WO2024062161A1 - Antibody discovery and development method - Google Patents

Abstract

A method to introduce a mutation to a variable domain of an antibody or its fragment by introducing at least one peptide linker to close proximity of the variable domain, wherein the peptide linker is encoded by a nucleic acid forming a G quadruplex (G4) structure and the G4 structure promotes activity of activation-induced cytidine deaminase (AID).

Description

ANTIBODY DISCOVERY AND DEVELOPMENT METHOD

TECHNICAL FIELD

The present invention relates to a method to introduce mutations to variable domains of an antibody or its fragment by introducing at least one peptide linker to close proximity of variable domains, wherein said peptide linker is encoded by a nucleic acid forming a G-quadruplex (G4) structure and said G4 structure promotes activity of activation-induced cytidine deaminase (AID). The present invention is advantageous in screening, discovering or maturating antibodies, their fragments or their variants with in vitro display methods.

BACKGROUND OF THE INVENTION

In vitro display methods enable precisely controlled selection and screening conditions for antibody discovery, and they make it possible to select binders against different types of target molecules from a large pool of variants. By forming a physical link between an antibody, or an alternative antibody format, displayed on the cell surface and the genetic information encoding it, it allows isolation of the genes that encode a protein with the desired binding function. With this information it is possible to further improve the antibodies by creating alternative constructs with improved biophysical properties. Multiple display technologies have been developed such as bacterial, baculovirus, yeast, ribosomal, phage and mammalian display (Bradbury et al. 2011; Elgundi et al. 2017)

Phage display is a widely used platform for antibody discovery, because of its low cost, adaptability, and efficiency (Kang and Lee 2021). It enables antibody production in vitro easily and with high diversity (Nagano and Tsutsumi 2021). In addition, phage display may be used for discovery of antibodies, antibody fragments and antibody variants against nearly any targets including small molecules and their chemical modifications, proteins and subtle changes in them, toxins, pathogens, non-immunogenic, highly conserved antigens, or even RNA molecules which are not receptive to nucleotide probes (Bradbury et al. 2011). Phage display has also had an important role in creating monoclonal antibodies with fully human sequences to reduce their immunogenicity in humans (Nagano and Tsutsumi 2021). In vitro display methods play also important role in antibody maturation to improve antibody features like affinity and specificity (Bradbury et al 2011).

Antibody libraries of over 1 x 10¹¹ clones have previously been created with the phage display system for antibody generation and characterization, proving its role as an efficient tool for antibody discovery (Almagro et al. 2019; Tiller et al. 2013). Phage display can also be used to narrow down libraries before reformatting them for other cell display systems (Parthiban et al. 2019). However, it has been shown that antibodies discovered with phage display may have more undesired biophysical characteristics than those derived from mammalian sources (Jain et al. 2017).

Mammalian cell display method enables controlled selection and screening conditions for antibody discovery, making it possible to select binders against different targets from a large pool of variants. Mammalian cell display can also overcome limitations of phage display related to post-translational modifications of the antibodies. Mammalian cell display may be used to engineer antibody fragments like scFvs, fragment antigen-binding regions (Fab) or even whole functional antibodies with the required post-translational modifications (Ho and Pastan 2009).

When the human immune system encounters a new foreign antigen, it has a limited number of B cells available to produce and present unique antibodies against the possible threat. To answer this limited spectrum of different antibodies, the immune system selects the B cells producing low affinity antibodies specific to the antigen, activates them and initiates somatic hypermutation of their antibodies. (Bowers et al. 2014.) Somatic hypermutation is a process where point mutations are purposefully generated within the immunoglobulin gene of activated proliferating B cell creating genetic diversity and leading to the affinity maturation of antibodies. As more antibodies against the target are produced, the antigen becomes limiting and the B cells producing antibodies with higher affinity have the selective advantage. Somatic hypermutation prefers antibody genes relative to other genes, which prevents harmful genomic mutations to the B cell. (Buerstedde et al. 2014.)

Somatic hypermutation requires the transcription of the antibody gene and the expression of the activation-induced cytidine deaminase (AID) protein, which is encoded by the AICDA gene. AID inflicts point mutations to the single stranded DNA by converting deoxycytidines to deoxyuridines by deamination, causing the base pair C-G to change to a U-G mismatch which is then converted to T-A. This regulated DNA damage may be repaired by other DNA repair mechanisms to promote other mutations resulting in amino acid substitutions in the antibodies (Bowers et al. 2011; Luo et al. 2020; Sheppard et al. 2018; Tang et al. 2020). Some sequence motifs, referred to as hotspots, can act as activators for AID by attracting the enzyme and thus mutations (Tang et al. 2020). Together with the error-prone DNA repair mechanisms, the process of deamination by AID can lead to various mutations in the DNA.

The wild-type AID enzyme has a nuclear localization signal (NLS) and a nuclear export signal (NES), catalytic domain which interacts with the enzyme substrate, and apolipoprotein B mRNA editing catalytic (APOBEC) protein-like domains enabling the cytosine deamination (Cervantes- Gracia et al. 2021; Luo et al. 2020). In B-cell specific gene diversification AID induces somatic hypermutation, immunoglobulin gene conversion and immunoglobulin class switch recombination (Buerstedde et al. 2014). Codon insertions and deletions have also been suggested to be generated via AID-mediated mutagenesis (Bowers et al. 2011).

Somatic hypermutation by AID may be coupled to mammalian cell display method (Bowers et al. 2011, Bowers et al 2014, W02008/103474A1; Luo et al. 2020). By first isolating the cells expressing antigen binding antibodies or displaying one known antibody on the cells and then transfecting them with AID, mutations and changes have been observed in the binding properties of antibodies by sequencing the antibody gene (Bowers et al. 2011). To affinity maturate the antibodies, the cells can go through multiple rounds of evolution consisting of transfection of AID, cultivation of the cells as the AID induces mutations during cell proliferation, and then enriching the cells displaying the antibody with improved properties by isolating them (Luo et al. 2020).

AID enzyme has been improved by re-engineering its protein structure, introducing point mutations (Wang et al. 2009), and optimizing the nucleic acid sequence to increase the mutation rate and number of mutation types. By removing the NES signal it has been possible to accumulate AID in the nucleus and increase the mutation rate. Also altering the way of implementing repeated transfections and antibiotic selection during rounds of antibody screening influences AID activity (Luo et al. 2020). Improving the enzyme activity would reduce the number of rounds AID must be introduced to the antibody genes and simplify the process before achieving antibodies with desired properties.

Single stranded DNA and RNA can form G-quadruplex (G4) structures in repetitive G-rich sequences, such as telomeric nucleic acids at linear chromosome ends (Nakanishi and Seimiya 2020). The ability of guanine to Hoogsteen hydrogen bond with itself leads to ring-like formations known as G-quartets of four guanines each forming hydrogen bonds with its two neighbours. These quartets can stack to form G4 structures stabilized by monovalent cations K+ and Na+. Different G4 structures are classified into parallel, antiparallel, and hybrid types according to the direction of the four strands (Lerner and Sale 2019; Mukundan and Phan 2013; Nakanishi and Seimiya 2020). G4 structures have also been shown to form even more complicated conformations containing longer loops and bulges in the structure (Mukundan and Phan 2013). G4 structures were first discovered in vitro but more evidence has arisen that they are also formed in living cells genome-wide, attracting interest because of their potential involvement in various biological processes (Lerner and Sale 2019; Nakanishi and Seimiya 2020). The role of G4s have been linked to regulating DNA replication by stalling DNA replication forks and affecting replication origin activity. G4s may also form during transcription and recruit transcription factors or inhibit the progression of RNA polymerase II. During translation, G4 structures formed in mRNA have shown to repress translation by inhibiting the progression of ribosomes or the recruitment of translation initiation factors (Nakanishi and Seimiya 2020). It seems that naturally formed G4 structures in cellular DNA may recruit AID to certain regions (Tang and McCarthy 2021; Honkonen 2021).

