Cross Reference To Related Applications

This application claims the benefit of U.S. provisional application serial number 61/299,380 filed January 29, 2010, which is incorporated by reference in its entirety.

Field of the Invention

The present invention relates to and provides rodent libraries, such as combinatorial antibody libraries which are suitable for selection of fully rodent antibodies, e.g. murine or rat antibodies. The invention also relates to synthetic nucleic acid sequences which encode individual or collections of rodent antibodies, i.e. nucleic acid sequences encoding rodent antibody libraries. Methods for generating and using such libraries are provided. In particular, the invention relates to the preparation of a library of rodent-derived antibody genes by the use of synthetic consensus and/or closest homologs to germline sequences which cover the structural repertoire of antibodies encoded in the genome.

Background of the Invention

Over the last several years, many humanized or fully human antibodies have been tested in clinical trials, and several of them have been approved for therapeutic application (Hoogenboom, H. R. (2005), Nat Biotechnol 23, 1 105-1 1 16; Carter.P.J. (2006). Nat Rev Immunol 6, 343-357). Although the human antibodies typically selected by phage display (Hoogenboom, 2005) are not likely to be immunogenic and therefore quickly neutralized in human, they usually suffer from rapid clearance in experimental animals due to induction of anti-human antibodies. The rapid clearance of human antibodies prevents long-term target-validation, mechanistic and ADME (PK/PD) preclinical studies in rodents. For these reasons, it is highly desirable to have available antibodies which are not immunogenic in the respective animal models. Most preferably, such antibodies are fully derived from the respective animal species. For example, an antibody used in murine animal experiments is ideally a fully murine antibody. Likewise, an antibody used in rat animal experiments is ideally a fully rat antibody.

For the generation of human antibodies synthetic combinatorial antibody libraries have been used, e.g. the HuCAL GOLD library described in Rothe et al., 2008, J Mol Biol 376, 1 182-1200. The selection by phage display of well expressed, target-selective high affinity-binders was demonstrated in these fully human library (Steidl et al. 2008. Mol Immunol 46(1 ):135-44). Synthetic combinatorial libraries offer several advantages over libraries derived from na ^' ive B-cell sources. Natural occurring gene sequences encoding antibody frameworks can be chosen and readily designed for high antibody-production yield in prokaryotic or eukaryotic hosts, e.g. by codon-optimization. Also, the synthetic nature of the libraries enables the implementation of other features, e.g. a modular design of the antibody frameworks with unique restriction sites. Such willfully introduced restriction sites at appropriate positions enable downstream optimization processes such as antibody maturation; for example, pools of binders can be rapidly optimized without knowledge of particular sequences using pre-built diversified complementarity determining region (CDR) cassettes. Mostly this optimization process is driven for a higher antigen affinity and then known as affinity maturation (Steidl et al. 2008. Mol Immunol 46(1 ):135-44) .

A similar approach was not deemed to be possible for rodent antibody libraries. Numerous reports demonstrate the belief in the scientific community that, e.g., mouse antibody libraries cannot be produced with sufficient complexity and/or with reasonably useful expression levels. Historically, mouse antibody-libraries were derived by PCR amplification of a VH-gene pool from spleens of immunized mice (Ward et al., 1989, Nature 341, 544-546), However, such libraries were fraught with problems, such as for example poor expression of the library members.

Only recently, the first synthetic mouse antibody library has been reported (Cobaugh et al., 2008, Journal of Molecular Biology 378, 622-633). But there, the library was focused towards peptide binders, based on single framework derived from antibody 26-10, and highly diversified in HCDR3 only.

WO 06/030238 describes the "surprising discovery that an antibody framework region based on a murine VH14 heavy chain and a murine VK2 light chain [...] is solubly expressed in a microorganism and is stable [....]"■ See lines 15-28 on page 3 of WO 06/030238. This summarizes all previous, failed attempts in the art to come up with similar murine or other rodent libraries. Notably, WO 06/030238 only reported on the soluble expression of one— and only one— particular VH/VL pair. Accordingly, no other rodent libraries were reported so far as having been successful. Further, no rodent antibody library comprising more than one specific VH/VL pair has been disclosed.

