Methods and means for producing immunoglobulin-like molecules

Abstract

The present invention relates to methods and means for producing immunoglobulin-like molecules. The present invention provides means and methods for producing one or more immunoglobulin-like molecules in a single host cell. Novel CH3 mutations capable of producing monospecific and / or bispecific immunoglobulin-like molecules of interest are also provided.

Description

[0001] This application is a divisional application of the Chinese patent application (Patent Application 1) with the application number 202111365429.7 and the invention title "Methods and Means for Generating Immunoglobulin-like Molecules" filed on November 17, 2021. Patent Application 1 is a divisional application of the Chinese patent application (Patent Application 2) with the application number 201710637343.2 and the invention title "Methods and Means for Generating Immunoglobulin-like Molecules" filed on July 28, 2017. Patent Application 2 is a divisional application of the Chinese patent application (Patent Application 3) with the application number 201380032190.9 and the invention title "Methods and Means for Generating Immunoglobulin-like Molecules". Patent Application 3 is the application that enters the Chinese national phase of the PCT international application PCT / NL2013 / 050293 filed on April 19, 2013. Technical Field

[0002] The present invention relates to the fields of molecular biology, medicine, and biotherapeutic agents. In particular, it relates to the field of therapeutic antibodies for treating various diseases. Background Art

[0003] Many currently used biotherapeutics are isolated recombinant human or humanized monoclonal antibodies that enhance the body's immune system's ability to neutralize or eliminate cells and / or molecules involved in the disease process or to destroy invading pathogens or infectious agents. Monoclonal antibodies bind to a single specific region or epitope of an antigen and are typically selected for therapeutic use based on desired functional properties, such as killing tumor cells, blocking receptor-ligand interactions, or virus neutralization. Currently, there are approximately 30 FDA-approved monoclonal antibodies, which are generally produced in large quantities and their biophysical and biochemical properties can be analyzed in detail to ensure batch-to-batch consistency, which helps to control regulatory acceptability. Despite these favorable properties, monoclonal antibodies also have many drawbacks, some of which relate to their monospecific nature and the complexity of diseases. Disease processes are often multifactorial in nature and involve the numerous or synergistic actions of disease-modulating factors or the upregulation of different receptors, including crosstalk between their signaling networks. Thus, blocking multiple different factors and pathways involved in pathology can lead to improved therapeutic outcomes. Due to their monospecific nature, monoclonal antibodies may only interfere with a single step in a complex disease process, which often does not result in optimal effects. In addition to not fully addressing multiple aspects of the disease process, it has become clear that targeting a single cellular protein or soluble protein or a single epitope of a pathogen is generally not sufficient to effectively treat a disease because the target epitope to which the monoclonal antibody binds and exerts the desired effect may no longer be available. For example, tumor cells often evade monoclonal antibody therapy by downregulating, mutating, or shielding the target epitope present on growth factor receptors. By activating alternative receptors and / or their ligands, tumor cells can then develop different pathways for continued growth and metastasis. Similarly, viruses and other pathogens frequently mutate and lose or shield target epitopes, thus evading monoclonal antibody therapy. Monoclonal antibodies that bind to a single epitope generally do not recruit the full range of effector mechanisms induced by polyclonal antibodies, which include: opsonization (enhancing phagocytosis of an antigen), steric hindrance (the antigen coated with antibodies is prevented from attaching to host cells or mucosal surfaces), toxin neutralization, aggregation or precipitation (aggregation and subsequent clearance caused by antibodies binding multiple soluble antigens), activation of complement, and antibody-dependent cellular cytotoxicity (antibodies are able to kill target cells through natural killer cells and neutrophils).

[0004] Polyclonal antibodies for therapeutic applications can be obtained from pooled human serum. Such serum-derived therapeutic polyclonal antibodies can be used, for example, to treat or prevent infections caused by viruses (such as rabies virus, cytomegalovirus, and respiratory syncytial virus), neutralize toxins (such as tetanus toxin and botulinum toxin), or prevent Rhesus D allograft immunization. The following facts have precluded a more widespread use of serum-derived polyclonal antibody preparations: the source plasma is only available for a limited range of targets, such as infectious diseases and toxins. In addition, in terms of both quantity and suitability, the product is highly dependent on the availability of donor blood, resulting in considerable variability between batches. Furthermore, screening techniques do not keep up with evolving viruses, and thus, immunoglobulin products carry the potential risk of transmitting infectious diseases. Finally, the long process of blood collection, screening, and immunoglobulin purification means that plasma-derived immunoglobulin production is expensive.

[0005] Mixtures of monoclonal antibodies can enhance the potency of monoclonal antibodies while avoiding the limitations associated with serum-derived polyclonal antibodies. In the art, combinations of two human or humanized monoclonal antibodies have been tested in preclinical models and clinical trials (such as mixtures of two monoclonal antibodies against the HER2 receptor, mixtures of two antibodies against the EGFR receptor, two monoclonal antibodies against rabies virus). In the art, it has been shown that combinations of two monoclonal antibodies can have additive or synergistic effects and recruit effector mechanisms not associated with either individual antibody. For example, mixtures of two monoclonal antibodies against EGFR or HER2 have been shown to be more potent in killing tumor cells based on activities including: enhancing receptor internalization, increasing blockade of downstream signaling pathways of the receptor, and enhancing immune effector-mediated cytotoxicity. For combination therapies based on two monoclonal antibodies, the component antibodies can be produced separately and combined at the protein level. A disadvantage of this approach is the high cost of developing the two antibodies separately in clinical trials and (partially) repeating the process with the combination. This results in an unacceptable cost for antibody-based combination therapies. Alternatively, the two recombinant cell lines producing the component monoclonal antibodies can be mixed in a fermenter, and the resulting antibody mixture can be purified as a single preparation (WO2004 / 061104). A disadvantage of this method is the poor control of its composition, and thus, the resulting recombinant polyclonal antibody preparation has poor reproducibility, especially considering that such composition may change over time as the cells are cultured.

[0006] In the past decade, bispecific antibodies have emerged as an alternative to the use of a combination of two antibodies. While a combination of two antibodies represents a mixture of two different immunoglobulin molecules that bind to different epitopes on the same or different targets, in a bispecific antibody, this is achieved by a single immunoglobulin molecule. By binding to two epitopes on the same or different targets, bispecific antibodies can have a similar effect compared to a combination of two antibodies that bind to the same epitope. In addition, different effects of these forms have also been observed in a single formulation due to the combination of two different monovalent binding regions in the IgG form of the bispecific antibody in a single molecule and the combination of two IgG antibodies as a mixture of two different bivalent binding molecules. From a technical and regulatory perspective, the development of a single bispecific antibody is less complex as manufacturing, preclinical testing, and clinical trials involve a single molecule. Thus, a less complex and cost-effective drug development process that simultaneously provides more effective antibody therapy has facilitated therapies based on single bispecific antibodies.

[0007] Bispecific antibodies based on the IgG form have been generated by a variety of methods and consist of two heavy chains and two light chains. For example, bispecific antibodies can be generated by fusing two antibody-secreting cell lines to produce a new cell line or by expressing two antibodies in a single cell using recombinant DNA technology. These methods produce a variety of antibodies because the respective heavy chains from each antibody may form monospecific dimers (also called homodimers), which contain two identical paired heavy chains with the same specificity, and bispecific dimers (also called heterodimers), which contain two different paired heavy chains with different specificities. In addition, the light and heavy chains from each antibody can pair randomly to form inappropriate, non-functional combinations. This problem is called heavy and light chain mispairing and can be solved by choosing to express the two heavy chains and a common light chain as a bispecific antibody. However, even when using a common light chain, expressing two heavy chains and a common light chain in a single cell will produce three different antibody species, namely two monospecific "parent" antibodies and the bispecific antibody, so the desired bispecific antibody needs to be purified from the resulting antibody mixture. Although techniques have been used to further increase the percentage of bispecific antibodies in the mixture of parent and bispecific antibodies and to reduce the percentage of heavy and light chain mispairing, there is still a need for bispecific forms that eliminate or minimize some of the disadvantages mentioned above.

[0008] In summary, the art provides a variety of techniques and methods for generating monoclonal antibodies, bispecific antibodies, mixtures of monoclonal antibodies, or mixtures of monospecific and bispecific antibodies that can subsequently be used in therapeutic applications in patients. However, as discussed above, each of these existing techniques and methods has its drawbacks and limitations. Accordingly, there is a need for improved and / or alternative techniques for generating biotherapeutic agents in the form of mixtures or in a bispecific manner that target multiple disease-modifying molecules. SUMMARY OF THE INVENTION

[0009] The present invention provides methods and means for improved and / or alternative techniques for generating biotherapeutic agents in the form of mixtures or in a bispecific manner that target multiple disease-modifying molecules, as well as products and uses obtained from these methods and means.

[0010] A variety of methods have been described in the art for promoting the formation of a desired bispecific antibody, thereby reducing the content of undesired antibodies in the resulting mixture.

[0011] For antibodies, it is well known that CH3-CH3 interactions are the major driving force for Fc dimerization (Ellerson JR et al., J. Immunol 1976 (Vol. 116) pp. 510-517; Deisenhofer J. Biochemistry 1981 (Vol. 20) pp. 2361-2370). It is also well known that when two CH3 domains interact with each other, they contact at a protein-protein interface that contains "contact" residues (also called contact amino acids, interface residues, or interface amino acids). The contact amino acids of the first CH3 domain interact with one or more contact amino acids of the second CH3 domain. In the three-dimensional structure of an antibody, the contact amino acids are typically within 5.5 angstroms (preferably within 4.5 angstroms) of each other. The interaction between contact residues from one CH3 domain and contact residues from a different CH3 domain can occur through, for example, van der Waals forces, hydrogen bonds, water-mediated hydrogen bonds, salt bridges, or other electrostatic forces, attractive interactions between aromatic side chains, disulfide bonds, or other forces known to those skilled in the art. It has previously been shown that approximately one-third of the contact amino acid side chains at the CH3 domain interface of human IgG1 may be the major contributors to domain folding and association. It can further be envisioned that other (adjacent) amino acid residues may influence the interactions in the protein-protein interface.

[0012] Methods that interfere with heavy chain dimerization have been applied in the art. Specific modifications are made in the CH3 domain to favor heterodimerization over homodimerization. Examples of such modifications of the CH3-CH3 interface include introducing complementary protrusion mutations and cavity mutations, also known as the "knob-into-hole" method, as described in, for example, WO1998 / 050431, Ridgeway et al. (1996), and Merchant et al. (1998).

[0013] Generally, the method involves introducing a protrusion at the interface of a first polypeptide and a corresponding cavity at the interface of a second polypeptide such that the protrusion can be placed in the cavity, thereby promoting the formation of heteropolymers and hindering the formation of homopolymers. The "protrusion" or "knob" is constructed by replacing a small amino acid side chain at the interface of the first polypeptide with a larger side chain (such as tyrosine or tryptophan). A compensatory "cavity" or "hole" of the same or similar size as the protrusion is created at the interface of the second polypeptide by replacing a large amino acid side chain with a smaller amino acid side chain (such as alanine or threonine). The protrusion and cavity can be prepared by synthetic means (such as altering the nucleic acid encoding the polypeptide) or by peptide synthesis.

[0014] When using the knob-into-hole technique alone, the proportion of the desired bispecific antibody in the mixture of the two parental antibodies and the bispecific antibody is at most 87%. By introducing additional disulfide bonds between the two CH3 domains at the CH3-CH3 interface, Merchant et al. successfully increased the proportion of the bispecific antibody to 95% of the mixture. Nevertheless, in order to use such bispecific antibodies as drugs, the bispecific antibodies must also be purified (isolated) from the homodimers and formulated into a pharmaceutically acceptable diluent or excipient. Due to the similarity in the physicochemical properties of the homodimers and heterodimers, purifying the heterodimers from such a mixture is a major challenge. An object of the present invention is to provide a method for producing bispecific antibodies in a single cell clone and the proportion of the bispecific antibody in the mixture. Thus, according to the present invention, the knob-into-hole technique can be used as a means (alone or in combination with other means) to achieve the further increased bispecific proportion in the mixture.

[0015] Another example of such modification of the CH3-CH3 interface is provided by the heterodimeric Fc technology, through engineering of the strand-exchange engineered domain (SEED) CH3 heterodimers, which support the design of bispecific and asymmetric fusion proteins. These SEED CH3 heterodimers are derivatives of the human IgG and IgA CH3 domains consisting of alternating segments of the CH3 sequences of human IgA and IgG, which results in the pairing of complementary human SEED CH3 heterodimers, so-called SEED-bodies (Davis JH. et al., Protein Engineering, Design & Selection 2010 (Vol. 23) pp. 195-202; WO2007 / 110205).

[0016] Another method for generating a given bispecific antibody of interest is based on the electrostatic engineering of the naturally charged contact residues within the CH3-CH3 interface, as described, for example, in EP01870459 or US2010 / 0015133, WO2007 / 147901, WO2010 / 129304, Gunasekaran et al. (2010) and WO2009 / 089004. These publications describe mutations in the heavy chain CH3 domain, where the contact residues of the naturally occurring charged amino acids are replaced with amino acid residues of the opposite charge (i.e., the charge reversal strategy). This creates a changed charge polarity across the Fc dimer interface such that co-expression of electrostatically matched Fc chains supports favorable attractive interactions, thus promoting the formation of the desired Fc heterodimers, while unfavorable repulsive charge interactions inhibit the formation of unwanted Fc homodimers.

[0017] According to the description, four unique charged residue pairs within the CH3-CH3 interface are involved in domain-domain interactions. These are D356 / K439’, E357 / K370’, K392 / D399’ and D399 / K409’ (numbering according to Kabat (1991), where residues in the first chain are separated from residues in the second chain by “ / ”, and where the prime (’) indicates the residue numbering in the second chain). Since the CH3-CH3 interface exhibits two-fold symmetry, each unique charge pair is represented twice in the intact IgG (i.e., the charge interactions K439 / D356’, K370 / E357’, D399 / K392’ and K409 / D399’ also exist at the interface). Using this two-fold symmetry, it has been demonstrated that a single charge reversal (e.g., K409D in the first chain, or D399’K in the second chain) results in a reduction in the formation of homodimers due to the repulsion of like charges. Combinations of different charge reversals further enhance this repulsive effect. It has been demonstrated that the expression of different CH3 domains containing different complementary charge reversals can drive heterodimerization, resulting in an increased proportion of bispecific species in the mixture.

[0018] Using the above method, bispecific antibodies can be produced in a single cell at a proportion of about 76% to about 96%. An object of the present invention is to provide a method for producing bispecific antibodies in a single cell and further increasing the percentage of the desired bispecific antibodies. According to the present invention, electrostatic engineering techniques can be used as one of the means, alone or in combination with other means (such as the knob-into-hole method), to achieve the further increase in the percentage of the desired (bispecific) antibodies.

[0019] In one aspect, the present invention provides a method for producing at least two different immunoglobulin-like (Ig-like) molecules from a single host cell, wherein each of the two immunoglobulin-like molecules comprises two CH3 domains capable of forming an interface, and the method comprises providing in the cell:

[0020] a) a first nucleic acid molecule encoding a first CH3 domain-comprising polypeptide chain,

[0021] b) a second nucleic acid molecule encoding a second CH3 domain-comprising polypeptide chain,

[0022] c) a third nucleic acid molecule encoding a third CH3 domain-comprising polypeptide chain, and

[0023] d) a fourth nucleic acid molecule encoding a fourth CH3 domain-comprising polypeptide chain,

[0024] At least two of the nucleic acid molecules have means for preferentially pairing the polypeptide chain having the first CH3 domain with the second polypeptide having the CH3 domain and the polypeptide having the third CH3 domain with the polypeptide having the fourth CH3 domain, and the method further comprises culturing the host cell and enabling the at least four nucleic acid molecules to be expressed and harvesting the at least two different immunoglobulin-like molecules from the culture.

[0025] It is often desirable to produce more than one (bispecific) antibody to, for example, more effectively interfere with multiple biological pathways involved in a disease process or interfere with the invasion, replication, and / or spread of a pathogen.

[0026] Mixtures of more than one bispecific antibody are also particularly useful for the treatment of certain diseases. For example, during treatment with antibodies or small molecule drugs, tumor cells develop resistance using many different strategies. Resistance may involve multiple cell surface receptors as well as soluble molecules, and it is thought to favor the development of antibody-based cancer therapies that simultaneously address multiple such disease-related molecules and molecules that evade them. In the case of more than 2 such disease-related target molecules or epitopes and molecules that evade them, mixtures of bispecific antibodies provide an innovative and attractive form of therapy. Preferably, such mixtures of bispecific antibodies are produced by a single cell to facilitate the drug development process, i.e., they are less complex from a regulatory perspective, cost-effective, and feasible from the perspective of drug manufacturing and clinical development. In a single cell-based approach, it is desirable to use a method that can be controlled and effectively produce bispecific antibodies, thereby reducing or even completely eliminating the need to separate a mixture of desired bispecific IgG molecules from undesired monospecific IgG molecules. In the prior art, mixtures of monospecific and bispecific antibodies have been produced by a single cell (WO2004 / 009618), but these mixtures represent a complex concoction of multiple different bispecific and monospecific antibody species. Another object of the present invention is to provide means and methods for producing a defined mixture of bispecific antibodies in a single cell. Preferably, regardless of the amount of monomeric by-products, the method provided results in a mixture of (bispecific) antibodies that accounts for at least 95%, at least 97%, or even more than 99% of the dimeric IgG molecules, as described hereinbelow. Typically, in cells that produce multiple intact IgG molecules, there may be half-molecules (monomeric by-products) that can be simply removed by size exclusion chromatography known in the art.

[0027] In one embodiment, the present invention provides a method for generating a defined mixture of at least two different immunoglobulin-like molecules rather than a single-purpose (bispecific) antibody in a single cell, wherein the formation of other undesired dimeric antibody species is reduced or even absent. The resulting mixture is well-defined and its composition is controlled by the design of the CH3 domain mutants. In addition, the regulation of the expression level and / or the different transfection ratios used for expression affects the composition of the mixture. In the method according to the present invention, a first nucleic acid molecule encodes a CH3 domain that preferentially pairs with a CH3 domain encoded by a second nucleic acid molecule, while a third nucleic acid molecule encodes a CH3 domain that preferentially pairs with a CH3 domain encoded by a fourth nucleic acid molecule. The present invention also provides a mixture of at least two different immunoglobulin-like molecules obtainable by the method of the present invention.

[0028] As used herein, the term "the first CH3 domain-containing polypeptide preferentially pairs with the second CH3 domain-containing polypeptide" means that substantially all resulting dimers containing the first CH3 domain-containing polypeptide and / or the second CH3 domain-containing polypeptide are dimers composed of a first CH3 domain-containing polypeptide paired with a second CH3 domain-containing polypeptide. Similarly, the term "the third CH3 domain-containing polypeptide and the fourth CH3 domain-containing polypeptide preferentially pair" means that substantially all resulting dimers containing the third CH3 domain-containing polypeptide and / or the fourth CH3 domain-containing polypeptide are dimers composed of a third CH3 domain-containing polypeptide paired with a fourth CH3 domain-containing polypeptide. As a result, when nucleic acid molecules encoding four different (A, B, C, D) CH3 domain-containing polypeptides are introduced into a single cell, a mixture of mainly two specific immunoglobulin-like molecules is produced rather than a mixture of 10 different immunoglobulin-like dimers (AA, AB, AC, AD, BB, BC, BD, CC, CD, and DD).

[0029] As explained in more detail below, in a preferred embodiment, the first polypeptide chain comprising a CH3 domain comprises the amino acid substitution T366K, and the second polypeptide chain comprising a CH3 domain comprises the amino acid substitution L351D. These amino acid alterations are preferred means for preferentially pairing the first polypeptide chain comprising a CH3 domain with the second polypeptide chain comprising a CH3 domain. The first polypeptide chain comprising a CH3 domain preferably further comprises the amino acid substitution L351K. In addition, the second polypeptide chain comprising a CH3 domain preferably further comprises an amino acid substitution selected from Y349E, Y349D and L368E, most preferably L368E. In another preferred embodiment, the third polypeptide chain comprising a CH3 domain comprises the amino acid substitutions E356K and D399K, and the fourth polypeptide chain comprising a CH3 domain comprises the amino acid substitutions K392D and K409D.

[0030] In the method according to the invention, each polypeptide chain comprising a CH3 domain preferably further comprises a variable region that recognizes a target epitope. This variable region, which is part of the polypeptide chain comprising a CH3 domain, preferably has a common light chain. In this case, only the VH of the variable region is different, while the VL in all variable regions is substantially the same. Thus, in a preferred aspect, a method according to the invention is provided which further comprises providing the host cell with a nucleic acid molecule encoding the common light chain. In a particularly preferred embodiment, the four variable regions of the four polypeptide chains comprising a CH3 domain each recognize a different target epitope. For example, if the heavy chain encoded by the first nucleic acid molecule further comprises a variable domain specific for antigen A, the heavy chain encoded by the second nucleic acid molecule further comprises a variable domain specific for antigen B, the heavy chain encoded by the third nucleic acid molecule further comprises a variable domain specific for antigen C, and the heavy chain encoded by the fourth nucleic acid molecule further comprises a variable domain specific for antigen D, then a mixture containing bispecific immunoglobulin-like molecules specific for AB and bispecific immunoglobulin-like molecules specific for CD will be produced. Due to the means for preferentially pairing the first polypeptide comprising a CH3 domain with the second polypeptide comprising a CH3 domain and the third polypeptide comprising a CH3 domain with the fourth polypeptide comprising a CH3 domain, the formation of monospecific antibodies (having AA, BB, CC or DD specificity) or bispecific antibodies having AC, AD, BC or BD specificity is reduced or even absent. Of course, additional nucleic acid molecules can be used, such as nucleic acid molecules encoding a fifth polypeptide chain comprising a CH3 domain and a sixth polypeptide chain comprising a CH3 domain, to produce a defined mixture comprising more than two different immunoglobulin-like molecules.

[0031] It should be noted that the ratio of the nucleic acids used in the method according to the present invention is not necessarily 1:1:1:1, and the ratio of the expressed immunoglobulin-like molecules is not necessarily 1:1. Means known in the art can be used to generate an antibody mixture with an optimal ratio. For example, by using different genetic elements (such as promoters, enhancers, and suppressors) or by controlling the copy number of the genomic integration sites of the DNA constructs encoding the antibodies, the expression level of the nucleic acid molecules can be regulated and thus the ratio of the generated immunoglobulin-like molecules can be regulated.