SUMMARY OF INVENTION

The current invention relates to a method to introduce a mutation to a variable domain of an antibody or its fragment by introducing at least one peptide linker to close proximity of the variable domain, wherein the peptide linker is encoded by a nucleic acid forming a G-quadruplex (G4) structure and the G4 structure promotes activity of activation-induced cytidine deaminase (AID).

The current invention relates to a method to introduce a mutation to a variable domain of an antibody or its fragment by introducing one peptide linker to close proximity of the variable domain, wherein the peptide linker is encoded by a nucleic acid forming a G4 structure and the G4 structure promotes activity of AID.

The current invention also relates to a method to introduce a mutation to a variable domain of an antibody or its fragment by introducing two peptide linkers to close proximity of the variable domain, wherein the peptide linkers are encoded by nucleic acids forming a G4 structure and the G4 structure promotes activity of AID.

The current invention also relates to a method to introduce a mutation to a variable domain of an antibody or its fragment by introducing three peptide linkers to close proximity of the variable domain, wherein the peptide linkers are encoded by nucleic acids forming a G4 structure and the G4 structure promotes activity of AID.

In some embodiments of the invention the nucleic acid forming a G4 structure is located upstream of a variable domain encoding nucleic acid, downstream of a variable domain encoding nucleic acid and/or between variable domains encoding nucleic acids.

In some embodiments of the invention the nucleic acid forming a G4 structure is located within 20 nucleotides, 15 nucleotides, 10 nucleotides, 5 nucleotides, 4 nucleotides, 3 nucleotides, 2 nucleotides or 1 nucleotide of a variable domain encoding nucleic acid.

In a preferred embodiment of the invention the nucleic acid forming a G4 structure is located adjacent to a variable domain encoding nucleic acid.

In some embodiments of the invention the nucleic acid forming a G4 structure increases a mutation rate of a variable domain at least 3%, 4%, 5%, 6% or 7% compared to a variable domain without the nucleic acid forming a G4 structure in the close proximity.

In some embodiments of the invention the nucleic acid forming a G4 structure increases a mutation rate most in a CDR domain closest to the nucleic acid forming a G4 structure.

In a preferred embodiment the CDR domain is CDR3. In some embodiments of the invention the length of the nucleic acid forming a G4 structure is 15 to 105 nucleotides.

In a preferred embodiment of the invention the length of the nucleic acid forming a G4 structure is 60 nucleotides.

In some embodiments of the invention the nucleic acid forming a G4 structure comprises a Guanine-rich (G-rich) sequence in a coding strand of the nucleic acid or in a template strand of the nucleic acid.

In a preferred embodiment of the invention the nucleic acid forming a G4 structure comprises at least 58%, 60%, 65%, 70%, 75% or 78% guanine nucleotides in a coding strand of the nucleic acid or in a template strand of the nucleic acid.

In another preferred embodiment of the invention the nucleic acid forming a G4 structure comprises at least 58%, 60%, 65%, 70%, 75% or 78% and at most 95% or 90% of guanine nucleotides in a coding strand of the nucleic acid or in a template strand of the nucleic acid.

In a preferred embodiment of the invention the nucleic acid forming a G4 structure comprises or consists a nucleic acid according to SEQ ID NO: 2.

In a preferred embodiment of the invention the variable domain is a variable heavy domain or/and a variable light domain.

In a preferred embodiment of the invention the variable light domain, the variable heavy domain and the peptide linker form a single-chain variable fragment (scFv).

In some embodiments of the invention the variable light domains, the variable heavy domains and the peptide linkers form a di-scFv or a tri-scFv.

In some embodiments of the invention the scFv, the di-scFv or the tri-scFv is connected to an Fc domain by a peptide linker, a hinge domain or their fragments. The current invention also relates to a method, wherein the nucleic acid encoding a variable domain of an antibody or its fragment and the nucleic acid forming a G4 structure are transfected to a cell.

In a preferred embodiment of the invention the nucleic acid encoding an activation-induced cytidine deaminase (AID) is transfected to a cell.

In some embodiments of the invention the nucleic acid encoding a variable domain of an antibody or its fragment and the nucleic acid forming a G4 structure are transfected stably to a cell and the nucleic acid encoding an AID is transfected transiently to a cell.

In a preferred embodiment of the invention the cell is a CHO cell.

In a preferred embodiment of the invention the cell expresses AID.

The current invention also relates to a vector comprising a nucleic acid sequence encoding an antibody variable heavy domain, a nucleic acid encoding an antibody variable light domain, and at least one nucleic acid forming a G4 structure and encoding a peptide linker.

In some embodiments of the invention the nucleic acid forming a G4 structure in the vector is located upstream or downstream of the variable light domain or the variable heavy domain encoding nucleic acid or between the variable heavy domain and the variable light domain encoding sequences.

In some embodiments of the invention the vector further comprises a nucleic acid encoding AID.

The current invention also relates to a kit comprising a vector comprising a nucleic acid encoding an antibody variable heavy domain, a nucleic acid encoding an antibody variable light domain, and at least one nucleic acid forming a G4 structure and a second vector encoding AID.

In a preferred embodiment of the invention the vector is a plasmid. The method of the present invention is advantageous in that mutations can be targeted to variable domains of an antibody by introducing the peptide linker according to the invention to close proximity of the variable domains. There is no need to change the sequences of the variable domains for example by codon optimization in order to target the mutations. The method can be used for introducing mutations to an antibody library containing different antibody variable domains. By using the method of the present invention in the antibody affinity maturation for in vitro display method, such as mammalian cell display, the size and the diversity of the created antibody library can be increased.

BRIEF DESCRIPTION OF DRAWINGS

Figure 1. Different variants of AIDs cloned to the pCEP4 vector. A) Variant was codon optimized for CHO cells (hAID/mAID CO). B) Variant as in A) with deletion of the NES (hAID/mAID del). C) Variant as in B) with point mutations K10E, T82I and E156G (Wang et al. 2009, Luo et al 2020) (hAID/mAID mut).

Figure 2. AID expression analysis by Western blot. Each cell sample transfected with different AID variants (codon optimized (CO), codon optimized with deletion (del), and codon optimized with deletion and point mutations (mut)) were analysed with western blot. The numbers indicate if the sample was taken three days (1) or a month (2) after transfection. The control sample (ctrl) used was from the same cell line without any AID transfection. The wild-type human and mouse AID variants were run as reference as they do not contain the FLAG-tag.

Figure 3. Positions of the two gates determined with the control cells scFv. The signal area produced by the stain antibody (y-axis) is presented in relation to forward scatter area (x-axis). The first gate is positioned where the signals are the highest and the second gate is positioned below it.

Figure 4. Examples of the two gates in cell sorting of AID transfected cells. The cell lines contained scFv_G4 (top left), scFv (top right), and scFv+E. Each cell line was transfected with mAID CO. The signal area produced by the stain antibody (y-axis) is presented in relation to forward scatter area (x-axis). Figure 5. Amino acid changes in each sequence position of VL domain (amino acids 1-112), VH domain (amino acids 133-250) and peptide linker (113-132). Complementarity-determining regions (CDRs) are marked with lines, so that CDR1-VL (CDRL1) is amino acids 24-39, CDR2- VL (CDRL2) is amino acids 55-61, CDR3-VL (CDRL3) is amino acids 94-100, CDR1-VH (CDRH1) is amino acids 163-167, CDR2-VH (CDRH2) is amino acids 182-198, and CDR3-VH (CDRH3) is amino acids 231-239. Potential G-quadruplex (G4) structure forming nucleic acid sequences are presented in the figure with boxes at the corresponding amino acid locations. These locations are 6-17, 10-18, 82-91, 106-138, 135-145, 139-148, 171-182. The number of mutations has been calculated in relation to all the sequencing reads in the sample group (y-axis) and are presented by the amino acid position in the sequence (x-axis). Background mutations derived from the control sample has been subtracted from the values.