Despite these discouraging references, the present inventors aimed for the generation of antibody libraries of a rodent nature, such as a fully murine and fully rat nature, respectively, and/or collectively. Such libraries could prove suitable for the selection of fully murine and rat antibodies with desired biophysical properties and target specificity. One of the main aims was to build a synthetic fully murine combinatorial antibody library (hereinafter HuCAL MOUSE) and a fully rat combinatorial antibody library (hereinafter HuCAL RAT), preferably comprising as many distinct VL/VH -frameworks as possible.

WO 01/92291 describes a method for the RNA trans-splicing mediated covalent intracellular fusion of transcripts of two different genes. In Example 9 a mouse library of dTS (double trans splicing) genes is constructed by comparing and the leader exons of 41 functional mouse VH genes. No mouse antibody libraries were however generated.

De Jaeger et al (FEBS Letters (1997) 403, 1 16-22) use a single-chain murine phage display library for the isolation of binders against an enzyme from Petunia hybrids. This library is however not a synthetic library. Furthermore the library of de Jaeger et al. is derived from immunized mice, i.e. the library was generated from nucleic acid material that has been pre-exposed to the antigen.

Sommavilla et al (J Immunol Methods (2010) 353, 31 -43) report the design and construction of a naive mouse antibody library. Like other attempts, the libraries of Sommavilla et al. only contain one VH and one VL germline gene. To the inventors' knowledge, neither a synthetic rodent antibody library nor any other reliable rodent antibody library with a diverse VH/VL composition previously has been disclosed which comprises members of more than one VH and/or more than one VL germline family. One of the breakthroughs in the instant invention is based on the observation that different frameworks impose distinct conformations on CDRs, and hence affect the range of antibody-structures capable of antigen binding. This is one of the main problems in the generation of rodent antibodies, and one of the reasons why up until now no rodent antibody library as contemplated by the present invention has been generated. For the first time it was possible overcome the prejudice in the prior art. Rational analysis of the naturally occurring rodent antibody sequences and sophisticated design of the libraries, led to the first synthetic rodent antibody libraries that are broadly useful for biomedical research.

Summary of the Invention

Since frameworks encoded by murine germline sequences are expected not to be immunogenic in mice, we searched in the IMGT database (Giudicelli, et al. (1997), Nucleic Acids Res 25, 206-21 1 ) for germline sequences representing classes of the most abundant murine antibodies. We tested six representative murine VL kappa germline sequences (3 for C57BL/6 mouse strain and 3 for Balb/c mouse strain) and five representative murine VH germline sequences (plus a variant of a mVH1 germline sequence, termed mVH1 _RV). Out of the 36 possible mVL-mVH combinations we identified five combinations which exhibited the following advantageous properties: (i) they are well displayed on the tip of filamentous phages in Fab-format, (ii) they are expressed in high amounts and in soluble form in Fab-format, as well as in IgG formats, (iii) they are thermodynamically stable, and (iv) they are identified frequently in the Kabat antibody-database.

For the five most favorable mVL-mVH combinations the L- CDR3 and the H-CDR3 regions were replaced by highly diversified L-CDR3 and H-CDR3 library cassettes, respectively, thereby achieving an overall library diversity of more than 1 x10 ¹⁰ For the purpose of affinity-maturation of binders obtained from initial panning rounds, L-CDR3 and H-CDR2 maturation cassettes were synthesized and cloned into auxiliary vector. Besides the modularity, other features of the HuCAL GOLD-system (Rothe et al., 2008, J Mol Biol 376, 1 182-1200), were implemented as well, e.g. the CysDisplay ^® technology (see WO 00/105950). Binders selected from the synthetic rodent antibody-library can for example be used for target validation in mouse. Furthermore, selected murine or rat antibodies exhibiting a similar binding mode as their HuCAL GOLD-derived target-specific counterparts may be used as non-immunogenic surrogates in respective animal experiments, for example for toxicology and pharmacology studies.

The present invention relates generally to synthetic rodent antibody libraries, preferably libraries comprising members of at least two VH germline families and/or at least two VL germline families. Specifically the present invention provides synthetic mouse and rat antibody libraries, although libraries of other rodent species are contemplated by the invention.

