Methods for screening antibody FC regions

Abstract

The present invention relates to a method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of: (a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface, preferably wherein each cell comprised in the plurality of cells displays an engineered antibody Fc region comprising at least one mutation; (b) contacting the cells in step (a) with an Fc receptor; (c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor, preferably wherein the ability of an engineered antibody Fc region to bind to the Fc receptor is determined by comparison with a reference antibody Fc region; and (d) sequencing cells from at least one of the populations obtained in step (c), preferably to identify mutations in the engineered antibody Fc region that (i) confer improved binding to the Fc receptor, (ii) do not affect binding to the Fc receptor, and / or (iii) confer reduced binding to the Fc receptor. Further provided herein are methods for predicting the Fc receptor binding properties of an engineered antibody Fc region, as well as methods for generating an engineered antibody Fc region having desired Fc receptor binding properties

Description

[0001] METHODS FOR SCREENING ANTIBODY FC REGIONS

[0002] BACKGROUND TO THE INVENTION

[0003] Monoclonal antibodies are ubiquitous biological drugs used to treat a range of indications including cancers, autoimmune conditions, and infectious diseases. The popularity of these molecules as drugs continues to rapidly grow with 16 candidates gaining approval in 2023 and 23 further candidates following in the advanced stages of regulatory approval (Crescioli et al., MAbs (2024) 16(l):2297450). Discovering mAbs against new druggable targets remains a critical area of research, alongside efforts to improve the effectiveness of existing mAbs through protein engineering techniques.

[0004] The efficacy of an antibody is determined by two key functional domains: the antigen-binding fragment (Fab) and the Fc domain. Variable domains of the Fab fragment exhibit high affinity and specificity for virtually any antigenic target. These binding events can enable the direct neutralization of the target or mark the target for subsequent immune engagement. To activate the immune system, antibodies leverage their constant domains (Fc) to bind Fc-receptors on the surfaces of immune cells and carry out a diverse array of complementary effector functions that promote disease control. Such functions include antibody-dependent cell cytotoxicity (ADCC), antibody-dependent cell phagocytosis (ADCP), trogocytosis, complement activation, degranulation, cytokine production, and the induction of DNA-based extracellular traps (Cottignies-Calamarte et al., Front Immunol (2023) 14:1037033). The precise program of effector responses mediated by an antibody is controlled by the Fc domain affinity for each Fc receptor as well as the expression patterns of these receptors on individual immune cell subsets.

[0005] In humans, circulating antibody constant domains are derived from one of 5 distinct isotype classes: IgM, IgD, IgG, IgE, and IgA (Vidarsson et al., Front Immunol (2014) 5:520). The most abundant class, IgG, is further divided into four subclasses (IgG 1, lgG2, lgG3, and lgG4) which each direct unique effector functions despite sharing >90% sequence identity (Yu et al., J Hematol Oncol (2020) 13(1):45). Among the subclasses, IgGl has the greatest abundance in circulation and is most widely used as the backbone of therapeutic monoclonal antibodies. Conversely, lgG3 possesses the most potent effector function of any subclass but suffers from poor stability (Warrender et al., Protein Sci (2023) 32(3):e4589), limiting its clinical developability. However, lgG2 and lgG4 constant domains have found application in anticancer (Rosner et al., Mol Cancer Ther (2019) 18(l):75-88) and immuno-silent applications (Rispens & Huijbers, Nat Rev Immunol (2023) 23(ll):763-778).

[0006] Fc domain-mediated effector functions are driven by the engagement of antibody constant domains with members of the FcR family, the Fc receptor-like family, and several non-canonical receptors (e.g. TRIM21) alongside soluble proteins like complement factor Clq. The most widely studied of these are the human FcyRs, comprising FcyRI (CD64), FcyRlla (CD32a), FcyRllb (CD32b), FcyRllc (CD32c), FcyRllla (CD16a), and FcyRlllb (CD16b). Except for high-affinity FcyRI, these receptors have low affinity for constant domains and require antibodyantigen complexes to induce cell responses. Inflammatory cellular activation by the FcyRs is balanced by the sole inhibitory receptor FcyRllb but it is the pattern of Fc receptors expressed on individual cell subsets that determines the strength and nature of the induced effector functions. For instance, FcyRllla is predominantly found on the surface of NK cells, which lack FcyRllb expression, where it induces potent ADCC responses. However, significant variation in the expression of Fc receptors is observed between individuals and is also dependent on the maturation state of the immune cell (Delidakis et al., Annu Rev Biomed Eng (2022) 6:24:249-274). Further, genetic variation within Fc receptor classes can result in high and low-affinity alleles for certain receptors, which may further influence the quality of the immune response induced (Kim et al., Nat Commun (2021) 12(1):288).

[0007] Alongside immune effector functions mediated by FcyRs, the FcRn neonatal receptor also plays an important role in determining antibody half-life in circulation. Endocytosed antibodies are rescued from degradation pathways by binding at acidic pH and returned to the cell surface for release at pH 7.4. This recycling mechanism accounts for the significantly longer serum half-lives of IgG antibodies as compared to other classes like IgA (21 days vs ~7 days) (Van Tetering et al., Antibodies (Basel) (2020) 9(4):70).

[0008] Manipulating the constant domain to tune therapeutic effector functions has been an important strategy to enhance antibody drug efficacy. Strategies have included mutating the amino acid sequences, engineering N-linked glycan structures, or applying both approaches (R. Liu et al., Antibodies (Basel) (2020) 9(4):64). Typically, binding affinity towards select receptors can be improved by orders of magnitude over wild-type levels to either augment or attenuate effector functions. For instance, greater than 100-fold enhanced binding to FcyRllla (V allele) can be achieved via the removal of fucose sugars from the glycan structure or through the introduction of point mutations to the constant domains (e.g. S239D / I332E / A330L) (Lazar et al., Proc Natl Acad Sci USA (2006) 103(ll):4005-10). Significant effort has also been expended engineering constant region interaction with the FcRn receptor to enhance drug half-life (e.g. M428L / N434S (Zalevsky et al., Nat Biotechnol (2010) 28(2):157-9), M252Y / S254T / T256E (Dall'Acqua et al., J Biol Chem (2006) 281(33):23514-24).

[0009] A key challenge in developing new constant region variants is achieving optimal binding profiles to Fc receptors other than those targeted during screening. For instance, retaining residual binding to the inhibitory FcyRllb receptor can reduce desired cytotoxic functions. This is despite various strategies having been explored to discover variants in glycosylated, aglycosylated, and asymmetrical constant regions (Damelang et al., Front Immunol (2024) 14:1304365). While most clinically approved constant region variants are in glycosylated backbones, aglycosylated variants potentially offer unique advantages for Fc receptor selectivity. Eliminating N-linked glycans appended to N297 via mutation of the N-X / S-T motif, or by expression in a host lacking glycosylation machinery, increases the overall flexibility of the constant domains. These changes in dynamic protein behavior are accompanied by significant reductions, but not complete ablation, of FcyR binding affinity (Delidakis et al., Annu Rev Biomed Eng (2022) 24:249-274). It is proposed that subsequent introduction of point mutations can fix conformations not readily achieved by glycosylated antibodies and reintroduce binding to Fc receptors with greater discrimination (Borrok et al., ACS Chem Biol (2012) 7(9):1596-602). Despite initial concerns around the immunogenicity, pharmacokinetic, and stability profiles of aglycosylated antibodies, at least three aglycosylated variants have been approved for clinical use (Atezolizumab, Mosunetuzumab, and Eptinezumab (Dhillon, Drugs (2020) 80(7):733-739; Mathieu et al., Clin Cancer Res (2023) 29(16):2973-2978; Tapia-Galisteo et al., J Hematol Oncol (2023) 16(1) :83).

[0010] The discovery of mutations such as S298G / T299A first demonstrated that glycans are not an essential prerequisite for FcyR engagement. Despite disrupting the N-glycan acceptor motif, this variant displays selective binding to FcyRI, FcyRlla, and FcyRllb receptors (Sazinsky et al., Proc Natl Acad Sci USA (2008) 105(51):20167-72). Subsequent work has since restored near wild-type binding across all the canonical FcyRs (Chen et al., ACS Chem Biol (2017) 12(5 ) :1335-1345). In parallel, efforts have focused on generating highly selective FcyR binding in aglycosylated constant regions for individual Fc receptors such as FcyRI (Jung et al., Proc Natl Acad Sci U S A (2010) 107(2) :604-9), FcyRlla over FcyRllb (Jung et al., Biotechnol Bioproc (2013) E 18, 625-636), FcyRllb (US 2024 / 0010742), and Clq (Lee et al., Nat Immunol (2017) 18(8):889-898). These approaches have relied on rational design based on structural information, or directed evolution, where random or systematic diversity is introduced to the constant region followed by high-throughput screening for FcyR binding.

[0011] However, a major bottleneck with experimental screening systems is the limited diversity space that can be physically explored. Transformation efficiencies into protein display systems (e.g. bacteria, yeast, or mammalian cells) do not routinely exceed 1E9 variants and typically represent only a small fraction of the possible combinatorial diversity. Further, deconvoluting the influence of mutations on overlapping FcyR receptor binding sites is a challenge that remains unsolved, and most engineered Fc domains have optimized binding to one or two select Fc receptors. The recent development of machine learning approaches to support applications in protein engineering offers a means to circumvent the physical limitations of variant library generation. Models trained on datasets of binding and non-binding variants can learn the high-dimensional features that govern protein-protein interactions and predict the sequences of variants with desirable properties outside of those sequences screened within the experimental datasets. However, there are no reports of machine-guided engineering of antibody constant regions. Accordingly, there is still a need in the art for novel methods for engineering the constant region of antibodies.

[0012] SUMMARY OF THE INVENTION

[0013] The present invention is characterized in the herein provided embodiments and claims. In particular, the present invention relates, inter alia, to the following embodiments:

[0014] 1. A method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of:

[0015] a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface;

[0016] b) contacting the cells in step (a) with an Fc receptor;

[0017] c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor; and d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region.

[0018] 2. The method according to embodiment 1, wherein steps (b) - (d) are repeated for one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B or FcRn.

[0019] 3. The method according to embodiment 1 or 2, wherein in step (d), the at least one cell population, or a sample thereof, is subjected to deep sequencing.

[0020] 4. The method according to embodiment 3, wherein deep sequencing comprises a step of determining the frequency of individual mutations in the engineered antibody Fc region in a population.

[0021] 5. The method according to embodiment 4, wherein the impact of an individual mutation on the binding of the engineered antibody Fc region to the Fc receptor is determined based on the frequency of said individual mutation in the population.

[0022] 6. The method according to any one of embodiments 1 to 5, wherein in step (d) mutations in the engineered antibody Fc region are identified that (i) confer improved binding to the Fc receptor, (ii) do not affect binding to the Fc receptor, and / or (iii) confer reduced binding to the Fc receptor.

[0023] 7. The method according to any one of embodiments 1 to 6, wherein the sequence information obtained in step (d) is used to train a machine learning model.

[0024] 8. The method according to embodiment 7, wherein the machine learning model is used to (a) predict the Fc receptor binding properties of an engineered antibody Fc region; and / or (b) generate an engineered antibody Fc region having desired Fc receptor binding properties.

[0025] 9. The method according to any one of embodiments 1 to 8, wherein each cell comprised in the plurality of cells displays an engineered antibody Fc region comprising at least one mutation

[0026] 10. The method according to any one of embodiments 1 to 9, wherein the plurality of cells has been obtained by introducing a DNA library encoding engineered antibody Fc regions into the cells.

[0027] 11. The method according to embodiment 10, wherein the DNA library encodes a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the library comprises a single mutation or a combination of two or more single mutations.

[0028] 12. The method according to embodiment 10 or 11, wherein the DNA library is a deep mutational scanning (DMS) library encoding a plurality of engineered antibody Fc regions, wherein each engineered antibody Fc region encoded in the DMS library comprises a single mutation, preferably wherein the DMS library encodes at least 70%, at least 80% or at least 90% of all possible single mutants of the antibody Fc region.

[0029] The method according to embodiment 10 or 11, wherein the DNA library is a combinatorial library encoding a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the combinatorial library comprises two or more mutations.

[0030] The method according to embodiment 13, wherein the two or more mutations are at positions that have been previously identified as being involved in the binding of one or more Fc receptor.

[0031] The method according to embodiment 13 or 14, wherein obtaining the combinatorial library comprises a step of assembling two or more DNA fragments encoding parts of an engineered antibody Fc region, preferably wherein each of the DNA fragments comprises at least one degenerate codon.

[0032] The method according to embodiment 15, wherein one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a first Fc receptor, and another one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a second Fc receptor.

[0033] The method according to any one of embodiments 13 to 16, wherein the two or more mutations are in positions 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446, preferably in positions 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering) of the antibody Fc region. The method according to any one of embodiments 1 to 17, wherein the ability of an engineered antibody Fc region to bind to the Fc receptor is determined by comparison with a reference antibody Fc region.

[0034] The method according to any one of embodiments 1 to 18, wherein cells in step (c) are separated by fluorescence-activated cell sorting (FACS).

[0035] The method according to embodiment 19, wherein the antibody Fc regions are labelled with a first fluorescent dye and the Fc receptor is labelled with a second fluorescent dye.

[0036] The method according to any one of embodiments 1 to 20, wherein the engineered antibody Fc regions have been derived from an aglycosylated human IgGl antibody Fc domain.

[0037] The method according to embodiment 21, wherein the aglycosylated human IgGl antibody Fc domain comprises mutations T299A / K326I / A327Y / L328G (EU numbering).

[0038] The method according to any one of embodiments 1 to 22, wherein the engineered antibody Fc regions are displayed on the surface of a yeast cell.

[0039] A method for identifying engineered antibody Fc regions with altered Fc receptor binding characteristics, the method comprising the steps of:

[0040] a) identifying amino acid residues of an antibody Fc region that are involved in binding to one or more Fc receptors;

[0041] b) introducing mutations in at least two positions of the antibody Fc region encoding the amino acid residues identified in step (a) to obtain an engineered antibody Fc region;

[0042] c) testing the engineered antibody Fc region obtained in step (b) for binding to one or more Fc receptors; and

[0043] d) determining the engineered antibody Fc region to have altered Fc receptor binding characteristics based on the outcome of the testing in step (c). 25. The method according to embodiment 24, wherein the amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors are identified using deep mutational scanning and / or alanine scanning.

[0044] 26. The method according to embodiment 24 or 25, wherein the engineered antibody Fc region obtained in step (b) is encoded by a combinatorial library encoding a plurality of engineered antibody Fc regions.

[0045] 27. The method according to embodiment 26, wherein the combinatorial library is obtained by assembling two or more DNA fragments encoding parts of an engineered antibody Fc region, preferably wherein each of the DNA fragments comprises at least one mutation, more preferably wherein each of the DNA fragments comprises at least one degenerate codon.

[0046] 28. The method according to any one of embodiments 24 to 27, wherein one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a first Fc receptor, and another one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a second Fc receptor.

[0047] 29. The method according to any one of embodiments 24 to 28, wherein mutations are introduced in at least two of the following positions of the antibody Fc region: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446, preferably in positions 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering). The method according to any one of embodiments 24 to 29, wherein testing the engineered antibody Fc region for binding to one or more Fc receptors involves a step of contacting the engineered antibody Fc region with one or more Fc receptors.

[0048] The method according to any one of embodiments 24 to 30, wherein the binding of the engineered antibody Fc region to one or more Fc receptors is tested by flow cytometry.

[0049] The method according to any one of embodiments 24 to 31, comprising an additional step of determining the amino acid sequence of the engineered antibody Fc region, preferably by sequencing, more preferably by deep sequencing.

[0050] A method for predicting the Fc receptor binding properties of an engineered antibody Fc region, the method comprising the steps of:

[0051] a) training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors;

[0052] b) providing a nucleic acid sequence or an amino acid sequence of the engineered antibody Fc region to the trained machine learning model; and

[0053] c) obtaining, from the machine learning model, the predicted Fc receptor binding properties of said engineered antibody Fc region provided in step (b).

[0054] A method for generating an engineered antibody Fc region having desired Fc receptor binding properties, the method comprising the steps of:

[0055] a) training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors; and

[0056] b) utilizing the trained machine learning model to generate a nucleic acid sequence or an amino acid sequence encoding an engineered antibody Fc region with the desired Fc receptor binding properties.

[0057] The method according to embodiment 33 or 34, wherein the binding data is obtained using the method according to any one of embodiments 1 to 32.

[0058] 36. The method according to any one of embodiments 33 to 35, wherein the machine learning model is trained with binding data that has been obtained with at least two different Fc receptors.

[0059] Accordingly, in a particular embodiment, the invention relates to a method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of: (a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface; (b) contacting the cells in step (a) with an Fc receptor; (c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor; and (d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region.

[0060] That is, the method according to the invention may be used to identify antibody Fc regions with specific binding properties to one or more Fc receptors. For instance, the method may be employed, without limitation, to identify antibody Fc regions that exhibit enhanced binding affinity to a first Fc receptor while demonstrating reduced binding affinity to a second Fc receptor. For instance, it is demonstrated in Example 3 that the method of the invention may be used to identify positions in the antibody Fc region that can alter (increase and / or decrease) the binding affinity to nine different Fc receptors (see Figs. 6 and 7).

[0061] Alternatively, the purpose of the method of the invention may be rephrased as a "method for identifying antibody Fc regions having desired Fc receptor binding characteristics" or as a "method for identifying mutations in an antibody Fc region that affect binding of an Fc receptor".