Figure 6. A summary of different combinations and locations of nucleic acids encoding a variable light domain, a variable heavy domain and a peptide linker (nucleic acid forms a G4 structure) present in the vector. Following combinations and locations are examples and should not be considered as limitations for other combinations and locations: (i) - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain - (ii) - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain - (iii) - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - optionally hinge domain - optionally Fc domain - (iv) - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - optionally hinge domain - optionally Fc domain - (v) - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain - (vi) - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain -.

DESCRIPTION OF THE INVENTION

Features and embodiments of the current invention are described by way of nonlimiting examples in the disclosure. The present disclosure should not be considered as limitation to particular methods, compounds, compositions, uses described in the disclosure. It should be understood that a skilled person may make apparent modifications and variations to the current invention and embodiments. Singular forms a, an, the used in the application refers one or more.

To practice the current invention and embodiments the skilled person may use common techniques and methods of biology, biochemistry, chemistry, molecular biology, microbiology, and immunology. Common techniques and methods are described in literature, for example in laboratory manuals and laboratory protocols. Such literature is for example Current Protocols in Cell Biology, Current Protocols in Immunology, Current Protocols in Molecular Biology, Current Protocols in Microbiology, Molecular cloning: A Laboratory Manual. The used technical and scientific terms have the meaning commonly understood by a skilled person based on scientific literature and technology dictionaries.

Antibodies and their fragments

Antibody refers to an immunoglobulin specifically binding to an epitope of an antigen. Antibody is a protein formed from four polypeptides. Antibody has two light chains and two heavy chains. Each light chain is formed from a variable light domain (VL) and a constant light domain (CL) and each heavy chain is formed from a variable domain (VH) and three constant domains (CHI, CH2, CH3). In the heavy chain a hinge domain connects VH and CHI domains to CH2 and CH3 domains. Heavy chains are connected to each other by disulfide bridges located in the hinge domains of the heavy chains. Light chains are connected to heavy chains by disulfide bridges between CHI and CL. Antigen binding fragment (Lab) is formed from VH, VL, CHI and CL domains. Crystallizable fragment is formed from two CH2 and two CH3 domains. Variable regions of heavy and light chains (VH and VL) form variable fragment (Ev). Both VH and VL domains have three complementarity determining regions (CDR1-VL, CDR2-VL, CDR3-VL, CDR1-VH, CDR2-VH and CDR3-VH). CDR sequences determine the binding specificity and affinity of an antibody or an antibody fragment to an antigen. In different antibodies CDR3-VH shows the most variability in both length and amino acid sequence and can be longer than the other CDRs. CDR3-VH thus usually plays a primary role in the antibody-antigen interactions. Antibody structural and functional features are reviewed for example by Chiu et al. 2019. The antibody may be a monoclonal antibody or a polyclonal antibody. Term antibody includes without limitation chimeric antibodies, humanized antibodies, bispecific antibodies, antibody fragments including without limitation nanobodies, camelid antibodies, antigen-binding fragments (Fab), variable fragments (Fv), variable heavy domains (VH), variable light domains (VL), bivalent Fab regions (F(ab’)2), single chain antibody fragments (scAb), single chain variable fragments (scFv), di-scFVs, tri-scFvs, bivalent scFv (sc(Fv)2) and antibody fragment comprising fusion proteins.

In the current invention an antibody fragment is preferably a single-chain variable fragment (scFv). The scFv comprises a variable light domain (VL) and a variable heavy domain (VH), which are connected by a peptide linker. Both VH and VL domains have three complementarity determining regions (CDR1-VL, CDR2-VL, CDR3-VL, CDR1-VH, CDR2-VH and CDR3-VH). Thus, the scFv contains a complete antigen binding site. The linker combining VH and VL domains is a flexible peptide linker. The ScFv may be combined to another scFv with another peptide linker forming a di-scFv. A third scFv may be combined to a di-scFv forming a tri-scFv. An scFv may also be combined to an Fc region with a linker forming a scFv-Fc.

In a preferred embodiment antibody fragments comprising an scFv is used in antibody discovery, screening or maturation with an in vitro display method. In a preferred embodiment antibody fragments comprising an scFv are used in antibody discovery, screening or maturation with a mammalian display method.

In the literature scFvs are characterized monovalent and small in size. ScFvs have a short in vivo half-life and low functional affinity. Despite any undesired properties for a therapeutic agent, they are exceptionally stable, cost effectively expressed and easily genetically engineered (Bradbury et al. 2011; Weisser and Hall 2009). Fusing a fragment crystallizable region (Fc) to an scFv increases its stability by lengthening its half-life, enhances its activity and is convenient in purification steps (Strube and Chen 2004; Weisser and Hall 2009). scFv and scFv-Fc fusion proteins are useful for antibody discovery purposes.

Nucleic acids, nucleotides Nucleic acids are biopolymers composed of nucleotides. The two main classes of nucleic acids are deoxyribonucleic acid (DNA) and ribonucleic acid (RNA). A nucleotide consists of a nitrogenous base, a sugar, and a phosphate group. In DNA, the sugar is deoxyribose, and in RNA, the sugar is ribose. In nucleic acids, the sugar and phosphate groups of nucleotides are connected to each other through phosphodiester linkages to form a sugar-phosphate backbone. The variable part of nucleic acid is the sequence of its bases, which carries the genetic information. In DNA, there are four bases: cytosine (C), guanine (G), adenine (A) and thymine (T). In RNA, the bases are cytosine (C), guanine (G), adenine (A) and uracil (U). By convention, the base sequence in DNA is written in the 5’ to 3’ direction, from the free 5 ’-phosphate group to the free 3 ’-OH group.

Naturally occurring DNA molecules are normally double-stranded, whereas RNA molecules are normally single-stranded. Double stranded DNA is made up of two strands having complementary sequences. The two strands run in opposite directions and are held together by hydrogen bonds between pairs of bases.

Gene is a DNA sequence which encodes for a specific protein. Gene expression is a process of manufacturing the corresponding protein. First, the information in DNA is transferred to a messenger RNA (mRNA) molecule by transcription. The template strand (noncoding strand) of the DNA is used as a template for transcription. The other DNA strand, nontemplate strand, is referred to as the coding strand because its sequence will be the same as that of the newly synthesized mRNA molecule (although with thymine replaced by uracil). Thus, the coding strand of DNA is the strand that codes for the gene of interest. The resulting mRNA is a singlestranded copy of the gene, which next must be translated into a protein molecule. In translation, the protein is synthesized according to instructions given by mRNA template.

In one embodiment of the present invention, the nucleic acid is DNA. In one embodiment, the nucleic acid is double-stranded DNA.

Antibody maturation

Antibody maturation methods may be divided into targeted mutagenesis and random mutagenesis. In targeted mutagenesis, the antibody maturation library contains mutation(s) in predefined location(s) with certain amino acid content. This can be achieved for example by using mutagenic oligonucleotides or with ready synthesized libraries. Random mutagenesis introduces mutations all over the gene and resulting mutations may not be evaluated beforehand. Error-prone PCR and Activation-Induced Cytidine Deaminase based mutagenesis represent examples of random mutagenesis. After the introduction of mutations, the library is introduced to the antigen and required number of selection rounds are performed to enrich target specific binders. Antibody maturation is used, when antibody properties are developed towards effective therapeutic antibody. Such properties are without limitation for example specificity, affinity, solubility, covalent integrity, conformational stability, colloidal stability, low polyreactivity, low immunogenicity, stability and tendency of self-interaction during production or administration.

In vitro display methods

In vitro display methods are utilized for selection and screening for discovery of binder molecules binding to different types of target molecules from a large pool of variants. Such target molecules may be without limitation small molecules, proteins, peptides, nucleic acids, and their modifications. Binder molecules may be without limitation proteins, peptides, preferably antibodies or antibody fragments. By forming a physical link between an antibody, or an antibody fragment, displayed on the cell surface and the genetic information encoding it, it allows isolation of the genes that encode a protein with the desired binding function. In vitro display methods include without limitation bacterial, baculovirus, yeast, ribosomal, phage and mammalian display methods described for example in Bradbury et al 2011 and Elgundi et al. 2017.