The present invention also provides synthetic rodent antibody libraries comprising more than 50 % of the natural mouse VH repertoire and/or more than 50 % of the natural mouse VL repertoire. Said libraries furthermore cover more than 50 % of the naturally occurring H-CDR3 lengths of the murine H-CDR3 repertoire.

The present invention also provides synthetic rodent antibody libraries in which essentially all VH- VL combinations of said library are efficiently displayed.

The present invention also provides synthetic rodent antibody libraries in which essentially all VH- VL combinations are well expressed in E. coli in Fab format.

The present invention also provides synthetic rodent antibody libraries in which wherein essentially all VH-VL combinations are well expressed in a mammalian system in IgG format.

The present invention also provides synthetic rodent antibody libraries in which all VH-VL combinations are thermally stable. The present invention also provides collections of nucleic acid molecules encoding the antibodies of said synthetic rodent antibody libraries.

The present invention also provides vectors encoding said nucleic acid molecules.

The present invention also provides recombinant host cells comprising said nucleic acid molecules or vectors.

The present invention also provides methods to isolate antibodies specific for an antigen, said method comprising the steps of:

(a) contacting the synthetic rodent antibody libraries of the present invention with an antigen;

(b) removing those members of the library which do not bind to (or are not specific for) the antigen; and

(c) recovering those members of the library bound to (or specific for) the antigen.

The present invention also provides an antibody isolated from a library contemplated by the present invention, including a modified isolate such as a humanized antibody. The present invention also provides an antibody variant derived from an antibody isolated from a library contemplated by the present invention, including antibody variants, wherein said antibody variant is a humanized antibody.

The selection of the frameworks was chosen to optimize the chance of obtaining antibodies which possess favourable biophysical properties and which are devoid of short comings of antibodies derived from synthetic libraries which have not undergone in-vivo maturation. Such favorable and desired biophysical properties include higher stability, higher expression levels and a low tendency for aggregation. Unless otherwise defined, all technical and scientific terms used herein in their various grammatical forms have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although methods and materials similar to those described herein can be used in the practice or testing of the present invention, the preferred methods and materials are described below. In case of conflict, the present specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and are not limiting.

Further features, objects, advantages, and aspects of the present invention are apparent in the claims and the detailed description that follows. It should be understood, however, that the detailed description and the specific examples, while indicating preferred aspects of the invention, are given by way of illustration only, since various changes and modifications within the spirit and scope of the invention will become apparent to those skilled in the art from this detailed description.

Brief Description of the Drawings

Figure 1 shows an alignment of framework region 3 of mVH1 and the variant mVH1_RV. This variant comprises K77R and T78V mutations. In this variant a stabilising salt bridge is formed between R77 and D100, which adds extra high stability.

Figure 2 shows framework region 4 for the heavy chains and the mouse library.

Figure 3 shows framework region 4 for the light chains and the mouse library.

Figure 4 shows the natural amino acid distribution of murine HCDR3 sequences for HCDR3 sequences of the length of 7, 8, 9, 10, 1 1 and 12 amino acids. Single amino acid distributions are aligned by grouping for the D-segment (leading to a right-centred alignment).

Figure 5 summarizes all lengths in one murine HCDR3 design. The weighted average was calculated for each position. Figure 6 shows the murine HCDR3 design after implementing all required modifications.

Figure 7 shows the natural amino acid distribution of murine LCDR3 sequences of the lenghth of 8 amino acids.

Figure 8 shows the murine LCDR3 design after implementing all required modifications.

Figure 9 shows the natural amino acid distribution of murine HCDR2 sequences for the three selected mVH mastergenes mVH1_RV, mVH2 and mVH5. In order to preserve prevalence of the germline sequences, the alternative residues present in the rearranged antibodies but not in the germline sequences were intentionally underrepresented in the design.

Figure 10 summarizes the position-dependent amino acid distribution of the HCDR2 of the three selected mVH mastergenes.

Figure 11 shows the murine HCDR2 design after implementing all required modifications.

Figure 12 depicts display of murine mastergenes in Fab-format on filamentous phage. Altogether 36 murine VL-VH combinations were tested and all combinations tested, with the exception of all mVL-B7 combinations, were well displayed on phages.

Figure 13 depicts soluble expression of the murine mastergenes in Fab-format. All murine VL- VH combinations tested, with the exception of all mVL-B7 combinations, were well expressed in E. coli.