[0032] The means for preferential pairing preferably may include engineered complementary interface pore mutations, disulfide bonds, charge mutations (including charge reversal mutations), or combinations thereof. Those skilled in the art will understand that the means for preferential pairing can be selected among certain types of mutations, i.e., all at least 4 nucleic acid molecules encoding polypeptide chains containing CH3 domains can all contain, for example, charge mutations as the means for preferential pairing. Additionally, in certain cases, unengineered wild-type CH3 can also be used to preferentially pair two wild-type polypeptide chains containing CH3 domains. In a particularly preferred embodiment, the means for preferential pairing includes at least one CH3 mutation selected from Table B, as explained in other parts of the present application. Thus, a preferred embodiment provides a method according to the present invention, wherein all 4 nucleic acid molecules have means for preferentially pairing the first polypeptide chain containing a CH3 domain with the second polypeptide chain containing a CH3 domain and the third polypeptide chain containing a CH3 domain with the fourth polypeptide chain containing a CH3 domain, wherein the means for preferentially pairing the first polypeptide chain containing a CH3 domain with the second polypeptide chain containing a CH3 domain is different from those for preferentially pairing the third polypeptide chain containing a CH3 domain with the fourth polypeptide chain containing a CH3 domain.

[0033] One aspect of the invention provides a method according to the invention, wherein the means for preferentially pairing the first polypeptide comprising a CH3 domain with the second polypeptide comprising a CH3 domain is different from the means for preferentially pairing the third polypeptide comprising a CH3 domain with the fourth polypeptide comprising a CH3 domain. "Different from" means that the means designed to preferentially pair the first polypeptide comprising a CH3 domain with the second polypeptide comprising a CH3 domain is such that preferential pairing of the first and second chains is favored. This design results in substantially no interaction between the polypeptide chain of the first polypeptide comprising a CH3 domain and the polypeptide chains of the third and / or fourth polypeptides comprising a CH3 domain. In other words, the dimerization between the first polypeptide comprising a CH3 domain and the third or fourth polypeptide is reduced to substantially zero, and so on. The third polypeptide comprising a CH3 domain and the fourth polypeptide comprising a CH3 domain can be wild-type or can comprise means for preferential pairing, wherein the means is different from the means for preferentially pairing the first CH3 domain with the second CH3 domain. Current research focuses on using, for example, knobs-into-holes technology or mutating (reversing) charged contacting amino acids present in the CH3 domain to generate single bispecific antibodies. However, prior to the present invention, it has not been possible to generate a defined mixture of at least two (bispecific) immunoglobulin-like molecules without significant co-generation of other dimer by-products.

[0034] The invention provides methods for efficiently and controllably generating a well-defined mixture of immunoglobulin-like molecules and having a high bispecific proportion in the mixture. In a system requiring two bispecifics, a proportion of at least 95%, at least 97% or higher (of the two) bispecifics is even obtained. This means that at most 5%, at most 3% or less of monospecific bivalent by-products are obtained. Notably, the amount of monomer by-products (i.e., half-molecules) is less important as these half-molecules can be easily separated from the dimers by virtue of their size differences.

[0035] In another preferred embodiment, the variable regions of the first and second CH3 domain-containing polypeptide chains recognize different target epitopes, while the variable regions of the third and fourth CH3 domain-containing polypeptide chains recognize the same target epitope. This will result in the production mainly of a bispecific immunoglobulin-like molecule and a monospecific immunoglobulin-like molecule. For example, if the variable regions of the first and second CH3 domain-containing polypeptide chains recognize different target epitopes, and if the variable regions of both the third and fourth CH3 domain-containing polypeptide chains recognize the same target epitope that is different from the target epitopes recognized by the first and second CH3 domains, a mixture of immunoglobulin-like molecules with AB or CC specificity will be formed. Accordingly, there is also provided a method according to the invention, wherein the target epitope recognized by the variable regions of the third and fourth CH3 domain-containing polypeptide chains is the same, but different from the target epitopes recognized by the variable regions of the first or second CH3 domain-containing polypeptide chain.

[0036] Alternatively, when the variable regions of the first and second CH3 domain-containing polypeptide chains recognize different target epitopes, and when the variable regions of both the third and fourth CH3 domain-containing polypeptide chains recognize the same epitope as the first or second CH3 domain-containing polypeptide chain, a mixture of immunoglobulin-like molecules with AB and AA, or AB and BB specificity will be formed. Accordingly, there is provided a method according to the invention, wherein the target epitope recognized by the variable regions of the third and fourth CH3 domain-containing polypeptide chains is the same as the target epitope recognized by the variable regions of the first or second CH3 domain-containing polypeptide chain. Another object of the invention is to provide a means and method for producing a defined mixture of bispecific and monospecific antibodies in a single cell culture. A non-limiting example of such a well-defined mixture is a mixture of a bispecific antibody with AB specificity and a monospecific antibody with AA specificity. Another example is a mixture of a bispecific antibody with AB specificity and a monospecific antibody with BB specificity. Another example is a mixture of a bispecific antibody with AB specificity and a monospecific antibody with CC specificity. Additionally, preferred means and methods are provided which produce a mixture of the desired antibodies that is at least 90%, more preferably at least 95% and most preferably at least 97% or even more than 99% of the desired antibodies.

[0037] In another embodiment, there is provided a method according to the invention, wherein the variable regions of the first and second polypeptide chains comprising CH3 domains recognize the same target epitope, while the variable regions of the third and fourth polypeptide chains comprising CH3 domains recognize a second target epitope, said second target epitope being different from the target epitope recognized by the first and second variable regions. This will result in the predominantly production of monospecific immunoglobulin-like molecules with AA specificity or BB specificity. The formation of bispecific immunoglobulin-like molecules is reduced or even avoided. In a plurality of embodiments, it is preferred to produce a mixture of monospecific antibodies in a single cell rather than a mixture of bispecific antibodies. For example, when crosslinking of two identical target molecules is desired, or when two targets are too far apart from each other to be bound by a single bispecific antibody. Producing a mixture of monospecific antibodies in a single cell may also be advantageous because the mixture can be regarded as a single therapeutic product. In the art, the therapeutic efficacy and safety of various monospecific antibodies have been demonstrated and market authorization has been obtained. Producing a mixture of monospecific antibodies in a single cell will thus facilitate testing the efficacy and safety of several such mixtures and will reduce the effort and cost of registration approval and manufacturing. However, there is currently no method available for producing a specific mixture of monospecific antibodies in a single cell and in which the formation of bispecific by-products is reduced to less than 5%. Another object of the present invention is to provide means and methods for producing such a well-defined mixture of homodimeric antibodies in a single cell and in which the formation of bispecific antibodies is reduced to less than 5%.

[0038] Thus, a method according to the invention is suitable for producing any desired mixture of bispecific and / or monospecific immunoglobulin-like molecules. Additionally, additional nucleic acid molecules, such as nucleic acid molecules encoding fifth and sixth (and seventh and eighth, etc.) polypeptide chains comprising CH3 domains, can be used to produce a defined mixture comprising more than two different immunoglobulin-like molecules.

[0039] Preferably, in a method according to the invention, at least two such CH3 domains are used, which comprise at least one combination of the mutations provided by the present invention. Through these mutations, a new specific interaction is formed between the two CH3 domains. These mutations according to the invention will be discussed in more detail below.

[0040] As used herein, the term "immunoglobulin-like molecule" refers to a protein molecule having at least one immunoglobulin (Ig) domain. The immunoglobulin-like molecule comprises a sequence having the function of at least the immunoglobulin CH3 domain, preferably a sequence comprising the CH3 domain of IgG1. A protein molecule having at least one CH3 domain can be further equipped with a specific binding moiety. Thus, the CH3 domain of the present invention comprising means for preferential pairing can be used to preferentially pair two CH3 domain-containing protein molecules to design binding molecules or mixtures of binding molecules of the desired heterodimer. The moiety that binds to the CH3 domain-containing protein molecule can be designed to be any binding agent, including but not limited to, single-chain Fvs, single-chain or tandem diabodies VHH, Fab, ankyrin repeat proteins or DART, TCR-like antibodies, MicroProteins, or In a preferred embodiment, the binding moiety is an antibody variable region (i.e., a VH / VL combination). The variable region, as part of the polypeptide chain containing the CH3 domain, preferably has a common light chain. In this case, only the VH of the variable region is different, while the VL of all variable regions is substantially the same.

[0041] Alternatively or additionally, other molecules can be engineered into the CH3 domain of the present invention, said other molecules including cytokines, hormones, soluble ligands, receptors, and / or peptides.

[0042] In a more preferred embodiment, the immunoglobulin-like molecule comprises a full-length Fc backbone. In a most preferred embodiment, the immunoglobulin-like molecule is an antibody. The variable regions of these antibodies preferably have a common light chain, but their VH regions can be different. As used herein, the term "antibody" refers to a protein molecule belonging to the immunoglobulin protein class that contains one or more domains that bind to an epitope on an antigen, wherein such domains are from the variable region of an antibody or are sequence homologous to the variable region of an antibody. Antibodies are known in the art and include a variety of isotypes, such as IgG1, IgG2, IgG3, IgG4, IgA, IgD, IgE, and IgM. The antibodies according to the present invention can be any of these isotypes, or functional derivatives and / or fragments thereof. In a preferred embodiment, the immunoglobulin-like molecules produced are antibodies of the IgG isotype because IgG antibodies have a longer half-life compared to antibodies of other isotypes.

[0043] Antibodies generated using the method according to the present invention can have sequences of any origin, including murine and human sequences. The antibodies can consist of sequences from only one source, such as fully human antibodies, or they can have sequences from more than one source, such as generating chimeric antibodies or humanized antibodies. Antibodies for therapeutic use are preferably as natural as possible for the subject to be treated (e.g., human antibodies for human subjects). The binding of antibodies can be expressed in terms of specificity and affinity. Specificity determines which antigen or its epitope binds to the binding domain. Affinity is a measure of the strength of binding to a specific antigen or epitope. Specific binding is defined as having an affinity (K -5 ) of at least 1×10 -7 M, more preferably 1×10 -9 M, and more preferably higher than 1×10 D M. Generally, monoclonal antibodies for therapeutic applications have an affinity of up to 1×10 -10 M or even higher. The term "antigen" as used herein refers to a substance or molecule that triggers the production of antibodies by the immune system when introduced into the body. Antigens and the like can be derived from pathogenic organisms, tumor cells, or other abnormal cells, can be derived from haptens, or even be their own structures. At the molecular level, an antigen is characterized by its ability to bind to the antigen-binding site of an antibody. A mixture of antigens can also be regarded as an "antigen", that is, those skilled in the art should understand that sometimes the lysate of tumor cells or virus particles can be represented as an "antigen", however, there are many antigenic determinants in such preparations of tumor cell lysates or virus particles. An antigen contains at least one, but often more, epitopes. The term "epitope" as used herein refers to the part of an antigen that is recognized by the immune system, particularly antibodies, B cells, or T cells. Although epitopes are generally considered to be derived from non-self proteins, recognizable sequences derived from the host can also be considered to be classified as epitopes.

[0044] The term "CH3 domain" is well known in the art. The structure of IgG has four chains, two light chains and two heavy chains; each light chain has two domains, variable light chain and constant light chain (VL and CL), and each heavy chain has four domains, variable heavy chain (VH) and three constant heavy chain domains (CH1, CH2, CH3). The regions of the CH2 and CH3 domains of the heavy chain are referred to as the Fc (fragment crystallizable) portion, Fc fragment, Fc scaffold or simply Fc. The IgG molecule is a heterotetramer having two heavy chains and two light chains, wherein the heavy chains are joined together by disulfide bonds (-S-S-) in the hinge region. The heavy chains dimerize through interactions at the CH3-CH3 domain interface and through interactions in the hinge region. The number of hinge disulfide bonds varies among immunoglobulin subclasses (Papadea and Check 1989). The Fc fragment of the immunoglobulin molecule is a dimer of the two C-terminal constant regions of the heavy chain (i.e., the CH2 and CH3 domains). Its physiological functions are to interact with the complement system and with specific receptors on various cell surfaces, etc. It is known that the interaction between the CH3 domains of two separate heavy chains plays an important role in driving heavy chain dimerization. Thus, the CH3 domain directs the association of antibody heavy chains, and it is known that the interface between CH3 domains contains more than 20 contact residues from each chain, which play a role in CH3-CH3 interactions (Deisenhofer J., Biochemistry 1981 (Vol. 20) pp. 2361-2370; Miller S., J. Mol. Biol. 1990 (Vol. 216) pp. 965-973; Padlan, Advances in Protein Chemistry 1996 (Vol. 49) pp. 57-133). The CH3 variants of the present invention can thus be used to associate with other antibody domains to produce bispecific or monospecific full-length antibodies. The specificity of the antibody defined by the combination of VH / VL generally does not affect the dimerization behavior of the heavy chain driven by the CH3 domain.

[0045] As used herein, the terms "contact residue", "contact amino acid", "interface residue" and "interface amino acid" generally refer to any amino acid residue present in the CH3 domain that can participate in interdomain contacts, such as can be calculated by techniques known in the art, including calculating the solvent accessible surface area (ASA) of CH3 domain residues in the presence and absence of a second chain (Lee and Richards J. Mol. Biol. 1971 (55) 379), wherein a difference in ASA is shown between these two calculations The residues were identified as contact residues. According to the EU numbering system, the identified contact residues include residues at positions 347, 349, 350, 351, 352, 353, 354, 355, 356, 357, 360, 364, 366, 368, 370, 390, 392, 394, 395, 397, 399, 400, 405, 407, 409, 439 (Table A).

[0046] Table A: List of CH3 domain interface residues

[0047] Interface residues in chain A Contact residues in chain B Q347 K360 Y349 S354, D356, E357, K360 T350 S354, R355 L351 L351, P352, P353, S354, T366 S354 Y349, T350, L351 R355 T350 D356 Y349, K439 E357 Y349, K370 K360 Q347, Y349 S364 L368, K370 T366 L351, Y407 L368 S364, K409 K370 E357, S364 N390 S400 K392 L398, D399, S400, F405 T394 T394, V397, F405, Y407 P395 V397 V397 T394, P395 D399 K392, K409 S400 N390, K392 F405 K392, T394, K409 Y407 T366, T394, Y407, K409 K409 L368, D399, F405, Y407 K439 D356

[0048] Contact residues within the CH3-CH3 interface can be charged amino acids or neutral amino acid residues. As used herein, the term "charged amino acid residue" or "charged residue" refers to an amino acid residue having a charged side chain. These can be positively charged side chains, such as those present in arginine (Arg, R), histidine (His, H), and lysine (Lys, K), or they can be negatively charged side chains, such as those present in aspartic acid (Asp, D) and glutamic acid (Glu, E). As used herein, the term "neutral amino acid residue" or neutral residue refers to all other amino acids that do not carry a charged side chain. These neutral residues include serine (Ser, S), threonine (Thr, T), asparagine (Asn, N), glutamine (GLu, Q), cysteine (Cys, C), glycine (Gly, G), proline (Pro, P), alanine (Ala, A), valine (Val, V), isoleucine (Ile, I), leucine (Leu, L), methionine (Met, M), phenylalanine (Phe, F), tyrosine (Tyr, Y), and tryptophan (Trp, T).

[0049] As used herein, the term "CH3-CH3 domain interface" or "CH3 interface", "CH3-CH3 pairing", "domain interface" or the simplified "interface" refers to the association between two CH3 domains of a polypeptide containing CH3 domains in isolation, which is the result of amino acid residue interactions, i.e., at least one interaction between the amino acids of the first CH3 domain and the amino acids of the second CH3 domain. Such interactions are, for example, through van der Waals forces, hydrogen bonds, water-mediated hydrogen bonds, salt bridges or other electrostatic forces, attractive interactions between aromatic side chains, formation of disulfide bonds, or other forces known to those skilled in the art.

[0050] As used herein, the means for preferentially pairing the first CH3 domain-containing polypeptide with the second CH3 domain-containing polypeptide and the third CH3 domain-containing polypeptide with the fourth CH3 domain-containing polypeptide can be any means known in the art. In one embodiment, at least one nucleic acid molecule encodes a CH3 domain having a large amino acid residue (i.e., a "knob" or "protrusion") (such as R, F, Y, W, I or L) at the contacting residue position, and at least one additional nucleic acid molecule encodes a CH3 domain having a small amino acid residue (i.e., a "hole" or "cavity") (such as G, A, S, T or V) at the complementary contacting residue position. Due to the spatial conformation of the contacting amino acids, the resulting CH3 domains will preferentially pair with each other. The knob-into-hole technology has been described in detail previously herein. In another embodiment of the invention, at least one nucleic acid molecule encodes a CH3 domain containing an amino acid now carrying a charge opposite to that of the wild type at a naturally charged contacting residue position (i.e., a naturally occurring K, H, R, D or E), and at least one additional nucleic acid molecule encodes a CH3 domain containing an amino acid now carrying a charge opposite to that of the wild type at a naturally charged complementary contacting residue. Due to the opposite charges of the contacting amino acids, the resulting engineered CH3 domains will preferentially pair with each other, while the pairing of identical CH3 domains will be reduced due to electrostatic repulsion. In one embodiment, CH3 mutations as described in EP01870459, WO2009 / 089004, Gunasekaran et al. (2010) are used. In one embodiment, the means for preferentially pairing the first CH3 domain-containing polypeptide with the second CH3 domain-containing polypeptide are "knob" and "hole" amino acid residues, and the means for preferentially pairing the third CH3 domain-containing polypeptide with the fourth CH3 domain-containing polypeptide are charge-engineered amino acids. Preferably, both of the means for preferentially pairing the first CH3 domain-containing polypeptide with the second CH3 domain-containing polypeptide and the third CH3 domain-containing polypeptide with the fourth CH3 domain-containing polypeptide are charge-engineered amino acids. In one embodiment, different amino acid residues are engineered to preferentially pair the first CH3 domain-containing polypeptide with the second CH3 domain-containing polypeptide compared to the amino acid residues engineered to preferentially pair the third CH3 domain-containing polypeptide with the fourth CH3 domain-containing polypeptide. In a particularly preferred embodiment, at least the first and second nucleic acid molecules encode CH3 domains having the new mutations provided by the present invention. As described in more detail below, the present invention provides new mutations of CH3 that are capable of producing certain desired bispecific immunoglobulin-like molecules without significant amounts of undesired (dimeric) by-products.The present invention also provides novel CH3 mutations that are capable of generating certain desired monospecific immunoglobulin-like molecules without significant amounts of undesired (dimeric) by-products. Thus, it is preferred to use at least one of these CH3 mutations according to the present invention.

[0051] As used herein, the terms "polypeptide", "polypeptide molecule" or "polypeptide chain" refer to a chain of amino acids covalently linked together by peptide bonds. Proteins are generally composed of one or more polypeptide molecules. Each polypeptide has a free amino group at one end, which is referred to as the amino terminus or N-terminus. The other end with its free carboxyl group is referred to as the carboxyl terminus or C-terminus. The polypeptides according to the present invention may have undergone post-translational modification processes and may be glycosylated, for example. Thus, the polypeptide chain containing the CH3 domain according to the present invention refers to a polypeptide chain that at least encompasses the Ig CH3 domain and that may have undergone post-translational modification processes.

[0052] As used herein, the term "nucleic acid molecule" is defined as a molecule comprising a nucleotide chain, more preferably a DNA and / or RNA chain. In one embodiment, double-stranded RNA is used. In other embodiments, the nucleic acid molecules of the present invention comprise other types of nucleic acid structures, such as, for example, DNA / RNA helices, peptide nucleic acids (PNAs), locked nucleic acids (LNAs) and / or ribozymes. Thus, the term "nucleic acid molecule" also encompasses chains containing non-natural nucleotides, modified nucleotides and / or non-nucleotide building blocks that exhibit the same function as natural nucleotides.

[0053] The present invention also provides a method for preparing a host cell for producing at least two different immunoglobulin-like molecules, the method comprising the steps of: introducing into the host cell nucleic acid sequences encoding at least first, second, third and fourth polypeptide chains containing the CH3 domain, wherein at least two of the nucleic acid sequences have means for preferentially pairing the first polypeptide chain containing the CH3 domain with the second polypeptide chain containing the CH3 domain and the third polypeptide chain containing the CH3 domain with the fourth polypeptide chain containing the CH3 domain, and wherein the nucleic acid sequences are introduced either successively or simultaneously.

[0054] Another aspect of the invention is to provide a method for preparing a heterodimeric immunoglobulin-like molecule, the method comprising the steps of: introducing into the host cell nucleic acid sequences encoding at least a first and a second polypeptide chain comprising a CH3 domain, wherein the first polypeptide chain comprising a CH3 domain comprises at least one substitution of a neutral amino acid residue with a positively charged amino acid residue, and wherein the second polypeptide chain comprising a CH3 domain comprises at least one substitution of a neutral amino acid residue with a negatively charged amino acid residue, and wherein the nucleic acid sequences are introduced sequentially or simultaneously. The method for preparing the host cell preferably further comprises the step of introducing into the host cell a nucleic acid sequence encoding a common light chain.

[0055] Also provided herein is a recombinant host cell comprising nucleic acid sequences encoding at least a first, a second, a third and a fourth polypeptide chain comprising a CH3 domain, wherein at least two of the nucleic acid molecules have means for preferentially pairing the first polypeptide chain comprising a CH3 domain with the second polypeptide chain comprising a CH3 domain and for preferentially pairing the third polypeptide chain comprising a CH3 domain with the fourth polypeptide chain comprising a CH3 domain. The invention also provides a recombinant host cell comprising nucleic acid sequences encoding at least a first and a second polypeptide chain comprising a CH3 domain, wherein the first polypeptide chain comprising a CH3 domain comprises at least one substitution of a neutral amino acid residue with a positively charged amino acid residue, and wherein the second polypeptide chain comprising a CH3 domain comprises at least one substitution of a neutral amino acid residue with a negatively charged amino acid residue.

[0056] The recombinant host cell according to the invention preferably further comprises a nucleic acid sequence encoding a common light chain.