Phage display method relates to a widely used platform for binder molecule discovery, preferably antibody or antibody fragment discovery. Phage display method is a widely used platform for binder molecule, especially antibody discovery, because of its low cost, adaptability, and efficiency. It enables producing antibodies or their fragments in vitro easily and with high diversity (reviewed recently by Kang and Lee 2021; Nagano and Tsutsumi 2021). The phage display method utilizes the expression of proteins, antibodies or their fragments on the surface of filamentous bacteriophage by fusing them with the phage’s coat protein. When an antibody or antibody fragment library is fused to the coat protein of the phages, they are displayed on the surface of the phages allowing in vitro selection by antigen-specificity and recovering their corresponding gene sequence. Generally, the selection is performed through repeated rounds of biopanning. The target molecules to which the antibodies or their fragments are intended to bind are immobilized on the well and incubated with the phage library displaying different variants of antibodies or their fragments. After washing, the phages with bound antibodies or their fragments remain and can be collected by elution. These phages can then be amplified in host bacteria and used again in multiple selection rounds enriching their activity. A review by Elgundi et al. 2017 summarizes the principle of phage display method.

Mammalian display relates to a widely used platform for binder molecule discovery, preferably antibody or antibody fragment discovery. Like phage display, mammalian cell display is based on expressing antibodies or their fragments on the cell surface by fusing the antibody gene to a transmembrane domain of a transmembrane protein, which results in surface display of the antibody. Mammalian display is summarized in article by Nguyen et al. 2020. Mammalian cell display can overcome limitations of phage display related to post-translational modifications of the antibodies and antibody fragments. Mammalian cell display may be used to engineer antibodies or antibody fragments like (scFvs), fragment antigen-binding regions (Fabs) with the required post-translational modifications (Ho and Pastan 2009).

Mammalian cell display allows the identification of antibodies with desired properties directly from the cells at an early stage of antibody discovery (Bowers et al. 2014). Antigen binding antibodies may be recognized and isolated from the antibody library by screening. Fluorescence activated cell sorting (FACS) is commonly used to identify antibody-displaying cells with desired affinities and high expression level. In addition to using mammalian cell display for antibody discovery, it has been shown that it can also be used as a tool to isolate proteins with better biophysical characteristics based on their display level achieved. When the display level is lower on the cell surface, the quality control machinery of the cell has removed antibodies that have aggregated or are prone to polyreactivity with other proteins resulting in lower surface expression level in comparison to antibodies with more suitable biophysical properties.

Mammalian cell display methods can be categorized as those utilizing transient expression and those that utilize stable cell lines in protein expression. In transient expression the gene is introduced to the cells in a vector designed for protein expression. The transient protein expression is rapid but inefficient as the expression vector will stay present in the cells for only a few days, making the method most suitable for single round of selection from smaller libraries before the recovery of the antibody genes. Stable expression methods integrate the gene into the host cell genome accompanied with an antibiotic selection gene or other selection marker, and the gene can thus be maintained in the cells for long periods of time. By utilizing this selection marker, the cells which have the antibody gene successfully integrated into their genome can be selected from the rest.

Activation-induced cytidine deaminase (AID) and its use in mammalian display

The term “Activation-induced cytidine deaminase” abbreviated “AID” refers to a protein encoded by AICDA gene. AID has a nuclear localization signal (NLS), a nuclear export signal (NES), a catalytic domain, and apolipoprotein B mRNA editing catalytic (APOBEC) protein-like domains. AID has enzyme activity deaminating cytosines. AID promotes point mutations to single-stranded DNA by converting deoxy cytidines to deoxyuridines.

The term “activity of AID” refers to AID’s enzyme activity for causing point mutations to single-stranded DNA by converting deoxycytidines to deoxyuridines by deamination. A mutation is a change in the DNA sequence of an organism. A point mutation is a mutation where a single base pair is altered. AID changes the base pair C-G to a U-G mismatch which is then converted to T-A. As the DNA repair mechanisms process this regulated DNA damage, these mutations can then induce other mutations and also lead to amino acid substitutions causing even more significant changes in the antibody. The DNA repair mechanism is known to be error- prone in somatic hypermutation and it has multiple pathways leading to different transitions or trans versions. The mutations can also spread around the targeted base pairs. Together with the error-prone DNA repair mechanisms, the process of deamination by AID can lead to various mutations in the DNA. When AID is targeted to the variable domain of the antibody gene, the mutations are created to positions in antigen-binding site and can alter antigen binding properties (Bowers et al. 2011).

G-quadruplex (G4) forming structures

Single stranded DNA and RNA may form “G-quadruplex” abbreviated “G4” structures in repetitive G-rich sequences (Nakanishi and Seimiya 2020). The ability of guanine to Hoogsteen hydrogen bond with itself leads to ring-like formations known as G-quartets of four guanines each forming hydrogen bonds with its two neighbours. These quartets can stack to form G4 structures stabilized by monovalent cations K+ and Na+. Different G4 structures are classified into parallel, antiparallel, and hybrid types according to the direction of the four strands (Lerner and Sale 2019; Mukundan and Phan 2013; Nakanishi and Seimiya 2020). G4 structures have also been shown to form even more complicated conformations containing longer loops and bulges in the structure (Mukundan and Phan 2013).

G4 structures were first discovered in vitro but more evidence has arisen that they are also formed in living cells genome-wide, attracting interest because of their potential involvement in various biological processes (Lerner and Sale 2019; Nakanishi and Seimiya 2020). The role of G4 structures have been linked to regulating DNA replication by stalling DNA replication forks and affecting replication origin activity. G4 structures may also form during transcription and recruit transcription factors or inhibit the progression of RNA polymerase II. During translation, G4 structures formed in mRNA have shown to repress translation by inhibiting the progression of ribosomes or the recruitment of translation initiation factors (Nakanishi and Seimiya 2020). It seems that G4 structures in cellular DNA may recruit AID to certain regions (Tang and McCarthy 2021; Honkonen 2021).

In the current invention at least one nucleic acid encoding a peptide linker has been designed to form a G4 structure very effectively. The nucleic acid encoding the peptide linker is located in a close proximity of an antibody variable domain. When the nucleic acid forms a G4 structure, it promotes Activation-induced cytidine deaminase (AID) activity and increases mutations in the close proximity of the G4 structure of the nucleic acid. When variable domains are located in the close proximity of the nucleic acid forming a G4 structure, mutations will occur in certain areas of the variable regions.

In the current invention there is at least one nucleic acid forming a G4 structure and encoding a peptide linker. In some embodiments there are one nucleic acid forming a G4 structure and encoding one peptide linker. In some embodiments there are multiple nucleic acids forming G4 structures and encoding peptide linkers. There may be two, three, four, five, six or seven nucleic acids forming G4 structures and encoding peptide linkers. Nucleic acids forming G4 structures and encoding peptide linkers may be similar or different with each other. In the current invention the nucleic acid forming a G4 structure and encoding a peptide linker is located in the close proximity of a variable domain. The nucleic acid forming a G4 structure and encoding a peptide linker is located upstream of a variable domain, downstream of a variable domain or between variable domains. Variable domains may be a variable heavy domain or a variable light domain. In a preferred embodiment the variable domains and the peptide linker form an scFv. In other embodiments two scFv form a di-scFv or three scFv form a tri-scFv. ScFv, di-scFv or tri-scFv may be connected to a Fc domain with a peptide linker or with some other connector for example antibody hinge domain.