Figure 14 shows that the soluble expression in E. coli and the display on filamentous phage correlates remarkably well.

Figure 15 and 16 show the expression of murine IgGs (lgG1 and lgG2a). Essentially all combinations tested show expression levels of well above 10 mg/L.

Figure 17 shows an overview of expression yields of all tested mouse framework combinations after washing on I MAC with and without isopropanol in a color code for selection criteria. Figure 18 illustrates thermal stability of murine mastergenes in Fab-format.The thermal stability of all tested murine VL-VH combinations, including the mVLB7 combinations, was reasonably high.

Figure 19 shows the design of the separate HCDR3 cassettes for the HCDR3 sequences of the lengths of 7, 8, 9, 10, 1 1 and 12 amino acids.

Figure 20 depicts the representation of the actual amino acids found in the LCDR3 of the murine library of the present invention, as compared to the predicted occurrence the these residues. "T" in the second row refers to trinucleotide mixtures used for the synthesis of the diversified oligonucleotides for the CDR cassette. Letters "F" and "P" indicate found and planned residue occurrence, respectively.

Figure 21 depicts the representation of the actual amino acids found in the HCDR3 of the murine library of the present invention, as compared to the predicted occurrence the these residues. "T" and "W" in the second row refers to trinucleotide mixtures and wobble nucleotides, respectively, used for the synthesis of the diversified oligonucleotides for the CDR cassette. Letters "F" and "P" indicate found and planned residue occurrence, respectively.

Figure 22 depicts the representation of the actual amino acids found in the HCDR2 maturation cassette of the murine VH1 RV sublibrary of the present invention, as compared to the predicted occurrence the these residues. "MMT" and "W" in the second row refers to trinucleotide mixtures and wobble nucleotides, respectively, used for the synthesis of the diversified oligonucleotides for the maturation cassette. Letters "F" and "P" indicate found and planned residue occurrence, respectively.

Figure 23 depicts the representation of the actual amino acids found in the HCDR2 maturation cassette of the murine VH2 sublibrary of the present invention, as compared to the predicted occurrence the these residues. "MMT" and "W" in the second row refers to trinucleotide mixtures and wobble nucleotides, respectively, used for the synthesis of the diversified oligonucleotides for the maturation cassette. Letters "F" and "P" indicate found and planned residue occurrence, respectively.

Figure 24 depicts the representation of the actual amino acids found in the HCDR2 maturation cassette of the murine VH5 sublibrary of the present invention, as compared to the predicted occurrence the these residues. "MMT" and "W" in the second row refers to trinucleotide mixtures and wobble nucleotides, respectively, used for the synthesis of the diversified oligonucleotides for the maturation cassette. Letters "F" and "P" indicate found and planned residue occurrence, respectively.

Figure 25 depicts the representation of the actual amino acids found in the HCDR2 maturation cassette of the rat library of the present invention, as compared to the predicted occurrence the these residues. "MMT" and "W" in the second row refers to trinucleotide mixtures and wobble nucleotides, respectively, used for the synthesis of the diversified oligonucleotides for the maturation cassette. Letters "F" and "P" indicate found and planned residue occurrence, respectively.

Figure 26 depicts the planned HCDR3 length distribution as compared to the HCDR3 length distribution actually found in the murine library of the present invention.

Figures 27-31 depict an overview of the all synthesized master genes of the murine library of the present invention. Kabat indicates the position of amino acid residue in VL- or VH-sequences. mVL denotes murine variable domain in the light chain (exclusively of the type kappa), and mVH denotes murine variable domain in the heavy chain. Murine VK germ-line genes encode mVL starting from position 1 (at the N-terminus of framework 1 ) to 95 (at the C-terminus of L-CDR3), and murine JK germ-line genes encode for the very C-terminal residue in L-CDR3 (constant position 96) and whole framework 4 in mVL (positions 97 - 1 08) . We have chosen murine JK2, since Tyr is the most frequent residue in position 96 of rearranged antibodies found in the IMGT database.

Figure 32 shows diversity and abundance of different frameworks combinations isolated from a test panning with the murine library of the present invention.

Figure 33 shows an overview of the obtained yields of mM-CSF specific Fab fragments after expression and purification from bacterial cells.