[0057] A "host cell" according to the present invention can be any host cell capable of expressing a recombinant DNA molecule, including: bacteria, such as Escherichia (e.g., E. coli), Enterobacter, Salmonalla, Bacillus, Pseudomonas, Streptomyces; yeast, such as S. cerevisiae, K. lactis, P. pastoris, Candida or Yarrowia; filamentous fungi (e.g., Neurospora, Aspergillus oryzae, Aspergillus nidulans and Aspergillus niger); insect cells, such as Spodoptera frugiperda SF-9 or SF-21 cells; and preferably mammalian cells, such as Chinese hamster ovary (CHO) cells, BHK cells, murine cells including SP2 / 0 cells and NS-0 myeloma cells; primate cells, such as COS and Vero cells, MDCK cells, BRL 3A cells, hybridomas, tumor cells, immortalized primary cells; human cells, such as W138, HepG2, HeLa, HEK293, HT1080 or embryonic retina cells such as PER.C6, etc. Generally, the expression system chosen will include a mammalian cell expression vector and a host such that the antibody can be glycosylated appropriately. Human cell lines (preferably PER.C6) can advantageously be used to obtain antibodies with a fully human glycosylation pattern. The conditions for cell growth or propagation (see, e.g., Tissue Culture, Academic Press, Kruse and Paterson, eds. (1973)) and the conditions for expressing the recombinant product can be somewhat different, and the process is generally optimized according to methods well known to those skilled in the art to increase the proportion of the product and / or the growth of the cells relative to each other. Generally, the principles, protocols and practical techniques for maximizing the productivity of mammalian cell cultures can be found in Mammalian Cell Biotechnology: a Practical Approach (edited by M. Butler, IRL Press, 1991). The expression of antibodies in recombinant host cells has been widely described in the art (see, e.g., EP0120694, EP0314161, EP0481790, EP0523949; U.S. Patent 4,816,567, WO00 / 63403).Nucleic acid molecules encoding light and heavy chains can exist as extrachromosomal copies and / or be stably integrated into the chromosome of a host cell, the latter being preferred.

[0058] Another aspect of the invention provides a culture of recombinant host cells according to the invention, or a culture of recombinant host cells obtainable or obtained by a method according to the invention, which culture produces at least two different immunoglobulin-like molecules or produces heterodimeric immunoglobulin-like molecules.

[0059] To obtain the expression of a nucleic acid sequence encoding a polypeptide containing a CH3 domain, it is well known to those skilled in the art that a sequence capable of driving such expression can be functionally linked to the nucleic acid sequence encoding the polypeptide containing the CH3 domain. Functional linkage is intended to describe that the nucleic acid sequence encoding the polypeptide containing the CH3 domain or its precursor is linked to a sequence capable of driving expression such that these sequences can drive the expression of the polypeptide containing the CH3 domain or its precursor. Useful expression vectors are available in the art, such as the pcDNA vector series from Invitrogen. In which the sequence encoding the polypeptide of interest is correctly inserted relative to the sequences controlling the transcription and translation of the encoded polypeptide, and the resulting expression cassette can be used to produce the polypeptide of interest, which is referred to as expression. Sequences capable of driving expression can include promoters, enhancers, etc., and combinations thereof. These should be able to function in a host cell to drive the expression of a nucleic acid sequence functionally linked to them. Promoters can be constitutive or regulatable and can be obtained from a variety of sources, including viral, prokaryotic or eukaryotic sources, or be artificially designed. The expression of the nucleic acid of interest can be from a native promoter or its derivatives, or from a completely heterologous promoter. Some well-known and very commonly used promoters for expression in eukaryotic cells include: promoters from viruses (such as adenovirus) such as the promoter of E1A, promoters from cytomegalovirus (CMV) such as the CMV immediate early (IE) promoter, promoters from Simian Virus 40 (SV40), etc. Suitable promoters can also be from eukaryotic cells, such as the metallothionein (MT) promoter, elongation factor 1α (EF-1α) promoter, actin promoter, immunoglobulin promoter, heat shock promoter, etc. Any promoter or enhancer / promoter capable of driving the expression of the sequence of interest in a host cell is suitable for the present invention. In one embodiment, the sequence capable of driving expression contains a region from the CMV promoter, preferably the region of nucleotides -735 to +95 of the CMV immediate early gene enhancer / promoter. Those skilled in the art will recognize that the expression sequences used in the present invention can be appropriately combined with elements capable of stabilizing or enhancing expression (such as insulators, matrix attachment regions, STAR elements (WO03 / 004704), etc.). This can enhance stability and / or expression levels.

[0060] The production of proteins in recombinant host cells has been widely described, for example, in Current Protocols in Protein Science, 1995, Coligan JE, Dunn BM, Ploegh HL, Speicher DW, Wingfield PT, ISBN 0-471-11184-8; Bendig, 1988. Cells are cultured to enable them to metabolize and / or grow and / or divide and / or produce the recombinant protein of interest. This can be accomplished by methods well known to those skilled in the art, including but not limited to providing nutrients to the cells. The methods include growth attached to a surface, suspension growth, or a combination thereof. Multiple culture conditions can be optimized by methods well known in the art to optimize protein production. The culture can be carried out, for example, in culture dishes, roller bottles, or bioreactors, using batch, fed-batch, continuous systems, hollow fibers, etc. To achieve large-scale (continuous) production of recombinant proteins by cell culture, it is preferred in the art to enable the cells to grow in suspension and preferably to enable the cells to be cultured without serum of animal or human origin or components of serum of animal or human origin. Since there are no additional animal or human proteins from the culture medium, it is easier to purify and safety is improved. At the same time, since synthetic media have the best reproducibility, the system is also reliable.

[0061] Immunoglobulin-like molecules are expressed in host cells and harvested from the cells or preferably from the cell culture medium by methods generally known to those skilled in the art. After harvesting, these immunoglobulin-like molecules can be purified by using methods known in the art. Such methods can include precipitation, centrifugation, filtration, size exclusion chromatography, affinity chromatography, cation and / or anion exchange chromatography, hydrophobic interaction chromatography, and so on. For mixtures of antibodies containing IgG molecules, protein A, or protein G, affinity chromatography can be appropriately used (see, for example, U.S. Patents 4,801,687 and 5,151,504).

[0062] The immunoglobulin-like molecules and / or mixtures thereof produced by the method according to the invention preferably have a common light chain. Accordingly, a method according to the invention is also provided which further comprises providing to the host cell a nucleic acid molecule encoding a common light chain. This is a light chain that is capable of pairing with at least two different heavy chains to form a functional antigen-binding domain. The functional antigen-binding domain is capable of specifically binding an antigen. Preferably, a common light chain is used that is capable of pairing with all heavy chains produced by the method according to the invention to form a functional antigen-binding domain, thereby avoiding the mismatching of non-matching heavy and light chains. In one aspect, only a common light chain having one identical amino acid sequence is used. Alternatively, those skilled in the art will recognize that "common" also refers to functionally equivalent light chains that do not have identical amino acid sequences. There are many variants of said light chain in which there are mutations (deletions, substitutions, additions) that do not substantially affect the formation of the functional binding region. Such variants are thus also capable of binding different heavy chains and forming a functional antigen-binding domain. Accordingly, the term "common light chain" as used herein refers to light chains that may be identical or have some amino acid sequence differences but the resulting antibody retains binding specificity after pairing with a heavy chain. For example, non-identical but still functionally equivalent light chains may be prepared or found by introducing and testing conservative amino acid changes and / or amino acid changes in regions that are not or only partially helpful for binding specificity when paired with a heavy chain. Combinations of certain common light chains and such functionally equivalent variants are encompassed by the term "common light chain". Reference may be made to the detailed description of the use of common light chains in WO2004 / 009618. Preferably, the common light chain used in the present invention is a germline-like light chain, more preferably a germline light chain, preferably a rearranged germline human κ light chain, and most preferably a rearranged germline human κ light chain IgVκ1-39 / Jκ or IGVκ3-20 / Jκ.

[0063] Alternatively, one skilled in the art may choose to use means for forcing the pairing of heavy and light chains as an alternative to using a common light chain and avoid the mismatching of non-matching heavy and light chains, such means being described in, for example, WO2009 / 080251, WO2009 / 080252, and / or WO2009 / 080253.

[0064] The present invention provides novel engineered CH3 domains, as well as novel combinations of CH3 mutations. Prior to the present invention, charged contacting amino acids of CH3 domains known to be involved in CH3-CH3 pairing were replaced with amino acids of opposite charge (charge reversal), thereby affecting CH3-CH3 pairing. The mutations according to the present invention are a creative alternative to this approach, as now uncharged or neutral CH3 amino acids in wild-type CH3 are replaced with charged residues. In this embodiment, the present invention does not exchange charged contacting amino acids with amino acids of opposite charge, but rather replaces uncharged CH3 amino acids with charged CH3 amino acids. The method of the present invention not only provides a method for effectively manipulating the dimerization of CH3 domains, but also has the advantage of generating at least one additional charge-charge interaction at the CH3 interface. Given this additional charge-charge interaction at the CH3-CH3 interface in addition to the existing charge pairs, dimers according to the present invention are generally more stable than wild-type dimers (wild-type dimers are defined as bispecific IgG (AB) without CH3 engineering compared to their parental homodimers (AA or BB)). In addition, it has surprisingly been possible to further increase the proportion of one or more desired immunoglobulin-like molecules in a mixture. As previously mentioned, known methods for preferentially generating bispecific antibodies in the art typically involve the production of some undesired dimeric by-products. For example, the proportion of the desired bispecific antibody using the knob-into-hole technique is at most 87%, while the electrostatic engineering method of replacing charged contacting amino acids with amino acids of opposite charge also results in a proportion as high as 96% (see, for example, Example 11). Quite surprisingly, the inventors have successfully introduced mutations to further increase the proportion of desired immunoglobulin-like molecules in a mixture. For example, Example 17 discloses a method using the mutations according to the present invention, in which the proportion of the desired bispecific antibody is increased to such an extent that no dimeric by-products are detected in the resulting mixture. Unpaired half-molecules consisting of only a single heavy chain paired with a common light chain are present to some extent in the mixture, but these are the result of imbalanced heavy chain expression and can be easily separated from the mixture by size exclusion chromatography. Thus, using such mutations according to the present invention, bispecific immunoglobulin-like molecules with a high proportion can be produced in a single cell and there are substantially no contaminating dimeric by-products present, which are particularly suitable for the production of pharmaceutical compositions.

[0065] Accordingly, a preferred embodiment of the present invention provides a method for generating heterodimeric immunoglobulin-like molecules from a single cell, wherein the immunoglobulin-like molecules comprise two CH3 domains capable of forming an interface, the method comprising providing in the cell:

[0066] a. A first nucleic acid molecule encoding a first polypeptide chain containing a CH3 domain,

[0067] b. A second nucleic acid molecule encoding a second polypeptide chain containing a CH3 domain, wherein the first polypeptide chain containing a CH3 domain comprises at least one substitution of a neutral amino acid residue with a positively charged amino acid residue, and wherein the second polypeptide chain containing a CH3 domain comprises at least one substitution of a neutral amino acid residue with a negatively charged amino acid residue. The method further comprises culturing the host cell and enabling the expression of the two nucleic acid molecules and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0068] The method preferably further comprises the step of providing the host cell with a nucleic acid molecule encoding a common light chain, which has the advantages outlined previously herein.

[0069] The amino acid at position 366 of one CH3 domain and the amino acid at position 351 of the second CH3 domain have been reported to be a pair of contacting residues at the CH3-CH3 interface, which means that in the three-dimensional conformation of the resulting immunoglobulin-like molecule, they are located close enough to each other to be able to interact with each other. Thus, the first CH3 domain will preferentially pair with the second CH3 domain.

[0070] In one embodiment, the threonine (T) at position 366 of the first CH3 domain is replaced with a first charged amino acid, and the leucine (L) at position 351 of the second CH3 domain is replaced with a second charged amino acid, wherein the first and second charged amino acids have opposite charges. If the first polypeptide chain containing a CH3 domain carrying a charged residue at position 366 further comprises a variable domain specific for antigen A, and if the second polypeptide chain containing a CH3 domain carrying an oppositely charged residue at position 351 further comprises a variable domain specific for antigen B, then a bispecific immunoglobulin-like molecule with AB specificity will be mainly formed. Thus, there is also provided a method according to the invention, wherein the means for preferentially pairing the first polypeptide chain containing a CH3 domain with the second polypeptide chain containing a CH3 domain or the means for preferentially pairing the third polypeptide chain containing a CH3 domain with the fourth polypeptide chain containing a CH3 domain is to replace the threonine at position 366 of the first or third CH3 domain with a first charged amino acid and replace the leucine at position 351 of the second or fourth CH3 domain with a second charged amino acid, wherein the first and second charged amino acids have opposite charges.

[0071] A preferred combination of mutations according to the invention is: threonine (T) at position 366 of the first CH3 domain-containing polypeptide is replaced by lysine (K), the polypeptide further comprising a variable domain (e.g., having specificity A), and leucine (L) at position 351 of the second CH3 domain-containing polypeptide is replaced by aspartic acid (D), the polypeptide further comprising a variable domain (e.g., having specificity B). This is designated the T366K / L351D paired mutation. As previously explained, the amino acid at position 366 of one CH3 domain and the amino acid at position 351 of the second CH3 domain have been reported to be a pair of contacting residues at the CH3-CH3 interface. The lysine introduced at position 366 and the aspartic acid introduced at position 351 have opposite charges, such that these amino acids electrostatically attract each other. Thus, the first CH3 domain will preferentially attract the second CH3 domain and will predominantly form an immunoglobulin-like molecule comprising the first CH3 domain having lysine at position 366, the first CH3 domain pairing with the second CH3 domain having aspartic acid at position 351. If the first CH3 domain-containing polypeptide has antigen A specificity and if the second CH3 domain-containing polypeptide has antigen B specificity, a bispecific immunoglobulin-like molecule having "AB" specificity will predominantly be formed. Note that in some embodiments, the specificities of the variable domains of both the first and second CH3 domain-containing polypeptide chains can be the same, which will result in the formation of a monospecific immunoglobulin-like molecule (e.g., having "AA" specificity). As described above, one of the advantages of the mutations according to the invention is the fact that new interactions are created between the newly introduced charged amino acid pairs rather than replacing existing charged amino acid interactions. This has not been previously disclosed or suggested. Accordingly, one aspect of the invention provides a method according to the invention for producing at least two different immunoglobulin-like molecules from a single host cell, wherein the first CH3 domain-containing polypeptide chain comprises the amino acid substitution T366K and the second CH3 domain-containing polypeptide chain comprises the amino acid substitution L351D. One embodiment provides a method for producing a heterodimeric immunoglobulin-like molecule from a single cell, wherein the immunoglobulin-like molecule comprises two CH3 domains capable of forming an interface, the method comprising providing in the cell:

[0072] - a first nucleic acid molecule encoding a first CH3 domain-containing polypeptide chain, and

[0073] - a second nucleic acid molecule encoding a second CH3 domain-containing polypeptide chain,

[0074] Wherein, the first polypeptide chain containing a CH3 domain comprises the amino acid substitution T366K, and wherein the second polypeptide chain containing a CH3 domain comprises the amino acid substitution L351D, and the method further comprises the steps of culturing the host cell and enabling the two nucleic acid molecules to be expressed and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0075] Using the amino acid substitutions according to the invention mentioned above, it is possible to produce heterodimeric immunoglobulin-like molecules from a single cell, such that the presence of contaminating homodimers is less than 5%, preferably less than 2%, more preferably less than 1%, or most preferably, whereby contaminating homodimers are substantially absent. Accordingly, one embodiment provides a method for producing heterodimeric immunoglobulin-like molecules from a single cell, wherein the immunoglobulin-like molecule comprises two CH3 domains capable of forming an interface, and wherein the presence of contaminating homodimers is less than 5%, preferably less than 2%, more preferably less than 1%, and most preferably contaminating homodimers are substantially absent, the method comprising providing in the cell:

[0076] - a first nucleic acid molecule encoding a first polypeptide chain containing a CH3 domain, and

[0077] - a second nucleic acid molecule encoding a second polypeptide chain containing a CH3 domain,

[0078] Wherein, the first polypeptide chain containing a CH3 domain comprises the amino acid substitution T366K, and wherein the second polypeptide chain containing a CH3 domain comprises the amino acid substitution L351D, and the method further comprises the steps of culturing the host cell and enabling the two nucleic acid molecules to be expressed and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0079] Preferably, there is provided a method according to the invention for producing at least two different immunoglobulin-like molecules, or a method according to the invention for producing heterodimeric immunoglobulin-like molecules, wherein the first polypeptide chain containing a CH3 domain further comprises the amino acid substitution L351K. More preferably, the second polypeptide chain containing a CH3 domain further comprises an amino acid substitution selected from Y349E, Y349D, and L368E. Most preferably, the second polypeptide chain containing a CH3 domain further comprises the amino acid substitution L368E.

[0080] Thus, in a preferred embodiment, the above-described T366K / L351’D mutation according to the present invention is further combined with the substitution of leucine (L) for glutamate (E) at position 368 in the second CH3 domain. For example, this is designated as the T366K / L351’D,L368’E mutation (although alternative designations are possible, such as T336K / L351D-L368E or T366K / L351D,L368E or T366K-L351D,L368E). As shown in Example 17, introducing such a mutation according to the present invention into a first CH3 domain-containing polypeptide specific for antigen A and a second CH3 domain-containing polypeptide specific for antigen B results in a particularly good proportion of bispecific immunoglobulin-like molecules with dual AB specificity. Using this mutant pairing, it is even possible to obtain bispecific antibodies without forming any detectable amount of homodimers. Thus, a particularly preferred embodiment provides a method for generating heterodimeric immunoglobulin-like molecules from a single cell, wherein the immunoglobulin-like molecules comprise two CH3 domains capable of forming an interface and wherein the presence of contaminating homodimers is less than 5%, preferably less than 2%, more preferably less than 1%, and most preferably essentially no contaminating homodimers are present, the method comprising providing in the cell:

[0081] - a first nucleic acid molecule encoding a first polypeptide chain containing a CH3 domain, and

[0082] - a second nucleic acid molecule encoding a second polypeptide chain containing a CH3 domain,

[0083] wherein the first polypeptide chain containing a CH3 domain comprises the amino acid substitution T366K, and wherein the second polypeptide chain containing a CH3 domain comprises the amino acid substitutions L351D and L368E, the method further comprising the steps of culturing the host cell and enabling expression of the two nucleic acid molecules and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0084] In yet another preferred embodiment, the threonine (T) at position 366 of the first CH3 domain is replaced with lysine (K), the leucine (L) at position 351 of the second CH3 domain is replaced with aspartic acid (D) and the tyrosine (Y) at position 349 of the second CH3 domain is replaced with glutamic acid (E). This is, for example, designated as the T366K / L351’D,Y349’E mutation, but other ways of denoting these mutations can include, for example, T366K-L351D:Y349E or T366K / L351D,Y349E or just T366K / L351DY349E. Residue Y349 is an adjacent residue to the residue at position 351 and may contribute to dimer interactions. According to computer simulation data, Y349E increases the stability of the heterodimer (lower computer simulation score) as well as the destabilization of the homodimer (higher computer simulation score) and glutamic acid (E) at position 349 is more favorable than aspartic acid (D). Thus, introducing a second amino acid substitution in the second CH3 domain-containing polypeptide that already contains an amino acid substitution at position 351 further favors heterodimerization.

[0085] Accordingly, a particularly preferred embodiment provides a method for generating a heterodimeric immunoglobulin-like molecule from a single cell, wherein the immunoglobulin-like molecule comprises two CH3 domains capable of forming an interface, and wherein the contaminating homodimer is less than 5%, more preferably less than 2%, even more preferably less than 1%, and most preferably substantially absent, the method comprising providing in the cell:

[0086] - a first nucleic acid molecule encoding a first polypeptide chain comprising a CH3 domain, and

[0087] - a second nucleic acid molecule encoding a second polypeptide chain comprising a CH3 domain,

[0088] wherein the first polypeptide chain comprising a CH3 domain comprises the amino acid substitution T366K, and wherein the second polypeptide chain comprising a CH3 domain comprises the amino acid substitutions L351D and Y349E, the method further comprising the steps of culturing the host cell and enabling the expression of the two nucleic acid molecules and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0089] In yet another preferred embodiment, the threonine (T) at position 366 of the first CH3 domain is replaced with lysine (K), the leucine (L) at position 351 of the second CH3 domain is replaced with aspartic acid (D), the tyrosine (Y) at position 349 of the second CH3 domain is replaced with glutamic acid (E), and the leucine (L) at position 368 of the second CH3 domain is replaced with glutamic acid (E). This is designated as the T366K / L351’D,Y349’E,L368’E mutation. These two residues, Y349 and L368, are residues that may contribute to dimer interactions. According to computer simulation data, Y349E and L368E increase the stability of the heterodimer (lower computer simulation score) and the instability of the BB dimer (higher computer simulation score), and glutamic acid (E) at positions 349 and 368 is more favorable than aspartic acid (D). Thus, introducing the second and third amino acid substitutions into the B chain that already contains the amino acid substitution at position 351 further favors heterodimerization. Accordingly, a particularly preferred embodiment provides a method for generating a heterodimeric immunoglobulin-like molecule from a single cell, wherein the immunoglobulin-like molecule comprises two CH3 domains capable of forming an interface, and wherein contaminating homodimers are less than 5%, more preferably less than 2%, even more preferably less than 1%, and most preferably substantially absent. The method comprises providing in the cell:

[0090] - a first nucleic acid molecule encoding a first polypeptide chain comprising a CH3 domain, and

[0091] - a second nucleic acid molecule encoding a second polypeptide chain comprising a CH3 domain, wherein the first polypeptide chain comprising a CH3 domain comprises the amino acid substitution T366K, and wherein the second polypeptide chain comprising a CH3 domain comprises the amino acid substitutions L351D, Y349E, and L368E. The method further comprises culturing the host cell and enabling expression of the at least two nucleic acid molecules and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0092] In yet another preferred embodiment, the threonine (T) at position 366 of the first CH3 domain is replaced by lysine (K), the leucine (L) at position 351 of the first CH3 domain is replaced by lysine (K), the leucine (L) at position 351 of the second CH3 domain is replaced by aspartic acid (D), and the leucine (L) at position 368 of the second CH3 domain is replaced by glutamic acid (E). This is designated as the T366K,L351K / L351’D,L368’E mutation. This mutation also enhances the proportion of the desired (bispecific) antibody, as shown in the examples. Similarly, these mutations also enable the production of bispecific antibodies without forming any detectable amount of homodimers. Accordingly, there is also provided a method for producing a heterodimeric immunoglobulin-like molecule from a single cell, wherein the immunoglobulin-like molecule comprises two CH3 domains capable of forming an interface, and wherein the contaminating homodimers are less than 5%, preferably less than 2%, more preferably less than 1%, and most preferably substantially absent, the method comprising providing in the cell:

[0093] - a first nucleic acid molecule encoding a first polypeptide chain comprising a CH3 domain, and

[0094] - a second nucleic acid molecule encoding a second polypeptide chain comprising a CH3 domain, wherein the first polypeptide chain comprising a CH3 domain comprises the amino acid substitutions T366K and L351K, and wherein the second polypeptide chain comprising a CH3 domain comprises the amino acid substitutions L351D and L368E, the method further comprising culturing the host cell and enabling the expression of the two nucleic acid molecules and harvesting the heterodimeric immunoglobulin-like molecule from the culture.