Figure 6 summarizes some combinations and locations of nucleic acids encoding a variable light domain, variable heavy domain and a peptide linker (nucleic acid forms a G4 structure) present in the vector. Following combinations and locations are examples and should not be considered as limitations for other combinations and locations:

(i) - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain -

(ii) - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain -

(iii) - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - optionally hinge domain - optionally Fc domain -

(iv) - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - optionally hinge domain - optionally Fc domain -

(v) - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain -

(vi) - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - a peptide linker - a variable light domain - a peptide linker - a variable heavy domain - optionally hinge domain - optionally Fc domain -

The nucleic acid forming a G4 structure and encoding a peptide linker is located in the close proximity of a variable domain. The close proximity refers to location within 20 nucleotides, 15 nucleotides, 10 nucleotides, 5 nucleotides, 4 nucleotides, 3 nucleotides, 2 nucleotides or 1 nucleotide of a variable heavy domain encoding nucleic acid or a variable light domain encoding nucleic acid. In a preferred embodiment of the invention the nucleic acid forming a G4 structure is located adjacent to a variable heavy domain encoding nucleic acid or a variable light domain encoding nucleic acid.

The nucleic acid forming a G4 structure and encoding a peptide linker increases the mutation rate of a variable domain at least 3%, 4%, 5%, 6% or 7% compared to a variable domain without the nucleic acid forming a G4 structure in the close proximity. In a preferred embodiment of the invention the mutation rate is at least 5%.

The nucleic acid forming a G4 structure and encoding a peptide linker increases the mutation rate most in the CDRs of the variable domains that are closest to the nucleic acid. In a preferred embodiment the closest CDR domain is CDR3.

The length of the nucleic acid nucleic acid forming a G4 structure and encoding a peptide linker is 15 to 105 nucleotides, which encodes a peptide linker of 5 to 35 amino acids. Preferably the length of the nucleic acid forming a G4 structure and encoding a peptide linker is 60 nucleotides, which encodes a peptide linker of 20 amino acids.

In the current invention a nucleic acid encoding a peptide linker was designed so that it functions optimally as a peptide linker connecting antibody fragments, preferably variable domains, and simultaneously the nucleic acid forms a G4 structure optimally. The peptide linker is based on an amino acid sequence with multiple Glycine residues. One of the nucleotide codons encoding Glycine is guanine-guanine-guanine (ggg). Therefore, most of the codons encoding Glycine residue in the nucleic acid sequence were changed to ggg and the guanine content of the nucleic acid encoding the peptide linker was increased. However, maximizing the guanine content of the nucleic acid encoding the peptide linker too much (i.e. linker containing only ggg codons or very few other nucleotides) is not desirable. Most likely such a nucleic acid is extremely hard to be synthesized and subcloned. Furthermore, extremely long nucleotide repeats in the gene would probably be suboptimal also in terms of transcription and translation.

For an optimal formation of the G4 structure in the nucleic acid encoding the peptide linker, the nucleic acid comprises a guanine-rich sequence in the coding strand. Guanine-rich sequence may be also present in the template strand. For an optimal formation of the G4 structure in the nucleic acid encoding the peptide linker, the nucleic acid comprises at least 58%, 60%, 65%, 70%, 75% or 78% guanines in the coding strand or in the template strand. For an optimal formation of the G4 structure in the nucleic acid encoding the peptide linker, the nucleic acid comprises at least 58%, 60%, 65%, 70%, 75% or 78% and at most 95% or 90% guanines in the coding strand or in the template strand. An exemplary nucleic acid sequence forming effectively a G4 structure and encoding a proper peptide linker is represented in SEQ ID NO: 2. The nucleic acid according to SEQ ID NO: 2 encodes a peptide linker according to SEQ ID NO: 1.

Sequences of nucleic acids and peptides or proteins related to the invention are summarized in Table 2. The sequences represent examples but should not be considered as limitations for other alternative sequences with a similar function. SEQ ID NO: 1 is an amino acid sequence of a peptide linker, which is encoded by a nucleic acid forming a G4 structure according to SEQ ID NO: 2. A nucleic acid according to SEQ ID NO: 3 also encodes the peptide linker but it does not form a G4 structure. SEQ ID NO: 4 is an amino acid sequence of a variable light domain, which is encoded by a nucleic acid according to SEQ ID NO: 5, which forms a G4 structure. A nucleic acid according to SEQ ID NO: 6 also encodes the variable light domain but it does not form a G4 structure. SEQ ID NO: 7 is an amino acid sequence of a variable heavy domain, which is encoded by a nucleic acid according to SEQ ID NO: 8, which forms a G4 structure. A nucleic acid according to SEQ ID NO: 9 also encodes the variable light domain but it does not form a G4 structure. SEQ ID NO: 10 is an amino acid sequence comprising a fragment of a hinge domain, which is encoded by a nucleic acid according to SEQ ID NO: 11. SEQ ID NO: 12 is an amino acid sequence of a Fc domain, which is encoded by a nucleic acid according to SEQ ID NO: 13. SEQ ID NO: 14 is an amino acid sequence of a scFv-Fc construct

Nucleic acid sequences encoding a variable domain of an antibody or its fragment and the nucleic acid forming a G4 structure and encoding a peptide linker are integrated into a nucleic acid of a vector. scFv, di-scFv or tri-scFv encoding nucleic acids may be integrated into a vector nucleic acid. Different combinations of nucleic acids represented in Figure 6 may be integrated into a vector nucleic acid.

The selected vector may be any vector used for transfecting a nucleic acid into a cell. Preferably the vector is a plasmid vector. The cell is preferably a CHO cell. The current invention also relates to a kit comprising with multiple vectors. Preferably, the kit comprises a first vector comprising a nucleic acid encoding an antibody variable heavy domain, a nucleic acid encoding an antibody variable light domain, and at least one nucleic acid forming a G4 structure and a second vector encoding AID.

EXAMPLES

1 Materials and methods

Cell culture

Flp-In CHO cells (ThermoFisher Scientific, USA) with integrated Bxbl landing pad 4 (LP4) were cultured in complete medium (IX F-12 nutrient mixture supplemented with 2 mM GlutaMax and 10 % FBS (ThermoFisher Scientific)). Blasticidin (InvivoGen, USA) was added to the cultures to 10 ml/ml as selection. Cells were incubated in 37 °C, 5 % CO2.

Transfection of the cells

A day before transfection, the cells were washed and dissociated by trypsination. After the trypsination was stopped the cells were counted with trypan blue. For the transfection cells were supplemented with complete medium and incubated in 37°C, 5 % CO2 overnight.

To stably transfect Flp-In CHO LP4 cells, Bxbl targeting vector (Bxb Tv) and Bxbl integrase expression vector Bxb Ev (ThermoFisher Scientific) were co-transfected by using Lipofectamine 3000 (ThermoFisher Scientific) according to the manufacturer’s instructions with modifications to the reagent amounts. The total amount of DNA was 5 mg, the amount of P3000 was 10 ml and lipofectamine 4.5 ml. The Bxb Ev to Bxb Tv DNA ratio was 1 :4. The cells were incubated in 37 °C, 5 % CO2 for three days before adding selection.

Plasmid construction

Cloning scFv-Fc genes with the G4 structure linker to the Bxbl targeting vector The fusion antibody fragment scFv-Fc was synthetised by GeneArt (ThermoFisher Scientific). The scFv-Fc composed of variable heavy domain and variable light domain, IGHV3-23*01 and IGKV2-28*01, respectively. Variable heavy domain and variable light domain were connected with a peptide linker to form an scFv. The Fc region of IGHG1 was combined to the scFv with a hinge domain. The VH domain and the VL domain of the scFv were chosen to contain G4 structures and acquired from Immunogenetics (IMGT) database. To create another variant of the gene with the same amino acid sequence, the G4 structures were removed manually by codon optimization with DNA analyser (Brazda et al. 2016). An additional G4 structure was generated to the nucleic acid encoding the linker by changing glycine encoding GGC, GGA and GGT codons to GGG codon in the linker encoding sequence. The scFv-Fcs were cloned to the targeting vector Bxb Tv without mammalian cell promoter and kanamycin resistance gene for bacterial plasmid amplification and attachment sites for Bxbl integrase and a resistance gene for puromycin between them.