[0062] The term "antibody Fc region" refers to the constant region of an antibody's heavy chain that is responsible for binding to Fc receptors on immune cells and interacting with various components of the immune system. This region does not participate in antigen binding but plays a crucial role in mediating immune effector functions such as antibody-dependent cellular cytotoxicity (ADCC), complement activation, and phagocytosis. The Fc region is critical for the antibody's ability to communicate with and recruit immune cells to target and eliminate pathogens or diseased cells.

[0063] In IgG antibodies, the Fc region comprises the CH2 and CH3 domains of the heavy chains. This region is highly conserved and is responsible for binding to Fc gamma receptors (FcyRs) on the surface of immune cells such as natural killer (NK) cells, macrophages, and neutrophils. Additionally, the Fc region interacts with the neonatal Fc receptor (FcRn), which plays a crucial role in regulating the antibody's half-life in circulation. Modifications to the Fc region can be engineered to enhance or diminish these interactions, thereby modulating the antibody's effector functions and pharmacokinetic properties. That is, in certain embodiments, the antibody Fc region is the antibody Fc region of an IgG-type antibody, preferably wherein the antibody Fc region comprises the CH2 and CH3 domains of the heavy chains.

[0064] An IgG-type antibody is a Y-shaped immunoglobulin G molecule composed of two identical heavy chains and two identical light chains. This antibody type, the most abundant in human serum, plays a critical role in recognizing and neutralizing pathogens and is classified into four subclasses (IgG 1, lgG2, lgG3, and lgG4), each with unique structural and functional properties, making them essential for therapeutic and diagnostic applications. Within the present invention, the IgG-type antibody may be an IgGl, lgG2, lgG3, or lgG4 antibody. In a particularly preferred embodiment, the antibody Fc region is the antibody Fc region of an IgGl antibody, more preferably a human IgGl antibody.

[0065] In certain embodiments, the antibody Fc region used in the method of the invention is a dimeric fragment of an IgG-type antibody comprising the CH2 and CH3 domains of the heavy chains, more preferably a dimeric fragment of an IgGl antibody comprising the CH2 and CH3 domains of the heavy chains, most preferably a dimeric fragment of a human IgGl antibody comprising the CH2 and CH3 domains of the heavy chains.

[0066] In certain embodiments, the antibody Fc region used in the method of the invention is a dimeric fragment of a human IgGl antibody comprising residues E216 to K447 (EU numbering) of the heavy chains, as shown in SEQ ID NO:1 below. Fc region of human IgGl antibody (wild type) (SEQ ID NO:1):

[0067] EPKSCDKTHTCPPCPAPELLGGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVH NAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTLPPSRD ELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCS VMHEALHNHYTQKSLSLSPGK

[0068] The starting point for the engineered antibody Fc regions that are screened in the method of the invention may be a wild type antibody Fc region, such as the antibody Fc region of a human IgGl antibody as shown above in SEQ ID NO:1. In such embodiments, the plurality of cells used in the first step of the method of the invention may be obtained by introducing mutations into said wild type antibody Fc region.

[0069] However, the starting point for the engineered antibody Fc regions that are screened in the method of the invention may be an antibody Fc region that has been modified. In a preferred embodiment, the antibody Fc region used as starting point in the method of the invention has been modified to prevent or remove glycosylation of the antibody Fc region. Glycosylation of the antibody Fc region may be prevented by introducing specific mutations into the antibody Fc region that eliminate the glycosylation site or prevent glycosylation of the antibody.

[0070] In IgG antibodies, in particular human IgGl antibodies, glycosylation takes place at residue N297 (EU numbering). Accordingly, introducing a point mutation at position N297 may eliminate the glycosylation site. Thus, in certain embodiments, the antibody Fc region used as starting point in the method of the invention may be the antibody Fc region of an IgG antibody comprising a mutation at position N297 of the heavy chain (EU numbering), preferably of a human IgG antibody comprising a mutation at position N297 of the heavy chain (EU numbering), more preferably of a human IgGl antibody comprising a mutation at position N297 of the heavy chain (EU numbering). In certain embodiments, the mutation is N297Q.

[0071] In certain embodiments, the antibody Fc region used as starting point in the method of the invention may comprise a mutation that prevents glycosylation of the antibody. In IgG-type antibodies, mutations at position T299 of the heavy chain (EU numbering) have been demonstrated to prevent glycosylation of the antibody Fc region. Thus, in certain embodiments, the antibody Fc region used as starting point in the method of the invention may be the antibody Fc region of an IgG antibody comprising a mutation at position T299 of the heavy chain (EU numbering), preferably of a human IgG antibody comprising a mutation at position T299 of the heavy chain (EU numbering), more preferably of a human IgGl antibody comprising a mutation at position T299 of the heavy chain (EU numbering). Most preferably, the antibody Fc region comprises the mutation T299A in the heavy chain (EU numbering).

[0072] In certain embodiments, the antibody Fc region further comprises a mutation at position S298 of the heavy chain (EU numbering). In certain embodiment, the mutation is S298G.

[0073] Alternatively, the antibody Fc region used as starting point in the method of the invention may be enzymatically deglycosylated using enzymes known in the art, such as PNGase F.

[0074] It is known that deglycosylated or aglycosylated antibody Fc regions exhibit altered binding to at least some Fc receptors compared to their glycosylated counterparts. Therefore, the deglycosylated or aglycosylated antibody Fc regions disclosed above may comprise further mutations that restore near wild-type levels of binding to canonical Fc receptors. In particular the mutations K326I / A327Y / L328G (EU numbering) have been shown to restore near wild type levels of binding to canonical Fc receptors in aglycosylated human IgGl antibody Fc regions. Thus, in certain embodiments, the antibody Fc region used as starting point in the method of the invention may be a deglycosylated or aglycosylated antibody Fc region of an IgG antibody comprising one or more mutations at position K326, A327 and / or L328 of the heavy chain (EU numbering), preferably a human IgG antibody comprising one or more mutations at position K326, A327 and / or L328 of the heavy chain (EU numbering), more preferably a human IgGl antibody comprising one or more mutations at position K326, A327 and / or L328 of the heavy chain (EU numbering).

[0075] In a preferred embodiment, the antibody Fc region used as starting point in the method of the invention may be an aglycosylated antibody Fc region of a human IgG antibody comprising one or more of the mutations K326I, A327Y and / or L328G of the heavy chain (EU numbering). In a more preferred embodiment, the antibody Fc region used as starting point in the method of the invention may be an aglycosylated antibody Fc region of a human IgGl antibody comprising one or more of the mutations K326I, A327Y and / or L328G of the heavy chain (EU numbering). In a most preferred embodiment, the antibody Fc region used as starting point in the method of the invention may be the antibody Fc region of a human IgGl antibody comprising the mutations T299A and one or more of the mutations K326I, A327Y and / or L328G of the heavy chain (EU numbering).

[0076] Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein the engineered antibody Fc regions have been derived from an aglycosylated human IgGl antibody Fc domain.

[0077] In a particular embodiment, the invention relates to the method according to the invention, wherein the aglycosylated human IgGl antibody Fc domain comprises mutations T299A / K326I / A327Y / L328G (EU numbering).

[0078] The sequence of a human IgGl antibody Fc region comprising the mutations T299A / K326I / A327Y / L328G (EU numbering) is listed in SEQ ID NO:2.

[0079] Human IgGl Fc region comprising mutationsT299A / K326l / A327Y / L328G (EU numbering) (SEQ ID NO:2):

[0080] EPKSCDKTHTCPPCPAPELLGGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVH NAKTKPREEQYNSAYRVVSVLTVLHQDWLNGKEYKCKVSNIYGPAPIEKTISKAKGQPREPQVYTLPPSRD ELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCS VMHEALHNHYTQKSLSLSPGK

[0081] Within the present invention, the antibody Fc region, such as any of the antibody Fc regions described herein above, needs to be displayed on the surface of a cell, such that it is accessible to exogenously added Fc receptors or soluble versions thereof. The term "displayed on the surface of a cell" refers to the presentation of molecules, such as proteins, peptides, or other biomolecules, on the exterior membrane or the cell wall of a cell, where they are accessible for interaction with external entities such as other cells, antibodies, or ligands. This surface display can be achieved through various mechanisms, includi ng genetic engineering, where the molecule of interest is fused to an anchor sequence, or through natural cellular processes that transport and integrate the molecule into the cell membrane or cell wall. Methods for displaying proteins on the surfaces of different types of cells are known in the art and available to the person skilled in the art.

[0082] In a preferred embodiment, the antibody Fc region, such as any of the antibody Fc regions described herein above, is displayed on the surface of a yeast cell, i.e. by yeast display. The term "yeast display" refers to a molecular biology technique in which proteins, peptides, or other biomolecules are genetically fused to cell wall proteins of yeast cells, resulting in their presentation on the exterior surface of the yeast. This method allows for the high-throughput screening and selection of molecules based on their binding interactions with external entities such as antibodies, ligands, or other target molecules. Yeast display is widely used in protein engineering, affinity maturation, and therapeutic discovery, providing a powerful platform for identifying and optimizing biomolecules with desired properties. Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein the engineered antibody Fc regions are displayed on the surface of a yeast cell.

[0083] In a preferred embodiment, yeast surface display is achieved by fusing the antibody Fc region to the yeast cell wall protein AGA2. AGA2 is a yeast cell wall protein that is commonly used in yeast display systems due to its ability to anchor proteins or peptides to the exterior of the yeast cell. It is part of the AGA1-AGA2 system, where AGA2 is covalently linked to the cell wall through its interaction with AGA1, another cell wall protein. By genetically fusing the protein or peptide of interest to AGA2, researchers can effectively display these molecules on the surface of the yeast cell. This system facilitates the high-throughput screening and selection of displayed molecules based on their binding interactions. The fusion of an antibody Fc region to the AGA2 protein is described in Example 1 below and has been previously described in the art (Powers et al., J Immunol Methods (2001) 251(l-2):123-35; Wozniak-Knopp et al., Protein Eng Des Sei (2010) 23(4):289-97; Zheng et al., J Immunol (1999) 163(7):4041-8). The construct comprising the antibody Fc region fused to the yeast AGA2 protein may comprise one or more peptide tags that may facilitate detecting cells that efficiently express the antibody Fc region on the cell surface. The peptide tag may be a commonly used epitope tag such as FLAG, HA, Myc, or His-tag, which can be recognized by specific antibodies or other detection reagents. These tags enable the use of flow cytometry, immunofluorescence, or other analytical techniques to identify and quantify the expression levels of the antibody Fc region on the yeast cell surface. Incorporating such peptide tags into the construct enhances the ability to monitor and optimize the display system, ensuring robust and reliable expression fordownstream applications.

[0084] In certain embodiments, the construct comprising the antibody Fc region fused to the yeast AGA2 protein comprises a FLAG tag and / or an HA tag.

[0085] A FLAG tag is a short, hydrophilic peptide sequence used as an epitope tag for the detection, purification, and characterization of recombinant proteins. The most commonly used FLAG tag sequence is DYKDDDDK (SEQ ID NO:3). An HA tag, or hemagglutinin tag, is a short peptide sequence derived from the influenza virus hemagglutinin protein. The most commonly used HA tag sequence is YPYDVPDYA (SEQ ID NO:4).

[0086] These tags are widely used in molecular biology and biotechnology for the detection, purification, and characterization of recombinant proteins. The FLAG tag and HA tag can be recognized by specific monoclonal antibodies, such as anti-FLAG or anti-HA antibodies, which allow for the convenient detection and analysis of tagged proteins through various techniques, including Western blotting, immunoprecipitation, immunofluorescence, and flow cytometry. The small size of the FLAG tag and HA tag minimize their impact on the structure and function of the fused protein, making them a versatile and valuable tool in protein research and development.

[0087] In certain embodiments, the construct comprising the antibody Fc region fused to the yeast AGA2 protein comprises the structure AGA2-HA-Fc-FLAG. Exemplary constructs that may be used in the method of the invention are shown below: AGA2-HA-Fc(human IgGl wildtype)-FLAG (SEQ ID N0:5):

[0088] MQLLRCFSIFSVIASVLAQELTTICEQIPSPTLESTPYSLSTTTILANGKAMQGVFEYYKSVTFVSNCGSHPST TSKGSPINTQYVFKLLQASGGGGSGGGGSGGGGSYPYDVPDYAGSEPKSCDKTHTCPPCPAPELLGGPSV FLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQYNSTYRVVSVLTVLH QDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTLPPSRDELTKNQVSLTCLVKGFYPSDIAVEW ESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNHYTQKSLSLSPGKGSD YKDDDDK

[0089] AGA2-HA-Fc(human IgGl T299A / K326I / A327Y / L328G)-FLAG (SEQ ID N0:6):

[0090] MQLLRCFSIFSVIASVLAQELTTICEQIPSPTLESTPYSLSTTTILANGKAMQGVFEYYKSVTFVSNCGSHPST TSKGSPINTQYVFKLLQASGGGGSGGGGSGGGGSYPYDVPDYAGSEPKSCDKTHTCPPCPAPELLGGPSV FLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQYNSAYRVVSVLTVLH QDWLNGKEYKCKVSNIYGPAPIEKTISKAKGQPREPQVYTLPPSRDELTKNQVSLTCLVKGFYPSDIAVEW ESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNHYTQKSLSLSPGKGSD YKDDDDK

[0091] In the first step of the method according to the invention, a plurality of cells displaying engineered antibody Fc regions on their cell surface is provided. This plurality of cells displaying engineered antibody Fc regions on their cell surface is preferably a plurality of cells of the same type, wherein at least some of the cells display an engineered antibody Fc region on their cell surface.

[0092] The plurality of cells may comprise any number of cells that can be conveniently handled in the method of the invention. In certain embodiments, the plurality of cells comprises at least 100, at least 1,000, at least 10,000, at least 100,000, at least 500,000, at least 1,000,000, at least 5,000,000 or at least 10,000,000 cells displaying antibody Fc regions on their cell surface.

[0093] At least one cell in the plurality of cells displaying engineered antibody Fc regions on their cell surface expresses an engineered antibody Fc region on its cell surface. That is, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be a mixture of cells displaying engineered antibody Fc regions and cells displaying the antibody Fc region that served as starting point for generating the engineered antibody Fc regions. However, it is preferred that all cells, or essentially all cells, in the plurality of cells display engineered antibody Fc regions on their cell surface.

[0094] An "engineered antibody Fc region" as used herein is an antibody Fc region comprising at least one genetic modification relative to the antibody Fc region that was used as starting point to generate the engineered antibody Fc region. The genetic modification may be, without limitation, an amino acid substitution, deletion and / or insertion. In certain embodiments, the genetic modification is an amino acid substitution. Amino acid substitutions may also be referred to herein as mutations or point mutations.

[0095] As discussed herein above, the starting point for generating the engineered antibody Fc regions may be a wild type antibody Fc region or an antibody Fc region that was previously engineered. Preferably, the starting point for generating the engineered antibody Fc regions may be an antibody Fc region that was engineered to prevent glycosylation of the antibody Fc region, as described herein above. In a preferred embodiment, the starting point for generating the engineered antibody Fc regions may be a human IgGl antibody Fc region comprising the mutations T299A / K326I / A327Y / L328G (EU numbering).

[0096] Preferably, at least 50%, at least 60%, at least 70%, at least 80%, at least 90%, at least 95%, at least 98% or at least 99% of the cells comprised in the plurality of cells display engineered antibody Fc regions on their cell surface, i.e., antibody Fc regions comprising at least one genetic modification relative to the antibody Fc region that was used as starting point. In a preferred embodiment, the invention relates to the method according to the invention, wherein each cell comprised in the plurality of cells displays an engineered antibody Fc region comprising at least one mutation.

[0097] Thus, in a particular embodiment, the invention relates to a method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of: (a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface, wherein each cell comprised in the plurality of cells displays an engineered antibody Fc region comprising at least one mutation; (b) contacting the cells in step (a) with an Fc receptor; (c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor; and (d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region.

[0098] In certain embodiments, the method according to the invention may be performed with at least 1,000,000, at least 2,000,000, at least 3,000,000, at least 4,000,000, at least 5,000,000, at least 6,000,000, at least 7,000,000 at least 8,000,000, at least 9,000,000 or at least 10,000,000 cells displaying engineered antibody Fc regions on their cell surface. In certain embodiments, the method according to the invention may be performed with between 1,000,000 and 1,000,000,000 cells, preferably with between 5,000,000 and 100,000,000 cells, more preferably between 5,000,000 and 10,000,000 cells displaying engineered antibody Fc regions on their cell surface.

[0099] Preferably, the plurality of cells used in the method according to the invention display at least 2, at least 10, at least 50, at least 100, at least 500, at least 1,000, at least 5,000, at least 10,000, at least 100,000, at least 1,000,000 different engineered antibody Fc regions.

[0100] The different engineered antibody Fc regions displayed by the plurality of cells are preferably variants of an initial antibody Fc region and may comprise one or more mutations relative to said initial antibody Fc region. In a preferred embodiment, the different engineered antibody Fc regions displayed by the plurality of cells are variants of the human IgGl antibody Fc region comprising the mutations T299A / K326I / A327Y / L328G (EU numbering) and comprise one or more additional mutations.