[0095] In yet another preferred embodiment, the threonine (T) at position 366 of the first CH3 domain is replaced with lysine (K), the leucine (L) at position 351 of the first CH3 domain is replaced with lysine (K), the leucine (L) at position 351 of the second CH3 domain is replaced with aspartic acid (D), the tyrosine (Y) at position 349 of the second CH3 domain is replaced with aspartic acid (D), and the arginine (R) at position 355 of the second CH3 domain is replaced with aspartic acid (D). This is designated as the T366K,L351K / L351’D,Y349’D,R355’D mutation. The T366K-L351K / L351’D-Y349’D pairing can be further improved by the R355’D mutation of the B chain, which results in a higher BB computer simulation score and a slight increase in the AB computer simulation score. Accordingly, there is further provided a method for generating heterodimeric immunoglobulin-like molecules from a single cell, wherein the immunoglobulin-like molecules comprise two CH3 domains capable of forming an interface, and wherein the contaminating homodimers are less than 5%, more preferably less than 2%, even more preferably less than 1%, and most preferably substantially absent, the method comprising providing in the cell:

[0096] - a first nucleic acid molecule encoding a first polypeptide chain comprising a CH3 domain, and

[0097] - a second nucleic acid molecule encoding a second polypeptide chain comprising a CH3 domain, wherein the first polypeptide chain comprising a CH3 domain comprises the amino acid substitutions T366K and L351K, and wherein the second polypeptide chain comprising a CH3 domain comprises the amino acid substitutions L351D, Y349D, and R355D, the method further comprising culturing the host cell and enabling the expression of the two nucleic acid molecules and harvesting the heterodimeric immunoglobulin-like molecules from the culture.

[0098] Table B provides an overview of mutations that can be introduced into the CH3 domain, which mutations are preferred means for preferential pairing to produce heterodimers or homodimers.

[0099] Table B:

[0100] Amino acid substitutions in CH3 Construct # Pairs preferentially with -(wild type) - Wild type E356K, D399K 1 Construct 2 or 3 K392D, K409D 2 Construct 1 K392D, K409D, K439D 3 Construct 1 K392D, D399K, K409D 4 Construct 4 E356K, E357K, K439D, K370D 5 Construct 5 T366W 6 Construct 7 T366S, L368A, Y407V 7 Construct 6 T366K 43 Constructs 63, 69, 70, 71, 73 L351D 63 Constructs 43, 68 T366K, L351K 68 Constructs 63, 69, 70, 71, 72, 75 L351D, L368E 69 Constructs 43, 68 L351E, Y349E 70 Constructs 43, 68 L351D, Y349E 71 Constructs 43, 68 L351D, R355D 72 Constructs 43, 68 L351D, Y349E, L368E 73 Construct 43 L351D, Y349D, R355D 75 Construct 68

[0101] Accordingly, provided herein is also a method according to the present invention for generating at least two different immunoglobulin-like molecules, or a method according to the present invention for generating a heterodimeric immunoglobulin-like molecule, wherein the means for preferentially pairing the first CH3 domain-containing polypeptide with the second CH3 domain-containing polypeptide and / or the means for preferentially pairing the third CH3 domain-containing polypeptide with the fourth CH3 domain-containing polypeptide comprise at least one combination of the mutations shown in Table B. Preferably, the means for preferentially pairing the first CH3 domain-containing polypeptide with the second CH3 domain-containing polypeptide and the means for preferentially pairing the third CH3 domain-containing polypeptide with the fourth CH3 domain-containing polypeptide comprise at least two combinations of the mutations shown in Table B.

[0102] The present invention also provides new combinations of CH3 mutations, which make it possible to generate a mixture of at least two monospecific immunoglobulin-like molecules in a single cell, wherein the contaminating bispecific immunoglobulin-like molecules are less than 5%, preferably more than 2%, even more preferably less than 1%, and most preferably even substantially absent. Thus, these mutations according to the invention are particularly suitable for generating mixtures of monospecific antibodies, which are advantageous in situations such as when a high cross-linking level of two identical target molecules is desired, when a sufficient density of antibodies on the target cell is required to recruit certain effector functions (such as complement-mediated lysis of tumor cells), or when the two targets are too far apart from each other such that they cannot be bound by a single bispecific antibody, or for simplifying the regulatory approval process. In these situations, it is generally desirable to optimize the production platform for such monospecific antibodies. As shown in Example 10, the present invention provides the recognition that when the lysine (K) at position 392 of the first polypeptide containing a CH3 domain (e.g., having specificity A) is replaced by aspartic acid (D), and when the aspartic acid (D) at position 399 of the first polypeptide containing a CH3 domain is replaced by lysine (K), and when the lysine (K) at position 409 of the first polypeptide containing a CH3 domain is replaced by aspartic acid (D), it becomes possible to generate a mixture of at least two different monospecific immunoglobulin-like molecules in a single cell, including monospecific immunoglobulin-like molecules having AA specificity, wherein the formation of bispecific by-products (bispecific immunoglobulin-like molecules) is reduced to less than 5%, or even less than 3% or even substantially undetectable. Thus, the combination of the above mutations (referred to herein as K392D, D399K, K409D) is particularly preferred for generating mixtures of monospecific immunoglobulin-like molecules. Those skilled in the art will understand that functional variants thereof, namely K392E, D399R, K409E, may result in similar effects. In addition, double mutants (including D399K and K409D substitutions) or other functional variants such as K392D and K409D, D399R and K409E, etc. may also result in similar effects.

[0103] The same applies to combinations of the following mutations, where the glutamic acid (E) at position 356 of the first CH3 domain-containing polypeptide is replaced by lysine (K), and where the glutamic acid (E) at position 357 of the first CH3 domain-containing polypeptide is replaced by lysine (K), and where the lysine (K) at position 439 of the first CH3 domain-containing polypeptide is replaced by aspartic acid (D), and where the lysine (K) at position 370 of the first CH3 domain-containing polypeptide is replaced by aspartic acid (D). The combination of mutations (referred to herein as E356K, E357K, K439D, K370D) is also particularly preferred for generating mixtures of monospecific immunoglobulin-like molecules. Those skilled in the art will understand that their functional variants, namely E356R, E357R, K439E, K370E, may result in similar effects. In addition, triple or double mutants (including E356K and K439D and E357K and K370D substitutions) or other functional variants may also result in similar effects. Accordingly, another embodiment provides a method for generating at least two different monospecific immunoglobulin-like molecules from a single host cell, wherein the two immunoglobulin-like molecules each comprise two CH3 domains capable of forming an interface, the method comprising providing in the cell:

[0104] a) a first nucleic acid molecule encoding a first polypeptide chain of a CH3 domain-containing polypeptide having A specificity,

[0105] b) a second nucleic acid molecule encoding a second polypeptide chain of a CH3 domain-containing polypeptide having B specificity,

[0106] wherein the first CH3 domain-containing polypeptide chain comprises K392D, D399K, K409D mutations and the second CH3 domain-containing polypeptide chain comprises a wild-type CH3 domain or comprises E356K, E357K, K439D, K370D mutations, the method further comprising culturing the host cell and enabling expression of the nucleic acid molecules and obtaining the at least two different immunoglobulin-like molecules from the culture.

[0107] An alternative embodiment provides a method for generating at least two different monospecific immunoglobulin-like molecules from a single host cell, wherein the two immunoglobulin-like molecules each comprise two CH3 domains capable of forming an interface, the method comprising providing in the cell:

[0108] a) a first nucleic acid molecule encoding a first polypeptide chain of a CH3 domain-containing polypeptide having A specificity,

[0109] b) a second nucleic acid molecule encoding a second polypeptide chain of a CH3 domain-containing polypeptide having B specificity,

[0110] Among them, the polypeptide chain of the first CH3 domain-containing polypeptide chain contains a wild-type CH3 domain or contains K392D, D399K, K409D mutations, and the polypeptide chain of the second CH3 domain-containing polypeptide chain contains E356K, E357K, K439D, K370D mutations. The method further includes the steps of culturing the host cell and enabling the nucleic acid molecule to be expressed and harvesting the at least two immunoglobulin-like molecules from the culture.

[0111] As shown in Example 10, two monospecific immunoglobulin-like molecules can be produced in a single cell, in which the formation of bispecific immunoglobulin-like molecules is substantially undetectable. Those skilled in the art can select a third nucleic acid molecule encoding a wild-type or modified polypeptide chain of the CH3 domain-containing polypeptide chain to be provided to the host cell to produce a mixture of three monospecific antibodies, and so on.

[0112] In one aspect of the present invention, there is provided a method according to the present invention for producing at least two different immunoglobulin-like molecules or for producing heterodimeric immunoglobulin-like molecules, wherein the polypeptide chains of the CH3 domain-containing polypeptide chain also each contain variable regions that recognize different target epitopes, and wherein the target epitopes are located on the same molecule. This generally enables more effective counteraction of the (biological) function of the target molecule compared to the case of targeting only one epitope. For example, the heterodimeric immunoglobulin-like molecule can simultaneously bind to two epitopes present on, for example, a growth factor receptor or a soluble molecule that is crucial for the proliferation of tumor cells, thereby effectively blocking multiple independent signal transduction pathways that lead to uncontrolled cell proliferation, and any combination of at least two immunoglobulin-like molecules can simultaneously bind to 2, or even 3 or 4 epitopes present on such a growth factor receptor or soluble molecule.

[0113] In a preferred embodiment, the target molecule is a soluble molecule. In another preferred embodiment, the target molecule is a membrane-bound molecule.

[0114] In another aspect of the invention, there is provided a method according to the invention for generating at least two different immunoglobulin-like molecules or for generating heterodimeric immunoglobulin-like molecules, wherein the polypeptide chains comprising the CH3 domain each further comprise a variable region that recognizes a target epitope, wherein the target epitope is located on different molecules. In this case, the different target molecules can each be a soluble molecule or a membrane-bound molecule. In one embodiment, the different target molecules are both soluble molecules. Alternatively, one target molecule is a soluble molecule and the second target molecule is a membrane-bound molecule. In another alternative, both of the target molecules are membrane-bound molecules. In one embodiment, the different target molecules are expressed on the same cell, while in other embodiments, the different target molecules are expressed on different cells. As a non-limiting example, any heterodimeric immunoglobulin-like molecule or any combination of at least two immunoglobulin-like molecules can be suitable for simultaneously blocking multiple membrane-bound receptors, neutralizing multiple water-soluble molecules (such as cytokines or growth factors directed against tumor cells) or neutralizing different virus serotypes or virus strains.

[0115] A preferred embodiment provides a method according to the invention for generating at least two different immunoglobulin-like molecules or for generating heterodimeric immunoglobulin-like molecules, wherein at least one of the target epitopes is located on a tumor cell. Alternatively or additionally, at least one of the target epitopes is located on the surface of an effector cell. This is suitable for, for example, recruiting T cells or NK cells for tumor cell killing. For example, at least one immunoglobulin-like molecule produced by the method according to the invention, by specifically binding to a target molecule located on an immune effector cell, is capable of recruiting immune effector cells, preferably human immune effector cells. In another embodiment, the immune effector cells are activated once the immunoglobulin-like molecule binds to the target molecule. The recruitment of effector mechanisms can, for example, encompass redirecting immunomodulatory cytotoxicity by administering an immunoglobulin-like molecule produced by the method according to the invention that is capable of binding to a cytotoxicity-triggering molecule (such as a T cell receptor or an Fcγ receptor), thereby activating downstream immune effector pathways. As used herein, the term "immune effector cell" or "effector cell" refers to a cell in the natural cell repertoire of the mammalian immune system that can be activated to affect the viability of a target cell. Immune effector cells include cells of the lymphoid lineage, such as natural killer (NK) cells, T cells (including cytotoxic T cells, or B cells), and cells of the myeloid lineage can also be considered immune effector cells, such as monocytes or macrophages, dendritic cells, and neutrophils. Thus, the effector cells are preferably NK cells, T cells, B cells, monocytes, macrophages, dendritic cells, or neutrophils.

[0116] Target antigens presented by immune effector cells can include CD3, CD16, CD25, CD28, CD64, CD89, NKG2D, and NKp46. Accordingly, there is also provided a method according to the invention for generating at least two different immunoglobulin-like molecules or for generating a heterodimeric immunoglobulin-like molecule, wherein the target epitope is located on a CD3, CD16, CD25, CD28, CD64, CD89, NKG2D, or NKp46 molecule.

[0117] The viability of target cells can include the ability of the cells to survive, proliferate, and / or interact with other cells.

[0118] Accordingly, in one aspect, the invention provides a method according to the invention for generating a heterodimeric immunoglobulin-like molecule, wherein the polypeptide chains containing the CH3 domain each further comprise a variable region that recognizes a target epitope. In one embodiment, the two variable regions of the polypeptide chains containing the CH3 domain each recognize the same target epitope but have different affinities. In another embodiment, the two variable regions of the polypeptide chains containing the CH3 domain each recognize different target epitopes. In another embodiment, the different target epitopes are located on the same target molecule, which can be a membrane-bound molecule or a soluble molecule. In another embodiment, the different target epitopes are located on different target molecules, which can be expressed on the same cell or different cells. Alternatively, the different target molecules can be soluble molecules, or one target molecule can be a soluble molecule while the second target molecule is a membrane-bound molecule. In a preferred embodiment, at least one target molecule of the heterodimeric immunoglobulin-like molecule is located on a tumor cell. In another preferred embodiment, at least one target molecule of the heterodimeric immunoglobulin-like molecule is located on an effector cell (i.e., an NK cell, T cell, B cell, monocyte, macrophage, dendritic cell, or neutrophil, and the target epitope can be located on a CD3, CD16, CD25, CD28, CD64, CD89, NKG2D, or NKp46 molecule).

[0119] In a preferred embodiment, there is provided a method according to the invention for generating at least two different immunoglobulin-like molecules or for generating a heterodimeric immunoglobulin-like molecule, wherein the at least two different immunoglobulin-like molecules are antibodies, most preferably IgG isotype antibodies, and even more preferably IgG1 isotype as described above.

[0120] Also provided are immunoglobulin-like molecules, heterodimeric immunoglobulin-like molecules, or mixtures of at least two immunoglobulin-like molecules obtainable by the method according to the invention. The (heterodimeric) immunoglobulin-like molecule or mixture of immunoglobulin-like molecules preferably comprises at least one CH3 mutation as described in Table B. Thus also provided herein are (heterodimeric) immunoglobulin-like molecules or mixtures of at least two immunoglobulin-like molecules comprising at least one mutation as described in Table B, and pharmaceutical compositions comprising at least one immunoglobulin-like molecule or mixture of at least two immunoglobulin-like molecules according to the invention. In one embodiment, the immunoglobulin-like molecule is a bispecific immunoglobulin-like molecule, such as a bispecific antibody. In another embodiment, the immunoglobulin-like molecule is a monospecific immunoglobulin-like molecule, such as a monospecific antibody. A preferred embodiment provides a mixture of at least two different immunoglobulin-like molecules obtainable by the method according to the invention, wherein the at least two different immunoglobulin-like molecules bind to different epitopes on the same antigen and / or to different epitopes on different antigens. Also provided are heterodimeric immunoglobulin-like molecules obtainable by the method according to the invention, wherein the heterodimeric immunoglobulin-like molecule binds to different epitopes on the same antigen and / or to different epitopes on different antigens. The advantages and preferred uses of such mixtures and antibodies are those described previously herein. The invention also provides a mixture of at least two different immunoglobulin-like molecules obtainable by the method according to the invention, wherein the at least two different immunoglobulin-like molecules comprise at least one heterodimeric immunoglobulin-like molecule. In one embodiment, two of the at least two different immunoglobulin-like molecules are both heterodimeric immunoglobulin-like molecules. Another preferred embodiment provides a heterodimeric antibody comprising two CH3 domains, wherein one of the two CH3 domains comprises the amino acid substitutions L351D and L368E, and wherein the other of the two CH3 domains comprises the amino acid substitutions T366K and L351K. As explained previously, these amino acid substitutions are preferred means for preferentially pairing the two CH3 domains. The amino acid substitutions L351D and L368E in one of the two CH3 domains and the amino acid substitutions T366K and L351K in the other of the two CH3 domains together are referred to as the 'DEKK mutation combination', 'DEKK variant', 'DEKK pair', 'DEKK-engineered CH3 domain', 'DEKK', or an alternative name using reference to DEKK. The CH3 domain carrying the amino acid substitutions L351D and L368E is also referred to as the 'DE side' and the CH3 domain carrying the amino acid substitutions T366K and L351K is also referred to as the 'KK side'.

[0121] There is also provided a pharmaceutical composition comprising an (heterodimeric) immunoglobulin-like molecule obtainable by the method of the present invention or a mixture of at least two immunoglobulin-like molecules. The (heterodimeric) immunoglobulin-like molecule or the at least two immunoglobulin-like molecules according to the present invention are preferably antibodies. The pharmaceutical composition may comprise the (heterodimeric) immunoglobulin-like molecule, a mixture comprising monospecific or bispecific immunoglobulin-like molecules, or a combination of monospecific and bispecific immunoglobulin-like molecules. In addition, the pharmaceutical composition according to the present invention comprises a pharmaceutically acceptable carrier. As used herein, such a "pharmaceutically acceptable carrier" includes any and all physiologically compatible substances such as solvents, salts, dispersion media, coatings, antibacterial and antifungal agents, isotonic agents, and absorption delaying agents, etc. Depending on the route of administration (e.g., intravenous, subcutaneous, intra-articular, etc.), the immunoglobulin-like molecule may be encapsulated within a material to protect the immunoglobulin-like molecule from the action of acids and other natural conditions that may inactivate the immunoglobulin-like molecule. In one aspect, there is provided a pharmaceutical composition comprising a mixture of at least two immunoglobulin-like molecules obtainable by any method according to the present invention, wherein the at least two different immunoglobulin-like molecules are produced by a recombinant host cell according to the present invention. In addition, there is provided a pharmaceutical composition comprising a heterodimeric immunoglobulin-like molecule obtainable by any method according to the present invention, wherein the heterodimeric immunoglobulin-like molecule is produced by a recombinant host cell according to the present invention.

[0122] There are also provided herein nucleic acid molecules encoding polypeptide chains containing CH3 domains with at least one mutation shown in Table B, and recombinant host cells comprising at least one nucleic acid molecule encoding a polypeptide chain containing a CH3 domain with at least one mutation shown in Table B.

[0123] The present invention is further illustrated by the following examples. These examples do not limit the present invention in any way, but are only used to illustrate the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0124] Figure 1 : A) Schematic diagram of the construct vector MV1057. The filled region is the region into which the antibody VH region was cloned. B) Schematic diagram of the phage display vector MV1043.

[0125] Figure 2 : Amino acid sequence of wild-type IgG1 Fc, present in the construct vector MV1057 (EU numbering scheme was applied).

[0126] Figure 3 : Nucleotide and amino acid sequences of the VH region for cloning into various constructs.

[0127] Figure 4 : Mass spectrometry data of transfection A, G, and H.

[0128] Figure 5 : Mass spectrometry data of transfection M and U.

[0129] Figure 6 : Mass spectrometry data of transfection O.

[0130] Figure 7 : Preventing homodimerization by replacing neutral amino acids with charged amino acids.

[0131] Figure 8 : Native MS spectrum (A) of transfected sample ZO(T366K / L351’D) and convoluted MS spectrum of transfected sample ZO(T366K / L351’D). The second / main peak represents the bispecific molecule (B).

[0132] Figure 9 : HADDOCK scores of experimentally verified mutant pairs.

[0133] Figure 10 : Cartoon diagrams of CH3-CH3 interface interactions; A) K409D:K392D / D399’K:E356’K, B) D399K:E356K / D399’K:E356’K, C) K409D:K392D / K409’D:K392’D.

[0134] Figure 11 : HADDOCK scores of various 366 / 351’ charge mutants.

[0135] Figure 12 : Cartoon diagrams of CH3-CH3 interface interactions; A) L351D / L351’D, B) L351D:S354A:R355D / L351’D:S354’A:R355’D.

[0136] Figure 13 : HADDOCK scores of additional charge mutations near position L351.

[0137] Figure 14 : HADDOCK scores of additional charge mutations near position T366 in chain A and position L351 in chain B.

[0138] Figure 15 : Cartoon diagram of CH3-CH3 interface interaction.

[0139] Figure 16 : HADDOCK scores of mutations near T366 / L351.

[0140] Figure 17 : Additional mutated HADDOCK scores near T366 / L351.

[0141] Figure 18 : Example of the nMS spectrum of a bispecific immunoglobulin obtained after co-expression of construct T366K,L351K with construct L351D (left hand figure) or L351D,Y349E (right hand figure), magnifying the single charge state of the intact IgG (half bodies not shown).

[0142] Figure 19A : Native MS results showing the relative abundances of AA, AB, BB, A, and B (total of all species 100%); Figure 19B : Same as above, but now without AB, for better understanding of the undesired species AA, BB, A, and B.

[0143] Figure 20 : Results of thermal stability determination. Squares: wild type; Triangles: charge reversal pair E356K:D399K / K392D:K409D; Circles: mutant CH3 combinations shown in each figure above.

[0144] Figure 21 : Results of the 10× freeze-thaw experiment. 1122 = first parental antibody BB; 1337 = second parental antibody AA; wild type = AA, AB, BB; CR = bispecific of charge reversal pair E356K:D399K / K392D:K409D; 3-6 and 9-12 = bispecific molecules from combinations 3-6 and 9-12 in Table 15.

[0145] Figure 22 : Serum stability results, determined by ELISA using fibrinogen as the coated antigen. A) ELISA data of IgG samples diluted to 0.5 μg / ml; B) ELISA data of IgG samples diluted to 0.05 μg / ml; Results were normalized to the time point at T = 0 days (100%). 1337 = second parental antibody AA; wild type = AA, AB, BB; CR = bispecific of charge reversal pair E356K:D399K / K392D:K409D; 3-6 and 9-12 = bispecific molecules from combinations 3-6 and 9-12 in Table 15.