The scFv genes without and with G4 structures (sample codes scFv and scFv_G4, respectively) were cloned to Bxb_Tv by restriction enzyme digestion with Notl and Nhel (New England Biolabs, USA) and ligation with T4 ligase according to manufacturer’s instructions (New England Biolabs). The inserts and vector’s concentrations, as all the sample concentrations in this study, were measured with DeNovix DS-11 FX (DeNovix, USA).

The ligation products were transformed into Competent 5-alpha E. coli cells (New England Biolabs) according to manufacturer’s instructions with half of the recommended cell volume. Selection was done on Invitrogen imMedia growth medium agar plates containing kanamycin (ThermoFisher Scientific). Plates were incubated in 37°C overnight. One colony from each plate was transferred to liquid cultivations in 5 ml Invitrogen imMedia growth medium containing kanamycin (ThermoFisher Scientific) which were incubated in 37°C, 250 rpm, overnight. Plasmids were extracted from the cells using QIAprep Spin Miniprep Kit (Qiagen, Germany) according to manufacturer’s instructions.

Cloning different versions of AIDs to the pCEP4 vector The human AID (hAID) gene was ordered from GeneArt in an episomal pCEP4 mammalian expression vector (ThermoFisher Scientific). Hygromycin resistance gene of the vector was switched to neomycin resistance gene to enable selection of the transfected cells.

In addition to the wild-type (wt) human AID (hAID) and mouse AID (mAID), 3 variations for each species (Luo et al. 2020) were designed to study the effects of different kinds of AID variants on mutations. The base for the altered AID variants were the wild type hAID and mAID. All the different AID variants had FLAG-tags flanked by Xhol and Kpnl restriction sites, except for the wt which were ordered from GeneArt in pCEP4 without the FLAG-tag.

The first AID variants were codon optimized for CHO cells (hAID/mAID CO) (figure 1 A). The second variations were made from both codon optimized AID variants by deletion of the NES (hAID/mAID del) (figure IB), and the third variations were made to the AID variants with deletion by introducing K10E, T82I and E156G point mutations (Wang et al. 2009, Luo et al 2020) (hAID/mAID mut) (figure 1C).

The different AID variants were cloned into the episomal pCEP4 vector with a neomycin resistance gene and a CMV promoter. The vectors containing different variants of AID were transformed into Stable Competent E. coli cells (New England Biolabs). The cells were cultured as described previously, but in 30°C and in medium containing ampicillin. The plasmids were extracted as described previously.

AID transfection and expression

The cells including scFv were prepared for transfection as described previously. The plasmids with different AID variants were transfected into the cells as described previously. After two days, the cells were split 1 :2 and culture was continued with 600 ng/ml geneticin (ThermoFisher Scientific) selection and the cultivation was continued in 37°C, 5 % CO2.

Cell samples for AID expression analysis were collected three days after the transfection. The samples were stored in -80°C until lysis and Western blot analysis. Cells were cultured a month in antibiotic selection, after which the second samples were collected. The frozen cell pellets were thawed and suspended to Phosphate Buffered Saline mixed with IX complete EDTA-free Protease Inhibitor Cocktail (Roche, Switzerland). To release the intracellular proteins in which AID is included, the cells were sonicated with Sartorius Labsonic M (Sartorius, Germany).

The samples were run on SDS-PAGE on Mini-Protean TGX gel (Bio-Rad) in IX Tris/Glycine/SDS Running Buffer (Bio-Rad). The DNA ladder used was Chameleon Duo Pre- Stained Protein Ladder (LI-COR Biosciences). The gels were transferred into 0.2 pm nitrocellulose membranes (Bio-Rad).

The membranes were analysed with horseradish peroxidase (HRP)-conjugated anti-FLAG-tag antibody (BioLegend, USA) (1 :1000). Bio-Rad Clarity Western ECL Substrate (Bio-Rad) was added to the membranes according to manufacturer’s instructions and the membranes were imaged.

AID expression was analysed by Western blot (Figure 2). Each cell sample transfected with different AID variants (codon optimized (CO), codon optimized with deletion (del), and codon optimized with deletion and point mutations (mut)) were analysed with western blot. The numbers indicate if the sample was taken three days (1) or a month (2) after transfection. The control sample (ctrl) used was from the same cell line without any AID transfection. The wildtype human and mouse AID variants were run as reference as they do not contain the FLAG-tag.

The results from the western blot (Figure 2) indicate that the AID expression decreases significantly over time and that the expression is highest after transfection. This was also demonstrated in the study by Luo et al. (2020), where the AID mRNA levels were 5-fold on day 5 after transfection compared to the day 12.

According to the membrane pictures, there were no significant difference in expression level between the hAID and mAID variants, or within the hAID variants. Comparing the expression levels of mAID variants with each other it seems that the codon optimized AID’s expression would be the highest and the AID with deletions would not express almost at all. The expression of the wild type hAID and mAID were included as reference samples. These results show that AID is expressed in each cell line after the transfection before the time of taking the samples. According to these results, the final FACS sorting was done the day after repeating the AID transfection in effort to improve the transfection efficiency and to maximise the probability of AID staying in the cells and inducing mutations when the cells divide.

Fluorescence-activated sorting of AID transfected cells

ScFv, scFv_G4 and scFv+E cell lines were transfected one day before single cell FACS sorting with the eight different AID variants using 400000 cells as described previously resulting in 24 AID transfected cell lines. The cells expressing scFv were used as control without AID transfections.

Antibody PE- Anti-Human IgG Fc (BioLegend) was diluted to concentration of 1.25 Lig/ml for staining the cells in concentration of 10 million cells/ml as described previously. The final cell concentration transferred to FACS tubes was 5 million cells/ml.

The single cell FACS sorting was performed with BD FACSMelody Cell Sorter (BD Biosciences, USA) according to manufacturer’s instructions. The cells were sorted to 96-well plates. The medium contained 1:3 conditioned medium and the rest fresh complete medium, 10 Lig/ml puromycin (InvivoGen), 600 Lig/ml geneticin (ThermoFisher Scientific) and IX Gibco Penicillin- Streptomycin (ThermoFisher Scientific). The two gates were determined using the control cells to acquire cells producing high signal and weakened signal, which may be an indication of mutations in the area where the stain is binding (Figure 3). Thirty cells were single cell sorted from each gate. The plates were incubated in 37°C, 5 % CO2 for two weeks. From each AID transfected cell line 1-4 clones were pooled and transferred to a 6-well plate. When possible, equal number of clones from both gates were chosen. The cells were split routinely and grown in 37°C, 5 % CO2 until the 6-well plates were confluent, before splitting the cells for transfection but with split ratio of 1 :4 and repeating the AID transfection as described. The cells were then split routinely and expanded with 20 ml of medium, 10 Lig/ml puromycin (InvivoGen) and 600 Lig/ml geneticin (ThermoFisher Scientific) and grown in 37°C, 5 % CO2 for three days.

Sequencing the AID mutated genomic DNA Genomic DNA was extracted from the cell pellets using Monarch Genomic DNA Purification Kit (New England Biolabs) according to manufacturer’s instructions. As a control, the genomic DNA was also extracted from the cells transfected with Bxb Tv scFv without AID transfections.

PCR amplification was used to attach the adapters to the target sequences for Illumina sequencing through Eurofins NGSelect Amplicons 2nd PCR protocol. The sequencing was performed for both scFv alternatives (with and without G4 structures) VL and VH domains separately.

DNA was amplified using NEBNext Ultra II Q5 Master Mix (New England Biolabs) with BioRad T100 Thermal Cycler (Bio-Rad), the quality of the PCR products was evaluated with agarose gel electrophoresis and the PCR products were purified and prepared for sequencing.

Fifty micrograms of the purified PCR product was amplified in a 2nd PCR step to attach Illumina adaptors and indeces to the amplicons. Illumina MiSeq instrument with MiSeq Reagent Kit v3 with 600 cycles was used to achieve 2 x 300 bp paired-end reads.

