substitutions 124E and 180E;
is selected from
and constructs wherein the amino acid residue numbering is according to the Kabat numbering system. 2. The construct of claim 1, wherein H1 comprises amino acid substitutions 124R and 172T, L1 comprises amino acid substitutions 133G, 174R and 176D, H2 comprises amino acid substitutions 124E and 172R, and L2 comprises amino acid substitutions 133G and 176R. . H1 contains amino acid substitutions 139W, 143E, 145T and 179E, L1 contains amino acid substitutions 116A, 124R, 135V and 178R, H2 contains amino acid substitutions 179K and L2 contains amino acid substitutions 124E, 135W, 160E and 180E. , the construct of claim 1. H1 contains amino acid substitutions 124A, 143F and 179K, L1 contains amino acid substitutions 124E, 133W, 176T, 178L and 180E, H2 contains amino acid substitutions 124W, 143E, 145T and 179E, and L2 contains amino acid substitutions 124R, 133A. , 176T and 178R. 2. The construct of claim 1, wherein H1 contains amino acid substitutions 143E, 145T and 179E, L1 contains amino acid substitutions 124R and 178R, H2 contains amino acid substitutions 143R, and L2 contains amino acid substitutions 124E and 133E. 2. The construct of claim 1, wherein H1 comprises amino acid substitutions 143E, 145T and 179E, L1 comprises amino acid substitutions 124R and 178R, H2 comprises amino acid substitutions 179R, and L2 comprises amino acid substitutions 124E, 178E and 180E. . 2. The construct of claim 1, wherein H1 contains amino acid substitutions 143E, 145T and 179E, L1 contains amino acid substitutions 124R and 178R, H2 contains amino acid substitutions 179K, and L2 contains amino acid substitutions 124E and 180E. when H1, H2, L1 and L2 are co-expressed in a cell or mammalian cell; or when H1, H2, L1 and L2 are co-expressed in a cell-free expression system; or H1, H2, L1 and L2 are co-produced or when H1, H2, L1 and L2 are co-produced via redox production, H1 preferentially pairs with L1 over L2 and H2 preferentially pairs with L2 over L1. The construct of any one of claims 1-7, wherein the construct is The pairing of the modified H1-L1 and H2-L2 heterodimer pairs is higher than each pairing observed in the corresponding H1-L1 and H2-L2 heterodimer pairs without amino acid modifications. The construct of any one of claims 1-7, which is large. The construct further comprises an Fc comprising a first CH3 sequence and a second CH3 sequence, wherein the first CH3 sequence is linked to the first heterozygous of claims 1-9, wherein said second CH3 sequence is bound to said second heterodimer with or without one or more linkers. A construct according to any one of paragraphs. 11. The construct of claim 10, wherein Fc is human Fc, human IgG1 Fc, human IgG Fc, human IgG2 Fc, human IgG3 Fc, or human IgG4 Fc. 12. The construct of claim 11, wherein said Fc is a heterodimeric Fc. 11. The Fc of claim 10, wherein the Fc comprises one or more modifications in at least one of the CH3 sequences that promote formation of a heterodimeric Fc with stability comparable to that of a wild-type homodimeric Fc. construction. said Fc is
i) a heterodimeric IgG1 Fc having the modification 351Y_405A_407V in the first Fc polypeptide and the modification 366L_392M_394W in the second Fc polypeptide;
ii) a heterodimeric IgG1 Fc having the modification 351Y_405A_407V in the first Fc polypeptide and the modification 366L_392L_394W in the second Fc polypeptide;
iii) a heterodimeric IgG1 Fc having the modification 350V_351Y_405A_407V in the first Fc polypeptide and the modification 350V_366L_392L_394W in the second Fc polypeptide;
iv) a heterodimeric IgG1 Fc having the modification 350V_351Y_405A_407V in the first Fc polypeptide and the modification 350V_366L_392M_394W in the second Fc polypeptide; or v) having the modification 350V_351Y_400E_405A_407V in the first Fc polypeptide; Heterodimeric IgGl Fc with modification 350V_366L_390R_392M_394W in the second Fc polypeptide
14. The construct of claim 13, wherein the numbering of residues in IgG1 Fc is according to the EU numbering system. 11. The construct of claim 10, wherein said Fc further comprises at least one CH2 sequence. 16. The construct of claim 15, wherein said CH2 sequence of said Fc comprises one or more modifications. 17. The construct of claim 16, wherein Fc comprises one or more modifications that facilitate selective binding of Fc gamma receptors. 11. The construct of claim 10, wherein said Fc is attached to said heterodimer by one or more linkers, or said Fc is attached to Hl and H2 by one or more linkers. 19. The construct of claim 18, wherein said one or more linkers are one or more polypeptide linkers. 19. The construct of claim 18, wherein said one or more linkers comprise one or more antibody hinge regions. 19. The construct of claim 18, wherein said one or more linkers comprise one or more IgGl hinge regions. 20. The construct of claim 19, wherein said one or more linkers comprise one or more modifications. 23. The construct of claim 22, wherein said one or more modifications promote selective binding of Fc gamma receptors. The construct of any one of claims 1-23, which is multispecific or bispecific. The construct of any one of claims 1-23, which is multivalent or bivalent. 26. The construct of any one of claims 1-25, conjugated to a therapeutic agent or drug moiety. An isolated polynucleotide or set of isolated polynucleotides comprising at least one sequence encoding a construct according to any one of claims 1-25. 28. A vector or set of vectors comprising one or more or a set of polynucleotides of claim 27. 29. An isolated cell comprising the polynucleotide or set of polynucleotides of claim 27, or the vector or set of vectors of claim 28. 30. The isolated cell of claim 29, which is a yeast cell, bacterial cell, hybridoma, Chinese Hamster Ovary (CHO) cell or HEK293 cell. A pharmaceutical composition comprising the construct of any one of claims 1-26 and a pharmaceutically acceptable carrier. Use of a construct according to any one of claims 1-26, or a pharmaceutical composition according to claim 31, in the manufacture of a medicament for the treatment of a disease or disorder in a subject. A method of obtaining a construct according to any one of claims 1-25 from a host cell culture, comprising:
(a) obtaining a host cell culture comprising at least one host cell comprising one or more nucleic acid sequences encoding said construct;
(b) recovering the construct from said host cell culture. A method for obtaining a construct according to any one of claims 1 to 25, comprising:
(a) obtaining H1, L1, H2 and L2;
(b) preferentially pairing H1 with L1 over L2, and preferentially pairing H2 with L2 over L1;
(c) obtaining a construct. A method of preparing a construct according to any one of claims 1 to 25, comprising the steps of:
a. obtaining a polynucleotide or set of polynucleotides encoding at least one construct;
b. Determining an optimal ratio of each of said polynucleotides or sets of polynucleotides for introduction into at least one host cell, said optimal ratio upon expression of H1, L1, H2 and L2. H1-L1 and H2-L2 heterodimer pairs formed upon expression of H1, L1, H2 and L2 compared to mismatched H1-L2 and H2-L1 heterodimer pairs formed the stage, determined by evaluating the amount of;
c. selecting a preferred optimal ratio, wherein transfection of at least one host cell with said polynucleotide or set of polynucleotides in said optimal ratio results in expression of said construct;
d. transfecting said at least one host cell with said optimal ratio of said polynucleotide or set of polynucleotides;
e. culturing said at least one host cell to express said construct.

Images