[0101] In certain embodiments, the engineered antibody Fc regions displayed by the plurality of cells may be variants of an initial antibody Fc region, such as the human IgGl antibody Fc region comprising the mutations T299A / K326I / A327Y / L328G (EU numbering), wherein all variants comprise a single mutation relative to the initial antibody Fc region. Such cell populations may be obtained by deep mutational scanning, as explained in more detail below. In certain embodiments, the different engineered antibody Fc regions displayed by the plurality of cells may be variants of an initial antibody Fc region, such as the human IgGl antibody Fc region comprising the mutations T299A / K326I / A327Y / L328G (EU numbering), wherein all variants comprise one or more mutations relative to the initial antibody Fc region. Such cell populations may be obtained, without limitation, with combinatorial DNA libraries, as explained in more detail below.

[0102] The plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained with any suitable method known in the art. Preferably, the plurality of cells displaying engineered antibody Fc regions on their cell surface is obtained by introducing a nucleic acid library encoding engineered antibody Fc regions into said cell.

[0103] Nucleic acid libraries, such as DNA libraries, encoding engineered antibody Fc regions can be generated by introducing one or more mutations into a template nucleic acid sequence encoding an initial antibody Fc region. These mutations may be introduced using various methods such as site-directed mutagenesis, error-prone PCR, or DNA shuffling, which allow for the systematic or random alteration of specific amino acids within the Fc region. Once the mutations are introduced, the resulting nucleic acid library represents a diverse pool of engineered Fc regions with potentially enhanced or altered properties. This library may then be cloned into an appropriate expression vector and introduced into host cells, such as yeast cells, to create a plurality of cells displaying the engineered Fc regions on their surface.

[0104] Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein the plurality of cells has been obtained by introducing a DNA library encoding engineered antibody Fc regions into the cells.

[0105] The skilled person is aware of various methods for introducing a DNA library into a cell, particularly a yeast cell. For example, transformation techniques such as electroporation, which uses an electrical field to increase the permeability of the cell membrane, can be employed to facilitate the uptake of the DNA library. Alternatively, chemical transformation methods, such as the lithium acetate method, can be used to induce competence in yeast cells, allowing them to take up exogenous DNA. Additionally, biolistic transformation, which involves the delivery of DNA-coated particles into cells using a gene gun, can also be utilized.

[0106] It is preferred herein that the DNA library encodes a plurality of engineered antibody Fc regions. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein the DNA library encodes a plurality of engineered antibody Fc regions.

[0107] That is, the DNA library may encode at least 2, at least 10, at least 50, at least 100, at least 500, at least 1,000, at least 5,000, at least 10,000, at least 100,000, at least 1,000,000, at least 10,000,000, at least 100,000,000 of at least 500,000,000 different engineered antibody Fc regions.

[0108] In certain embodiments, the DNA library may encode a plurality of engineered antibody Fc regions, wherein each engineered antibody Fc region comprises a single point mutation relative to the antibody Fc region that was used as starting point for generating the engineered antibody Fc regions. In certain embodiments, the DNA library may encode a plurality of engineered antibody Fc regions, wherein each engineered antibody Fc region comprises one or more mutations relative to the antibody Fc region that was used as starting point for generating the engineered antibody Fc regions. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein each engineered antibody Fc region encoded in the library comprises a single mutation or a combination of two or more single mutations.

[0109] In certain embodiments, the plurality of cells used in the method of the invention may be obtained by introducing a deep mutational scanning (DMS) library into a population of cells. The term "deep mutational scanning (DMS)" refers to a high-throughput experimental technique used to systematically introduce and evaluate the effects of a comprehensive set of mutations across a protein or protein domain. This method involves creating a library of variants, each containing specific mutations, and subjecting these variants to selective pressures to assess their functional impact. By coupling the mutational data with high-throughput sequencing and analysis, DMS provides detailed insights into the structurefunction relationships, stability, and activity of the protein, facilitating the identification and optimization of variants with desired properties for therapeutic, diagnostic, or industrial applications.

[0110] A DMS library may be generated as explained in Example 2. That is, mutagenesis may be achieved using custom oligo pools containing degenerate NNK codons tiled across each position of the antibody Fc domain. The "NNK" codon scheme, where "N" represents any nucleotide (A, T, C, or G) and "K" represents G or T, allows for the encoding of all 20 amino acids while minimizing the occurrence of stop codons. By incorporating NNK codons at desired positions within the oligos, a diverse pool of DNA sequences can be generated, each containing different amino acid substitutions at the targeted sites. These mutated sequences may then be transformed into E. coli cells and the DMS library may be extracted from the E. coli cells.

[0111] Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein the DNA library is a deep mutational scanning (DMS) library encoding a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the DMS library comprises a single mutation.

[0112] More specifically, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DMS library into a plurality of cells, wherein at least two codons in the nucleic acid sequence encoding the antibody Fc region have been mutagenized using degenerate codons, preferably degenerate NNK codons.

[0113] In certain embodiments, the DMS library may be generated by nicking mutagenesis as described by Wrenbeck et al. (Nat Methods (2016) 13(ll):928-930): In short, nicking mutagenesis is a one-pot saturation mutagenesis method performed on double-stranded plasmid DNA (dsDNA) without the need for uracil-containing single-stranded DNA (ssDNA) templates. This technique involves the use of nicking endonucleases, such as Nt.BbvCI and Nb.BbvCI, to create single-strand breaks at specific sites within the plasmid. The nicked strand is then selectively degraded by exonuclease III, generating an ssDNA template. Mutagenic oligonucleotides are annealed to the ssDNA template, and the mutant strand is synthesized using a DNA polymerase. The newly synthesized strand is then ligated to form a double- stranded plasmid with the desired mutations. This process can be repeated to introduce multiple mutations, allowing for the creation of comprehensive single-site or multi-site saturation mutagenesis libraries.

[0114] Alternatively, DMS libraries may be generated wherein two or more specific positions within the protein are simultaneously mutagenized using dual-site or multi-site deep mutational scanning. This approach involves designing oligonucleotides with one or more degenerate codons at the targeted positions, allowing for the exploration of the combined effects of mutations at these sites. Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein the DNA library is a deep mutational scanning (DMS) library encoding a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the DMS library comprises a two or more mutations.

[0115] In certain embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DMS library into a plurality of cells, wherein at least 10%, at least 20%, at least 30% at least 40%, at least 50%, at least 60%, at least 70%, at least 80, at least 90%, at least 95% or all codons in the nucleic acid sequence encoding the antibody Fc region have been mutagenized using degenerate codons, preferably degenerate NNK codons.

[0116] In certain embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DMS library into a plurality of cells, wherein at least 10%, at least 20%, at least 30% at least 40%, at least 50%, at least 60%, at least 70%, at least 80, at least 90%, at least 95% or all codons in the nucleic acid sequence encoding residues 216 to 447 (EU numbering) of the human IgGl antibody Fc region have been mutagenized using degenerate codons, preferably degenerate NNK codons.

[0117] Instead of performing deep mutational scanning of the entire antibody Fc region, deep mutational scanning may also be restricted to specific parts of the antibody Fc region that is known or suspected to be involved in Fc receptor binding. For example, it has been reported in Examples 3 and 4 that most of the residues of the human IgGl Fc region involved in Fc receptor binding are residues 232 - 368 (EU numbering).

[0118] Thus, in certain embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DMS library into a plurality of cells, wherein at least 10%, at least 20%, at least 30% at least 40%, at least 50%, at least 60%, at least 70%, at least 80, at least 90%, at least 95% or all codons in the nucleic acid sequence encoding residues 232- 368 (EU numbering) of the human IgGl antibody Fc region have been mutagenized using degenerate codons, preferably degenerate NNK codons.

[0119] It is to be understood that, although theoretically possible when using degenerate NNK codons, obtaining all possible single mutants may not be practically achievable; see, e.g., Fig.4. Thus, in certain embodiments, the invention relates to the method according to the invention, wherein the DMS library encodes at least 70%, at least 80%, at least 90%, at least 95 or at least 98% of all possible single mutants.

[0120] In embodiments where the human IgGl antibody Fc region includes the T299A mutation, mutagenesis of this specific position may be avoided to prevent the formation of glycosylated antibody Fc regions.

[0121] In certain embodiments, the invention relates to the method according to the invention, wherein the plurality of cells display at least 10%, at least 20%, at least 30% at least 40%, at least 50%, at least 60%, at least 70%, at least 80, at least 90%, at least 95%, or at least 98% of all possible single mutation variants of an antibody Fc region. In certain embodiments, the invention relates to the method according to the invention, wherein the plurality of cells display at least 10%, at least 20%, at least 30% at least 40%, at least 50%, at least 60%, at least 70%, at least 80, at least 90%, at least 95%, or at least 98% of all single mutation variants in residues 216 to 447 (EU numbering) of the human IgGl antibody Fc region. In certain embodiments, the invention relates to the method according to the invention, wherein the plurality of cells display at least 10%, at least 20%, at least 30% at least 40%, at least 50%, at least 60%, at least 70%, at least 80, at least 90%, at least 95%, or at least 98% of all single mutation variants in residues 232- 368 (EU numbering) of the human IgGl antibody Fc region. Preferably, the initial antibody Fc region used as starting point is the human IgGl antibody Fc region comprising the mutations T299A / K326I / A327Y / L328G (EU numbering).

[0122] In certain embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446 (all according to EU numbering).

[0123] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446 (all according to EU numbering).

[0124] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyRl binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 233-238, 263, 265-266, 270, 273, 301, 325, 328-329, 332, 334-336, 338, 362, 366, 405-407, 428 (all according to EU numbering).

[0125] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 233-238, 263, 265-266, 270, 273, 301, 325, 328-329, 332, 334-336, 338, 362, 366, 405-407, 428 (all according to EU numbering).

[0126] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyR2A-H binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 237-238, 251, 262-263, 265-266, 275-276, 278, 287, 292, 297, 301, 304, 307, 314, 322, 324-325, 328-329, 332, 334-336, 338, 341, 346, 374, 378, 407, 428-429 (all according to EU numbering).

[0127] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 237-238, 251, 262-263, 265-266, 275-276, 278, 287, 292, 297, 301, 304, 307, 314, 322, 324-325, 328-329, 332, 334-336, 338, 341, 346, 374, 378, 407, 428-429 (all according to EU numbering).

[0128] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 233-238, 263, 265-266, 270, 273, 301, 325, 328-329, 332, 334-336, 338, 362, 366, 405-407, 428 (all according to EU numbering).

[0129] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyR2A-R binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 238, 250-251, 263, 265-266, 278, 287, 297, 301-302, 304, 310, 313, 324-325, 328-329, 332-334, 336, 338, 374, 407 (all according to EU numbering).

[0130] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 238, 250-251, 263, 265-266, 278, 287, 297, 301-302, 304, 310, 313, 324-325, 328-329, 332-334, 336, 338, 374, 407 (all according to EU numbering).

[0131] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyR2B binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 237-238, 251, 257, 263, 265-266, 270, 273, 297-298, 301-302, 324-326, 328-329, 332, 334, 336, 338, 366, 405-407 (all according to EU numbering).

[0132] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 237-238, 251, 257, 263, 265-266, 270, 273, 297-298, 301-302, 324-326, 328-329, 332, 334, 336, 338, 366, 405-407 (all according to EU numbering).

[0133] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyR3A-F binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering). Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, 338 (all according to EU numbering).

[0134] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyR3A-V binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 237-238, 242, 251, 263, 265-266, 270, 297, 301, 304, 307, 312-313, 324-325, 328-329, 332, 334, 336, 338, 407 (all according to EU numbering).

[0135] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 237-238, 242, 251, 263, 265-266, 270, 297, 301, 304, 307, 312-313, 324-325, 328-329, 332, 334, 336, 338, 407 (all according to EU numbering).

[0136] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcyR3B binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 232-238, 257, 263, 265-267, 269-273, 292, 295, 297, 300-302, 322, 324-329, 331-339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 376, 378, 380, 395, 405-407, 409, 428, 431-432, 435 (all according to EU numbering).

[0137] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 232-238, 257, 263, 265-267, 269-273, 292, 295, 297, 300-302, 322, 324-329, 331-339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 376, 378, 380, 395, 405-407, 409, 428, 431-432, 435 (all according to EU numbering).

[0138] In certain embodiments, the method according to the invention may be used to identify engineered antibody Fc regions with modified FcRn (pH6) binding properties. In such embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding human IgGl antibody Fc regions into a plurality of cells, wherein one or more of the following amino acid positions of the human IgGl antibody Fc region have been mutagenized, preferably with a degenerate codon, more preferably with a degenerate NNK codon: 250-254, 257, 288, 310, 314, 338, 428, 435, 446 (all according to EU numbering).

[0139] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may comprise mutations in one or more of the following amino acid positions: 250-254, 257, 288, 310, 314, 338, 428, 435, 446 (all according to EU numbering).

[0140] In a particular embodiment, the invention relates to the method according to the invention, wherein each cell in the plurality of cells displays an engineered antibody Fc region, and wherein each engineered antibody Fc region comprises a (different) point mutation.

[0141] Such cell populations comprising a plurality of single mutant variants of an antibody Fc region, preferably wherein the single mutations span the entire antibody Fc region, may be used to identify specific positions and / or mutations that affect binding of the antibody Fc region to one or more Fc receptors.

[0142] Instead of introducing single mutations into an antibody Fc region as described above, multiple mutations may be introduced simultaneously into an antibody Fc region to identify combinatorial effects of single mutations. Thus, in a particular embodiment, the invention relates to a method wherein the DNA library is a combinatorial library encoding a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the combinatorial library comprises two or more mutations. The term "combinatorial library" is to be understood in the broadest sense and encompasses any DNA library encoding engineered antibody Fc regions that comprise a combination of two or more mutations.

[0143] In certain embodiments, the engineered antibody Fc regions displayed by the plurality of cells may comprise on average about 2, 3, 4, 5, 6, 7, 8, 9, 10 or more mutations.

[0144] Two or more mutations may be introduced randomly into an antibody Fc region, for example by error-prone PCR.

[0145] However, it is preferred herein that the two or more mutations are introduced in a targeted manner. For example, two or more mutations may be introduced into an antibody Fc region at positions that are known or suspected to be involved in Fc receptor binding or have been identified in previous screening approaches, e.g., by using DMS as described herein, to be involved in Fc receptor binding. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein the two or more mutations are at positions that have been previously identified as being involved in the binding of one or more Fc receptor.

[0146] In certain embodiments, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding antibody Fc regions into a plurality of cells, wherein each antibody Fc region encoded in the DNA library comprises at least two mutations. These mutations are preferably introduced using degenerate codons.

[0147] In a preferred embodiment, the plurality of cells displaying engineered antibody Fc regions on their cell surface may be obtained by introducing a DNA library encoding engineered human IgG antibody Fc regions into a plurality of cells, wherein in each engineered human IgG antibody Fc region encoded in the DNA library, at least two of the following amino acid positions have been mutagenized, preferably with a degenerate codon: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446 (all according to EU numbering).

[0148] Accordingly, the engineered antibody Fc regions displayed by the cells comprised in the plurality of cells may be human IgG antibody Fc regions comprising mutations in at least two of the following amino acid positions: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446 (all according to EU numbering).

[0149] The degenerate codons used for mutagenesis of the targeted amino acid positions may be NNK codons encoding all 20 canonical amino acids. However, the degenerate codons may also be designed in a way that they only encode a subset of the 20 canonical amino acids. For example, the degenerate codons may be designed such that they only encode amino acids that have been identified in a prior DMS screen. The use of such degenerate codons is shown in Fig.9

[0150] The skilled person is aware of various methods for simultaneously introducing two or more point mutations into a single nucleic acid molecule in a site-directed manner. Techniques such as site-directed mutagenesis, oligonucleotide-directed mutagenesis, and CRISPR / Cas9-mediated genome editing allow for precise and controlled introduction of multiple mutations at specific sites within the nucleic acid sequence. Alternatively, oligonucleotides comprising two or more degenerate codons may be custom synthesized and ordered from commercial vendors.

[0151] In certain embodiments, a DNA library encoding a plurality of engineered antibody Fc regions may be obtained by site-directed mutagenesis. This technique involves the use of synthetic oligonucleotides designed to introduce specific mutations at targeted positions within the antibody Fc region. By employing site-directed mutagenesis, multiple variants of the antibody Fc region can be generated, each containing precise amino acid substitutions, insertions, or deletions. For example, multiple mutations may be simultaneously introduced by amplifying a nucleic acid encoding an antibody Fc region with one or more oligonucleotides comprising mutated or degenerate codons at specific positions.

[0152] It is preferred herein that the DNA library encoding engineered antibody Fc regions is assembled from two or more (pools of) DNA fragments, preferably wherein at least two of these DNA fragments encode a part of an engineered antibody Fc region. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein obtaining the combinatorial DMS library comprises a step of assembling two or more DNA fragments encoding parts of an engineered antibody Fc region.

[0153] That is, in certain embodiments, obtaining the DNA library encoding engineered antibody Fc regions may comprise a step of assembling 2, 3, 4, 5, or more (pools of) DNA fragments encoding parts of the antibody Fc region. The skilled person is aware of methods such as Golden Gate assembly, overlap extension PCR and Gibson assembly, which can be utilized for this purpose. Golden Gate assembly uses type Ils restriction enzymes to create specific overhangs that facilitate the directional and seamless assembly of multiple DNA fragments in a single reaction. Similarly, Gibson assembly involves the use of exonuclease, DNA polymerase, and DNA ligase to create overlapping ends, extend the DNA fragments, and seal the nicks, respectively, enabling the seamless joining of multiple DNA fragments in a single reaction. By employing these advanced assembly techniques, diverse libraries of engineered antibody Fc regions can be efficiently generated. The skilled person is capable of designing specific overhangs of the DNA fragments that enable seamless assembly of the fragments.