Figure 34 shows the results of the determination of the affinity constants for some mM-CSF specific Fab fragments as determined in an ELISA assay.

Figure 35 shows the design of the two groups of rat VL mastergenes Figure 36 shows framework 4 region of the rat VH mastergene Figure 37 shows framework 4 region of the rat VL mastergene

Figure 38 shows display of rat mastergenes in Fab-format on filamentous phage. All four tested rVL-rVH framework combinations were clearly displayed on phages.

Figure 39 depicts soluble expression of the rat mastergenes in Fab-format. The rVL1 - combinations expressed particularly well. Shown are results from two individual experiments (labeled as "NVS" and "MOR"), as well as the average of these two experiments.

Figure 40 shows the expression of rat lgG2b. There is not much difference among the four rat combinations tested.

Figure 41 shows the expression of rat lgG2c. Best expression is observed with the rVH5- combinations.

Figure 42 shows an overview of expression yields of all tested Fab rat framework combinations after washing on IMAC with and without isopropanol in a color code for selection criteria.

Figure 43 illustrates thermal stability of murine mastergenes in Fab- and IgG format. The thermal stability of all tested rat VL-VH combinations was reasonably high.

Figure 44 shows the variability introduced in the HCDR2 of the nal ^' ve HuCAL rat library.

Figure 45 demonstrates specificity of four rat Fab fragments for GM-CSF. All four Fab fragments were highly specific.

Figure 46 demonstrates that binders isolated from the murine library of the present invention maintain their functional activity after conversion from Fab format into full length immunoglobulin format. Detailed Description of the Invention

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this invention pertains.

The terms "about" or "approximately" in the context of numerical values and ranges refers to values or ranges that approximate or are close to the recited values or ranges such that the invention can perform as intended, such as having a desired number or percentage of sequence homology, as is apparent to the skilled person from the teachings contained herein. This is due, at least in part, to the varying culture conditions and the variability of biological systems. Thus, these terms encompass values beyond those resulting from systematic error. These terms make explicit what is implicit.

All ranges set forth herein in the summary and description of the invention include all numbers or values thereabout or therebetween of the numbers of the range. The ranges of the invention expressly denominate and set forth all integers, decimals and fractional values in the range. The term "about" can be used to describe a range.

The term "antibody" as used herein includes whole antibodies and any antigen binding fragment (i. e., "antigen-binding portion") or single chains thereof. A naturally occurring "antibody" is a glycoprotein comprising at least two heavy (H) chains and two light (L) chains inter-connected by disulfide bonds. Each heavy chain is comprised of a heavy chain variable region (abbreviated herein as VH) and a heavy chain constant region. The heavy chain constant region is comprised of three domains, CH1 , CH2 and CH3. Each light chain is comprised of a light chain variable region (abbreviated herein as VL) and a light chain constant region. The light chain constant region is comprised of one domain, CL. The VH and VL regions can be further subdivided into regions of hypervariability, termed complementarity determining regions (CDR), interspersed with regions that are more conserved, termed framework regions (FR). Each VH and VL is composed of three CDRs and four FRs arranged from amino-terminus to carboxy-terminus in the following order: FR1 , CDR1 , FR2, CDR2, FR3, CDR3, FR4. The variable regions of the heavy and light chains contain a binding domain that interacts with an antigen. The constant regions of the antibodies may mediate the binding of the immunoglobulin to host tissues or factors, including various cells of the immune system (e.g., effector cells) and the first component (C1 q) of the classical complement system. The terms "antigen binding portion" or "fragment" of an antibody are used equivalently in the present application. These terms refer to one or more fragments of an intact antibody that retain the ability to specifically bind to a given antigen. Antigen binding functions of an antibody can be performed by fragments of an intact antibody. Examples of binding fragments encompassed within the term "antigen binding portion" of an antibody include a Fab fragment, a monovalent fragment consisting of the VL, VH, CL and CH1 domains; a F(ab)2 fragment, a bivalent fragment comprising two Fab fragments linked by a disulfide bridge at the hinge region; an Fd fragment consisting of the VH and CH1 domains; an Fv fragment consisting of the VL and VH domains of a single arm of an antibody; a single domain antibody (dAb) fragment (Ward et al., 1989 Nature 341 :544-546), which consists of a VH domain; and an isolated complementarity determining region (CDR). Preferred antigen binding portions or fragments of antibodies are Fab fragments.