[0146] Figure 23: nMS results of the ratio experiment with transfection ratios of 1:5 to 5:1. A) The DEKK mutant combination with specific "A" on the DE side and specific "B" on the KK side; B) The DEKK mutant combination with specific "C" on the DE side and specific "B" on the KK side; C) The mutated charge-reversal combination with specific "A" on the E356K:D399K side and specific "B" on the K392D:K409D side.

[0147] Figure 24 : nMS results of transfection #1-11 from Table 20.

[0148] Figure 25 : HADDOCK scores of dimers with different CH3-modified vectors. Gray bars: Desired species AB and CD; black bars: Undesired species AA, BB, CC, DD, AC, BC, AD, BD.

[0149] Figure 26 : SDS-PAGE of transfection #1-11 from Table 20. Control samples DE / KK, DE / DE, and KK / KK are also included.

[0150] Figures 27A - 27C : nMS of transfection #9 (A, B) and transfection #11 (C).

[0151] Figure 28 : nMS of gel filtration samples 1516:1516 (A), 1337:1337 (B), and 1516:1337 (C).

[0152] Figure 29 : Serum levels (pK study) of samples of the DEKK-modified antibody and its two parental antibodies. Detailed Description

[0153] Examples

[0154] Example 1: Amino Acid Substitutions to Generate a Variety of Different CH3 Domains

[0155] Multiple known amino acid substitutions that promote heterodimer formation and multiple alternative amino acid substitutions that have not been previously reported or tested but were selected to promote homodimer formation were introduced into the construct vector (construct vector MV1057; Figure 1 of A) to obtain a wide variety of immunoglobulin-like molecules with different CH3 domains, such that the pairing of CH3 domains (including immunoglobulin-like molecules) was preferentially promoted or inhibited. As Figure 2As shown, the construct vector MV1057 contains a nucleic acid sequence encoding the normal wild-type IgG1 Fc portion. Table 1 lists the amino acid substitutions introduced into this wild-type Fc, resulting in a series of seven constructs. All constructs were generated at Geneart. Constructs 1, 2, and 3, or their alternatives, have been previously described as driving heterodimerization (EP01870459, WO2009 / 089004), as were constructs 6 and 7 (WO98 / 50431). Constructs 4 and 5 are new and were designed to promote homodimerization.

[0156] Table 1

[0157]

[0158] Example 2: Cloning of VH into constructs containing CH3 mutations

[0159] Several antibody VH regions with known specificities and known to be able to pair with the human IGKV1-39 light chain were used for cloning into these constructs. As previously described, all CH3 variants can be used to associate with other antibody domains to generate bispecific or monospecific full-length antibodies. The antibody specificity defined by the VH / VL combination will not affect the heavy chain dimerization behavior driven by the CH3 domain. Throughout the study, a model VH / VL combination was used, where all VLs were based on the germline human IGKV1-39 and the VH was varied. Figure 3 The full sequences and specificities of the antibody VH regions used throughout the study are provided. The MF coding refers to the internal Merus name for different VHs. For example, VH MF1337 is specific for tetanus toxoid, MF1025 is specific for porcine thyroglobulin, and MF1122 is specific for bovine fibrinogen. The VH region present in the phage display vector MV1043 ( Figure 1 of B) was digested with the restriction enzymes Sfil and BstEII (New England Biolabs / catalog number #R0123L and R0162L / according to the manufacturer's instructions) to release the VH fragment from this vector. The vector MV1057 was digested with Sfil and BstEII according to standard methods (according to the manufacturer's instructions). The fragments and the vector were gel-purified (Promega / catalog number #V3125 / according to the manufacturer's instructions) to isolate the cut vector and the VH gene insert. The two were combined by ligation, and then the ligation product was transformed into E. coli DH5α (Invitrogen / catalog number #12297-016 / according to the manufacturer's instructions). After overnight screening, single colonies were picked and the vectors containing the correct insert were identified by sequencing.

[0160] Example 3: Transfection and Expression of Full IgG in HEK293T Cells

[0161] According to standard methods, various plasmids encoding recloned VH variants and also encoding the common light chain huIGKVl-39 were transfected into HEK293T cells such that IgG could be expressed (de Kruif et al. Biotech Bioeng. 2010). After transfection, the expression level of IgG in the supernatant was measured using the ForteBIO Octet-QK system, which is based on Bio-Layer Interferometry (BLI) and enables real-time quantification and kinetic characterization of biomolecular interactions; for details, see www.fortebio.com . When the measured expression level exceeded 5 μg / ml, IgG was purified using protein A affinity purification.

[0162] Example 4: Purification of IgG

[0163] The culture supernatant was purified using a protein A column (GE Healthcare / catalog number #11-0034-95 / according to the manufacturer's instructions), eluted with 0.1 M citrate buffer pH 3.0, and immediately neutralized with an equal volume of 1.0 M Tris-HCL pH 8.0, or directly rebuffered to PBS using a desalting column. Alternatively, IgG could be purified using protein A beads (agarose beads CL-4B, GE healthcare catalog number #170780-01).

[0164] Example 5: Antigen-Specific ELISA’s

[0165] Antigen-specific ELISAs were performed to determine the binding activity to the antigen, and capture ELISAs were performed to demonstrate the binding activity of the bispecific antibody. The complex was detected using a biotinylated secondary antibody (de Kruif et al. BiotechBioeng. 2010).

[0166] Example 6: SDS-PAGE

[0167] According to standard methods, the purified IgG mixture was analyzed by SDS-PAGE ( 4-12% bis-tris gel / Invitrogen / catalog number #NP0323BOX) under reducing and non-reducing conditions, and protein staining in the gel was performed using colloidal blue (PageBlue TM Protein Staining Solution / Fermentas / catalog number #RO571).

[0168] Example 7: Enzymatic Deglycosylation of IgG1

[0169] Due to the heterogeneity in the glycosylation of IgG, the protein was deglycosylated to produce a single product with different masses suitable for mass spectrometry analysis. For every 10 μg of IgG1, one unit of N-glycosidase F (PNGase F; Roche Diagnostics, Mannheim, Germany) was incubated overnight at 37 °C. Buffer exchange was performed using a 10 kDa MWCO centrifugal filter column (Millipore) to remove the initial purification buffer (0.1 M citrate buffer pH 3.0 / 1.0 M Tris-HCL pH 8.0) and re-buffer to PBS. A similar buffer exchange step was performed to remove the separated glycan chains and the buffer was changed to 150 mM ammonium acetate at pH 7.5. The filter was washed with 200 μL of 150 mM ammonium acetate at pH 7.5 for 12 minutes at 4 °C and 11,000 rpm. After washing, 50 μL of deglycosylated IgG was loaded onto the filter and 450 μL of 150 mM ammonium acetate at pH 7.5 was added, followed by another round of centrifugation for 12 minutes at 4 °C and 11,000 rpm. In total, the centrifugation was repeated 5 times, with fresh 150 mM ammonium acetate buffer at pH 7.5 added each time to a total volume of 500 μL. After the last centrifugation step, approximately 25 μL of the remaining buffer-exchanged deglycosylated IgG1 was collected and transferred to a microcentrifuge tube in preparation for mass spectrometry analysis.

[0170] Example 8: Native Mass Spectrometry

[0171] Use mass spectrometry to identify different IgG species in a purified IgG mixture and determine at what ratios these IgG species are present. Briefly, 2 to 3 μl of IgG at a concentration of 1 μM in 150 mM ammonium acetate, pH 7.5, was loaded onto an in-house made gold-coated borosilicate capillary (pulled using a Sutter P-97 puller [Sutter Instruments Co., Novato, CA, USA] and an Edwards Scancoat six sputter-coater [Edwards Laboratories, Milpitas, CA, USA]) for analysis on an LCT 1 mass spectrometer (Waters Corp., Milford, MA, USA), and the mass spectrometer was tuned for optimal performance in high mass detection (Tahallah et al., RCM 2001). A capillary voltage of 1300 V was used, and the sampling cone voltage was 200 V; however, these settings were adjusted when a higher resolution “signal-to-noise ratio” was required. The source backing pressure was increased to promote collision cooling to approximately 7.5 mbar. Proteins at a concentration of 1 μM in 5% formic acid were sprayed to determine IgG1 under denaturing conditions.

[0172] Example 9: Data processing and quantification

[0173] The acquired spectra were processed using MassLynx 4.1 software (Waters Corp., Milford, MA, USA). Minimum smoothing was employed, followed by spectral focusing. The masses of the species were calculated using each charge state in a series. The corresponding intensities for each charge state were assigned and summed through MassLynx. This method enables the relative quantification of all species in a single sample. Alternatively, peak quantification was performed using the area under the curve (AUC) method known in the art. All analyses were repeated three times to calculate the standard deviation of the IgG mass and its relative abundance.

[0174] Example 10: Mixtures of two or three monospecific antibodies from single cells

[0175] A number of antibodies with known specificities and known to be capable of pairing with the human IGKV1-39 light chain ( Figure 3)The paired antibody VH regions were used for recloning into the wild-type construct vector MV1057, or into construct 4 or construct 5 in Table 1, thereby obtaining vectors I to III (Table 2). Subsequently, the resulting vectors I, II, and III were transfected into cells individually to demonstrate the formation of only intact monospecific antibodies, or co-transfected with one or two other construct vectors into cells to obtain a mixture of two or three monospecific antibodies. Vectors I, II, and III each contain nucleic acid sequences encoding a common human light chain and an Ig heavy chain with different CH3 regions and different VH specificities. Table 3 shows the transfection protocols and results.

[0176] Table 2: VH specificities inserted in different constructs

[0177]

[0178] Table 3: Transfection protocols and results

[0179]

[0180] nd = not done.

[0181] It was observed that transfection A, G, and H resulted in the formation of only homodimers, obtaining 100% bivalent monospecific AA, BB, or CC from cells transfected with any one of vectors I, II, or III ( Figure 4 ). Although this was expected and had also been demonstrated previously in transfection A, it is now in fact shown for the first time that homodimerization of CH3-modified Ig heavy chains (transfections G and H) is reported, the heavy chains containing the triple amino acid substitution of construct 4 (i.e., K392D, D399K, K409D) or the quadruple amino acid substitution of construct 5 (i.e., E356K, E357K, K439D, K370D).

[0182] Next, co-expression experiments of two vectors were carried out in single cells. Interestingly, transfection of M and N showed that wild-type and CH3-modified Ig heavy chains could be co-expressed with a common light chain in a single cell, resulting in a mixture of two monospecific antibodies, without the presence of unwanted bispecific antibodies in the mixture, but with as little as 4% to 5% of contaminating "other molecules". "Other molecules" are defined as all molecules that do not have the mass of a complete IgG, including half-molecules composed of a single heavy chain and a light chain pair. Importantly, part of the "other" does not include bispecific products. In transfection M, when equal ratios of vector DNA were transfected, the ratio of AA:BB was close to 1:1. However, transfection of N resulted in a ratio of AA:CC of almost 10:1. Therefore, this transfection (transfection U) was repeated with adjusted ratios of DNA. In fact, a ratio of 1:5 of vector DNA I:III equalized the ratio of AA:CC antibody products in the mixture, tending to approach a ratio of 1:1. Thus, transfection M and U demonstrated that two different substantially pure monospecific antibodies could be expressed in a single cell without unwanted by-products (i.e., without the substantial presence of AC or half-molecules A or C). Figure 5 ). The new CH3 modifications of constructs 4 and 5 are significantly different from wild-type CH3, such that no heterodimerization occurs between wild-type and 4 or wild-type and 5, which is beneficial for the application of large-scale production of monospecific antibody mixtures from single cells.

[0183] Similar to these results, transfection of two different CH3-modified Ig heavy chains (constructs 4 and 5) was also expected to result only in a mixture of two different monospecific antibodies, without the presence of other unwanted species. It was inferred that the CH3 modification of construct 4 was significantly different from the CH3 modification of construct 5, such that no heterodimerization occurred. In this case, co-expression of the CH3-modified heavy chains of constructs 4 and 5 together with the wild-type CH3 heavy chain in a single cell resulted in only 3 monospecific antibodies.

[0184] In fact, it was observed that this was the case, that is, it was found that a mixture of three pure monospecific antibodies could also be obtained by expressing three different Ig heavy chains (designed to form homodimers rather than heterodimers) together with a common light chain in a single cell, and there were no contaminants in the mixture (transfection O). Figure 6 ). As shown in Table 3, using equal ratios of vector DNA used during transfection O, a ratio of AA:BB:CC antibodies of 1:1:1 was not obtained. Transfection using a changed vector DNA ratio (1:1:10, transfection V) demonstrated that the ratio of AA:BB:CC in the mixture could be controlled towards the expected ratio.

[0185] In summary, these experiments show that two or three substantially pure monospecific antibodies can be expressed in a single cell without undesirable by-products, facilitating the large-scale production of mixtures of therapeutic monospecific antibodies.

[0186] Example 11: Mixture of two bispecific antibodies from a single cell

[0187] Given the reported use of CH3-modified heavy chains for the production of a single bispecific antibody elsewhere, this experiment was designed to investigate whether it is feasible to produce a mixture of 2 different bispecific antibodies from a single cell.

[0188] Antibody VH regions with known specificities and known to be capable of pairing with the human IGKV1-39 light chain ( Figure 3 ) were recloned into vectors containing constructs 1 to 3 or 6 to 7 in Table 1, resulting in vectors IV to X (Table 4). Subsequently, vectors IV to X were transfected into cells individually to demonstrate the limitation in the formation of intact monospecific antibodies, or co-transfected with another construct vector into cells to obtain bispecific antibodies or a mixture of two bispecific antibodies, each of vectors IV to X containing a nucleic acid sequence encoding a common human light chain and an Ig heavy chain with different CH3 regions and different VH specificities. Table 5 shows the transfection protocols and results.

[0189] Table 4: VH specificities inserted in different constructs

[0190]

[0191] Table 5:

[0192]

[0193] Previously, it has been demonstrated that CH3-modified Ig heavy chains encoded by constructs 1 and 2 are still capable of forming homodimers when expressed individually in a single cell (WO2009 / 089004). However, WO2009 / 089004 also reported that CH3 domains engineered to contain triple-charge pair mutations (such as the CH3 domain present in construct 3) are no longer able to form homodimers when expressed individually.

[0194] In this study, these findings were only partially confirmed. In fact, in addition to a high proportion of unpaired half-molecules, the results of transfection of B, C, and D also demonstrated the presence of full IgG, thus demonstrating partial homodimerization in the CH3 domains encoded by constructs 1 and 2. In addition to the unpaired half-molecules, transfection of E and F also led to the production of full IgG, thus demonstrating that the triple-charge mutation of construct 3 did not completely restrict homodimerization. It was further demonstrated that the "knot" and "hole" CH3 variants of constructs 6 and 7 also formed homodimers (18% of the homodimers were "knot-knot" and 42% were "hole-hole").

[0195] CH3 variants that preferably completely avoid homodimerization when expressed alone are desired in order to avoid or minimize unwanted by-products (homodimers) when co-expressed with a second CH3 variant for heterodimerization.

[0196] Interestingly, this experiment demonstrated for the first time that a mixture of bispecific antibodies could also be expressed in a single cell with little homodimer in the mixture. Transfection of K and L clearly showed that the expected bispecific species BC + AB were indeed obtained (38% + 47% in transfection K and 16% + 60% in transfection L). In both transfections, a relatively high percentage of unwanted half-molecules was observed (15% half-molecule A + half-molecule C in transfection K and 24% half-molecule A + half-molecule C in transfection L). The relatively high percentage of remaining half-molecules was attributed to a small amount of matching heavy chains in vector IV due to unbalanced heavy chain expression in the matching pairs. Therefore, in transfections S and T, the transfection was repeated with adjusted ratios of vector DNA (2:1:1). This resulted in equal amounts of IgG heavy chains that formed matching pairs and the absence of half-IgG molecules, a pure mixture of bispecific IgG with as little as 3% homodimer BB. Ideally, this low proportion of contaminating monospecific products should be reduced to essentially 0. Therefore, it is desirable to find additional CH3 mutants that can result in a minimum of contaminating monospecific antibodies in a mixture of bispecific antibodies.

[0197] This study demonstrated for the first time that a substantially pure mixture of two bispecific antibodies that recognize three different target epitopes could be produced in a single cell and that there was a minimum of monospecific antibodies in the mixture.

[0198] Example 12: Multiple Mixtures

[0199] As has been demonstrated, it is technically feasible to generate a mixture of two bispecific antibodies that recognize three epitopes from a single cell, or a mixture of two or three monospecific antibodies from a single cell. We next explored the feasibility of controllably generating a variety of other mixtures. The VH region of a fourth antibody, which has a known specificity and is known to be able to pair with the human IGKV1-39 light chain, was used for recloning into vectors containing constructs 1 to 3 or 7 in Table 1, thereby generating vectors I’, ΙI’, ΙII’ or X’ (the ’ indicates different specificities, compared to the corresponding vector numbers). Subsequently, the resulting vectors I’ to ΙII’, X’ and IV to IX were transfected into cells in combination with other construct vectors to obtain a variety of mixtures of bispecific and / or monospecific antibodies, each of the vectors I’ to ΙII’, X’ and IV to IX containing nucleic acid sequences encoding a common human light chain and Ig heavy chains with different CH3 regions and different VH specificities. The various mixtures obtained from a single cell include a mixture of two bispecific antibodies that recognize four epitopes, a mixture of two bispecific antibodies and one monospecific antibody, or a mixture of one bispecific and one monospecific antibody. Table 6 shows the transfection protocols and expected results.

[0200] Table 6

[0201]

[0202]

[0203] Although, in theory, the generation of all mixtures should be feasible, previous work by others has shown that the large-scale production of classical knot-in-hole variants is limited by instability problems. Therefore, it is expected that the mixtures generated by transfection of ZA, ZB, ZL, ZM and ZN will have problems when transferred to larger-scale production.

[0204] Since it has been reported that knot-in-hole variants are unstable and it cannot be excluded that CH3 domains containing “knots” or “holes” form dimers with charged variants or wild-type CH3 domains, the current set of constructs present in Table 1 does not allow the generation of all theoretical mixtures from a single cell on a larger scale. Therefore, it is desirable to design new CH3 variants that are engineered to preferentially form only homodimers or heterodimers and that do not form homodimers or heterodimers with constructs 1 to 5 in Table 1 to enable co-expression in a single cell.

[0205] Example 13: Identification of new charge pair mutants

[0206] The aim of this study was to engineer the IgG CH3 domain such that when different IgG heavy chains are co-expressed in a single cell, only heterodimers or only homodimers are produced, where the newly engineered CH3 domain does not form homo- or heterodimers with known engineered CH3 domains or the wild-type CH3 domain. Thus, as a first step in identifying new engineered CH3 domains that meet the criteria, many of the interfacial contact residues in the IgG CH3 domain were scanned individually or in groups to find replacements that could cause like-chain repulsion through electrostatic interactions, i.e., replacements that reduce homodimer formation. The aim was to obtain a series of such residues that, when replaced with charged residues, cause like-chain repulsion such that when different IgG heavy chains are co-expressed, these mutations can be used to drive homo- and / or heterodimer formation, thereby stabilizing the resulting full-length IgG and producing it in high proportion. Subsequently, by engineering matching pairs of CH3 residues (CH3 region) in one or more IgG heavy chains, the identified replacements were used to generate bispecific antibodies or mixtures of bispecific and / or monospecific antibodies. In addition, newly identified charge mutant pairs can be combined with existing pairs such that multiple nucleic acid molecules encoding different heavy chains all carrying different complementary CH3 mutations can be used for expression in cells, thereby preferentially obtaining mixtures of only monospecific antibodies or only bispecific antibodies, or defined mixtures of monospecific and bispecific antibodies. The residues to be tested in this study were previously identified contact residues (Deisenhofer J., 1981; Miller S., 1990; Padlan, 1996, Gunasekaran, 2010). The basic principle of the method was to engineer repulsive charges into each available pair of contact residues. The samples were then analyzed on non-reducing SDS-PAGE to identify pairs that reduce dimer formation, such as the appearance of a band of approximately 72 kD. According to this method, since the repulsive electrostatic interaction between a non-matching pair may or may not be sufficient to produce a sufficient amount of half-molecules for detection, all available pairs were screened as single mutations or in combination with a single other mutation, and the mutations were combined.

[0207] According to Table 7, amino acid replacements were introduced into the construct vector MV1057 by Geneart, and the constructs were expressed by transfection into HEK293T cells according to standard methods. IgG expression levels were measured in Octet. When production could not be achieved twice, the mutation was considered detrimental to expression and the mutation was not further investigated.

[0208] Table 7: List of amino acid replacements in the various constructs prepared (EU numbering)

[0209]

[0210]

[0211]

[0212] The supernatant containing >5 μg / ml IgG was analyzed by SDS-PAGE, and IgG was purified using Protein A. The proteins were stained with Coomassie Blue. The homodimer was visible as a band at approximately 150 kDa. The smaller band at approximately 75 kDa indicated the presence of half-molecules (see negative controls: K392D and K409D). The blot is shown in Figure 7 in.

[0213] As shown in the right column of Table 7, the SDS-PAGE gel results were analyzed and scored. Many residues were considered promising for further testing in combination, including residues Q347, S354, Y349, L351, K360, T366, T394, and V397. This selection was based on high scores in inhibiting homodimer formation and the availability of contacting residues that could be modified without causing problems, e.g., other non-complementary charges. For example, residues F405 and Y407 are known to have multiple interactions at the CH3-CH3 interface, including interactions with charged residues, and after introducing multiple charge mutations in these interacting residues, the interactions may be problematic (see Table A). New constructs were generated in vector MV1057 (Table 8), and the VH regions of antibodies with known specificities and known to be able to pair with the human IGKV1-39 light chain were used for recloning into vectors containing these new constructs (see Table 9) such that the combinations could be further tested. Table 10 shows the transfection protocols and results.

[0214] Table 8:

[0215]

[0216] Table 9: VH specificities inserted in different constructs

[0217]

[0218] Table 10:

[0219]

[0220] CH3 variant combinations were expressed and analyzed by SDS-PAGE (data not shown) and native mass spectrometry (MS). The results are summarized in Table 10. ZO transfection resulted in the highest proportion of heterodimers (69% AC) in the mixture. Interestingly, in ZO transfection, no AA homodimers were present, while a small proportion of CC homodimers (7%) were included. Mass spectrometry analysis revealed that the remaining proteins in the mixture consisted of half-A molecules, which may have originated from unequal expression of A and C heavy chains. The original MS data from the transfected sample ZO is shown in Figure 8 in.