Sequencing data analysis

The sequencing returned the minimum of 60000 reads per sample. The forward and reverse sequences from each sequencing sample were merged by Eurofins to eliminate faulty sequencing data resulting in sequence mismatch. The use of sequencing quality score Q30 was tested to rule out sequences with insufficient quality. Sequences of both VH and VL were annotated for mutations using custom annotation by Pipebio (Denmark). The annotation compared the sequencing data to the original reference sequence and marked any amino acid changes as mutations. The sequences were then clustered with 100 % sequence identity to combine sequences with the same mutations which led to acquiring the number of mutations in each sequence position.

The number of amino acid changes for each position in the sequences in relation to all sequencing reads was calculated for each sample using Microsoft Excel. The synonymous mutations were not included in the analysis, as they do not lead to amino acid changes which were examined here.

The number of mutations in total were compared between sample groups with G4 structures and without them by counting the sum of mutations in the same position from each sample in the sample group, and again by calculating each position’s amino acid changes in relation to all the sequencing reads within the sample group. The mutations from VH and VL control samples including scFv were subtracted from the values of each position to exclude background. Six mutations chosen amongst the mutations with the highest values were considered as mutations of interest and they were collected manually from Pipebio for closer examination.

G4 structures attract more mutations to their neighboring areas

The Illumina sequencing was performed for both scFv alternatives’ (with G4 structures and without them) VL and VH domains separately. Sequences of both VH and VL were annotated for mutations causing amino acid changes using custom annotation by Pipebio and clustered with 100 % sequence identity. The number of amino acid changes for each position in relation to all sequencing reads within the sample group, with G4 structures or without them, was calculated. Background mutations derived from control sample scFv (without AID transfection) were subtracted from the values and the two sample groups were compared. Mutation rate in samples with G4 structures was 9,2% in variable light domain and 16,7% in variable heavy domain. In samples without G4 structures mutation rate was 1,7% in variable light domain and 4,0% in variable heavy domain. Thus, the difference between samples with G4 structures and without G4 structures was 7,5% in variable light domain and 12,7% in variable heavy domain. The synonymous mutations were left out of the analysis as the interest was in mutations causing amino acid changes.

Figure 5 summarizes amino acid changes caused by mutations in the nucleic acid sequence in each sequence position of VL domain (amino acids 1-112), VH domain (amino acids 133-250) and peptide linker (113-132). In all VL samples the number of mutations increases from amino acid position 75 onwards, while in VH samples the number of mutations is increased until amino acid position 182. In the sequences with G4 structures, the end of the VL domain and the beginning of VH domain are located near the linker which contains an optimized sequence forming G4 structure. The presence of the G4 structure in the nucleic acid encoding the peptide linker may explain the difference in mutation rates between the sample groups representing G4 structures forming linker and linker not forming such structures. The G4 structures may attract AID by being a preferable substrate for the enzyme over linear DNA (Qiao et al. 2017). Therefore, more mutations could occur in the neighboring regions when more enzyme gathers to deaminate cytosines resulting in DNA repair mechanisms to activate (Liu and Schatz 2009).

Six amino acid changes were chosen based on their higher prevalence within all sequencing reads and collected manually from Pipebio for closer inspection in effort to confirm some of the mutations’ origin. These mutations are summarized in table 1.

The results suggest that G4 structures might have an important role in mutation targeting. The individual mutations with highest prevalence were found in samples containing G4 structures and all the mutations in these samples were in the nearing regions of G4 structures.

The most important finding of the study is that the location of mutations can be directed to a certain area and increase the sheer mutation rate by introducing strong G4 structures to the linker. Furthermore, it should be noted that the approach is universal since it can be used with different variable domain gene segments.

REFERENCES

Almagro, J. C., Pedraza-Escalona, M., Arrieta, H. I. & Perez-Tapia, S. M. (2019) Phage Display Libraries for Antibody Therapeutic Discovery and Development. Antibodies 8:44.

Bowers, P. M., Horlick, R. A., Neben, T. Y., Toobian, R. M., Tomlinson, G. L.,

Dalton, J. L., Jones, H. A., Chen, A., Altobell III, L., Zhang, X., Macomber, J. L., Krapf, I. P., Wu, B. F., McConnell, A., Chau, B., Holland, T., Berkebile, A. D., Neben, S. S., Boyle, W. J., & King, D. J. (2011) Coupling Mammalian Cell Surface Display with Somatic Hypermutation for the Discovery and Maturation of Human Antibodies. Proc Natl Acad Sci U S A 108:20455- 20460. Bowers, P. M., Horlick, R. A., Kehry, M. R., Neben, T. Y., Tomlinson, G. L., Altobell, L., Zhang, X., Macomber, J. L., Krapf, I. P., Wu, B. F., McConnell, A. D., Chau, B., Berkebile, A. D., Hare, E., Verdino, P. & King, D. J. (2014) Mammalian Cell Display for the Discovery and Optimization of Antibody Therapeutics. Methods 65:44-56.

Bradbury, A. R. M., Sidhu, S., Dubel, S. & McCafferty, J. (2011) Beyond Natural Antibodies: The Power of in Vitro Display Technologies. Nat Biotechnol 29:245-254.

Brazda, V., Kolomaznik, J., Lysek, J., Haronikova, L., Coufal, J. & St'astny, J. (2016) Palindrome Analyser - A New Web-Based Server for Predicting and Evaluating Inverted Repeats in Nucleotide Sequences. Biochem Biophys Res Commun 478:1739-1745.

Buerstedde, J-M., Alinikula, J., Arakawa, H., McDonald, J. J. & Schatz, D. G. (2014) Targeting Of Somatic Hypermutation By Immunoglobulin Enhancer And Enhancer-Like Sequences. PLoS Biol 12:el001831.

Cervantes-Gracia, K., Gramalla-Schmitz, A., Weischedel, J. & Chahwan, R. (2021) APOBECs Orchestrate Genomic and Epigenomic Editing across Health and Disease. Trends Genet 37:1028-1043.

Chiu, M. L., Goulet, D. R., Teplyakov, A. & Gilliland, G. L. (2019) Antibody Structure and Function: The Basis for Engineering Therapeutics. Antibodies 8, 55.

Elgundi, Z., Reslan, M., Cruz, E., Sifhiotis, V. & Kayser, V. (2017) The State -of-Play and Future of Antibody Therapeutics. Adv Drug Deliv Rev 122: 2-19.

Honkonen, E. (2021) Role of G-quadruplexes in targeting of somatic hypermutation. University of Turku. Institute of Biomedicine. Master’s Thesis.

Jain, T., Sun, T., Durand, S., Hall, A., Houston, N. R., Nett, J. H., Sharkey, B., Bobrowicz, B., Caffiy, I., Yu, Y., Cao, Y., Lynaugh, H., Brown, M., Baruah, H., Gray, L. T., Krauland, E. M., Xu, Y., Vasquez, M. & Wittrup, K. D. (2017) Biophysical Properties of the Clinical-Stage Antibody Landscape. Proc Natl Acad Sci U S A 114:944-949.

Kang, S. H. & Lee, C-H. (2021) Development of Therapeutic Antibodies and Modulating the Characteristics of Therapeutic Antibodies to Maximize the Therapeutic Efficacy. Biotechnol Bioprocess Eng 26:295-311.

Lerner, L. K. & Sale, J. E. (2019) Replication of G Quadruplex DNA. Genes 10:95.

Liu, M. & Schatz, D. G. (2009) Balancing AID and DNA Repair during Somatic Hypermutation. Trends Immunol 30:173-181.

Luo, R., Zhao, Y., Fan, Y., An, L., Jiang, T., Ma, S. & Hang, H. (2020) High Efficiency CHO Cell Display-Based Antibody Maturation. Sci Rep 10:8102.

Mukundan, V. T. & Phan, A. T. (2013) Bulges in G-Quadruplexes: Broadening the Definition of G-Quadruplex-Forming Sequences. J Am Chem Soc 135:5017-5028.

Nagano, K. & Tsutsumi, Y. (2021) Phage Display Technology as a Powerful Platform for Antibody Drug Discovery. Viruses 13:178.