[0154] The two or more (pools of) DNA fragments encoding parts of an engineered antibody Fc region may be amplified by PCR or may be custom synthesized, for example as single stranded DNA fragments, as described in Example 4. Preferably, all DNA fragments in a pool of DNA fragments have the same length and encode the same fragment of the antibody Fc region, but may comprise one or more mutations.

[0155] To obtain a DNA library encoding a plurality of engineered antibody Fc regions, it is preferred that at least one of the DNA fragments from which the DNA library is assembled comprises a degenerate codon. More preferably, each of the DNA fragments from which the DNA library is assembled comprises a degenerate codon. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein obtaining the combinatorial DMS library comprises a step of assembling two or more DNA fragments encoding parts of an engineered antibody Fc region, wherein each of the DNA fragments comprises at least one degenerate codon.

[0156] In certain embodiments, obtaining the DNA library encoding engineered antibody Fc regions comprises a step of assembling two DNA fragments encoding parts of the antibody Fc region, wherein each fragment comprises at least one degenerate codon. In certain embodiments, obtaining the DNA library encoding engineered antibody Fc regions comprises a step of assembling two DNA fragments encoding parts of the antibody Fc region, wherein at least one or both of the fragments comprise two or more degenerate codons.

[0157] In certain embodiments, obtaining the DNA library encoding engineered antibody Fc regions comprises a step of assembling three DNA fragments encoding parts of the antibody Fc region, wherein each fragment comprises at least one degenerate codon. In certain embodiments, obtaining the DNA library encoding engineered antibody Fc regions comprises a step of assembling three DNA fragments encoding parts of the antibody Fc region, wherein at least one, at least two or all of the fragments comprise two or more degenerate codons.

[0158] In certain embodiments, the engineered antibody Fc regions encoded in the library are derived from human IgGl antibody Fc regions. In such embodiments, the DNA fragments encoding parts of the human IgGl antibody region may comprise degenerate codons at one or more of the following amino acid positions: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446 (all according to EU numbering); preferably at one or more of the following amino acid positions: 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering). In certain embodiments, the DNA library is obtained by assembling at least three DNA fragments encoding parts of a human IgGl antibody Fc region, wherein each DNA fragment comprises at least one degenerate codon. For example, a first DNA fragment may comprise degenerate codons at one or more of the amino acid positions: 237, 238, 251, and / or 257; a second DNA fragment may comprise degenerate codons at one or more of the amino acid positions: 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, and / or 307; and / or a third DNA fragment may comprise degenerate codons at one or more of the amino acid positions: 324, 325, 326, 328, 329, 332, 334, 336, and / or 338.

[0159] To increase the diversity of the library, it may be possible to assemble two or more DNA fragments encoding parts of the antibody Fc region, wherein a first fragment comprises at least one degenerate codon at amino acid positions that have been identified as being involved in binding of a first Fc receptor, and a second fragment comprises at least one degenerate codon at an amino acid position that has been identified as being involved in binding of a second Fc receptor.

[0160] Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a first Fc receptor, and another one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a second Fc receptor.

[0161] The concept of shuffling DNA fragments comprising mutations identified as being involved in binding to a specific Fc receptor is illustrated in Fig. 10. That is, an antibody Fc region may be assembled from multiple DNA fragments, each covering a different part of the antibody Fc region. For each of the fragments, multiple pools of DNA fragments may be prepared, with each pool containing one or more mutations at positions identified as being involved in binding to a specific Fc receptor. These pools of DNA fragments may then be freely combined to create a combinatorial DNA library encoding engineered antibody Fc regions. By combining mutations at positions identified as being involved in binding to different Fc receptors, engineered antibody Fc regions with novel binding characteristics can be obtained. Specifically, this approach allows for the development of engineered antibody Fc regions with altered binding characteristics toward two or more different Fc receptors.

[0162] Positions that are involved in binding to a specific Fc receptor can be identified, inter alia, by deep mutational scanning as described herein.

[0163] In Example 3 (Figures 6 / 7), the following residues of the human IgG antibody Fc region have been identified to be involved in the binding of Fc receptors:

[0164] FcyRl: 233, 234, 235, 236, 237, 238, 263, 265, 266, 270, 273, 301, 325, 328, 329, 332, 334, 335, 336, 338, 362, 366, 405, 406, 407, and 428.

[0165] FcyR2A-H: 237, 238, 251, 262, 263, 265, 266, 275, 276, 278, 287, 292, 297, 301, 304, 307, 314, 322, 324, 325, 328, 329, 332, 334, 336, 338, 341, 346, 374, 378, 407, 428, and 429.

[0166] FcyR2A-R: 238, 250, 251, 263, 265, 266, 278, 287, 297, 301, 302, 304, 310, 313, 324, 325, 328, 329, 332, 334, 336, 338, 374, and 407.

[0167] FcyR2B: 237, 238, 251, 257, 263, 265, 266, 270, 273, 297, 298, 301, 302, 324, 325, 326, 328, 329, 332, 334, 336, 338, 366, 405, 406, and 407.

[0168] FcyR3A-F: 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, 338, 407, and 429.

[0169] FcyR3A-V: 237, 238, 242, 251, 263, 265, 266, 270, 297, 301, 304, 307, 312, 313, 324, 325, 328, 329, 332, 334, 336, 338, and 407.

[0170] FcyR3B: 232, 233, 234, 235, 236, 237, 238, 257, 263, 265, 266, 267, 269, 270, 271, 272, 273, 292, 295, 297, 300, 301, 302, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 376, 378, 380, 395, 405, 406, 407, 409, 428, 431, 432, and 435.

[0171] FcRn: 250, 351, 252, 253, 254, 257, 288, 310, 314, 338, 428, 435, and 446.

[0172] Accordingly, assembling the combinatorial library may comprise a step of assembling at least two DNA fragments encoding parts of an antibody Fc region, wherein the first fragment comprises one or more mutation that has been identified as being involved in the binding of any one of FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B or FcRn, and the second fragment comprises one or more mutation that has been identified as being involved in the binding of another one of FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B or FcRn.

[0173] In certain embodiments, assembling the combinatorial library may comprise a step of assembling at least two DNA fragments encoding parts of an antibody Fc region, wherein at least one fragment comprises one or more mutation that has been identified as being involved in the binding of any one of FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B or FcRn, and another fragment comprises one or more mutation that has been identified as being involved in the binding of another one of FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B or FcRn.

[0174] In a particular embodiment, the invention relates to the method according to the invention, wherein the DNA library is a combinatorial library encoding a plurality of engineered human IgG antibody Fc regions, wherein each engineered human IgG antibody Fc region encoded in the combinatorial library comprises two or more mutations, and wherein the two or more mutations in the human IgG antibody Fc region are in amino acid positions: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446; preferably in amino acid positions: 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, 338 (all according to EU numbering). In the second step of the method of the invention, the plurality of cells provided in the first step is contacted with an Fc receptor. Contacting an Fc receptor with a plurality of cells typically involves incubating the cells in a liquid containing the Fc receptor under conditions that facilitate binding interactions. This can be achieved by adding a purified Fc receptor protein to the cell culture. The incubation conditions, such as temperature, time, and buffer composition, may be optimized to promote specific binding between the engineered antibody Fc regions displayed on the cell surface and the Fc receptors. The skilled person is capable of identifying suitable conditions under which binding of the antibody Fc region to the Fc receptor can be achieved.

[0175] The Fc receptor that is contacted with the plurality of cells is preferably a soluble Fc receptor. Soluble Fc receptors can be produced recombinantly and purified to high purity, allowing for precise control over the binding interactions during the screening process. Using a soluble Fc receptor facilitates uniform and consistent exposure of the cells to the receptor, enhancing the accuracy and reliability of the assays. The soluble Fc receptor may be tagged with a detectable marker, such as a fluorescent or biotin label, to enable easy identification and quantification of binding events. This approach allows for the efficient selection and enrichment of cells displaying antibody Fc regions with desired binding properties, thereby streamlining the development and optimization of antibodies.

[0176] In a particular embodiment, the Fc receptor that is contacted with the plurality of cells is a biotinylated soluble Fc receptor. The skilled person is aware of methods to produce biotinylated soluble Fc receptors. Moreover, biotinylated soluble Fc receptors are commercially available in purified form.

[0177] The biotinylated soluble Fc receptor may be linked to a detectable marker via a streptavidin molecule. Streptavidin has a high affinity for biotin, forming a strong and stable complex that facilitates the attachment of the Fc receptor to various detectable markers, enabling the visualization, quantification, and / or isolation of cells that have bound the Fc receptor.

[0178] The detectable marker that is attached to the Fc receptor (either directly or indirectly via, e.g., a linker or a biotin-streptavidin pair) may be a fluorescent dye, which provides bright and stable fluorescence for flow cytometry and other fluorescence-based assays. Other detectable markers may include enzymes for colorimetric detection, magnetic beads for magnetic-activated cell sorting (MACS), or other fluorophores suitable for high-sensitivity detection.

[0179] In certain embodiments, the detectable marker is a fluorescent dye, such as, without limitation, phycoerythrin, fluorescein isothiocyanate (FITC), allophycocyanin (APC), Alexa Fluor 488, Alexa Fluor 647, or Texas Red.

[0180] The soluble Fc receptor may be contacted with the plurality of cells at any suitable concentration that allows sufficient binding of the engineered antibody Fc regions displayed by the cells to the Fc receptor. In certain embodiments, 5-10 million cells displaying engineered antibody Fc regions may be contacted with an Fc receptor at a concentration ranging from 0.01 - 100 pg / mL, preferably 0.05 - 50 pg / mL, more preferably 0.08 - 10 pg / mL. Contacting may be carried out at room temperature for a sufficient amount of time. In certain embodiments, cells are contacted with the Fc receptor for one hour at room temperature.

[0181] The Fc receptor that is contacted with the cells, in particular the soluble Fc receptor, may be any Fc receptor known in the art. The term "Fc receptor" as used herein refers to a family of cell surface receptors that bind to the Fc region of antibodies, mediating various immune responses. Fc receptors include, but are not limited to, Fc gamma receptors (FcyRs) such as FcyRl (CD64), FcyR2 (CD32), and FcyR3 (CD16); Fc alpha receptors (FcaRs) such as FcaRI (CD89); Fc epsilon receptors (FceRs) such as FceRI and FceRII (CD23); Fc mu receptors (FcpRs); and the neonatal Fc receptor (FcRn). These receptors play crucial roles in processes such as phagocytosis, antibody-dependent cellular cytotoxicity (ADCC), degranulation, and immune complex clearance.

[0182] However, the term "Fc receptor" may also be understood to encompass other proteins that interact with the antibody Fc region, such as complement component lq (or Clq).

[0183] In certain embodiments, the Fc receptor is an FcyRl receptor. FcyRl, also known as CD64, is a high-affinity Fc gamma receptor that specifically binds to the Fc region of IgG antibodies. This receptor is predominantly expressed on the surface of immune cells such as macrophages, monocytes, and dendritic cells. FcyRl plays a pivotal role in the immune system by mediating various effector functions, including phagocytosis, where it facilitates the engulfment and destruction of antibody-coated pathogens. Additionally, FcyRl is involved in the release of inflammatory mediators, enhancing the immune response.

[0184] In certain embodiments, the Fc receptor is an FcyR2 receptor. FcyR2 receptors, also known as CD32, include three main subtypes: FcyR2A, FcyR2B, and FcyR2C, each with distinct functions and expression patterns.

[0185] FcyR2A (CD32A) is an activating receptor expressed on various immune cells, including neutrophils, macrophages, and platelets. FcyR2A plays a significant role in mediating phagocytosis and inflammatory responses. FcyR2A exists in two common polymorphic forms: FcyR2A-H (histidine at position 131) and FcyR2A-R (arginine at position 131). These polymorphic variants differ in their affinity for IgG subclasses and their ability to mediate immune responses, with FcyR2A-R generally having a higher affinity for IgG subclasses compared to FcyR2A-H.

[0186] FcyR2B (CD32B) is an inhibitory receptor predominantly expressed on B cells, but also found on other immune cells such as macrophages and dendritic cells. FcyR2B plays a crucial role in regulating immune responses and maintaining immune tolerance by delivering inhibitory signals that counteract activation signals from other Fc receptors. This helps prevent overactivation of the immune system and maintains homeostasis.

[0187] FcyR2C (CD32C) is an activating receptor with an extracellular domain identical to FcyR2B but is less commonly expressed and has a more restricted distribution. FcyR2C is primarily found on natural killer (NK) cells and some monocytes. It mediates distinct immune functions from FcyR2B, due to differences in the cytoplasmic tail responsible for signalling.

[0188] In certain embodiments, the Fc receptor is an FcyR3 receptor. FcyR3 receptors, also known as CD16, include two main subtypes: FcyR3A and FcyR3B, each with distinct functions and expression patterns. FcyR3A (CD16A is an activating receptor expressed on natural killer (NK) cells, macrophages, and some T cells. FcyR3A plays a crucial role in mediating antibody-dependent cellular cytotoxicity (ADCC), where it facilitates the targeted killing of antibody-coated cells. FcyR3A exists in two polymorphic forms: FcyR3A-F (phenylalanine at position 176) and FcyR3A-V (valine at position 176). These polymorphic variants differ in their affinity for IgG subclasses, with FcyR3A-V generally having a higher affinity for IgGl and lgG3 compared to FcyR3A-F. This difference in affinity can influence the effectiveness of therapeutic antibodies that rely on ADCC for their mechanism of action.

[0189] FcyR3B (CD16B) is primarily expressed on neutrophils and is anchored to the cell membrane via a glycosylphosphatidylinositol (GPI) linkage. FcyR3B plays a role in immune complex clearance and inflammation. Unlike FcyR3A, FcyR3B does not mediate ADCC but is involved in the binding and clearance of immune complexes, contributing to the regulation of immune responses and prevention of excessive inflammation.

[0190] In certain embodiments, the Fc receptor is FcRn. The neonatal Fc receptor (FcRn) is a unique Fc receptor that plays a crucial role in regulating the half-life and transport of IgG antibodies. Unlike other Fc receptors, FcRn is not primarily involved in mediating immune effector functions such as ADCC. Instead, FcRn binds to the Fc region of IgG antibodies at acidic pH levels, typically found in endosomes, and protects them from lysosomal degradation. This binding allows FcRn to recycle IgG antibodies back to the cell surface, where they are released at neutral pH, thereby extending their half-life in the circulation.

[0191] Thus, in a particular embodiment, the Fc receptor that is contacted with the plurality cells in step (b) of the method of the invention is an FcyRl receptor, an FcyR2 receptor, an FcyR3 receptor or an FcRn receptor. In certain embodiments, the Fc receptor that is contacted with the plurality cells in step (b) of the method of the invention is an FcyRl receptor, an FcyR2A receptor, an FcyR2B receptor, an FcyR2C receptor, an FcyR3A receptor, an FcyR3B receptor or an FcRn receptor. In certain embodiments, the Fc receptor that is contacted with the plurality cells in step (b) of the method of the invention is an FcyRl receptor, an FcyR2A-H receptor, an FcyR2A-R receptor, an FcyR2B receptor, an FcyR2C receptor, an FcyR3A-F receptor, an FcyR3A- V receptor, an FcyR3B receptor or an FcRn receptor. In certain embodiments, the Fc receptor that is contacted with the plurality cells in step (b) of the method of the invention is an FcyRl receptor, an FcyR2A-H receptor, an FcyR2A-R receptor, an FcyR2B receptor, an FcyR3A-F receptor, an FcyR3A-V receptor, an FcyR3B receptor or an FcRn receptor.

[0192] In certain embodiments, two or more of the Fc receptors mentioned above are simultaneously contacted with the plurality of cells. In such embodiment, it is preferred that each Fc receptor is labeled with a different detectable marker.

[0193] In a subsequent step, the cells that have been contacted with the Fc receptor are separated according to their ability to bind to the Fc receptor. This separation can be achieved using various techniques, such as flow cytometry or magnetic-activated cell sorting (MACS). In flow cytometry, cells are labeled with fluorescent markers that indicate the presence of bound Fc receptors, allowing for the high-throughput sorting and quantification of cells based on their fluorescence intensity. In MACS, cells are labeled with magnetic beads conjugated to antibodies or streptavidin that bind to the Fc receptor, enabling the magnetic separation of cells that have successfully bound the Fc receptor. These sorted cells can then be collected and further analyzed to identify and isolate those displaying antibody Fc regions with desired binding properties.

[0194] The separation step may be repeated one or more times to reduce the risk of false positive results.

[0195] In a particular embodiment, the invention relates to the method according to the invention, wherein cells in step (c) are separated by fluorescence-activated cell sorting (FACS). The term "FACS" stands for Fluorescence-Activated Cell Sorting, a specialized type of flow cytometry. FACS is technique used to analyze and sort a heterogeneous mixture of cells based on their fluorescent characteristics. The process involves labeling cells with fluorescent markers that bind to specific cellular components, such as proteins, nucleic acids, or other molecules of interest.

[0196] In the method of the present invention, the cells comprised in the plurality of cells are separated into two or more cell populations according to their ability to bind to an Fc receptor.