[0221] Surprisingly, although transfection with ZO led to a substantial amount of bispecific product, transfection with the reverse-charged pair of ZP (the L351K / T366’D pair of ZO versus T366K / L351’D) did not result in a similar outcome, with only 52% of the bispecific product observed, with substantial amounts of two homodimers (30% AA and 13% CC). The explanation for this is that the negatively charged D is structurally close to T, and thus, T366D may not be sufficient to effectively repel itself, and thus, T366D still forms homodimers, as actually observed.

[0222] It is conceivable that minor variants of the newly discovered T366K / L351’D pair (e.g., by testing all substitutions, including the newly constructed T366R and L351E) may lead to a similar percentage of BsAb.

[0223] Example 14: HADDOCK for Designing New CH3 Mutants to Drive Effective Heterodimerization

[0224] As described in Example 13, the newly discovered charge pair T366K / L351’D increased the proportion of heterodimers in the mixture (69%), with a small fraction being the undesired CC homodimer (7%) (L351D / L351’D), and having a substantial fraction of semi-A molecules (24%) “contaminating” the mixture. In this example, an in silico method was used to further deepen the understanding of the amino acid residues involved in CH3 interface interactions, thereby testing complementary substitutions in the relative CH3 regions and discovering new CH3 pairs containing complementary substitutions, which further improve effective heterodimerization while avoiding the effective formation of homodimers of the two heavy chains.

[0225] HADDOCK (High Ambiguity Driven protein-protein DOCKing) is an information-driven flexible docking method for generating models of biomolecular complexes. The difference between HADDOCK itself and ab initio docking methods lies in the fact that HADDOCK encodes information from the identified or predicted protein interfaces from ambiguous interaction restraints (AIRs) to drive the docking process (de Vries et al., 2010).

[0226] The input to the HADDOCK web server consists of protein structure files, which can be crystal structures, NMR structure ensembles, or modeled structures. After docking or refinement, HADDOCK returns a so-called HADDOCK score, which is a weighted average of the VanderWaals energy, electrostatic energy, buried surface area energy, and desolvation energy. Although it is often difficult to obtain a direct translation of experimental data, the HADDOCK score can be interpreted as representing the binding energy or affinity. In addition, HADDOCK provides structure files for the "top four" structures resulting from a docking run. These structure files can be downloaded and visualized, enabling a detailed analysis of the interactions of individual residues.

[0227] In this example, the interaction between the CH3 domains of the IgG1 heavy chain was studied. The high-resolution crystal structure of the Fc portion of IgG (structure 1L6X) was used as the starting structure. (http: / / www.resb.org / pdb / explore.do?structureId=116x; Idusogie, E.E. et al., J. I. 2000 (164) 4178 - 4184).

[0228] In Example 13, it was found that co-transfection of vectors XIII and XVI led to the formation of a CC homodimer contaminant (Table 10). HADDOCK was used to search for additional mutations of the T366K / L351’D pair that avoid homodimerization.

[0229] The output of HADDOCK consists of a group of calculated energies, the HADDOCK score (a weighted average of the energies), and four structure files corresponding to the four lowest energy structures found by the program. The HADDOCK score is used to compare different structures; the other energies are only used to obtain an indication of what is happening in the structure (e.g., good electrostatic interactions, smaller buried surface, higher VanderWaals energy). The lower the HADDOCK score, the better. For each mutant pair, the scores for the AA, AB, and BB dimers were calculated.

[0230] The group of mutant pairs from Example 12 was analyzed in HADDOCK to see if the calculated energies were correlated with the experimental data. Table 11 shows all the theoretical energies visible in Figure 9 the [text not provided].

[0231] Table 11:

[0232]

[0233] * Due to the high VanderWaals energy score, this value is generally high and may be attributed to the steric clash of T366W / T366’W.

[0234] For the two wild-type CH3 domains, since the A and B CH3 regions are identical, the HADDOCK scores for AA, AB, and BB are the same. In most other cases, the AB pair has the lowest score, as expected. For the T366K / L351D pair, the BB score is slightly better than the AB score (-210.6 vs. -212.5), however, the difference is within the computational error. The heterodimer structures of these pairs are visible using HADDOCK. For example, construct combinations 1-2, 1-1, and 2-2 are shown in Figure 10 It is clearly visible from these visualizations that salt bridges are formed in the heterodimer ( Figure 10 left-hand A figure), while there is electrostatic repulsion between residues of the same chain ( Figure 10 B and C, middle and right-hand figures). Thus, the higher HADDOCK scores for homodimers can be explained by the electrostatic repulsion of the interfacial residues of the mutants. These residues bend away from each other and do not interact with residues on the other chain, resulting in a decrease in affinity.

[0235] Table 11 and Figure 9 confirm the results observed in Example 13. The T366K / L351’D AC heterodimer and the L351D / L351’D CC homodimer are formed with similar energies, explaining the presence of heterodimers and homodimers in the mixture. On the other hand, although the T366K semi-A molecules are present, the T366K / T366’K AA homodimer is barely detectable in the mixture. Table 11 and Figure 9 actually show that the HADDOCK score of the T366K / T366’K AA homodimer is higher than that of the AC heterodimer, so the formation of this homodimer is energetically unfavorable.

[0236] Example 15: 366 / 351 Variation

[0237] In Example 13, it was hypothesized that alternative T366K / L351’D mutant charge pairs could be designed, which could have similar results in terms of the percentage of bispecific antibodies in the mixture. Alternatives could include replacing T366R, T366D, T366E, L351E, L351K, and L351R. The proportion of the L351D / L351’D CC homodimer can be reduced by generating variants of the 366 / 351 pair. In HADDOCK, all possible mutant pairs were analyzed, and the resulting scores are shown in Table 12 and visualized in Figure 11 .

[0238] Table 12

[0239]

[0240] When looking at the HADDOCK scores, it was observed that some of the mutations had a similar "pattern" when compared to T366K / L351'D. For most substitutions, the AA homodimer was found to have a higher HADDOCK score than the AB heterodimer, but the BB homodimer seemed to be as favorable as the AB heterodimer. Although residue 351 is known to be the "neighbor" of itself on the other chain, i.e., at the CH3-CH3 interface, residue 351 of chain A pairs with residue 351 of chain B, when forming the BB dimer, the negative effect of this same charge is almost absent. Looking at the L351D / L351'D structure, the above is explained by aspartates that bend away from each other, the stabilizing effect of arginine that is naturally present at least at position 355, and the same stabilizing effect of the negative charge of serine that is naturally present at position 354 (see Figure 12 A). Mutations of these residues (S354A and R355D) only provided a small improvement. As is clearly visible from Figure 12 B, the backbone-hydrogen of A354 leads to the stabilization of the homodimer. For the bispecific molecules, the T366R / L351'E pair with the lowest HADDOCK score from this series seems to be the most favorable.

[0241] Example 16: Mutations near T366K / L351'D

[0242] In the series of HADDOCK analyses in this example, the T366K / L351'D or T366K / L351'E pair was used as the starting structure. Additional mutations on chain B were used to calculate the HADDOCK scores and energies to identify additional mutations that could further increase the predicted percentage of bispecificity of these chains A and B. When studying the structure of the CH3 domain using an observer for visualizing protein structures at the molecular level (YASARA, www.yasara.org), the distances between residues can be calculated. While doing so, it was observed that two residues, Y349 and L368, are neighboring residues that have positive or negative effects on dimer interactions, and in this example, in addition to the L351D mutation, these residues were also mutated to study the results of dimer formation regarding homo- and heterodimers (see Figure 13 ). Both residues seem to increase the stability of the heterodimer (lower HADDOCK score), while increasing the instability of the BB dimer (higher HADDOCK score). Glutamate (E) at positions 349 and 368 seems to be more favorable than aspartate (D). Therefore, introducing a second amino acid substitution in chain B that already contains an amino acid substitution at position 351 seems to be more favorable for heterodimerization.

[0243] In the next set of HADDOCK analyses, the T366K / L351’D pair was used again as the starting structure. In addition to the replacements in chain B that further increase heterodimerization (i.e., Y349D / E and L368E), additional mutations were added to chain A that already contained the T366K replacement. As Figure 14 shown, there are several mutant pairs that seem to favor the formation of bispecific heterodimers. In the T366K-L351K / L351’D-Y349’D pair, all four mutant residues are involved in heterodimer pairing, but this is not the case for T366K-L351K / L351’E-L368’E, where K351 is not directly involved in binding. However, the HADDOCK score for this latter heterodimer was -228.9, significantly lower than -214.2 for T366K / L351’E-L368’E, which can be explained by the hydrogen-bonding interactions of K at position 351 (see Figure 15 ). The T366K-L351K / L351’D-Y349’D pair could be further improved by the R355’D mutation in chain B, which led to a higher BB-HADDOCK score but also a slightly higher AB HADDOCK score. Overall, compared to the single T366K mutation in chain A, the additional L351K led to a lower AB score and similar AA and BB scores. In theory, this would result in a higher amount of bispecific heterodimers in the sample.

[0244] It is Figure 11 apparent that R at position 366 drives heterodimerization more effectively than K. Therefore, some of the HADDOCK analyses shown in Figure 13 were repeated, but now using T366R instead of T366K in chain A. It was demonstrated that combining R366 in chain A with the double mutations in chain B was unfavorable ( Figure 16 ). This may be attributed to the larger size of this residue, which interferes with other interfacial interactions even though all the expected salt bridges with R366 are present in the structure. Similarly, for R366, the HADDOCK score for the AA homodimer was lower than that for K366, which also had an adverse effect on the formation of heterodimers. Therefore, R366 in the interface was not used for further HADDOCK analyses.

[0245] A total of 14 best-performing pairs were selected based on HADDOC predictions (see Table 13 and Figure 17 ). In some pairs, the R355D replacement was included to eliminate the stabilizing effect of the naturally occurring R355 on the L351 / L351’D interaction.

[0246] Table 13:

[0247]

[0248] Example 17: Using the CH3 mutants predicted by HADDOCK, bispecific in vitro expression

[0249] The analysis in Example 16 showed that some CH3 variants with additional mutations around the T366K / L351’D pair produced mixtures with a higher proportion of bispecific components and potentially a lower proportion of homodimer components. These best-performing pairs were selected for production and further analysis. In addition, constructs T366R and L351E were also generated. Table 14 lists the constructs generated, which were used for recloning the VH regions of antibodies known to be specific and known to be able to pair with the human IGKV1-39 light chain. The expression of IgG containing each construct was previously reported in Example 13, and this expression was repeated with the constructs listed in Table 14. The aim was to evaluate which of the constructs could homodimerize in the absence of a matching heterodimerization partner. Ideally, a high percentage of half-molecules and a low percentage of homodimers would form. As a control, constructs containing previously reported charge mutations and constructs containing previously reported knobs-into-holes mutations were also used for recombinant cell expression as full-length IgG. The protein A-purified supernatant was analyzed by SDS-PAGE, and the results were analyzed and scored as shown in Table 14.

[0250] Table 14:

[0251]

[0252] The co-expression results of the common light chain with two different heavy chains carrying the amino acid substitutions shown in Table 14 or a heavy chain carrying the amino acid substitutions of the previous construct are shown in Table 15. The expression of two different heavy chains containing the amino acid substitutions T366K and L351’D:L368’E respectively resulted in approximately 87% bispecific AB heterodimers in the mixture, where no AA or BB homodimers were present (Combination nr. 3 in Table 15). Approximately 12% of the half-molecules (Half A) containing the T366K substitution were observed. In addition, when an additional amino acid substitution L351K was introduced into the first heavy chain, an increase in the percentage of bispecific AB heterodimers was found. For example, the co-expression of two different heavy chains containing the amino acid substitutions T366K:L351K and L351’D:L368’E respectively resulted in approximately 92% bispecific AB heterodimers, while AA and BB homodimers were essentially absent in the mixture (Combination nr. 12 in Table 15). Combinations 10 and 11 also resulted in a favorable distribution, with a high percentage of heterodimers and almost no homodimers. The absence of homodimers is advantageous because the portion containing the intact IgG molecule consists only of AB heterodimers. For purification and subsequent therapeutic applications, the half-molecules can be removed by standard methods (e.g., size exclusion chromatography). Therefore, in the production process of bispecific antibodies, the application of these newly identified charge mutants offers more advantages than the known charge mutants and knobs-into-holes mutants that do not exclude the presence of "contaminating" homodimeric antibodies. In addition, compared with the previously described E356K:D399K / K392’D:K409’D and E356K:D399K / K392’D:K409’D:K439’D charge reversal pairs, the additional advantage of the T366K / L351’D:L368’E and T366K:L351K / L351’D:L368’E charge pairs is that the previously described charge variants are based on the reversal of the existing charges within the CH3-CH3 interface, while the newly identified charge variants add additional charge pairs (charge-charge interactions) to the CH3-CH3 interface. Introducing additional charge pairs into the CH3-CH3 interface can further increase the stability of the interface and thereby increase the stability of the intact antibody. The same applies to the mutations used in Combinations nr. 4, 5, 6, 9, 10, and 11, which also result in a favorable proportion of bispecific heterodimers in the mixture, where the proportion of AA and BB homodimers is extremely low.

[0253] Table 15:

[0254]

[0255] *Chain A carries the specificity of MF1337 (=tetanus toxoid); **Chain B carries the idiotype of MF1122 (=fibrinogen)

[0256] Native MS

[0257] Native MS was performed on all bispecific samples. The obtained profiles were analyzed in the following two ways to determine the relative proportions of the species present: by peak height and by peak area. Peak area is the more scientifically correct way of analysis, but since all previous analyses in other studies were based on peak height, both methods were included in the analysis for comparison purposes. The difference between the methods was within the measurement error, and therefore, only the peak area values were used for further measurements. Two representative spectra are shown in Figure 18 . The results summary is shown graphically in Figure 19, and the values can be found in Table 15. In approximately half of the samples, the total contaminants of monospecific IgG were below 5%, and in only 3 cases was this value >10%, while for wt IgG, approximately 50% monospecific IgG was expected to be found in the mixture.

[0258] Ten groups of 2 different heavy chain combinations were selected from Table 15 for further analysis. These ten combinations included combinations 1, 2, 3, 4, 5, 6, 9, 10, 11, and 12 (Table 15). The selection of these ten was based on the determination by nMS of a low percentage of homodimers present in the mixture, but also on their overall physicochemical properties, including yield, SDS-PAGE, and the number of mutations present in the CH3 domain.

[0259] Example 18: IgG Stability Analysis

[0260] In this study, a series of CH3 mutant pairs could be further analyzed for the stability of the Fc portion of the IgG molecule, which resulted in a high proportion of bispecific heterodimers and a very low amount (<5%) of parental IgG in the intact IgG fraction. The CH3 domains of the mutations used to promote heavy chain heterodimerization could have an unexpected destabilizing effect on the Fc region of IgG, which could lead to undesirable properties such as reduced in vivo half-life, reduced effector function, and / or increased immunogenicity. The newly identified charge pairs were compared with wild-type bispecifics and bispecifics containing previously identified charge mutations (chain A contains construct 1, chain B contains construct 2). In this study, all bispecific molecules contained the same heavy and light chain variable regions, ensuring that the observed effects were caused by mutations in the Fc portion of the molecule rather than by changes in the variable regions.

[0261] A series of stability studies were performed on these bispecifics. These studies included spectroscopic analysis (UV-Vis absorption, fluorescence, and light scattering) and microscopic analysis (optical and fluorescence microscopy stained with Nile red), thus providing information on the aggregation state of the CH3 variants.

[0262] The UV-Vis absorption spectra were recorded at 25 °C using a Cary 300 Bio spectrophotometer with a double beam and two monochromators. The spectra from 250 nm to 400 nm were monitored using a 1 cm path length. Absorbance at 320 nm and longer wavelengths provides information on the aggregation state of IgG.

[0263] The intrinsic fluorescence spectra were monitored at 25 °C using a FluoroMax spectrofluorometer. The fluorescence method was optimized. Fluorescence emission provides information on conformational and aggregation properties. Using a FluoroMax spectrofluorometer, synchronous scanning with an integration time of 0.01 s was performed between 400 nm and 750 nm (λ em = λ eχ ) to monitor the 90° light scattering spectra at 25 °C. The excitation and emission slits were optimized. For example, right-angle light scattering can distinguish IgG samples without and with 5% dimers.

[0264] For fluorescence microscopy stained with Nile Red, an ethanol solution of Nile Red was added to the sample just before measurement. The sample was filled in a microscope slide and analyzed by fluorescence microscopy. The particles were counted. The lower limit of the size of the particles observable by fluorescence microscopy is about 0.5 μm.

[0265] Applying stress (such as temperature, pH, mechanical stress, or denaturants) to proteins can cause conformational changes (such as unfolding) and / or aggregation. As previously reported, charge-engineered bispecific antibodies lower the melting temperature of the modified CH3 (Gunasekaran 2010). These studies aim to distinguish the new charge mutants of the present invention from the existing known charge mutants.

[0266] Using a Protein A biosensor and by exploiting FcRn that binds to IgG, thermal stability studies were explored using Octet. Using a PCR machine, samples were incubated at 4, 50, 55, 60, 65, 70, and 75 °C for 1 hour at a concentration of 100 μg / ml (in PBS) to detect the thermal stability of CH3-modified IgG. Next, the samples were slowly cooled to 25 °C over 15 minutes and held at that temperature for 2 hours, and then stored overnight at 4 °C. The precipitated antibody was removed by centrifugation, and subsequently, the total IgG concentration of the soluble antibody was determined by Octet using a Protein A biosensor (diluted 1 / 10 in PBS). Using Octet, an assay for measuring the binding of CH3-modified IgG to FcRn was explored. A Protein L biosensor was used to bind the light chain of IgG to the sensor and then incubated with FcRn in solution, or an anti-penta-histidine biosensor was used to bind the histidine-tagged FcRn protein and then incubated with the target IgG. These methods can be more sensitive than using a Protein A biosensor and can also be used for thermal stability studies. In addition, the serum stability of all samples was analyzed. Briefly, (modified) IgG samples were incubated in human serum at 37 °C, and control samples were kept at 4 °C. After 1, 2, 3, and 4 weeks, the samples were centrifuged to remove precipitated IgG. Subsequently, the samples were titrated in an antigen-specific ELISA to determine the relative amount of functional IgG. Purified control antibodies freshly incorporated into human serum were used as a reference.

[0267] Example 19: Stability Analysis

[0268] In previous experiments, a high percentage of bispecific antibodies was obtained by co-expression of two different heavy chains containing CH3 mutations and a common light chain (Example 17).

[0269] Eight groups of two different heavy chain combinations were selected from Table 15 for further analysis. These eight combinations included combinations 3, 4, 5, 6, 9, 10, 11, and 12 (Table 15). In this study, these eight combinations were analyzed with a focus on the stability of the Fc portion of IgG. As controls, wild-type bispecifics (i.e., without CH3 mutations) and / or bispecifics based on previously reported CH3 charge mutations were included. Note that for wild-type bispecifics, the two heavy chains and the common light chain were co-expressed without using means for preferential heterodimerization. Thus, these "wild-type bispecifics" represent a mixture of AA, AB, and BB. In this study, all bispecifics were designed to carry the same VH / VL combination, ensuring that the observed effects were due to mutations in the Fc portion of the molecule rather than changes in the Fab portion.

[0270] Suppose that mutations pairs used to promote the heterodimeric pairing of two different heavy chains may be associated with unexpected structural or other destabilizing effects on the Fc region of IgG. Due to the presence of these mutations, this will subsequently lead to undesired problems, further restricting clinical development, for example, reducing the in vivo half-life, reducing effector function, and / or increasing immunogenicity.

[0271] Thermal stability

[0272] Applying stress, such as increasing or decreasing the temperature, can lead to conformational changes (e.g., unfolding) and / or protein aggregation. Using a PCR machine, bispecific molecules from combinations 3 to 6 and 9 to 12 (Table 15), wild-type bispecific molecules, and bispecific molecules obtained when using constructs 1 and 2 (E356K:D399K / K392D’:K409D’ combination, also known as the “charge reversal” pair) were incubated at 4, 60, 62.5, 65, 67.5, 70, and 72.5 °C at a concentration of 100 μg / ml (in PBS) for 1 hour to determine the thermal stability of CH3-modified IgG. Next, the samples were slowly cooled to 25 °C over a period of 15 minutes and held at this temperature for 2 hours, after which they were stored overnight at 4 °C. The next day, the precipitated antibodies were removed by centrifugation (18,000 rpm, 4 °C, 20 minutes), and subsequently, the total IgG concentration of the soluble antibodies was determined by Octet using a Protein A biosensor (diluted 1 / 10 in PBS). The results are shown in Figure 20 In comparison with the wild-type bispecific antibody (squares), a decrease in thermal stability was observed for the control CH3-modified bispecific antibody (charge reversal E356K:D399K / K392D’:K409D’ combination (triangles)). Bispecific molecules from combinations 3 to 6 and 9 to 12 (diamonds) also showed a decrease in thermal stability compared to the wild type. However, compared to the control CH3-modified bispecific antibody, three combinations showed a significant improvement in stability. The bispecifics of combinations 9, 10, and 11 were significantly more stable than other CH3-modified (charge reversal) bispecifics and as stable as the wild-type bispecific measured at the highest temperature.

[0273] Freeze-thaw stability

[0274] To determine the stability of CH3-modified IgG molecules upon repeated freezing and thawing, bispecific molecules from combinations 3 to 6 and 9 to 12 (Table 15), as well as wild-type bispecific molecules and bispecific molecules obtained using constructs 1 and 2 (E356K:D399K / K392D’:K409D’ combination, also known as the charge-reversal pair), were exposed to ten subsequent freeze-thaw cycles, which were conducted as follows: The samples were placed at -80 °C for at least 15 minutes until they were completely frozen. Subsequently, the samples were thawed at room temperature. When the samples were completely thawed, the freeze-thaw cycle was repeated. After 10 freeze-thaw cycles, the precipitated antibodies were removed by centrifugation (18,000 rpm, 4 °C, 20 minutes), and thereafter, the total IgG concentration of the soluble antibodies was determined by Octet using a Protein A biosensor (diluted 1 / 10 in PBS). The freeze-thaw stability test was repeated three times. The results are shown in Figure . It was observed that the stability of the control charge-reversed CH3-modified bispecific antibodies showed a slight decrease compared to the wild-type bispecific. In contrast, the stability of the bispecific molecules from combinations 3, 4, and 9 showed a slight improvement compared to the wild-type bispecific molecules. In summary, it can be concluded that for the CH3-modified variants, the stringent conditions of the freeze-thaw cycles do not cause significant stability problems.