Nakanishi, C. & Seimiya, H. (2020) G-Quadruplex in Cancer Biology and Drug Discovery. Biochem Biophys Res Commun 531 :45-50.

Nguyen, A. W., Le, K. & Maynard, J. A. (2020) Engineering Antibodies on the Surface of CHO Cells. Methods Mol Biol 2070:397-422.

Parthiban, K., Perera, R. L., Sattar, M., Huang, Y., Mayle, S., Masters, E., Griffiths, D., Surade, S., Leah, R., Dyson, M. R. & McCafferty, J. (2019) A Comprehensive Search of Functional Sequence Space Using Large Mammalian Display Libraries Created by Gene Editing. mAbs 11 :884-898. Qiao, Q., Wang, L., Meng, F-L., Hwang, J. K., Alt, F. W. & Wu, H. (2017) AID Recognizes Structured DNA for Class Switch Recombination. Mol Cell 67:361-373.

Shepard, H. M., Phillips, G. E., Thanos, C. D. & Feldmann, M. (2017) Developments in Therapy with Monoclonal Antibodies and Related Proteins. Clin Med (Fond) 17:220-232.

Strube, R. W. & Chen, S-Y. (2004) Enhanced Intracellular Stability of SFv-Fc Fusion Intrabodies. Methods 34:179-183.

Tang, C., Bagnara, D., Chiorazzi, N., Scharff, M. D. & MacCarthy, T. (2020) AID Overlapping and Pol | Hotspots Are Key Features of Evolutionary Variation Within the Human Antibody Heavy Chain (IGHV) Genes. Front Immunol 11 :788.

Tang, C. & MacCarthy, T. (2021) Characterization of DNA G-Quadruplex Structures in Human Immunoglobulin Heavy Variable (IGHV) Genes. Front Immunol 12:671944.

Tiller, T., Schuster, I., Deppe, D., Siegers, K., Strohner, R., Herrmann, T., Berenguer, M., Poujol, D., Stehle, J., Stark, Y., HeBling, M., Daubert, D., Felderer, K., Kaden, S., Kolln, J., Enzelberger, M. & Urlinger, S. (2013) A Fully Synthetic Human Fab Antibody Library Based on Fixed VH/VL Framework Pairings with Favorable Biophysical Properties. mAbs 5:445-470.

Wang, M., Yang, Z., Rada, C. & Neuberger, M. S. (2009) AID Upmutants Isolated Using a High-Throughput Screen Highlight the Immunity/Cancer Balance Limiting DNA Deaminase Activity. Nat Struct Mol Biol 16:769-776.

Weisser, N. E. & Hall, J. C. (2009) Applications of Single-Chain Variable Fragment Antibodies in Therapeutics and Diagnostics. Biotechnol Adv 27:502-520. TABLE 1

Examples of some amino acid changes present in different locations of variable light domain and variable heavy domain.

TABLE 2

Summary of amino acid sequences and nucleic acid sequences. Abbreviations in the table: (T) sequence type; (a) amino acid sequence; (n) nucleic acid sequence; (SP) species; (h) homo sapiens (a) artificial

Claims

1. A method to introduce a mutation to a variable domain of an antibody or its fragment by introducing at least one peptide linker to close proximity of the variable domain, wherein the peptide linker is encoded by a nucleic acid forming a G quadruplex (G4) structure and the G4 structure promotes activity of activation-induced cytidine deaminase (AID).

2. A method according to claim 1, wherein a second peptide linker is introduced to close proximity of the variable domain, wherein the second peptide linker is encoded by a nucleic acid forming a G4 structure and the G4 structure promotes activity of AID.

3. A method according to claim 2, wherein a third peptide linker is introduced to close proximity of the variable domain, wherein the third peptide linker is encoded by a nucleic acid forming a G4 structure and the G4 structure promotes activity of AID.

4. A method according to any one of claims 1 to 3, wherein the nucleic acid forming a G4 structure is located upstream of a variable domain encoding nucleic acid, downstream of a variable domain encoding nucleic acid and/or between variable domains encoding nucleic acids.

5. A method according to any one of claims 1 to 4, wherein the nucleic acid forming a G4 structure is located within 20 nucleotides, 15 nucleotides, 10 nucleotides, 5 nucleotides, 4 nucleotides, 3 nucleotides, 2 nucleotides or 1 nucleotide of a variable domain encoding nucleic acid.

6. A method according to any one of claims 1 to 4, wherein the nucleic acid forming a G4 structure is located adjacent to a variable domain encoding nucleic acid.

7. A method according to any one of claims 1 to 6, wherein the nucleic acid forming a G4 structure increases a mutation rate of a variable domain at least 3%, 4%, 5%, 6% or 7% compared to a variable domain without the nucleic acid forming a G4 structure in the close proximity.

8. A method according to any one of claims 1 to 7, wherein the nucleic acid forming a G4 structure increases a mutation rate most in a CDR domain closest to the nucleic acid forming a G4 structure.

9. A method according to claim 8, wherein the CDR domain is CDR3.

10. A method according to any of claims 1 to 9, wherein the length of the nucleic acid forming a G4 structure is 15 to 105 nucleotides.

11. A method according to claim 10, wherein the length of the nucleic acid forming a G4 structure is 60 nucleotides.

12. A method according to any of claims 1 to 11 , wherein the nucleic acid forming a G4 structure comprises a guanine-rich (g-rich) sequence in a coding strand of the nucleic acid or in a template strand of the nucleic acid.

13. A method according to any of claims 1 to 12, wherein the nucleic acid forming a G4 structure comprises at least 58%, 60%, 65%, 70%, 75% or 78% guanine nucleotides in a coding strand of the nucleic acid or in a template strand of the nucleic acid.

14. A method according to any of claims 1 to 13, wherein the nucleic acid forming a G4 structure comprises or consists a nucleic acid sequence according to SEQ ID NO: 2.

15. A method according to any of claims 1 to 14, wherein the variable domain is a variable heavy domain or/and a variable light domain.

16. A method according to claim 15, wherein the variable light domain, the variable heavy domain and the peptide linker form a single-chain variable fragment (scFv).

17. A method according to claim 15, wherein the variable light domains, the variable heavy domains and the peptide linkers form a di-scFv or a tri-scFv.

18. A method according to claims 16 or 17, wherein the scFv, di-scFv or tri-scFv is connected to an Fc domain by a peptide linker or a hinge domain or their fragments.

19. Use of a method according to any of claims 1 to 18 in an in vitro display method, a phage display method or a mammalian display method.

20. A method according to any of claims 1 to 18, wherein the nucleic acid encoding a variable domain of an antibody or its fragment and the nucleic acid forming a G4 structure are transfected to a cell.

21. A method according to claim 20, wherein a nucleic acid encoding an activation-induced cytidine deaminase (AID) is transfected to a cell.

22. A method according to claim 21, wherein the nucleic acid encoding a variable domain of an antibody or its fragment and the nucleic acid forming a G4 structure are transfected stably to a cell and the nucleic acid encoding an AID is transfected transiently to a cell.

23. A method according to any of claims 20 to 22, wherein the cell is a CHO cell.

24. A method according to any of claims 20 to 23, wherein the cell expresses an AID.

25. A vector comprising a nucleic acid sequence encoding an antibody variable heavy domain, a nucleic acid sequence encoding an antibody variable light domain, and at least one nucleic acid sequence forming a G4 structure and encoding a peptide linker.

26. A vector according to claim 25, wherein the nucleic acid forming a G4 structure is located upstream or downstream of the variable light domain or variable heavy domain encoding nucleic acid or between the variable heavy domain and the variable light domain encoding nucleic acids.

27. A vector according to claims 25 or 26, wherein the vector further comprises a nucleic acid encoding AID.

28. A vector according to any of claims 25 to 27, wherein the vector is a plasmid.

29. A kit comprising the vector according to claims 25 or 26 and a second vector encoding

AID.