Furthermore, although the two domains of the Fv fragment, VL and VH, are coded for by separate genes, they can be joined, using recombinant methods, by an artificial peptide linker that enables them to be made as a single protein chain in which the VL and VH regions pair to form monovalent molecules (known as single chain Fv (scFv); see, e.g., Bird et al., 1988 Science 242:423- 426; and Huston et al., 1988 Proc. Natl. Acad. Sci. 85:5879-5883). Such single chain antibodies include one or more "antigen binding portions" of an antibody. These antibody fragments are obtained using conventional techniques known to those of skill in the art, and the fragments are screened for utility in the same manner as are intact antibodies.

Antigen binding portions can also be incorporated into single domain antibodies, maxibodies, minibodies, intrabodies, diabodies, triabodies, tetrabodies, v-NAR and bis-scFv (see, e.g., Hollinger and Hudson, 2005, Nature Biotechnology, 23, 9, 1 126-1 1 36). Antigen binding portions of antibodies can be grafted into scaffolds based on polypeptides such as Fibronectin type I I I (Fn3) (see U.S. Pat. No. 6,703, 199, which describes fibronectin polypeptide monobodies).

Antigen binding portions can be incorporated into single chain molecules comprising a pair of tandem Fv segments (VH-CH1 -VH-CH1 ) which, together with complementary light chain polypeptides, form a pair of antigen binding regions (Zapata et al., 1995 Protein Eng. 8(1 0):1057- 1062; and U.S. Pat. No. 5,641 ,870). As used herein, the term "affinity" refers to the strength of interaction between antibody and antigen at single antigenic sites. Within each antigenic site, the variable region of the antibody "arm" interacts through weak non-covalent forces with antigen at numerous sites; the more interactions, the stronger the affinity.

As used herein, the term "high affinity" for an antibody, such as an IgG antibody, refers to an antibody having a K _D of 10 ^"8 M or less, 10 ^"9 M or less, or 10 ^"10 M, or 10 ^{"1 1} M or less for a target antigen. However, "high affinity" binding can vary for other antibody isotypes. For example, "high affinity" binding for an IgM isotype refers to an antibody having a KD of 10 ^"7 M or less, or 10 ^"8 M or less.

The term "binding specificity" as used herein refers to the ability of an individual antibody combining site to react with only one antigenic determinant. The combining site of the antibody is located in the Fab portion of the molecule and is constructed from the hypervariable regions of the heavy and light chains. Binding affinity of an antibody is the strength of the reaction between a single antigenic determinant and a single combining site on the antibody. It is the sum of the attractive and repulsive forces operating between the antigenic determinant and the combining site of the antibody.

Specific binding between two entities means a binding with an equilibrium constant (K _A ) of at least 1 x 10 ⁷ M ^"1 , 10 ⁸ M ^"1 , 10 ⁹ M ^"1 , 10 ¹⁰ M ^"1 , or 1 0 ¹¹ M ^"1 . The phrase "specifically (or selectively) binds" to an refers to a binding reaction that is determinative of the presence of a cognate in a heterogeneous population of proteins and other biologies. In addition to the equilibrium constant (K _A ) noted above, an antibody of the invention typically also has a dissociation rate constant (Kd) of about 1 x 10 ^"2 s ^"1 , 1 x 10 ^"3 s ^"1 , 1 x 10 ^"4 s ^"1 , 1 x 10 ^"4 s ^"1 , or lower, and binds to a target antigen with an affinity that is at least two-fold greater than its affinity for binding to a non-specific antigen. The phrases "an antibody recognizing an antigen" and "an antibody specific for an antigen" are used interchangeably herein with the term "an antibody which binds specifically to an antigen". The terms "cross-block", "cross-blocked" and "cross-blocking" are used interchangeably herein to mean the ability of an antibody or other binding agent to interfere with the binding of other antibodies or binding agents to the same target in a standard competitive binding assay.