[0275] In vitro serum stability

[0276] The bispecific molecules from combinations 3 to 6 and 9 to 12 (Table 15), as well as wild-type bispecific molecules and charge-reversed bispecific molecules, were incubated in 10% human serum at 37 °C to determine the stability of CH3-modified IgG in serum maintained at 37 °C. Control samples were maintained in human serum at 4 °C. After 1, 2, or 5 days, the precipitated antibodies were removed by centrifugation. Subsequently, the samples were titrated in a fibrinogen-specific ELISA to determine the relative amount of functional IgG. Purified control antibodies freshly incorporated into human serum were used as a reference.

[0277] The fibrinogen ELISA data showed that all samples were present quite stably in 10% human serum at 37 °C for 5 days. At lower IgG concentrations, the stability of the bispecific molecules from combinations 4 and 5 seemed to be slightly worse, especially at T = 1 and T - 2, but at the end point of this experiment, there were only minimal differences (see ​ ).

[0278] Example 20: Other stability tests

[0279] Another set of analytical methods was used to evaluate the stability of variant IgG. Bispecific molecules from combinations 3 to 6 and 9 to 12 (Table 15), as well as wild-type bispecifics (AA, AB, and BB), each parental antibody (AA and BB), and bispecific molecules obtained when using constructs 1 and 2 (E356K:D399K / K392D’:K409D’ combination (charge-reversal pair)) were used as samples in these stability assays. All IgG were diluted to 0.2 mg / ml and several stress conditions were applied (2 days at 50 °C, 2 weeks at 40 °C, 5× freeze-thaw) with the aim of being able to distinguish between different samples. It was noted that these high stress levels led to a situation where one of the parental antibodies used in all bispecifics (BB parental, carrying two 1122 Fabs) became unstable. At 2 days at 50 °C, aggregation of the protein was detected by UV absorption. This indicates that these stress conditions cannot distinguish between the instability of Fab and CH3 in the bispecifics, and the data obtained from the 50 °C incubation should be used with caution.

[0280] The results are summarized in Table 16. The analytical methods used included:

[0281] - Fluorescence microscopy using Nile red (“Nile red particles” in Table 16); to observe the amount of particles >0.5 μm after addition of the Nile red dye.

[0282] - UV spectroscopy at 350 nm (“UV 350 nm”); changes in absorption at wavelengths >320 nm give information about the protein aggregation state.

[0283] - 90° light scattering at 400 nm (“LS 400 nm”); a sensitive technique to observe changes in protein aggregation (e.g., differences between monomers and IgG dimers).

[0284] - Intrinsic fluorescence; the maximum fluorescence wavelength and intensity of aromatic residues in a protein change when the environment changes (e.g., upon unfolding).

[0285] - 1,8-ANS fluorescence spectroscopy; 1,8-ANS binds to cationic groups through electrostatic interactions to form ion pairs and can detect changes in protein structure and / or conformation.

[0286] UV-VIS spectroscopy

[0287] The UV-Vis absorption spectra were measured at 25 °C using a double-beam, two-monochromator Cary 300Bio spectrophotometer from Varian in different quartz cuvettes (e.g., black low-volume Hellma cuvettes with a path length of 1.0 cm and transparent Hellma cuvettes of 0.2 cm × 1.0 cm). Using a path length of 1.0 cm, the spectra between 220 nm and 450 nm were monitored. Absorption around 280 nm provides information on protein concentration. The region between 320 nm and 450 nm can provide information on the aggregation state of the sample.

[0288] 90° light scattering

[0289] 90° light scattering spectroscopy was developed to study protein aggregation and the method was carried out as described in Capelle, 2005; Demeule, 2007a. Using a FluoroMax spectrofluorometer (Spex, Instruments S.A., Inc. U.K.), the 90° light scattering spectra were monitored at 25 °C by running a synchronous scan with an integration time of 0.01 s between 400 nm and 750 nm (λ em = λ eχ ). Different slit settings were tested to find the optimal conditions. After optimization, the same slit settings were used for all measurements.

[0290] Steady-state fluorescence emission

[0291] Fluorescence emission of tryptophan, tyrosine, and phenylalanine residues gives information on the local environment of these fluorophores. Changes or differences in hydrophobicity and / or rigidity are determined. Generally, more hydrophobic and rigid environments result in an increase in fluorescence intensity and a blue shift of the emission maximum. Intrinsic fluorescence spectroscopy can provide information on the current state of the protein and monitor changes in physical and chemical properties. More information on tyrosine and tryptophan fluorescence can be found in Lakowicz's book [Lakowicz, 2006].

[0292] The fluorescence emission and excitation spectra in different quartz cuvettes were recorded at 25 °C. The samples were excited at different wavelengths. The integration time and slit settings were optimized. After optimization, the same integration time and slit settings were applied to all samples.

[0293] Fluorescence microscopy stained with Nile red

[0294] The Nile red staining method was developed to visualize protein aggregation and the method was carried out as described in Demeule et al., 2007b.

[0295] Microscopic observations were carried out on a Leica DM RXE microscope (Leica Microsystems GmbH, Wetzlar, Germany) equipped with a mercury lamp. Images were acquired using a Sony NEX-5 camera and its firmware. The objective lenses were 10×, 20×, and 40×. For microscopic examination, a fixed distance of 0.1 mm was used between the slide and the coverslip. The size of the 4x4 grid was 1 mm × 1 mm, corresponding to 0.1 μL.

[0296] 1,8-ANS fluorescence spectroscopy

[0297] 1-Anilinonaphthalene-8-sulfonic acid (1,8-ANS) is an uncharged small hydrophobic fluorescent probe (Mw 299.34 Da) used for studying membrane surfaces and proteins.

[0298] 1,8-ANS is essentially non-fluorescent in water and becomes significantly fluorescent only when bound to a membrane (quantum yield -0.25) or a protein (quantum yield -0.7). This property of 1,8-ANS makes it a sensitive indicator of protein folding, conformational changes, and other processes where the modified probe is exposed to water. References on 1,8-ANS can be found on the ​ Internet homepage of Molecular Probes.

[0299] The fluorescence emission spectra of 1,8-ANS were recorded using a FluoroMax spectrometer. A direct comparison of 1,8-ANS fluorescence between IgGs was not performed. Each IgG can have a different number of 1,8-ANS binding sites, so comparison is not possible. In principle, the lower the fluorescence, the fewer 1,8-ANS molecules are bound to the antibody. Changes in 1,8-ANS fluorescence intensity and emission wavelength caused by stress were evaluated.

[0300] Table 16: Overview of different forced degradation results on each IgG sample after dilution to 0.2 mg / ml. The color of the cells indicates the change between T = 0 and after applying stress: dark gray = large change, light gray = small change, colorless = no change (= stable).

[0301] * "Combination #" refers to the combination of mutations listed in Table 15; ** very small particles by fluorescence microscopy, the relevance of which is unknown; 2d 4°C = 2 days at 4°C; 2d 50°C = 2 days at 50°C; 2w 4°C = 2 weeks at 4°C; 2w 40°C = 2 weeks at 40°C; T0 = start of the experiment; 5FT = 5 freeze-thaw cycles

[0302]

[0303]

[0304] In summary, these data indicate that the IgG samples are significantly stable. Several harsh stress conditions (e.g., 2 days at 50 °C) are required to generate a measurable difference among the test samples. Under these conditions, the samples of Combinations #9 and #10 appear to aggregate more than the other samples.

[0305] For the stability among proteins, the most differentiating factors are freeze-thaw cycles and temperature rise. Considering the very stringent stress factor of incubation at 50 °C, the T366K / L351E, Y349E (Combination #4) and T366K, L351K / L351D, Y349E (Combination #11) variants are the two most stable proteins within the group, followed by T366K, L351K / L351D, Y349D (Combination #10) and T366K, L351K / L351D, L368E (Combination #12).

[0306] Example 21: Nature MS of ratio experiment; transfection ratios are from 1:5 to 5:1

[0307] More detailed ratio experiments were conducted to learn more about the CH3-mutated IgG, particularly the behavior of the T366K:L351K / L351D’:L368E’ combination (hereinafter referred to as KK / DE or DEKK) in skewed transfection mixtures.

[0308] The previously used antibody VH regions known to be able to pair with the common human IGKV1-39 light chain were recloned into constructs 1, 2, 68, and 69 to generate Vectors I to V in Table 17. Subsequently, Vectors I to V, each containing a nucleic acid sequence encoding the common human light chain and an Ig heavy chain with different CH3 regions and different antigen specificities, were transfected into cells at different transfection ratios shown in Table 18. The results are shown in ​ in.

[0309] Table 17:

[0310]

[0311] Table 18:

[0312]

[0313] * This sample was not measured due to a technical error.

[0314] ​A and B show that in all cases, for the DEKK mutation combinations, when there is an excess of A or C (A or C on the "DE side", B on the "KK side"), AB or BC is formed, but the remaining A or C exists as a mixture of homodimers and monomers. However, when there is an excess of B (B on the "KK side", A or C on the "DE side"), there are significant differences. AB or BC is still formed, but the remaining B basically does not exist as a homodimer and only forms monomers. The percentage was measured again by peak height Nota bene: Peaks detected in the range of 2% or lower are below the threshold that can be accurately measured by the applied nMS technique. Therefore, measurements of <2% are considered to be within the noise level of the analysis and are thus ignored. Surprisingly, an excess of B only results in a higher percentage of monomer B. Especially at A:B ratios of 1:3 and 1:5, a higher percentage of monomer B was observed in the absence of the homodimer BB ( ​ of A and B), indicating that the CH3 mutation on the KK side is not conducive to homodimerization. The absence of homodimers provides a key advantage because this "KK side" of the DEKK combination can be selected to introduce specificity, which may have known adverse effects when it exists as a homodimer (for example, it is known that cMET or CD3 antibodies have undesirable adverse side effects when present as bivalent homodimers in therapeutic compositions).

[0315] For different DE:KK ratios, the findings observed were contrary to the control charge-reversal CH3 mutations in vectors IV and V. ​ C shows that in all cases, for the E356K:D399K / K392D’:K409D’ mutation combination, when there is an excess of A (A on the "K392D:K409D side"), the remaining A exists as a mixture of homodimers and monomers, but similarly, in all cases, when there is an excess of B (B on the "E356K:D399K side"), the remaining B exists as a mixture of homodimers and monomers. Even at higher ratios of 1:3 and 1:5, monomer B was not observed, but homodimers were present, indicating that the E356K:D399K side is not as unfavorable for homodimerization as the KK side of the DEKK combination.

[0316] In summary, the DEKK mutation combination provides a significant benefit over charge-reversal CH3 mutations in that one of the chains of the heterodimer does not form a homodimer.

[0317] Example 22: Use of various mixtures of the DEKK combination

[0318] Since the DEKK mutant combination has been shown to drive the formation of highly pure bispecific IgG molecules (“AB”), we next explored the feasibility of controllably generating more complex antibody mixtures (e.g., “AB and AA” or “AB and AC” mixtures) from a single cell. The previously used model Fabs were introduced into vectors containing “DE constructs” or “KK constructs”, and multiple combinations of these vectors were co-expressed to generate mixtures, thereby demonstrating the versatility of the technique. The selection of the model Fabs MF1337 (tetanus toxoid), MF1122 (fibrogen), and MF1025 (thyroglobulin) was based on their overall stable performance, good expression levels, and the quality differences between the IgGs containing these Fabs (see Table 19).

[0319] Table 19:

[0320] ​ ​ ​ ​ ​ ​ 146747.03 +1842.05 ​ ​ 144904.98 0 ​ ​ 144259.87 -645.11

[0321] *MF = Merus Fab, name, e.g., MF1337 and 1337, both used interchangeably.

[0322] Table 20: Transfection protocol:

[0323]

[0324] SDS-PAGE analysis demonstrated that most samples consisted mainly of full IgG, and in some cases, a small percentage of half-molecules was present. Additionally, on non-reducing gels, many samples showed two bands at ca. 150 kDa, reflecting the presence of two different IgG species in the samples. Similarly, on reducing gels, two heavy chain bands were visible in some samples (data not shown).

[0325] Native MS was performed on all samples, and the percentage of the observed species was calculated based on the peak heights (species % observed in Table 20). The results are shown in ​ . In all eight samples co-expressing three heavy chains, two main peaks corresponding to the expected species were observed. In two of these samples (transfections 2 and 4) and transfection 11, a small amount of contaminating DE-DE homodimer was observed. In most samples, a very small amount of half-molecules (less than 2%) was detected, which, as previously mentioned, is not a problem as they can be easily separated from the full-length IgG moieties. After nMS, it was found that the IgG mass observed in sample 11 corresponded to a different species than expected, presumably due to a transfection error, i.e., in sample 11, apparently 1025-DE was co-transfected with 1337-KK instead of 1122-KK.

[0326] The IgG samples were further tested in sandwich ELISA to confirm the presence of functionality with the desired specificity. The ELISA plates were coated with fibrinogen or thyroglobulin and detected using fluorescein-labeled thyroglobulin or tetanus toxoid. The detection antigen was labeled with fluorescein according to the manufacturer's instructions (Pierce NHS-Fluorescein Antibody Labeling Kit, catalog number #53029). Subsequently, the fluorescein-labeled antigen could be detected by FITC-conjugated anti-fluorescein antibody (Roche diagnostics, catalog number #11426346910).

[0327] The results (OD450 values) of the bispecific ELISA are summarized in Table 21. The gray cells represent the expected species for each transfection. Generally speaking, the results were in line with the expected results, except as indicated in italics or bold. In transfections 1 to 3, the supposedly "negative" wells for species BC (tr. #1 and 2) or AC (tr. #3) demonstrated significant background signals. It is known from previous studies that bispecific ELISAs can suffer from higher background levels. These background levels can also be caused by potential hemibodies present in the samples. Clearly, the bispecific ELISA results did confirm that an error occurred in transfection 11, as species AC (bold value) was detected instead of BC.

[0328] Table 21: OD450 values from bispecific ELISA

[0329]

[0330] Example 23: Improved mixture of two bispecific antibodies (AB and CD) that recognize 4 different epitopes from a single cell

[0331] In Example 12, it was hypothesized that the mixtures expected to be produced by transfection ZA or ZB would become problematic upon transfer to large-scale production because the knot-in hole variants were reported to be unstable and it could not be excluded that the CH3 domains containing "knots" or "holes" would dimerize with the charge-modified CH3 domains. Since it was demonstrated in the above examples that the discovery of new charge pair mutations preferentially drives heterodimerization while hardly forming homodimers, the polypeptide chains containing CH3 domains with these new charge pair mutations can be expressed in cells together with the previously known charge-modified polypeptide chains containing CH3 domains or potentially with the SEED bodies, and may result in the preferential formation of only two bispecific molecules.

[0332] As can be seen from the above embodiments, by cloning cells, the DEKK mutation combination is excellent for generating one bispecific (AB) or two bispecifics (AB plus AC), and the dimerization of the heavy chain in the cells is driven by the CH3 domain. However, using only one set of vectors with complementary CH3 mutations limits the number of possible types of mixtures that can be produced. If a second "orthogonal" set of vectors is used in combination with DEKK, it will be possible to generate more complex mixtures of IgG and / or bispecifics, for example, "AB and CD" or "AB and CC" mixtures. When combining two sets of vectors, an important condition is that the heavy chains expressed by the two sets of vectors modified with two different CH3s cannot form "cross" dimers, that is, the heavy chain produced by one set of vectors dimerizes with the heavy chain expressed by the other set of vectors to form a full IgG.

[0333] To test the potential formation of "cross" dimers, a computational analysis using HADDOCK was performed to further understand whether possible pairing would occur between the wild-type CH3 domain and the CH3 domain containing DE- or KK mutations. Similarly, the potential pairing between the wild-type CH3 domain and the CH3 domain containing E356K, D399K or K392D, K409D mutations, and the potential pairing between the wild-type CH3 domain and the CH3 domain containing the knot-in-hole mutation and any combination of the above were analyzed. The combinations of CH3 mutants analyzed in HADDOCK are listed in Table 22, and the HADDOCK scores obtained are summarized in ​ the

[0334] Table 22: CH3 variants analyzed in HADDOCK, and single-letter codes are assigned to the heavy chains carrying each CH3 variant. *For consistency issues, the wild-type chains are named "C" and "D"; **When combined with the knot-in-hole variant, the charge-reversed variants are named "A" and "B", and when combined with the DE / KK variant, they are named "C" and "D".

[0335]

[0336] ​ It shows that based on these HADDOCK predictions, when co-transfected in a single cell, combining the CH3 combination of DEKK with the charge-reversed CH3 combination is most likely to successfully form the desired combination of two bispecifics (AB and CD) without contaminating by-products (especially AC, AD, BC, BD). By ​As can be seen, these undesired bispecific species AC, AD, BC, and BD have relatively high HADDOCK scores, while the desired AB and CD species have the lowest HADDOCK scores. Of course, when the DEKK or charge-reversed CH3 combination is introduced into constructs carrying the same specificities (e.g., "C" on the DE side, "C" on the KK side, "A" on the E356K, D399K side, and "B" on the E356K, D399K side, or "A" on the DE side, "B" on the KK side, "C" on the E356K, D399K side, and "C" on the E356K, D399K side), this will result in mainly CC and AB being produced upon co-expression in cells.

[0337] Conversely, when looking at the predictions for co-expressing DEKK with the wild type, it can be seen that the HADDOCK scores of both AC and AD are lower than the HADDOCK score of CD, indicating that AC and AD are very likely to be contaminants when attempting to produce a mixture of AB and CD by co-expressing a vector encoding the CH3 combination of DEKK with a vector encoding the wild-type CH3. Finally, it is predicted that co-expressing DEKK or a charge-reversed variant with a pore-forming variant results in undesired bispecific variants with relatively low HADDOCK scores, i.e., there is a higher likelihood of producing these undesired species upon co-expression.

[0338] It is thus inferred that combining the CH3 combination of DEKK with the charge-reversed CH3 combination (E356K, D399K / K392’D, K409D’) is ideally suited to obtain a substantially pure mixture of "AB and CD" and / or "AB and CC" antibodies.

[0339] Next, the above was put into practice to generate a mixture of two bispecifics that recognize 4 targets / epitopes (AB and CD) and a mixture of one bispecific antibody and one monospecific antibody that recognize 3 targets / epitopes (AB and CC). These mixtures were generated using 4 different VHs, all of which are capable of pairing with the common light chain IGVK1-39, but each VH / VL combination has a different specificity. There should be a sufficient mass difference (i.e., >190 Da) between the (expected) species to enable native MS analysis. Four individual VHs were selected, and these VHs should have such masses that the expected species can be identified and isolated by nMS upon co-transfection. In addition, the mass differences of the 4 selected VHs should be large enough to identify most of the possible contaminants in the mixture in addition to the two desired species. The selected VHs are listed in Table 23.

[0340] Table 23:

[0341]

[0342] Four different VHs were cloned into vectors containing either the "DE" or "KK" construct or the charge-reversal construct, as shown in Table 24, and several co-transfections were performed. NB: As usual, all vectors also contained nucleic acids encoding the common light chain IGKV1-39. As previously described, when combining two vector sets, an important condition is that the heavy chains expressed by the two differently CH3-modified vector sets do not form "cross" dimers, i.e., the heavy chains produced by one vector set do not dimerize with the heavy chains expressed by the other vector set to form full IgG. Control transfections were performed to test for this potential formation of "cross" dimers between the heavy chains containing the charge-reversal mutation and the heavy chains containing the DE or KK mutations.

[0343] Table 24:

[0344]

[0345] Table 25 further provides an overview of the expected species and the masses of possible contaminants in transfections #9 to 11 in Table 24.

[0346] Table 25: For each of transfections #9 to 11, the species were sorted by mass, and the mass differences were calculated using the masses above. Grey cells: expected (and desired) species; italic: mass differences too small to be separated in nMS analysis. *Species: single-letter for half-body; two-letter code for full IgG.

[0347]

[0348] All purified protein samples obtained from transfections #1 to #11, including three control samples ( ​ ), were analyzed on SD-PAGE. In addition, nMS analysis was performed on the protein samples from transfections #9 to #11 to identify all species in the samples. As ​ can be seen, transfections #3 and #4 resulted in the expected mismatch between the "KK" construct and "E356K:D399K" or "K392D:K409D", and the amount of half-bodies in the protein samples from these transfections exceeded the amount of full IgG molecules. Transfections #7 and #8 resulted in protein samples in which half-bodies and full IgG were present in approximately equal amounts. However, it could not be inferred from SDS-PAGE whether the full IgG represented a DE / DE dimer, a DE / E356K:D399K dimer, or a DE / K392D:K409D dimer. Significantly, almost no half-bodies were observed in the protein samples from transfections #9 to #11.

[0349] The nMS analysis of transfections #9 and #11 showed that in ​Among them, the percentages of the expected species and the contaminating species are calculated by peak height. It has been demonstrated that for transfection #9, the expected species "AB and CD" accounted for 97% (30% AB and 67% CD) in the mixture, while there was only as little as approximately 3% of the contaminating BD( ​ , ​ ). For transfection #11, the expected species "AB and CC" accounted for 94% (33% AB and 61% CC) in the mixture, while there was only as little as approximately 6% of the contaminating BC (4.1%) and AC (1.8%)( ​ ). These data indicate that when the second "orthogonal" vector set is used in combination with DEKK, it is indeed possible to generate more complex mixtures of IgG and / or bispecifics, such as "AB and CD" or "AB and CC" mixtures. Combining the charge-reversal construct with the DEKK construct results in only a very limited formation of "cross" dimers. By adjusting the transfection ratio, it is expected that these low percentages of contaminating by-products can be even further reduced.

[0350] Example 24: Single-dose pharmacokinetic study in mice

[0351] In this study, the pharmacokinetic parameters of three different IgG batches were determined and compared to investigate the pharmacokinetic (pK) behavior of bispecific antibodies carrying the DEKK mutation combination in their CH3 region. The three IgG batches included 1) wild-type anti-tetanus toxoid parental antibody 1337:1337 (two MF1337Fabs on a wild-type Fc backbone); 2) wild-type anti-tetanus toxoid parental antibody 1516:1516 (two MF1516Fabs on a wild-type Fc backbone); 3) CH3-modified bispecific anti-tetanus toxoid antibody 1516:1337, which carried the DEKK mutation combination in its Fc region (MF1516 Fab on the DE side and MF1337 Fab on the KK side).