[0197] The ability of a cell expressing an engineered antibody Fc region on the cell surface to bind to an Fc receptor may be determined by labeling the Fc receptor with a detectable marker, such as a fluorescent dye or a magnetic bead. Cells that have bound the Fc receptor can then be identified and quantified using techniques such as fluorescence-activated cell sorting (FACS) or magnetic-activated cell sorting (MACS). In FACS, cells are passed through a flow cytometer where they are individually interrogated by a laser, and the fluorescence intensity is measured to determine the extent of Fc receptor binding. Cells with higher fluorescence intensity are sorted into one population, while those with lower or no fluorescence are sorted into another. In MACS, cells are incubated with Fc receptors conjugated to magnetic beads, and those that bind the Fc receptor are separated using a magnetic field. These methods allow for the precise separation and enrichment of cell populations based on their Fc receptor binding capabilities.

[0198] The skilled person is capable of defining one or more fluorescence intensity thresholds, or gates, to separate the plurality of cells into distinct populations based on their ability to bind to an Fc receptor in step (c) of the method. By setting these gates in a FACS assay, cells can be accurately categorized according to their binding efficiency. This process can be performed without the need for a reference antibody Fc region. Instead, arbitrary thresholds can be defined to identify the best-binding and / or weakest-binding engineered antibody Fc regions, facilitating the selection and optimization of antibody variants with desired binding properties. For example, to identify the best-binding candidates in a library of engineered antibody Fc regions, the cells with the highest fluorescence intensity may be isolated, such as the top 1%, 2%, 3%, 4%, 5%, 6%, 7%, 8%, 9%, or 10% of cells with the highest fluorescence intensity. This approach enables the efficient identification and enrichment of antibody variants with superior binding characteristics.

[0199] In certain embodiments, the ability of an engineered antibody Fc region to bind to an Fc receptor is determined by comparing said binding to that of a reference antibody Fc region to the same Fc receptor.

[0200] Thus, in a particular embodiment, the invention relates to a method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of: (a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface; (b) contacting the cells in step (a) with an Fc receptor; (c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor, wherein the ability of an engineered antibody Fc region to bind to the Fc receptor is determined by comparison with a reference antibody Fc region; and (d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region.

[0201] For example, in step (c), the plurality of cells may be separated into a first population of cells that express engineered antibody Fc regions which bind more efficiently to an Fc receptor than cells expressing a reference antibody Fc region, and a second population of cells that express engineered antibody Fc regions which bind less efficiently to the same Fc receptor than cells expressing the reference antibody Fc region.

[0202] Alternatively, the plurality of cells in step (c) may be separated into three populations of cells: a first population of cells expressing engineered antibody Fc regions that bind more efficiently to an Fc receptor than cells expressing a reference antibody Fc region; a second population of cells expressing engineered antibody Fc regions that bind to the Fc receptor with similar efficiency as cells expressing the reference antibody Fc region; and a third population of cells expressing engineered antibody Fc regions that bind less efficiently to the same Fc receptor than cells expressing the reference antibody Fc region.

[0203] The reference antibody Fc region may be any suitable antibody Fc region. For example, the reference antibody Fc region may be a wildtype antibody Fc region. Alternatively, the reference antibody Fc region may be an antibody Fc region that served as starting point for the engineered antibody Fc regions displayed by the plurality of cells.

[0204] In certain embodiments, the reference antibody Fc region is the wildtype human IgG antibody Fc region (SEQ ID NO:1). In certain embodiments, the reference antibody Fc region is the human IgGl Fc region comprising mutationsT299A / K326l / A327Y / L328G (EU numbering) (SEQ ID NO:2). In certain embodiments, the reference antibody Fc region may be an engineered antibody Fc regions comprised in the plurality of cells provided in step (a) of the method. That is, one of the engineered antibody Fc regions comprised in the plurality of cells may be determined to be the reference antibody Fc region for the purpose of identifying engineered antibody Fc regions with particularly high / low binding affinities. Forthat purpose, an engineered antibody Fc region having similar binding characteristics as a wildtype antibody Fc region may be selected as the reference antibody Fc region. An engineered antibody Fc regions comprised in the plurality of cells may be determined to be the reference antibody Fc region without knowing the amino acid sequence of said engineered antibody Fc region.

[0205] In certain embodiments, two different fluorescent dyes may be used in parallel to identify engineered antibody Fc regions with the desired binding properties. That is, not only the soluble Fc receptor may be labelled with a fluorescent dye, but also the antibody Fc region may be labelled with a fluorescent dye. Labelling the antibody Fc region with a fluorescent dye may be achieved with a fluorescently labeled antibody that specifically binds to the antibody Fc region. In certain embodiments, the engineered antibody Fc region may be labeled with a fluorescently labeled antibody that specifically binds to a peptide tag that is fused to the antibody Fc region. In certain embodiments, the peptide tag is a FLAG tag. With this approach, it can be confirmed that the engineered antibody Fc region is displayed by the cell and bound by soluble Fc receptor.

[0206] Accordingly, the invention relates to the method according to the invention, wherein the engineered antibody Fc regions are labelled with a first fluorescent dye and the Fc receptor is labelled with a second fluorescent dye.

[0207] Preferably, the two fluorescent dyes are compatible, i.e., can be used simultaneously without significant spectral overlap or interference, allowing for accurate and distinct detection of each dye in a multi-color assay.

[0208] In the last step of the method, the cells separated in step (c) are subjected to sequencing to identify mutations in the engineered antibody Fc regions. For that, at least one cell from at least one population of cells obtained in step (c) may be subjected to sequencing. This may be achieved by isolating DNA from said cell and sequencing the part of the DNA encoding the engineered antibody Fc region or at least a fragment thereof comprising at least one mutation.

[0209] The identified mutations may be linked to a specific functional outcome. For example, mutations identified in a population of cells that were determined in step (c) to bind to an Fc receptor with high efficiency may be associated with improved binding to said Fc receptor. Conversely, mutations identified in a population of cells that were determined to bind to the Fc receptor with low efficiency may be associated with reduced binding to said Fc receptor. Mutations identified in a population of cells that were determined to bind to the Fc receptor with similar efficiency as a reference antibody Fc region may be associated with unaltered binding to said Fc receptor.

[0210] Accordingly, in a particular embodiment, the invention relates to a method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of: (a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface; (b) contacting the cells in step (a) with an Fc receptor; (c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor, preferably wherein the ability of an engineered antibody Fc region to bind to the Fc receptor is determined by comparison with a reference antibody Fc region; and (d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region that (i) confer improved binding to the Fc receptor, (ii) do not affect binding to the Fc receptor, and / or (iii) confer reduced binding to the Fc receptor.

[0211] Mutations in the engineered antibody Fc regions may be identified by any suitable method known in the art. These methods include, but are not limited to, deep sequencing, which allows for high-throughput and precise identification of mutations across large libraries of engineered antibody Fc regions. Sanger sequencing can also be employed for smaller libraries or for validation of specific mutations.

[0212] In a preferred embodiment, an entire population of cells, or a sample thereof, may be subjected to deep sequencing to identify mutations that affect the binding of an engineered antibody Fc region to an Fc receptor. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein in step (d), the at least one cell population, or a sample thereof, is subjected to deep sequencing.

[0213] The term "deep sequencing" refers to a high-throughput DNA sequencing technique that allows for the comprehensive and detailed analysis of genetic material by generating a large volume of sequence data. This method provides high coverage of the target DNA, enabling the detection of rare variants, low-frequency mutations, and subtle genetic changes with high accuracy and sensitivity. Deep sequencing is typically performed using next-generation sequencing (NGS) platforms, which can process millions of DNA fragments in parallel, resulting in a thorough and exhaustive examination of the genetic landscape.

[0214] In step (c) of the method according to the invention, cells expressing engineered antibody Fc regions are separated into multiple cell populations based on their ability to bind to an Fc receptor. Samples from these cell populations may be subjected to deep sequencing to identify mutations responsible for these binding characteristics. This approach enables the precise identification of mutations that enhance or alter the binding properties of the antibody Fc regions. For example, in step (c), a population of cells exhibiting high binding affinity to a specific Fc receptor may be isolated. Subjecting this cell population to deep sequencing can identify mutations in the antibody Fc region that are likely to enhance the affinity of the antibody Fc region for the Fc receptor. Conversely, cell populations with low binding affinity to the Fc receptor may be isolated and sequenced to identify mutations that reduce or impair binding, providing valuable insights into regions of the Fc domain critical for receptor interaction.

[0215] With this approach, mutations can be identified that confer improved binding of an engineered antibody Fc region to an Fc receptor, mutations that do not affect the binding of an engineered antibody Fc region to an Fc receptor, and / or mutations that reduce the binding of an engineered antibody Fc region to an Fc receptor.

[0216] Deep sequencing of a population of cells expressing engineered antibody Fc regions typically involves a step of determining the frequency of individual mutations. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein deep sequencing comprises a step of determining the frequency of individual mutations in the engineered antibody Fc region in a population.

[0217] That is, the more frequent a mutation is in a population of cells that has been isolated based on their binding characteristics to a specific Fc receptor, the more likely it is that this mutation has a significant impact on the binding affinity to said Fc receptor. By analyzing the frequency of individual mutations, researchers can identify key mutations that enhance or alter the binding properties of the antibody Fc regions.

[0218] The frequency of individual mutations in a population of cells may be determined as follows: First, the total number of sequencing reads containing a codon at a specific position may be counted. Second, the number of reads encoding a specific amino acid residue at that position is counted. The frequency of that amino acid at that position is then calculated by dividing the number of reads encoding that amino acid by the total number of reads forthat position. This process may be repeated for one or more other amino acids at that position and / or for each position of interest within the engineered antibody Fc region. By analyzing these frequencies, mutations that occur more frequently in the population may be identified, indicating a significant impact on the binding characteristics to the Fc receptor.

[0219] Preferably, the frequency of individual mutations is determined before and after the separation of the cells in step (c) of the method, as described by Fowler et al., (Nat Methods (2010) 7(9):741-6). That is, the frequency of an individual mutation may be determined in the initial plurality of cells displaying engineered antibody Fc regions and in a population of cells that has been isolated in step (c) based on their ability to bind to an Fc receptor. This comparative analysis allows to identify mutations that are enriched or depleted in the selected population, providing insights into which mutations enhance or impair binding to the Fc receptor. Moreover, the enrichment factor ("positional enrichment score") may provide additional clues on how strong the impact of a mutation on the binding of the Fc receptor is.

[0220] Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein the impact of an individual mutation on the binding of the engineered antibody Fc region to the Fc receptor is determined based on the frequency of said individual mutation in the population.

[0221] Within the present invention and in the context of engineered antibody Fc regions, the term "individual mutation" refers to an amino acid exchange. That is, when determining the frequency of an individual mutation, all reads that encode the same amino acid at a specific position may be combined, even if the codon usage is different.

[0222] The method described hereinabove may be used to identify engineered antibody Fc regions with altered binding characteristics toward a specific Fc receptor. Alternatively, the method may be employed to identify specific mutations in an antibody Fc region that affect the binding to a specific Fc receptor.

[0223] This workflow may be repeated one or more times with additional Fc receptors. Specifically, a plurality of cells displaying engineered antibody Fc regions may be individually contacted with multiple Fc receptors. For example, a plurality of cells displaying engineered antibody Fc regions may be separated into multiple samples and each sample may be contacted with (different) Fc receptor.

[0224] Alternatively, a population of cells that has been isolated in step (c) based on their ability to bind to a first Fc receptor may subsequently be contacted with a second Fc receptor to identify engineered antibody Fc regions with specific binding characteristics for both the first and second Fc receptors. This process can be repeated for additional Fc receptors, allowing for the identification of antibody Fc regions with tailored binding properties across multiple Fc receptors.

[0225] In a preferred embodiment, the method of the invention is used to identify individual mutations in antibody Fc regions that contribute to the binding of one or more Fc receptors. Accordingly, the method may be repeated multiple times with different Fc receptors to identify, for each tested Fc receptor, a pool of mutations that affect its binding. This approach allows for the comprehensive mapping of mutations that influence the binding properties of antibody Fc regions across various Fc receptors.

[0226] The method of the invention may be performed with any Fc receptor known in the art. In certain embodiments, the method of the invention is performed with one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2, FcyR3 and FcRn. In certain embodiments, the method of the invention is performed with one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2A, FcyR2B, FcyR2C, FcyR3A, FcyR3B and FcRn. In certain embodiments, the method of the invention is performed with one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR2C, FcyR3A-F, FcyR3A-V, FcyR3B and FcRn. In certain embodiments, the method of the invention is performed with one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B and FcRn.

[0227] Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein steps (b) - (d) are repeated for one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B and FcRn.

[0228] Accordingly, in a particular embodiment, the invention relates to a method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of: (a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface; (b) contacting the cells in step (a) with at least two different Fc receptors, preferably selected from the group consisting of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B and FcRn; (c) separating cells that have been contacted with an Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor; and (d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region.

[0229] The sequence information obtained in step (d) of the method of the invention, such as the deep sequencing data, may be used to train a machine learning model. That is, in a particular embodiment, the invention relates to a method wherein the sequence information obtained in step (d) is used to train a machine learning model. To train the machine learning model, sequencing data must be converted into a numerical format. This can be achieved through one-hot encoding, where each nucleotide (A, C, G, T) is represented as a binary vector, or through more complex embedding techniques like protein language models, which capture contextual relationships within the sequence. Once encoded, the data is fed into the machine learning model, which undergoes a training process involving data preparation, model architecture design, parameter optimization, validation, and testing. This comprehensive process enables the model to learn complex patterns and make accurate predictions, facilitating tasks such as the identification and optimization of antibody Fc regions.

[0230] A "machine learning model" refers to an artificial intelligence (Al) algorithm that is trained to learn from data and make predictions or decisions based on that data. These models can range from simple linear regression models to more complex algorithms like decision trees and support vector machines. In a preferred embodiment, the machine learning algorithm is a deep learning algorithm. A "deep learning model" refers to a specific type of machine learning model that uses multiple layers of neural networks to analyze and learn from large amounts of data. These models are capable of automatically identifying complex patterns and relationships within the data, making them highly effective for tasks such as image recognition, natural language processing, and predictive analytics. Deep learning models are characterized by their ability to improve performance as they are exposed to more data.

[0231] The term "training a machine learning model" refers to the process of teaching an algorithm to recognize patterns and make predictions based on a large dataset. This process can be applied to both traditional machine learning models and deep learning models. This may involve several steps: collecting relevant data, such as sequence information from deep sequencing; preparing the data by cleaning and organizing it; selecting or designing the model architecture, which may include choosing the type of algorithm or the number of layers and neurons in the case of neural networks; feeding the training data into the model and adjusting its parameters to minimize errors using an optimization algorithm; validating the model's performance on a separate validation set to ensure accurate learning; testing the model on a new test set to evaluate its performance on unseen data; and finally, deploying the trained model for practical use. In the case of deep learning models, the architecture typically involves multiple layers of neural networks, which can automatically identify complex patterns and relationships within the data. This comprehensive process enables both traditional machine learning models and deep learning models to learn complex patterns and make accurate predictions, facilitating tasks such as the identification and optimization of antibody Fc regions. In the context of this invention, the machine learning model is preferably a deep learning model due to its superior ability to handle large datasets and complex patterns, thereby enhancing the accuracy and efficiency of the predictive tasks.

[0232] The dataset used for training the machine learning algorithm is preferably a deep sequencing dataset or a plurality of deep sequencing datasets. In certain embodiments, the machine learning algorithm is trained with multiple deep sequencing datasets, each generated using the method of the invention, but with different Fc receptors in the contacting step. This approach allows the model to learn from a diverse set of binding interactions, enhancing its ability to predict binding characteristics across various Fc receptors. By incorporating multiple datasets, the machine learning algorithm can identify common patterns and unique features associated with different Fc receptors, thereby improving its predictive accuracy and robustness.

[0233] Preferably, the deep sequencing datasets used for training the machine learning algorithm are obtained using the method of the invention, wherein the following Fc receptors are used in the contacting step: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B, and / or FcRn.

[0234] The machine learning model may be used for various tasks. For example, the machine learning model may be used for predicting the Fc receptor binding properties of an engineered antibody Fc region. This predictive capability allows for the efficient screening and identification of antibody Fc regions with enhanced or desired binding characteristics without the need for extensive experimental validation. By leveraging the model's predictions, candidates that are most likely to exhibit desired binding properties may be prioritized, thereby streamlining the development process. Alternatively, the machine learning model may be used for generating engineered antibody Fc regions with desired Fc receptor binding properties. By inputting specific binding requirements into the model, suggestions for mutations or combinations of mutations that are predicted to achieve the desired binding profile may be obtained. This generative capability enables the rational design of antibody Fc regions tailored to specific needs, such as increased affinity for a particular Fc receptor or reduced binding to another.

[0235] Furthermore, the machine learning model can be utilized to explore the effects of multiple simultaneous mutations, providing insights into synergistic or antagonistic interactions between different mutations. This allows for a more comprehensive understanding of the mutational landscape and the identification of optimal mutation combinations that maximize binding efficacy.

[0236] Overall, the machine learning model serves as a powerful tool for the rapid and cost-effective development of engineered antibody Fc regions with tailored binding properties and enhanced therapeutic potential.

[0237] Accordingly, in a particular embodiment, the invention relates to the method according to the invention, wherein the machine learning model is used to (a) predict the Fc receptor binding properties of an engineered antibody Fc region; and / or (b) generate an engineered antibody Fc region having desired Fc receptor binding properties.

[0238] In another aspect, the invention relates to a method for identifying engineered antibody Fc regions with altered Fc receptor binding characteristics, the method comprising the steps of:

[0239] a) identifying amino acid residues of an antibody Fc region that are involved in binding to one or more Fc receptors;

[0240] b) introducing mutations in at least two positions of the antibody Fc region encoding the amino acid residues identified in step (a) to obtain an engineered antibody Fc region;

[0241] c) testing the engineered antibody Fc region obtained in step (b) for binding to one or more Fc receptors; and

[0242] d) determining the engineered antibody Fc region to have altered Fc receptor binding characteristics based on the outcome of the testing in step (c).