[0352] Based on specificity, parental antibodies 1337:1337 and 1516:1516 were selected to be included in the DEKK bispecific products because, based on previous studies, no pre-dose serum responses against these antibodies were present in several mouse strains. NB: The presence of a pre-dose serum response would of course invalidate this study. Additionally, there was sufficient mass difference between the parental antibodies to enable the identification of the 1337:1337 (wt Fc), 1516:1337 (DEKK Fc), and 1516:1516 (wt Fc) species by nMS. Three IgG batches were prepared as described previously, but the DNA used for transfection was prepared using an endotoxin-free maxiprep kit to ensure that the amount of endotoxin was as low as possible. Subsequently, the protein concentration, aggregation level, endotoxin level, and percentage of bispecific product of these batches were tested. It was demonstrated that the acceptance criteria for subsequent use of the IgG batches were met in the pK study, i.e., IgG concentration > 0.3 mg / mL after gel filtration, aggregation level < 5%, endotoxin level < 3 EU / mg protein, and the DEKK batch contained > 90% bispecific IgG.

[0353] Native mass spectrometry of the gel-filtered samples indicated that the expected species were present at high percentages. In sample 1516:1337, a small amount of the DE:DE homodimer was detected, which was estimated to be ca. 2% ( ​ ). It can thus be concluded that the 3 IgG batches are suitable for use in the pK study.

[0354] Three groups of female C57BL / 6J mice (Harlan, The Netherlands) were dosed with 1 mg / kg human IgG (5 ml / kg immunoglobulin solution / kg body weight) to compare the pK parameters between the three batches. At the time of dosing, the animals were 7 to 8 weeks old and weighed approximately 18 to 20 grams. Blood samples were collected before dosing and at 15 minutes, 60 minutes, 2 hours, 4 hours, 8 hours, 24 hours, 48 hours, 96 hours, 168 hours, 268 hours, and 336 hours after dosing. Serum samples were prepared and stored at < -20 °C until analysis. Each group consisted of 3 subgroups of 4 mice, i.e., 12 mice / group. Samples were taken from each mouse at 6 time points. The welfare of the animals was maintained in accordance with the basic principles governing the use of animals in experiments in the European Community (Directive 86 / 609 / EEC) and Dutch legislation (Animal Experimentation Act, 1997). This study was also conducted in accordance with the standards for humane care and use of laboratory animals promulgated by the Office of Laboratory Animal Welfare of the National Institutes of Health, USA, under identification number 45859-01 (expiration date: April 30, 2015).

[0355] Group 1 mice received full-length monospecific IgG 1516:1516 antibody (triangle); Group 2 mice received full-length monospecific IgG 1337:1337 antibody (square); Group 3 mice received full-length bispecific IgG 1516:1337 antibody with a DEKK-modified CH3 region (1516 on the DE side, 1337 on the KK side) (diamond);

[0356] Using a quantitative human IgG ELISA (ZeptoMetrix, NY USA; ELISA kit nr. 0801182), monoclonal human antibodies in mouse sera were quantitatively analyzed by ELISA assay. Briefly, the ELISA assay is based on the principle that human monoclonal antibodies bind to anti-human IgG coated in a 96-well ELISA plate. Subsequently, the bound antibodies were visualized using a polyclonal anti-human IgG antibody conjugated to horseradish peroxidase (HRP). The optical density (OD) of each well is proportional to the amount of antibody in the serum sample. The results are shown in ​ where significantly similar serum levels were observed for both the bispecific full-length IgG antibody carrying the DEKK mutant combination and its parental monospecific antibody. It was thus concluded that the CH3 mutations present in the DEKK bispecific antibody neither altered stability nor half-life, and the DEKK variant behaved like wild-type IgG.

[0357] ​

[0358] Deisenhofer J., Biochemistry 1981(20)2361 - 2370;

[0359] Miller S., J. Mol. Biol. 1990(216)965 - 973;

[0360] Padlan, Advances in Protein Chemistry 1996(49)57 - 133

[0361] Ellerson JR., et al., J. Immunol 1976(116)510 - 517;

[0362] Lee and Richards J. Mol. Biol. 1971(55)379.

[0363] Gunasekaran et al J. Biol. Chem. 2010(285)19637 - 19646

[0364] De Vries Nature Protocols 2010(5)883

[0365] Kabat et al., (1991)

[0366] Mammalian Cell Biotechnology:a Practical Approach (M. Butler, ed., IRL Press, 1991

[0367] Merchant Nature biotechnology 1998(16)677

[0368] Ridgeway Protein Engineering 1996(9)617 - 621.

[0369] Davis JH. Et al., Protein Engineering, Design & Selection 2010(23)195 - 202

[0370] Papadea and Check. Crit Rev Clin Lab Sci. 1989;27(1):27 - 58.

[0371] Tissue Culture, Academic Press, Kruse and Paterson, editors(1973)

[0372] Ionescu et al., J.Pharm.Sci. 2008(97)1414)

[0373] Current protocols in Protein Science 1995, coligan JE et al., Wingfield PT, ISBN 0 - 471 - 11184 - 8, Bendig 1988.

[0374] Capelle, M.A.H., Brugger, P., Arvinte, T. Vaccine 23(2005), 1686 - 1694.

[0375] Demeule, B., Lawrence, M.J., Drake, A.F., Gurny, R., Arvinte, T. Biochim.Biophys.

[0376] Acta 1774(2007a), 146 - 153.

[0377] Demeule, B., Gurny, R., Arvinte, T., Int.J.Pharm 329(2007b), 37 - 45.

[0378] Lakowicz, J.R., Principles of fluorescence spectroscopy;Second EditionKluwer Academic / Plenum Publishers, New York, Boston, Dordrecht, London, Moscow, (2006)ISBN 0 - 306 - 46093 - 9.

[0379] The following corresponds to the original claims in the parent application and is hereby incorporated herein as part of the specification:

[0380] 1. A method for producing at least two different immunoglobulin - like molecules from a single host cell, wherein each of the two immunoglobulin - like molecules comprises two CH3 domains capable of forming an interface, the method comprising providing in the cell:

[0381] a. a first nucleic acid molecule encoding a first polypeptide chain containing a CH3 domain,

[0382] b. a second nucleic acid molecule encoding a second polypeptide chain containing a CH3 domain,

[0383] c. a third nucleic acid molecule encoding a third polypeptide chain containing a CH3 domain, and

[0384] d. a fourth nucleic acid molecule encoding a fourth polypeptide chain containing a CH3 domain, wherein at least two of the nucleic acid molecules have means for preferentially pairing the first polypeptide chain containing a CH3 domain with the second polypeptide chain containing a CH3 domain and the third polypeptide chain containing a CH3 domain with the fourth polypeptide chain containing a CH3 domain, the method further comprising culturing the host cell and enabling expression of the at least four nucleic acid molecules and harvesting the at least two different immunoglobulin - like molecules from the culture.

[0385] 2. The method of claim 1, further comprising providing to the host cell a nucleic acid molecule encoding a common light chain.

[0386] 3. The method of claim 1 or 2, wherein the first polypeptide chain containing a CH3 domain comprises the amino acid substitution T366K, and the second polypeptide chain containing a CH3 domain comprises the amino acid substitution L351D.

[0387] 4. The method according to item 3, wherein the first polypeptide chain containing the CH3 domain further comprises the amino acid substitution L351K.

[0388] 5. The method according to item 3 or 4, wherein the second polypeptide chain containing the CH3 domain further comprises an amino acid substitution selected from Y349E, Y349D, and L368E.

[0389] 6. The method according to item 5, wherein the second polypeptide chain containing the CH3 domain further comprises the amino acid substitution L368E.

[0390] 7. The method according to any one of items 1 to 6, wherein the third polypeptide chain containing the CH3 domain comprises the amino acid substitutions E356K and D399K, and the fourth polypeptide chain containing the CH3 domain comprises the amino acid substitutions K392D and K409D.

[0391] 8. The method according to any one of items 1 to 7, wherein each of the polypeptide chains containing the CH3 domain further comprises a variable region that recognizes a target epitope.

[0392] 9. The method according to item 8, wherein the four variable regions of the four polypeptide chains containing the CH3 domain each recognize a different target epitope.

[0393] 10. The method according to item 8, wherein the variable region of the first polypeptide chain containing the CH3 domain and the variable region of the second polypeptide chain containing the CH3 domain recognize different target epitopes, while the variable region of the third polypeptide chain containing the CH3 domain and the variable region of the fourth polypeptide chain containing the CH3 domain recognize the same target epitope.

[0394] 11. The method according to item 10, wherein the target epitope recognized by the variable region of the third polypeptide chain containing the CH3 domain and the variable region of the fourth polypeptide chain containing the CH3 domain is the same as the target epitope recognized by the variable region of the first polypeptide chain containing the CH3 domain or the variable region of the second polypeptide chain containing the CH3 domain.

[0395] 12. The method according to item 10, wherein the target epitope recognized by the variable region of the third polypeptide chain containing the CH3 domain and the variable region of the fourth polypeptide chain containing the CH3 domain is different from the target epitope recognized by the variable region of the first polypeptide chain containing the CH3 domain or the variable region of the second polypeptide chain containing the CH3 domain.

[0396] 13. The method according to item 8, wherein the variable regions of the first polypeptide chain containing a CH3 domain and the variable regions of the second polypeptide chain containing a CH3 domain recognize the same target epitope, and the variable regions of the third polypeptide chain containing a CH3 domain and the variable regions of the fourth polypeptide chain containing a CH3 domain recognize a second target epitope different from the target epitope recognized by the first variable region and the second variable region.

[0397] 14. The method according to any one of items 8 to 13, wherein the target epitope is located on the same target molecule.

[0398] 15. The method according to item 14, wherein the target molecule is a soluble molecule.

[0399] 16. The method according to item 14, wherein the target molecule is a membrane-bound molecule.

[0400] 17. The method according to any one of items 8 to 13, wherein the target epitopes are located on different target molecules.

[0401] 18. The method according to item 17, wherein the different target molecules are expressed on the same cell.

[0402] 19. The method according to item 17, wherein the different target molecules are expressed on different cells.

[0403] 20. The method according to item 17, wherein the different target molecules are soluble molecules.

[0404] 21. The method according to item 17, wherein one target molecule is a soluble molecule and the second target molecule is a membrane-bound molecule.

[0405] 22. The method according to any one of items 1 to 21, wherein the at least two different immunoglobulin-like molecules are antibodies.

[0406] 23. The method according to item 1, wherein the means for preferential pairing includes engineered complementarity-determining region pore mutations, disulfide bonds, charge mutations, or combinations thereof.

[0407] 24. The method according to item 23, wherein the means for preferential pairing is selected from Table B.

[0408] 25. The method according to item 1, wherein all 4 of the nucleic acid molecules have means for preferentially pairing the first polypeptide containing a CH3 domain with the second polypeptide containing a CH3 domain and for preferentially pairing the third polypeptide containing a CH3 domain with the fourth polypeptide containing a CH3 domain, and wherein the means for preferentially pairing the first polypeptide containing a CH3 domain with the second polypeptide containing a CH3 domain are different from those for preferentially pairing the third polypeptide containing a CH3 domain with the fourth polypeptide containing a CH3 domain.

[0409] 26. The method according to any one of items 8 to 13, 17 to 19 or 21, wherein at least one of the target epitopes is located on a tumor cell.

[0410] 27. The method according to any one of items 8 to 13, 17 to 19 or 21, wherein at least one of the target epitopes is located on an effector cell.

[0411] 28. The method according to item 27, wherein the effector cell is a NK cell, a T cell, a B cell, a monocyte, a macrophage, a dendritic cell or a neutrophil.

[0412] 29. The method according to item 27 or 28, wherein the target epitope is located on a CD3, CD16, CD25, CD28, CD64, CD89, NKG2D or NKp46 molecule.

[0413] 30. A mixture of at least two different immunoglobulin-like molecules obtainable by the method according to any one of items 1 to 29.

[0414] 31. The mixture according to item 30, wherein the at least two immunoglobulin-like molecules bind to different epitopes on the same antigen and / or different epitopes on different antigens.

[0415] 32. The mixture according to item 30 or 31, wherein the at least two different immunoglobulin-like molecules include at least one heterodimeric immunoglobulin-like molecule.

[0416] 33. The mixture according to any one of items 30 to 32, wherein two of the at least two different immunoglobulin-like molecules are heterodimeric immunoglobulin-like molecules.

[0417] 34. A recombinant host cell comprising nucleic acid sequences encoding at least a first polypeptide chain containing a CH3 domain, a second polypeptide chain containing a CH3 domain, a third polypeptide chain containing a CH3 domain, and a fourth polypeptide chain containing a CH3 domain, wherein at least two of said nucleic acid sequences have means for preferentially pairing the first polypeptide containing a CH3 domain with the second polypeptide containing a CH3 domain and the third polypeptide containing a CH3 domain with the fourth polypeptide containing a CH3 domain.

[0418] 35. The recombinant host cell according to item 34, wherein said host cell further comprises a nucleic acid sequence encoding a common light chain.

[0419] 36. A pharmaceutical composition comprising at least two different immunoglobulin-like molecules according to any one of items 30 to 33 and a pharmaceutically acceptable carrier.

[0420] 37. The pharmaceutical composition according to item 36, wherein said at least two different immunoglobulin-like molecules are produced by a recombinant host cell according to item 34 or 35.

[0421] 38. A method for preparing a host cell for producing at least two different immunoglobulin-like molecules, said method comprising introducing into said host cell nucleic acid sequences encoding at least a first polypeptide chain containing a CH3 domain, a second polypeptide chain containing a CH3 domain, a third polypeptide chain containing a CH3 domain, and a fourth polypeptide chain containing a CH3 domain, wherein at least two of said nucleic acid sequences have means for preferentially pairing the first polypeptide containing a CH3 domain with the second polypeptide containing a CH3 domain and the third polypeptide containing a CH3 domain with the fourth polypeptide containing a CH3 domain, and wherein said nucleic acid sequences are introduced successively or simultaneously.

[0422] 39. The method according to item 38, further comprising the step of introducing into said host cell a nucleic acid sequence encoding a common light chain.

[0423] 40. A culture of a recombinant host cell according to item 34 or 35 or a culture of a recombinant host cell obtained by the method according to item 38 or 39, which produces at least two different immunoglobulin-like molecules.

Claims

1. A method for generating at least two different immunoglobulin-like molecules from a single host cell, wherein each of the two immunoglobulin-like molecules comprises two CH3 domains capable of forming an interface, the method comprising providing in the cell: a. a first nucleic acid molecule encoding a first polypeptide chain comprising a CH3 domain, b. a second nucleic acid molecule encoding a second polypeptide chain comprising a CH3 domain, c. a third nucleic acid molecule encoding a third polypeptide chain comprising a CH3 domain, and d. a fourth nucleic acid molecule encoding a fourth polypeptide chain comprising a CH3 domain, Among them, at least two of the nucleic acid molecules having means for preferentially pairing the first polypeptide chain comprising a CH3 domain with the second polypeptide chain comprising a CH3 domain and the third polypeptide chain comprising a CH3 domain with the fourth polypeptide chain comprising a CH3 domain, the method further comprising culturing the host cell and enabling expression of the at least four nucleic acid molecules and harvesting the at least two different immunoglobulin-like molecules from the culture.

2. The method of claim 1, further comprising providing to the host cell a nucleic acid molecule encoding a common light chain.

3. The method of claim 1 or 2, wherein the first polypeptide chain comprising a CH3 domain comprises the amino acid substitution T366K, and the second polypeptide chain comprising a CH3 domain comprises the amino acid substitution L351D.

4. The method of claim 3, wherein the first polypeptide chain comprising a CH3 domain further comprises the amino acid substitution L351K.

5. The method of claim 3 or 4, wherein the second polypeptide chain comprising a CH3 domain further comprises an amino acid substitution selected from Y349E, Y349D, and L368E.

6. The method of claim 5, wherein the second polypeptide chain comprising a CH3 domain further comprises the amino acid substitution L368E.

7. The method of any one of claims 1 to 6, wherein the third polypeptide chain comprising a CH3 domain comprises the amino acid substitutions E356K and D399K, and the fourth polypeptide chain comprising a CH3 domain comprises the amino acid substitutions K392D and K409D.

8. The method of any one of claims 1 to 7, wherein each of the polypeptide chains comprising a CH3 domain further comprises a variable region that recognizes a target epitope.

9. The method of claim 8, wherein the four variable regions of the four polypeptide chains comprising a CH3 domain each recognize a different target epitope.

10. The method of claim 8, wherein the variable region of the first polypeptide chain comprising a CH3 domain and the variable region of the second polypeptide chain comprising a CH3 domain recognize different target epitopes, while the variable region of the third polypeptide chain comprising a CH3 domain and the variable region of the fourth polypeptide chain comprising a CH3 domain recognize the same target epitope.

11. The method according to claim 10, wherein the target epitopes recognized by the variable regions of the third polypeptide chain containing a CH3 domain and the fourth polypeptide chain containing a CH3 domain are the same as the target epitopes recognized by the variable region of the first polypeptide chain containing a CH3 domain or the variable region of the second polypeptide chain containing a CH3 domain.

12. The method according to claim 10, wherein the target epitopes recognized by the variable regions of the third polypeptide chain containing a CH3 domain and the fourth polypeptide chain containing a CH3 domain are different from the target epitopes recognized by the variable region of the first polypeptide chain containing a CH3 domain or the variable region of the second polypeptide chain containing a CH3 domain.

13. The method according to claim 8, wherein the variable regions of the first polypeptide chain containing a CH3 domain and the second polypeptide chain containing a CH3 domain recognize the same target epitope, while the variable regions of the third polypeptide chain containing a CH3 domain and the fourth polypeptide chain containing a CH3 domain recognize a second target epitope different from the target epitope recognized by the first variable region and the second variable region.

14. The method according to any one of claims 8 to 13, wherein the target epitopes are located on the same target molecule.

15. The method according to claim 14, wherein the target molecule is a soluble molecule.

16. The method according to claim 14, wherein the target molecule is a membrane-bound molecule.

17. The method according to any one of claims 8 to 13, wherein the target epitopes are located on different target molecules.

18. The method according to claim 17, wherein the different target molecules are expressed on the same cell.

19. The method according to claim 17, wherein the different target molecules are expressed on different cells.

20. The method according to claim 17, wherein the different target molecules are soluble molecules.

21. The method according to claim 17, wherein one target molecule is a soluble molecule and the second target molecule is a membrane-bound molecule.

22. The method according to any one of claims 1 to 21, wherein the at least two different immunoglobulin-like molecules are antibodies.

23. The method according to claim 1, wherein the means for preferential pairing includes engineered complementarity-determining region pore mutations, disulfide bonds, charge mutations, or combinations thereof.

24. The method according to claim 23, wherein the means for preferential pairing is selected from Table B.

25. The method according to claim 1, wherein all 4 of the nucleic acid molecules have means for preferentially pairing the first polypeptide chain containing a CH3 domain with the second polypeptide chain containing a CH3 domain and the third polypeptide chain containing a CH3 domain with the fourth polypeptide chain containing a CH3 domain, and the means for preferentially pairing the first polypeptide chain containing a CH3 domain with the second polypeptide chain containing a CH3 domain is different from the means for preferentially pairing the third polypeptide chain containing a CH3 domain with the fourth polypeptide chain containing a CH3 domain.

26. The method according to any one of claims 8 to 13, 17 to 19 or 21, wherein at least one of said target epitopes is located on a tumor cell.

27. The method according to any one of claims 8 to 13, 17 to 19 or 21, wherein at least one of said target epitopes is located on an effector cell.

28. The method according to claim 27, wherein said effector cell is an NK cell, a T cell, a B cell, a monocyte, a macrophage, a dendritic cell or a neutrophil.

29. The method according to claim 27 or 28, wherein said target epitope is located on a CD3, CD16, CD25, CD28, CD64, CD89, NKG2D or NKp46 molecule.

30. A mixture of at least two different immunoglobulin-like molecules obtainable by the method according to any one of claims 1 to 29.

31. The mixture according to claim 30, wherein said at least two immunoglobulin-like molecules bind to different epitopes on the same antigen and / or different epitopes on different antigens.

32. The mixture according to claim 30 or 31, wherein said at least two different immunoglobulin-like molecules comprise at least one heterodimeric immunoglobulin-like molecule.

33. The mixture according to any one of claims 30 to 32, wherein two of said at least two different immunoglobulin-like molecules are heterodimeric immunoglobulin-like molecules.

34. A recombinant host cell comprising nucleic acid sequences encoding at least a first polypeptide chain comprising a CH3 domain, a second polypeptide chain comprising a CH3 domain, a third polypeptide chain comprising a CH3 domain and a fourth polypeptide chain comprising a CH3 domain, wherein at least two of said nucleic acid sequences have means for preferentially pairing the first polypeptide chain comprising a CH3 domain with the second polypeptide chain comprising a CH3 domain and the third polypeptide chain comprising a CH3 domain with the fourth polypeptide chain comprising a CH3 domain.

35. The recombinant host cell according to claim 34, wherein said host cell further comprises a nucleic acid sequence encoding a common light chain.

36. A pharmaceutical composition comprising at least two different immunoglobulin-like molecules according to any one of claims 30 to 33 and a pharmaceutically acceptable carrier.

37. The pharmaceutical composition according to claim 36, wherein said at least two different immunoglobulin-like molecules are produced by a recombinant host cell according to claim 34 or 35.

38. A method for preparing a host cell for producing at least two different immunoglobulin-like molecules, said method comprising introducing into said host cell nucleic acid sequences encoding at least a first polypeptide chain comprising a CH3 domain, a second polypeptide chain comprising a CH3 domain, a third polypeptide chain comprising a CH3 domain and a fourth polypeptide chain comprising a CH3 domain, wherein at least two of said nucleic acid sequences have means for preferentially pairing the first polypeptide chain comprising a CH3 domain with the second polypeptide chain comprising a CH3 domain and the third polypeptide chain comprising a CH3 domain with the fourth polypeptide chain comprising a CH3 domain, wherein said nucleic acid sequences are introduced successively or simultaneously.

39. The method according to claim 38, further comprising the step of introducing into said host cell a nucleic acid sequence encoding a common light chain.

40. A culture of the recombinant host cell of claim 34 or 35, or a culture of the recombinant host cell obtained by the method according to claim 38 or 39, which produces at least two different immunoglobulin-like molecules.

Images