[0243] That is, the invention relates to a method for identifying engineered antibody Fc regions having altered binding characteristics to one or more Fc receptors, the method comprising a first step of identifying amino acid residues in an antibody Fc region that are involved in Fc receptor binding, and a second step of introducing mutations at two or more of the positions identified in the first step. With this approach, the information obtained in the first step may be used to generate more focused combinatorial libraries that may be screened in the second step of the method.

[0244] In the first step, the amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors may be identified with any suitable method known in the art, preferably any method that can be used to introduce a single mutation into an antibody Fc region.

[0245] In a particular embodiment, the invention relates to the method according to the invention, wherein the amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors are identified using deep mutational scanning and / or alanine scanning.

[0246] That is, in certain embodiments, amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors may be identified using alanine scanning, as known in the art. Alanine scanning is a mutagenesis technique used to systematically substitute amino acid residues in a protein with alanine to assess the functional significance of each residue. By replacing specific amino acids with alanine, which is small and non-reactive, the impact of each substitution on the protein's structure, stability, and / or activity can be determined, thereby identifying critical residues involved in binding, catalysis, or other functions.

[0247] However, it is preferred that the amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors are identified using deep mutational scanning (DMS), as described in more detail elsewhere herein. The advantage of deep mutational scanning over alanine scanning is that DMS allows for the comprehensive evaluation of all possible amino acid substitutions at each position, not just alanine. This provides a more detailed and nuanced understanding of the functional impact of each residue, enabling the identification of both beneficial and detrimental mutations. Additionally, DMS can be performed in a high-throughput manner, allowing for the simultaneous assessment of thousands of mutations, which significantly accelerates the identification of critical residues involved in Fc receptor binding.

[0248] The residues of the antibody Fc region identified in step (a) are preferably involved in binding to at least one Fc receptor, preferably one or more of the Fc receptors FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B, and / or FcRn. Identifying residues of the antibody Fc region that are involved in binding to more than one Fc receptor may be achieved by screening deep mutational scanning libraries with two or more different Fc receptors.

[0249] The amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors may be identified by deep sequencing, in particular by subjecting a population of cells expressing mutated antibody Fc regions with altered Fc receptor binding characteristics to deep sequencing, as described elsewhere herein.

[0250] In certain embodiments, amino acid residues of an antibody Fc region that are involved in binding to one or more Fc receptors may be identified by detecting a loss of binding to one or more Fc receptors when the wild-type amino acid residue is replaced with another amino acid residue (detrimental or loss-of-function mutations).

[0251] Additionally or alternatively, amino acid residues of an antibody Fc region that are involved in binding to one or more Fc receptors may be identified by improved binding to one or more Fc receptors when the wild-type amino acid residue is replaced with another amino acid residue (beneficial orgain-of-function mutations).

[0252] Deep mutational scanning may be used to identify mutations that improve the binding to one or more Fc receptor, and / or mutations that reduce or eliminate the binding the binding to one or Fc receptor in a single experiment, as described elsewhere herein. In the second step of the method, two or more mutations are introduced at positions of an antibody Fc region that have been identified in step (a) to be involved in Fc receptor binding. Preferably, an engineered antibody Fc receptor is obtained by introducing a combinatorial library into a suitable cell line, i.e., a cell line for displaying engineered antibody Fc receptors on the cell surface.

[0253] Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein the engineered antibody Fc region obtained in step (b) is encoded by a combinatorial library encoding a plurality of engineered antibody Fc regions.

[0254] The combinatorial library may be any of the combinatorial disclosed elsewhere herein.

[0255] That is, in a particular embodiment, the invention relates to the method according to the invention, wherein the combinatorial library is obtained by assembling two or more DNA fragments encoding parts of an engineered antibody Fc region, preferably wherein each of the DNA fragments comprises at least one mutations, more preferably wherein each of the DNA fragments comprises at least one degenerate codon.

[0256] In a particular embodiment, the invention relates to the method according to the invention, wherein one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a first Fc receptor, and another one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a second Fc receptor.

[0257] In a particular embodiment, the invention relates to the method according to the invention, wherein mutations are introduced in at least two of the following positions of the antibody Fc region: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446, preferably in positions 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering).

[0258] Preferably, designing the combinatorial library takes into account the results from the deep mutational scanning in the first step of the method. That is, the combinatorial library may comprise specific mutations that have been shown in deep mutational scanning to have an impact on binding to one or more Fc receptors. Preferably, various mutations that are suspected to have an impact on binding to one or more Fc receptors may be encoded in the same position using degenerate codons, as described herein.

[0259] In a particular embodiment, the invention relates to the method according to the invention, wherein testing the engineered antibody Fc region for binding to one or more Fc receptors involves a step of contacting a cell displaying the engineered antibody Fc region with one or more Fc receptors.

[0260] That is, the method comprises at least one step of contacting cells displaying engineered antibody Fc regions with one or more Fc receptors. Preferably, also the first step of identifying amino acid residues in the antibody Fc region that are involved in binding to one or more Fc receptors comprises a step of contacting cells displaying engineered antibody Fc regions with one or more Fc receptors. The engineered antibody Fc regions may be displayed on the surface of a cell, preferably a yeast cell, as described elsewhere herein. Moreover, cells may be contacted with Fc receptors as described elsewhere herein.

[0261] In a particular embodiment, the invention relates to the method according to the invention, wherein the binding of the engineered antibody Fc region to one or more Fc receptors is tested by flow cytometry, as described herein.

[0262] Preferably, a plurality of engineered antibody Fc region is tested simultaneously in the method of the invention. In such embodiments, populations of cells may be isolated by flow cytometry according to their ability to bind to one or more Fc receptor. These populations of cells may then be subjected to deep sequencing, as described elsewhere herein, to identify mutations or combinations of mutations that are involved in Fc receptor binding. Thus, in a particular embodiment, the invention relates to the method according to the invention, comprising an additional step of determining the amino acid sequence of the engineered antibody Fc region, preferably by sequencing, more preferably by deep sequencing.

[0263] The machine learning models used in the methods of the invention are preferably trained with sequencing data that has been obtained with the methods described herein above. However, it is to be understood that any type of binding data may be used to train a machine learning model for the purpose of predicting the Fc receptor binding properties of an engineered antibody Fc region and / or generating an engineered antibody Fc region having desired Fc receptor binding properties.

[0264] That is, in a particular embodiment, the invention relates to a method for predicting the Fc receptor binding properties of an engineered antibody Fc region, the method comprising the steps of:

[0265] a) training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors;

[0266] b) providing a nucleic acid sequence or an amino acid sequence of the engineered antibody Fc region to the trained machine learning model; and

[0267] c) obtaining, from the machine learning model, the predicted Fc receptor binding properties of said engineered antibody Fc region provided in step (b).

[0268] The first step involves training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors. This binding data may include information on the binding affinity, specificity, or other relevant characteristics of the antibody Fc regions when interacting with different Fc receptors. The training process involves feeding this data into the machine learning model, which may be, without limitation a convolutional neural network (CNN), transformer, or variational auto-encoder (VAE). The model learns to recognize patterns and correlations between the sequence data of the antibody Fc regions and their binding properties to the Fc receptors. The term "binding data" refers to comprehensive information that encompasses both the binding properties of engineered antibody Fc regions for one or more Fc receptors and the corresponding sequence data. In other words, binding data includes sequence information that is directly associated with specific Fc receptor binding properties. Preferably, the binding data has been obtained with different Fc receptors, such as any of the Fc receptors disclosed herein.

[0269] In the second step, a nucleic acid sequence or an amino acid sequence encoding an engineered antibody Fc region is provided to the trained machine learning model. This sequence data serves as the input for the model, which has been trained to analyze and predict binding properties based on such sequences. The sequence data can be preprocessed and encoded appropriately to match the format used during the training phase.

[0270] In the final step, the trained machine learning model generates predictions regarding the Fc receptor binding properties of the engineered antibody Fc region provided in step (b). These predicted properties may indicate how the antibody Fc region will interact with one or more Fc receptors.

[0271] That is, the machine learning model may be used to predict the binding properties of an engineered antibody Fc region of interest.

[0272] Alternatively, a machine learning model may be used to generate engineered antibody Fc regions having desired Fc receptor binding properties. That is, in a particular embodiment, the invention relates to a method for generating an engineered antibody Fc region having desired Fc receptor binding properties, the method comprising the steps of:

[0273] a) training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors; and

[0274] b) utilizing the deep machine learning model to generate a nucleic acid sequence or an amino acid sequence encoding an engineered antibody Fc region with the desired Fc receptor binding properties. The first step of the method is identical to the first step of the predictive method described above.

[0275] In the second step, the trained machine learning model is utilized to generate a nucleic acid sequence or an amino acid sequence encoding an engineered antibody Fc region with the desired Fc receptor binding properties. By inputting specific binding requirements into the model, sequences that are predicted to exhibit the desired binding profile can be obtained. This generative capability allows for the rational design of antibody Fc regions tailored to specific needs, such as increased affinity for a particular Fc receptor or reduced binding to another. The model can also explore the effects of multiple simultaneous mutations, providing insights into combinatorial effects that may not be apparent through single-mutation studies.

[0276] The binding data used in the methods described hereinabove is preferably obtained by deep sequencing. That is, the binding data is preferably a deep sequencing dataset. The deep sequencing data preferably comprises sequence information of engineered antibody Fc regions and their corresponding binding properties to one or more Fc receptors. This may include high-throughput sequencing reads, as well as quantitative measurements of binding affinity, specificity, and other relevant characteristics. By integrating such datasets, the machine learning model can learn the relationships between specific sequence variations and their impact on Fc receptor binding, enabling precise predictions and the generation of optimized antibody Fc regions with desired binding properties.

[0277] More preferably, the deep sequencing datasets used as binding data in the method of the invention has been obtained with the method of the present invention. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein the binding data is obtained using the method according to the invention.

[0278] Regardless of the method with which the binding data has been obtained, it is preferred that the binding data comprises information from interactions with at least two different Fc receptors. This ensures that the machine learning model is trained with a diverse set of binding interactions, enhancing its ability to predict binding characteristics across various Fc receptors. Thus, in a particular embodiment, the invention relates to the method according to the invention, wherein the machine learning model is trained with binding data that has been obtained with at least two different Fc receptors.

[0279] Preferably, the binding data comprises information from interactions with at least two different Fc receptors selected from the group of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B, and FcRn.

[0280] All numeric ranges provided herein are inclusive of narrower ranges; delineated upper and lower range limits are interchangeable to create further ranges not explicitly delineated. The number of significant digits conveys neither limitation on the indicated amounts nor on the accuracy of the measurements.

[0281] In this document, the terms "a" or "an" are used to include one or more than one and the term "or" is used to refer to a nonexclusive "or" unless otherwise indicated.

[0282] The term "about," as used herein, means approximately, in the region of, roughly, or around. When the term "about" is used in conjunction with a numerical range, it modifies that range by extending the boundaries above and below the numerical values set forth. In general, the term "about" is used herein to modify a numerical value above and below the stated value by a variance of 10%. Therefore, about 50% means in the range of 45%-55%. Numerical ranges recited herein by endpoints include all numbers and fractions subsumed within that range (e.g. 1 to 5 includes 1, 1.5, 2, 2.75, 3, 3.90, 4, and 5). It is also to be understood that all numbers and fractions thereof are presumed to be modified by the term "about."

[0283] BRIEF DESCRIPTION OF THE DRAWINGS

[0284] Fig. 1: Schematic illustrating the antibody Fc surface display approach. Image created with BioRender.com.

[0285] Fig. 2: Flow cytometry analysis demonstrating Fc-receptor binding in the yeast display system of the wild-type human IgGl (top), and N297Q (bottom) Fc domain. Y-axis indicates Fc- receptor binding. X-axis indicates surface expression. Fc-receptor concentrations: FcgRl and FcgR2A-H (1.67 pg / mL), FcgR3A-V (0.278pg / mL).

[0286] Fig. 3: Fc-receptor binding profiles of variants with known functionality. Top shows Fc-receptor binding (geometric mean fluorescence intensity) normalized by FLAG surface expression across a range of concentrations. Bottom shows area under the titration curve for each variant.

[0287] Fig.4: Heatmap of variant frequency in the naive yeast Fc DMS library. Rows represent the 20 canonical amino acids, and columns represent Fc domain residues (EU numbering). Tile color intensity indicates variant frequency; grey tiles denote unobserved variants. Vertical black lines separate the hinge, CH2, and CH3 domains.

[0288] Fig. 5: Overview of the Fc DMS screening approach. The Fc library is evaluated by FACS for Fc-receptor binding (e.g., FcyR3A-F), isolating binding and non-binding populations through multiple rounds of sorting, followed by deep sequencing. Image created in part with BioRender.com.

[0289] Fig. 6: Positional enrichment analysis of Fc DMS data. Lollipop plots show enrichment of mutations at each position in the FcyRl, FcyR2A-H, FcyR2A-R, and FcyR2B non-binding populations compared to the naive library. The x-axis indicates Fc domain residues (EU numbering) and the y-axis indicates enrichment of non-wild-type amino acids. Grey shadings show previously described regions of wild-type IgGl Fc:Fcy receptor interaction (Sazinsky et al., 2008). Fisher's exact test with Bonferroni-adjusted p-value < 0.05 to determine significance.

[0290] Fig. 7: Positional enrichment analysis of Fc DMS data continued. Lollipop plots show enrichment of mutations at each position in the FcyR3A-F, FcyR3A-V, FcyR3B, FcRn pH6, and FcRn pH7 non-binding populations compared to the naive library. The x-axis indicates Fc domain residues (EU numbering) and the y-axis indicates enrichment of non-wild-type amino acids. Grey shadings show previously described regions of wild-type IgGl Fc:Fcy receptor interaction (Sazinsky et al., 2008). Fisher's exact test with Bonferroni-adjusted p-value < 0.05 to determine significance.

[0291] Fig. 8: Heatmap illustrating Fc residue importance based on DMS analysis. Columns indicate amino acid positions (EU numbering), rows indicate Fc receptors used for DMS screening. Dark tiles show "high-leverage" residues significantly enriched in the non-binding population, key for Fc-receptor engagement. Grey tiles show "low-leverage" residues not significantly enriched. Fisher's exact test with Bonferroni-adjusted p-value < 0.05 to determine significance. Mutagenesis targeted high-leverage residues. The Fc domain was divided into three fragments to fit the length constraints of IDT DNA Ultramers.

[0292] Fig.9: Degenerate codons utilized in each (FcyR3A-F fragment) of the Fc combinatorial library. Codons were based on amino acid frequencies in the binding population from DMS analysis. Residues not indicated were left as wild-type.

[0293] Fig. 10: Golden Gate assembly approach: 27 fragments combined in a mixed digestion-ligation reaction. Each fragment pool had 70% wild-type and 30% distributed among eight mutational fragments.

[0294] Fig. 11: Mutational distribution across the Fc domain. The x-axis shows Fc domain residues (EU numbering), and the y-axis shows the frequency of non-wild-type amino acids. Grey shadings indicate previously described regions of wild-type IgGl Fc:FcyR3A interaction.

[0295] Fig. 12: Spearman correlation of actual and expected mutational frequency at each residue. Rho and p-values indicated.

[0296] Fig. 13: Sorting the Fc combinatorial library for surface expression and Fc-receptor binding, a, Flow cytometry plots showing surface expression in the Fc combinatorial library (top), 299A-IYG variant (middle), and N297Q variant (bottom). The X-axis indicates surface expression (FLAG tag detection), and the Y-axis indicates side scatter area. The Fc combinatorial library was sorted for FLAG+ cells to enrich Fc domains that express and fold properly, b, Flow cytometry plots of the surface expression-enriched Fc combinatorial library stained with each Fc-receptor in the panel. The X-axis indicates surface expression (FLAG tag detection), and the Y-axis indicates Fc-receptor binding. The surface expression-enriched library was sorted, isolating binding and non-binding populations for each Fc receptor. EXAMPLES

[0297] The invention will be more fully understood by reference to the following examples. They should not, however, be construed as limiting the scope of the invention. It is understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application and scope of the appended claims.

[0298] Example 1: Construction and validation of yeast-based Fc expression platform

[0299] To enable the screening of antibody Fc-variant libraries against Fc receptors, the inventors first sought to develop an antibody Fc surface display system that allows for the efficient coupling of genotype and phenotype at high throughput. Yeast were selected as expression hosts as previous studies have shown isolated Fc domains remain functional even without a Fab domain (Powers et al., J Immunol Methods (2001) 251(l-2):123-35; Wozniak-Knopp et al., Protein Eng Des Sei (2010) 23(4):289-97; Zheng et al., J Immunol (1999) 163(7) :4041-8).

[0300] Human IgGl Fc sequences were ordered as custom, 5. cerevisiae codon-optimized gene fragments (Twist Bioscience) and cloned into the yeast surface display vector (pYDl) by Gibson assembly (Gibson et al., Nat Methods (2009) 6(5):343-5). The pYDl plasmid was modified by creating an Aga2 fusion with a cassette encoding a human IgGl Fc domain, expression tags, and stop codon to create pYDl-hlgGlFc wildtype and variant plasmids (HA tag-Fc-FLAG-Stop) (Fig. 1). HA and FLAG tags were additionally included to enable the detection of Fc surface display. Successful integration of the antibody Fc sequences was verified by Sanger sequencing.

[0301] Subsequently, pYDl-hlgGlFc wildtype and variant plasmids were transformed into 5. cerevisiae yeast (EBY100; ATCC, MYA-4941) as previously described (Benatuil et al., Protein Eng Des Sei (2010) 23(4):155-9). The cells were then grown in SD-CAA medium [20g / L glucose (Sigma, G8270), 8.56g / L NaH2PO4-H2O (Roth, K300.1), 6.77g / L Na2HPO4-2H2O (Sigma, 1.06580.0500), 6.7g / L yeast nitrogen base without amino acids (Sigma, Y0626) and 5g / L casamino acids (Gibco, 223120) in deionized water] for 2 days at 30°C with mild agitation. To induce antibody Fc surface display, the cells were transferred into SG-CAA medium [20g / L galactose (Sigma, G0625), 8.56g / L NaH2PO4-H2O (Roth, K300.1), 6.77g / L Na2HPO4-2H2O (Sigma, 1.06580), 6.7g / L yeast nitrogen base without amino acids (Sigma, Y0626) and 5g / L casamino acids (Gibco, 223120) in deionized water] and incubated at 30°C with mild agitation for 24 hours.

[0302] Subsequently, 5-10 million yeast cells were centrifuged at 6000xg for 3 minutes and washed twice in Wash Buffer [0.5% bovine serum albumin (Sigma, A2153), 2mM EDTA (Biosolve Chemicals, 05142391), and 0.1% Tween20 (Sigma, P9416) in PBS (Pan Biotech, P04-53500)]. Streptavidin-PE-Fc-receptor detection reagents were prepared by mixing streptavidin-PE (Sigma, 42250) with a biotinylated Fc-receptor (ACROBiosystems: FcyRl (FCA-H82E8), FcyR2A-H (CDA-H82E6), FcyR2A-R (CDA-H82E5), FcyR2B (CDB-H82E0), FcyR3A-F (CDA-H82E8), FcyR3A-V (CDA-H82E9), FcyR3B (CDB-H82E4), FcRn (FCM-H82W7)) at a 1:4 molar ratio. The yeast cells were then resuspended in the streptavidin-PE-Fc-receptor detection reagent at a concentration ranging from 0.08-10 pg / mL and incubated shaking at room temperature (RT) for 1 hour in the dark. Cells were washed twice, treated with 1:100 diluted anti-FLAG AF647 (BioLegend, 637316), and incubated with shaking at RT for 30 minutes in the dark. Following this incubation, the cells underwent two additional washes before final resuspension in Wash Buffer for flow cytometric analysis on a BD FACSAria Fusion, BD FACSDiscover S8, or Sony MA900 instrument.

[0303] Yeast expressing wild-type human IgGl Fc were first stained with soluble FcyRl, FcyR2A-H, and FcyR3A-V for flow cytometry analysis (Fig.2). Robust surface expression alongside binding to each of the three receptors was observed, confirming surface display and structural integrity of the Fc domain in the system. To ensure the Fc surface display system accurately represented expected Fc-receptor binding profiles, the inventors generated a panel of four non-glycosylated (aglycosylated) human IgGl Fc variants with known Fc-receptor binding profiles. Each Fc-variant was displayed on the surface of yeast and stained with soluble Fc-receptors to assess their respective binding patterns. Importantly, the Fc-receptor binding profile of each variant was largely recapitulated in the yeast display system. N297Q, an aglycosylated Fc variant expected to retain binding to FcyRl, but to exhibit minimal binding to the low-affinity Fc receptors such as FcyR2A-H, and FcyR3A-V (Chen et al., J Mol Biol (2017) 429(16):2528-2541; Tao & Morrison, J Immunol (1989) 143(8):2595-601; Walker et al., Mol Immunol (1989) 26(4):403-ll), behaved as expected (Fig. 2, Fig. 3). Similarly, S298G / T299A exhibited robust binding to FcgRl and FcyR2A-H, but weakly bound FcyR3A-V as described previously (Fig. 3) (Chen et al., J Mol Biol (2017) 429(16):2528-2541). DTT-IYG bound FcyRl and FcyR3A-V, but very weakly to FcyR2A-H as expected (Chen et al., J Mol Biol (2017) 429(16):2528-2541). Finally, 299A-IYG maintained Fc-receptor binding similarto the wild-type IgGl across all Fc receptors tested, consistent with previous work (Fig. 3) (Chen et al., J Mol Biol (2017) 429(16):2528-2541). Collectively, these data establish a yeast platform for the high-throughput functional screening of antibody Fc variants.

[0304] Example 2: Screening of antibody Fc deep mutational libraries

[0305] After establishing a robust platform for antibody Fc surface display, the inventors systematically mapped Fc sequence-function relationships for an aglycosylated human IgGl Fc domain (299A-IYG), which retains near-IgGl wild-type levels of binding to canonical Fc receptors (Fig. 3) (Chen et al., J Mol Biol (2017) 429(16):2528-2541). An aglycosylated Fc background was used to eliminate species-level differences in antibody glycosylation introduced by the yeast system that may impact Fc-receptor engagement. Given that antibody Fc glycosylation is a primary determinant of Fc function (Jennewein & Alter, Trends Immunol (2017) 38(5):358-372), removing this variable allows for a controlled analysis of the sequence-level impact of Fc mutations on antibody function.

[0306] Previous work using alanine scanning mutagenesis mapped the impact of each solvent-exposed residue in the human IgGl Fc domain on Fc-receptor binding (Shields et al., J Biol Chem (2001) 276(9):6591-604). While alanine scanning provides valuable insights, it is limited to characterizing the effects of substituting each residue with alanine. Moreover, the residues important for Fc-receptor binding in an aglycosylated Fc background remain undefined. Thus, building on this initial work, the inventors performed deep mutational scanning (DMS) of the 299A-IYG aglycosylated Fc domain to survey the functional impact of mutating each residue in the Fc to every other amino acid (Fowler et al., Nat Methods (2010) 7(9):741-6); Fowler & Fields, Nat Methods (2014) ll(8):801-7). DMS provides a comprehensive map of sequence space, enabling a deep understanding of the sequence-function relationships within the antibody Fc domain to be ascertained.

[0307] Single-site deep mutational scanning (DMS) libraries of an aglycosylated human IgGl Fc domain (299A-IYG) (Chen et al., J Mol Biol (2017) 429(16):2528-2541), were generated using a nicking mutagenesis approach as described previously (Wrenbeck et al., Nat Methods (2016) 13(ll):928-930). Mutagenesis was achieved using custom oligo pools ordered from Integrated DNA Technologies (IDT) containing degenerate NNK codons tiled across each position of the 299A-IYG Fc domain spanning from residue 216 to 447 (EU numbering). The alanine at position 299 was kept constant to preserve the aglycosylated framework of the Fc region. DNA fragments encoding the entire human IgGl antibody Fc domain cannot be covered with a single Illumina sequencing read. Thus, two libraries - library 1 and library 2- were individually constructed to cover the first and second halves of the Fc domain respectively. Each DMS library was built separately, but all subsequent steps were performed with pooled sequences from both libraries. Following mutagenesis, the libraries were transformed into E. coli DH5-alpha ElectroMAX (ThermoFisher, 11319019) via electroporation for selection and expansion. Antibody Fc DMS library plasmid was extracted from E. coli according to the manufacturer's instructions (Zymo, D4200). The antibody Fc DMS library plasmid was drop dialyzed on a 0.025pM membrane (Sigma, VSWP02500) for 2 hours using nuclease-free water (Invitrogen, AM9930), and transformed into 5. cerevisiae yeast (EBY100; ATCC, MYA-4941) as described above. The inventors created a DMS library in which NNK degenerate codons were tiled across each position in the 299A-IYG aglycosylated Fc domain spanning residues 216 to 447 (EU numbering), allowing substitution with any of the 20 conventional amino acids. Deep sequencing of the yeast Fc DMS library revealed ~90% coverage of all possible single-site amino acid mutations (Fig. 4).

[0308] Between 5-10 million yeast cells containing antibody Fc DMS library plasmids were prepared and labeled with all eight Fc receptors by the same method as described in Example 1. The inventors used fluorescence-activated cell sorting (FACS) to isolate Fc-receptor binding and non-binding populations through multiple rounds of sorting (Fig. 5). These populations of yeast cells were centrifuged at 3000 RPM for 5 minutes to remove the buffer. The cells were then resuspended in SD-CAA medium and grown for two days at 30°C with mild agitation. To maximize population purity, up to three successive sorting rounds were conducted for each Fc-receptor.

[0309] Example 3: Analysis of antibody Fc DMS libraries by deep sequencing

[0310] Collected populations were deep sequenced to quantify Fc-variant frequency in the binding and non-binding populations compared to the naive (unsorted) library. The antibody Fc plasmid library was extracted from yeast cells according to the instructions of the manufacturer (Zymo, D2004). Subsequently, two consecutive PCR reactions were performed to prepare the library for deep sequencing. The first PCR mixture contained NEBNext Ultra II Q5 Master Mix (New England Biolabs, M0544X), 80ng of library template DNA, and custom primers (Table 1) for library 1 (first half of the Fc domain) and library 2 (second half of the Fc domain) which possessed an 8 base pair unique molecular identifier (UMI) and Illumina partial adaptor sequences (Table 1). PCR 1 conditions were: initial denaturation at 98°C for 5 minutes, followed by 5 cycles of denaturation at 98°C for 30 seconds, annealing at 64°C (Library 1) or 62°C (Library 2) for 90 seconds, extension at 72°C for 30 seconds, and a final extension at 72°C for 5 minutes. 2pL of Exonuclease I (New England Biolabs, M0293L) was added to each PCR product, followed by incubation at 37°C for 75 minutes to remove excess primers. Postexonuclease treatment, PCR products underwent further purification using SPRIselect beads (Beckman Coulter, B23318) according to the manufacturer's instructions. For the second PCR, the mixture contained NEBNext Ultra II Q5 Master Mix (New England Biolabs, M0544X), lOng of library template DNA, and Nextera forward and reverse indexing primers (Illumina, 20027215). The cycling conditions for the second PCR amplification were identical to the first PCR except for two modifications: the annealing temperature was lowered to 61°C and the number of cycles was reduced to 25. PCR products of the expected size were excised and gel extracted according to the manufacturer's instructions (Zymo, D4002). The purified amplicons were pooled to assess sample purity using a Bioanalyzer. After quality control, the pooled library samples underwent paired-end sequencing using a MiSeq v3 600-cycle kit (Illumina, MS-102-3003). Table 1. Primers used in this study

[0311] Name Description Sequence (5'> 3')

[0312] EBI020 Fc elms library 1 NGS primerfwd GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNN NGGCCCATATGATGTGCCCGATTATGCG (SEQ ID N0:7)

[0313]

[0314] EBI024 Fc elms library 1 NGS primer rev TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNN AGCTTTAGAAATGGTCTTTTCGAT (SEQ ID N0:8) EBI025 Fc dms library 2 NGS primerfwd GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNN NGTTTCTAATATCTATGGTCCAGCACCA (SEQ ID N0:9)

[0315]

[0316] EBI023 Fc dms library 2 NGS primer rev TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNN CTTGTCATCATCGTCCTTGTAATC (SEQ ID NQ:10) EBI038 Fc combi lib fragl 2nd strand synthesis TGCGTCGTCTCAAGGTAAC (SEQ ID NO:11)

[0317]

[0318]

[0319] EBI039 Fc combi lib frag2 2nd strand synthesis TGCGTCGTCTCACTTTACCG (SEQ ID N0:12)

[0320] EBI040 Fc combi lib frag32nd strand synthesis TGCGTCGTCTCAAACCCTTG (SEQ ID N0:13)

[0321]

[0322]

[0323] EBI054 Amplify Fc combi library for HR fwd TACCCATACGACGTTCCAGACTACGCAGGATCCGAACCTAAGT CTTGCGATAAG (SEQ ID N0:14)

[0324] EBI055 Amplify Fc combi library for HR rev AACTGGTGGAGTAGTTTTGTAGTTGTTTTCTGGCTGACCATTAG ATTCCCATTC (SEQ ID N0:15) EBI090 Fc combi NGS fwd no spacer GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNN NACTTGTCCACCATGTCCAG (SEQ ID N0:16) EBI091 Fc combi NGS fwd lbp spacer GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNN NTACTTGTCCACCATGTCCAG (SEQ ID N0:17) EBI092 Fc combi NGS fwd 2bp spacer GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNN NGTACTTGTCCACCATGTCCAG (SEQ ID N0:18) EBI093 Fc combi NGS fwd 3bp spacer GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNN NCGAACTTGTCCACCATGTCCAG (SEQ ID N0:19) EBI094 Fc combi NGS rev no spacer TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNN CAGATGGGTAAAAACCCTTGAC (SEQ ID NQ:20) EBI095 Fc combi NGS rev lbp spacer TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNN TCAGATGGGTAAAAACCCTTGAC (SEQ ID N0:21) EBI096 Fc combi NGS rev 2bp spacer TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNN GTCAGATGGGTAAAAACCCTTGAC (SEQ ID NO:22) EBI097 Fc combi NGS rev 3bp spacer TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNN CGACAGATGGGTAAAAACCCTTGAC (SEQ ID NO:23)

[0325]

[0326] Preprocessing of deep sequencing data

[0327] Sequencing reads were paired, quality trimmed, and merged in Geneious (v2023.0.4) using the BBTools suite, with a quality threshold of Qphred R>20. Antibody Fc sequences were then extracted and further processed in R (version 4.2.1). Sequences sharing identical UMIs were collapsed into single consensus sequences to correct for PCR amplification bias, after which antibody Fc sequences were translated into amino acid sequences.

[0328] Deep mutational scanning analysis

[0329] Using the 299A-IYG wild-type sequence as a reference, amino acid substitutions in the deep sequencing data were counted, and variants lacking mutations or with more than one mutation were removed from the dataset. Variant count matrices, both pre- and post-sort, were then analyzed using R (version 4.2.1) for variant enrichment. Variants with fewer than five reads in the pre-sort (unselected) library were assigned a read count of 0 to reduce noise. A pseudocount of 1 was then added to each variant to prevent division by zero in subsequent steps. Pre- and post-selection matrices were normalized by their total read counts to account for differences in sequencing depth. Enrichment scores for individual variants were calculated by dividing the normalized post-selection frequency of each variant by its pre-selection frequency as described previously (Fowler et al., Nat Methods (2010) 7(9):741-6). For positional enrichment scores, the inventors specifically assessed the change in frequency of non-wild type amino acids at each position in the Fc domain. The normalized post-selection frequency of non-wild type (mutated from 299A-IYG) amino acids at each position was divided by their pre-selection frequency, isolati ng the effect of selective pressures on the variation of each residue as previously described (Fowler et al., Nat Methods (2010) 7(9):741-6). Fisher's exact tests were performed to assess the statistical significance of the enrichment observed at each position.

[0330] Analysis of the non-binding population, which contains variants that lose Fc-receptor binding compared to the 299A-IYG base variant, revealed sites on the Fc domain critical for Fc-receptor engagement. Residues significantly positively enriched in each Fc-receptor nonbinding population are likely crucial for interaction with that receptor. Across all Fey receptors evaluated, most positively enriched residues were located in the CH2 domain, highlighting its key role in recognition by Fey receptors (Fig. 6, Fig. 7). Specifically, a significant positive enrichment of variants with mutations in or near the lower hinge and the B / C, C' / E, and F / G loops was observed (Fig. 6, Fig. 7). These regions represent structural elements within the IgGl Fc domain involved in Fc-receptor engagement (Sazinsky et al., Proc Natl Acad Sci USA (2008) 105(51):20167-72; Sondermann et al., Nature (2000) 406(6793):267-73). Additionally, residues outside these areas were identified as critical for Fc binding, emphasizing the value of the unbiased screening approach. For instance, residue L251 located between the lower hinge and B / C loop was found to be critical for the binding of FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V and FcRn at pH 6 (Fig. 6, Fig. 7). Furthermore, several residues in the CH3 domain were important for binding, most commonly for FcRn and FcyR3B (Fig. 7).

[0331] Example 4: Design and construction of Antibody Fc combinatorial libraries

[0332] Results from the antibody Fc DMS screen were used to guide antibody Fc combinatorial library design. Specifically, residues observed to be significantly positively enriched in the nonbinding population for each receptor - which are likely crucial for interaction with that receptor - were identified through positional enrichment analysis as described above. All remaining residues were left as wild-type. The inventors performed this segregation of residues independently for each Fc-receptor, resulting in a map of residue importance spanning residue 232 - 368 (EU numbering) of the Fc domain for each Fc-receptor evaluated (Fig. 8). Residues 369 - 447 (EU numbering) had few high-leverage residues across the panel of Fc-receptors, so this region was left as wild-type enabling facile Illumina-based sequencing of the library.

[0333] To guide mutagenesis towards variants likely to retain or enhance Fc-receptor binding activity, the inventors targeted high-leverage residues with custom degenerate codons reflecting the amino acid residue distributions present in the DMS binding populations, ratherthan standard NNK degenerate codons (Fig.9). The selection of degenerate codons was based on minimizing the mean squared error between the amino acid frequencies determined by a given degenerate codon, and the amino acid frequency observed at that position within the population of binders (Mason et al., Nat Biomed Eng (2021) 5(6):600-612). In instances where the degenerate codon could not encode the wild-type amino acid, the codon was altered to ensure the inclusion of the wild type residue. After generating these libraries in silica, the inventors divided the Fc domain into three different fragments for ease of assembly and to provide enhanced control over library diversity. Library fragments were designed based on the DMS data from each of the eight Fc-receptors, resulting in a total of 24 library fragments which were ordered as single-stranded DNA Ultramers (IDT). All fragments included BsmBI recognition sites for scarless Golden Gate assembly (Engler et al., PLoS One (2008) 3(ll):e3647), with standardized overhangs for each fragment type. Assembly fidelity was optimized using the NEB ligase fidelity viewer tool (https: / / ligasefidelity.neb.com / viewset / run.cgi).

[0334] All 24 single-stranded library fragments were converted to double-stranded DNA through a single-cycle PCR amplification (Table 1), purified via gel extraction, and randomly combined in a single Golden Gate cloning reaction along with three 299A-IYG wild-type fragments (Fig. 10).

[0335] To maintain a controlled edit distance from the 299A-IYG wild type the inventors incorporated 70% of the 299A-IYG fragments (70% 299A-IYG fragment 1, 70% 299A-IYG fragment 2, and 70% 299A-IYG fragment 3) and distributed the remaining 30% evenly among the eight mutational sublibrary fragments. Double-stranded fragment inserts were mixed with 2000ng of a custom entry vector (based on pYTK090, addgene, 65197) in a 3:1 (insert:vector) molar ratio. Then, using the NEBridge Golden Gate Assembly Kit BsmBI-v2 (New England Biolabs, E1602L), all library fragments were randomly assembled into full-length Fc gene segments in a single digestion-ligation reaction. For large-scale Golden Gate reactions, the inventors proportionally scaled the reaction volumes provided by the manufacturer using 26.67pL of the NEB Golden Gate Enzyme Mix and 53.34uLT4 DNA Ligase Buffer. The assembly consisted of 65 cycles of alternating temperatures (42°C for 1 minute, then 16°C for 1 minute), concluded by a final step at 60°C for 15 minutes. The assembled library was then transformed into E. coli MC1061F cells (Biosearch Technologies, 60514-1), resulting in approximately 6xl08transformants with an estimated 93% rate of correctly assembled Fc sequences. Plasmid DNA from the Fc combinatorial library was then extracted from E. coli cells according to the instructions of the manufacturer (Zymo, D4200). Deep sequencing of the library revealed a median edit distance of 6 from the 299A-IYG wild-type Fc domain. Mutations were commonly located in or near the lower hinge and the B / C, C' / E, and F / G loops as expected (Fig. 11). The inventors additionally observed strong and significant correlations between the location of mutations present in the actual library and the expected location of mutations based on the library design, both overall and on a per-fragment basis (Fig. 12).

[0336] The assembled combinatorial Fc library was then PCR amplified (Table 1), and a custom yeast display vector (based on pYDl) containing residues 369 to 417 of the Fc domain (EU numbering), was linearized using the restriction enzyme EcoRI (New England Biolabs, R3101L). Both the library insert and digested vector were gel-purified according to the instructions of the manufacturer (Zymo, D4002). Importantly, the Fc combinatorial library insert contained 81 base pairs of overlapping sequences with the linearized destination vector to enable assembly via homologous recombination in yeast. The library insert and linearized vector were co-transformed into 5. cerevisiae yeast (EBY100; ATCC, MYA-4941) for assembly via homologous recombination using a previously described protocol with minor changes (Benatuil et al., Protein Eng Des Sei (2010) Apr;23(4):155-9). In brief, EBY100 was grown overnight in YPD media [20g / L glucose (Sigma, G8270), 20g / L vegetable peptone (Sigma, 19942), and lOg / L yeast extract (Sigma, Y1625) in deionized water] at 30°C. The next day, cells from this culture were diluted into 500mL of YPD to reach an ODeoo of 0.3 and grown until an ODeoo of approximately 1.6 was reached. 5mL of Tris-DTT [IM Tris pH8 (Thermo Scientific, J22638.K2), IM dithiothreitol (Sigma, D0632)] and 25mL of LiAc-TE [TE Buffer (Invitrogen, AM9849), 2M lithium acetate (Sigma, 517992)] were then added to the cells and incubated for another 15 minutes at 30°C with mild agitation to condition the cells, followed by three washes in cold Electroporation Buffer [0.6g Tris base (Sigma, T1503), 91.09g sorbitol (Sigma, S6021), 73.5mg CaCl2 dihydrate (Sigma, 1.02382.0500) in deionized water], Electrocom petent EBY100 yeast cells were transformed with 83.33pg of the Fc library insert and 16.67pg of the linearized vector using 1mm electroporation cuvettes (Cell Projects, EP-101). Postelectroporation, cells were allowed to recover for 1 hour in YPD at 30°C with mild agitation, then transferred to a selective SD-CAA medium for incubation overnight. After overnight incubation at 30°C with mild agitation, the library concentration was adjusted back to an ODeoo of 1 and subjected to another overnight growth phase. Colony-forming unit (CFU) plating was performed to assess library diversity, yielding approximately 4.1xl08transformants after three days. Example 5: Screening and deep sequencing of antibody Fc combinatorial libraries

[0337] The inventors next aimed to isolate and characterize library subpopulations with distinct binding profiles to each Fc receptor. However, flow cytometry analysis revealed that only approximately 16% of the Fc combinatorial library exhibited surface expression, reflecting stable Fc variants capable of proper folding (Fig. 13). By contrast, approximately 63% of the wild-type (299A-IYG) and N297Q variants displayed surface expression. Thus, to enrich for surface-expressing Fc variants and eliminate those that were not functional, the inventors sorted the naive Fc combinatorial library for expression, resulting in over lxlO8FLAG-positive cells, which comprised the expression-enriched library.

[0338] The antibody Fc combinatorial plasmid library was extracted from yeast cells according to the instructions of the manufacturer (Zymo, D2004). Subsequently, two consecutive PCR reactions were performed to prepare the library for deep sequencing. The first PCR mixture contained NEBNext Ultra II Q5 Master Mix (New England Biolabs, M0544X), 80ng of template DNA, and custom-designed primers which possessed an 8 base pair unique UMI, a l-3bp heterogeneity spacer (Fadrosh et al., Microbiome (2014) Feb 24;2(1):6), and Illumina partial adaptor sequences (Table 1). 2pL of Exonuclease I was added to each PCR product, followed by incubation at 37°C for 75 minutes to remove excess primers. Post-exonuclease, PCR products underwent a clean-up procedure using SPRIselect beads (Beckman Coulter, B23318) according to the instructions of the manufacturer. For the second PCR, the mixture contained NEBNext Ultra II Q5 Master Mix (New England Biolabs, M0544X), lOng of template DNA, and Nextera forward and reverse indexing primers (Illumina, 20027215). PCR products underwent another round of clean-up using SPRIselect beads (Beckman Coulter, B23318). The purified amplicons were pooled and sample size and purity were verified on a fragment analyzer (Advanced Analytical Technologies) using a DNF-473 standard sensitivity next-generation sequencing fragment analysis kit. he pooled library samples then underwent paired-end sequencing using a MiSeq v3 600-cycle kit (Illumina, MS-102-3003) or NovaSeq SP 500 kit (Illumina). Example 6: Deep learning for predictive and generative modeling of Fc sequence

[0339] Combinatorial antibody Fc libraries were used to generate training, validation, and test data to develop machine-learning models that predict antibody Fc-receptor binding profiles and that generate antibodies with specific functional properties based on their Fc sequence.

[0340] Initially, data are preprocessed to format them suitably for machine learning applications. Sequences in the antibody Fc combinatorial library containing novel, unintended N-linked glycosylation motifs (N-X-T / S) introduced during mutagenesis are excluded to mitigate the potential impacts of N-linked glycosylation on Fc function. The filtered, aglycosylated antibody Fc sequences are then one-hot or categorically encoded depending on the training architecture and then used as inputs for model training. Multiple training architectures are evaluated, including convolutional neural networks (CNNs), transformers, and variational auto-encoders (VAEs), among others. For CNNs, the sequences are presented as two-dimensional matrices (one-hot encoded), while for other models, including transformers and VAEs, sequences are categorically encoded into one-dimensional vectors. All models are built in Python and incorporate various frameworks such as PyToch, TensorFlow, and Scikit-Learn, along with data processing and visualization libraries including numpy, pandas, matplotlib, and seaborn. The models are tasked with predicting Fc receptor binding activity (enhanced, wild type level, or abrogated) based on Fc sequence. Special attention is given to balancing the training data, ensuring fair representation across classes, and performance is assessed using metrics such as accuracy, Fl score, and MCC on cross-validation sets during training and held-out test data after the training has been completed.

[0341] Beyond prediction, the inventors extended their computational approach to generative modeling to design novel Fc sequences with specific binding profiles.

[0342] Using pre-trained models and / or by training models ab initio such as variational autoencoders- and transformer-based architectures, the inventors reverse-engineer the sequence-to-function relationship. The inventors use language and / or autoregressive modeling during training to capture the latent and long-range interactions between the amino acids in the Fc sequence for different Fc-receptor binding profiles. By capturing hidden relationships, these models not only understand high-order sequence patterns but also can generate novel Fc sequences meeting predefined binding criteria. Generated Fc sequences are assessed computationally for their novelty, diversity, and predicted binding profiles, ensuring their alignment with the desired biological functionalities.

[0343] As an additional validation, both predictive and generative models are iteratively refined, with generated sequences being fed back into the predictive models. This cyclical approach enables a robust exploration of the sequence space guided by empirical Fc-receptor binding data, providing a holistic computational-experimental framework for the rational design of Fc regions with customized receptor binding profiles.

Claims

CLAIMS1. A method for screening engineered antibody Fc regions having modified Fc receptor binding properties, the method comprising the steps of:a) providing a plurality of cells displaying engineered antibody Fc regions on their cell surface;b) contacting the cells in step (a) with an Fc receptor;c) separating cells that have been contacted with the Fc receptor in step (b) into two or more cell populations according to their ability to bind to the Fc receptor; and d) sequencing cells from at least one of the populations obtained in step (c) to identify mutations in the engineered antibody Fc region.

2. The method according to claim 1, wherein steps (b) - (d) are repeated for one or more Fc receptors selected from the group consisting of: FcyRl, FcyR2A-H, FcyR2A-R, FcyR2B, FcyR3A-F, FcyR3A-V, FcyR3B or FcRn.

3. The method according to claim 1 or 2, wherein in step (d), the at least one cell population, or a sample thereof, is subjected to deep sequencing.

4. The method according to claim 3, wherein deep sequencing comprises a step of determining the frequency of individual mutations in the engineered antibody Fc region in a population.

5. The method according to claim 4, wherein the impact of an individual mutation on the binding of the engineered antibody Fc region to the Fc receptor is determined based on the frequency of said individual mutation in the population.

6. The method according to any one of claims 1 to 5, wherein in step (d) mutations in theengineered antibody Fc region are identified that (i) confer improved binding to the Fc receptor, (ii) do not affect binding to the Fc receptor, and / or (iii) confer reduced binding to the Fc receptor.

7. The method according to any one of claims 1 to 6, wherein the sequence information obtained in step (d) is used to train a machine learning model.

8. The method according to claim 7, wherein the machine learning model is used to (a) predict the Fc receptor binding properties of an engineered antibody Fc region; and / or (b) generate an engineered antibody Fc region having desired Fc receptor binding properties.

9. The method according to any one of claims 1 to 8, wherein each cell comprised in the plurality of cells displays an engineered antibody Fc region comprising at least one mutation10. The method according to any one of claims Ito 9, wherein the plurality of cells has been obtained by introducing a DNA library encoding engineered antibody Fc regions into the cells.

11. The method according to claim 10, wherein the DNA library encodes a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the library comprises a single mutation or a combination of two or more single mutations.

12. The method according to claim 10 or 11, wherein the DNA library is a deep mutational scanning (DMS) library encoding a plurality of engineered antibody Fc regions, wherein each engineered antibody Fc region encoded in the DMS library comprises a single mutation, preferably wherein the DMS library encodes at least 70%, at least 80% or at least 90% of all possible single mutants of the antibody Fc region.

13. The method according to claim 10 or 11, wherein the DNA library is a combinatoriallibrary encoding a plurality of engineered antibody Fc regions, preferably wherein each engineered antibody Fc region encoded in the combinatorial library comprises two or more mutations.

14. The method according to claim 13, wherein the two or more mutations are at positions that have been previously identified as being involved in the binding of one or more Fc receptor.

15. The method according to claim 13 or 14, wherein obtaining the combinatorial library comprises a step of assembling two or more DNA fragments encoding parts of an engineered antibody Fc region, preferably wherein each of the DNA fragments comprises at least one degenerate codon.

16. The method according to claim 15, wherein one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a first Fc receptor, and another one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a second Fc receptor.

17. The method according to any one of claims 13 to 16, wherein the two or more mutations are in positions 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446, preferably in positions 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering) of the antibody Fc region.

18. The method according to any one of claims 1 to 17, wherein the ability of an engineered antibody Fc region to bind to the Fc receptor is determined by comparison with areference antibody Fc region.

19. The method according to any one of claims 1 to 18, wherein cells in step (c) are separated by fluorescence-activated cell sorting (FACS).

20. The method according to claim 19, wherein the antibody Fc regions are labelled with a first fluorescent dye and the Fc receptor is labelled with a second fluorescent dye.

21. The method according to any one of claims 1 to 20, wherein the engineered antibody Fc regions have been derived from an aglycosylated human IgGl antibody Fc domain.

22. The method according to claim 21, wherein the aglycosylated human IgGl antibody Fc domain comprises mutations T299A / K326I / A327Y / L328G (EU numbering).

23. The method according to any one of claims 1 to 22, wherein the engineered antibody Fc regions are displayed on the surface of a yeast cell.

24. A method for identifying engineered antibody Fc regions with altered Fc receptor binding characteristics, the method comprising the steps of:a) identifying amino acid residues of an antibody Fc region that are involved in binding to one or more Fc receptors;b) introducing mutations in at least two positions of the antibody Fc region encoding the amino acid residues identified in step (a) to obtain an engineered antibody Fc region;c) testing the engineered antibody Fc region obtained in step (b) for binding to one or more Fc receptors; andd) determining the engineered antibody Fc region to have altered Fc receptor binding characteristics based on the outcome of the testing in step (c).

25. The method according to claim 24, wherein the amino acid residues of the antibody Fc region that are involved in binding to one or more Fc receptors are identified using deep mutational scanning and / or alanine scanning.

26. The method according to claim 24 or 25, wherein the engineered antibody Fc region obtained in step (b) is encoded by a combinatorial library encoding a plurality of engineered antibody Fc regions.

27. The method according to claim 26, wherein the combinatorial library is obtained by assembling two or more DNA fragments encoding parts of an engineered antibody Fc region, preferably wherein each of the DNA fragments comprises at least one mutation, more preferably wherein each of the DNA fragments comprises at least one degenerate codon.

28. The method according to any one of claims 24 to 27, wherein one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a first Fc receptor, and another one of the two or more DNA fragments encoding a part of the engineered antibody Fc region comprises one or more mutations at positions that have been identified as being involved in binding of a second Fc receptor.

29. The method according to any one of claims 24 to 28, wherein mutations are introduced in at least two of the following positions of the antibody Fc region: 232, 233, 234, 235, 236, 237, 238, 242, 250, 251, 252, 253, 254, 257, 262, 263, 265, 266, 267, 269, 270, 271, 272, 273, 275, 276, 278, 287, 288, 292, 295, 297, 298, 300, 301, 302, 304, 307, 310, 312, 313, 314, 322, 324, 325, 326, 327, 328, 329, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 346, 348, 362, 364, 366, 368, 373, 374, 376, 378, 380, 395, 405, 406, 407, 409, 428, 429, 431, 432, 433, 435, and / or 446, preferably in positions 237, 238, 251, 257, 263, 265, 266, 267, 270, 295, 297, 301, 302, 304, 307, 324, 325, 326, 328, 329, 332, 334, 336, and / or 338 (all according to EU numbering).

30. The method according to any one of claims 24 to 29, wherein testing the engineered antibody Fc region for binding to one or more Fc receptors involves a step of contacting the engineered antibody Fc region with one or more Fc receptors.

31. The method according to any one of claims 24 to 30, wherein the binding of the engineered antibody Fc region to one or more Fc receptors is tested by flow cytometry.

32. The method according to any one of claims 24 to 31, comprising an additional step of determining the amino acid sequence of the engineered antibody Fc region, preferably by sequencing, more preferably by deep sequencing.

33. A method for predicting the Fc receptor binding properties of an engineered antibody Fc region, the method comprising the steps of:a) training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors;b) providing a nucleic acid sequence or an amino acid sequence of the engineered antibody Fc region to the trained machine learning model; andc) obtaining, from the machine learning model, the predicted Fc receptor binding properties of said engineered antibody Fc region provided in step (b).

34. A method for generating an engineered antibody Fc region having desired Fc receptor binding properties, the method comprising the steps of:a) training a machine learning model using binding data obtained from a plurality of engineered antibody Fc regions and one or more Fc receptors; andb) utilizing the trained machine learning model to generate a nucleic acid sequence or an amino acid sequence encoding an engineered antibody Fc region with the desired Fc receptor binding properties.

35. The method according to claim 33 or 34, wherein the binding data is obtained using the method according to any one of claims 1 to 32.

36. The method according to any one of claims 33 to 35, wherein the machine learning model is trained with binding data that has been obtained with at least two different Fc receptors.

Images