The ability or extent to which an antibody or other binding agent is able to interfere with the binding of another antibody or binding molecule to the same target, and therefore whether it can be said to cross-block according to the invention, can be determined using standard competition binding assays. One suitable assay involves the use of the Biacore technology (e.g. by using the BIAcore 3000 instrument (Biacore, Uppsala, Sweden)), which can measure the extent of interactions using surface plasmon resonance technology. Another assay for measuring cross-blocking uses an ELISA- based approach.

The term "epitope" means a protein determinant capable of specific binding to an antibody. Epitopes usually consist of chemically active surface groupings of molecules such as amino acids or sugar side chains and usually have specific three dimensional structural characteristics, as well as specific charge characteristics. Conformational and non-conformational epitopes are distinguished in that the binding to the former but not the latter is lost in the presence of denaturing solvents.

The term "chimeric antibody" is an antibody molecule in which (a) the constant region, or a portion thereof, is altered, replaced or exchanged so that the antigen binding site (variable region) is linked to a constant region of a different or altered class, effector function and/or species, or an entirely different molecule which confers new properties to the chimeric antibody, e.g., an enzyme, toxin, hormone, growth factor, drug, etc.; or (b) the variable region, or a portion thereof, is altered, replaced or exchanged with a variable region having a different or altered antigen specificity. For example, a mouse antibody can be modified by replacing its constant region with the constant region from a human immunoglobulin. Due to the replacement with a human constant region, the chimeric antibody can retain its specificity in recognizing the antigen while having reduced immunogenicity in human as compared to the original mouse antibody.

A "humanized" antibody is an antibody that retains the reactivity of a non-human antibody while being less immunogenic in humans. This can be achieved, for instance, by retaining the non- human CDR regions and replacing the remaining parts of the antibody with their human counterparts (i.e., the constant region as well as the framework portions of the variable region). See, e.g., Morrison et al., Proc. Natl. Acad. Sci. USA, 81 :6851 -6855, 1984; Morrison and Oi, Adv. Immunol., 44:65-92, 1988; Verhoeyen et al., Science, 239:1534-1536, 1988; Padlan, Molec. Immun., 28:489- 498, 1991 ; and Padlan, Molec. Immun., 31 :169-217, 1994. Other examples of human engineering technology include, but is not limited to Xoma technology disclosed in US 5,766,886.

The term "human antibody", as used herein, is intended to include antibodies having variable regions in which both the framework and CDR regions are derived from sequences of human origin. Furthermore, if the antibody contains a constant region, the constant region also is derived from such human sequences, e.g., human germline sequences, or mutated versions of human germline sequences. The human antibodies of the invention may include amino acid residues not encoded by human sequences (e.g., mutations introduced by random or site-specific mutagenesis in vitro or by somatic mutation in vivo).

The term "rodent antibody", as used herein, is intended to include antibodies having variable regions in which both the framework and CDR regions are derived from sequences of rodent origin. For example both, the framework and CDR regions may be derived from sequences of mouse origin, or both the framework and CDR regions may be derived from sequences of rat origin. Furthermore, if the antibody contains a constant region, the constant region also is derived from such rodent sequences, e.g., rodent germline sequences, or mutated versions of rodent germline sequences. The rodent antibodies of the invention may include amino acid residues not encoded by rodent sequences (e.g., mutations introduced by random or site-specific mutagenesis in vitro or by somatic mutation in vivo).

The term "rodent" is art recognized and includes the following species: mouse, rat, squirrel, chipmunk, gopher, porcupine, beaver, hamster, gerbil, guinea pig, degu, chinchilla, prairie dog, and groundhog. In certain aspects of the present invention said mouse is selected from the laboratory strains Balb/c and C57BL/6. The term "isolated antibody" refers to an antibody that is substantially free of other antibodies having different antigenic. An isolated antibody that specifically binds to an antigen may, however, have cross-reactivity to other antigens. Moreover, an isolated antibody may be substantially free of other cellular material and/or chemicals.

The term "isotype" refers to the antibody class (e.g., IgM, IgE, IgG such as lgG1 or lgG4) that is provided by the heavy chain constant region genes. Isotype also includes modified versions of one of these classes, where modifications have been made to alter the Fc function, for example, to enhance or reduce effector functions or binding to Fc receptors.

The term "Kassoc" or "Ka", as used herein, is intended to refer to the association rate of a particular antibody-antigen interaction, whereas the term "Kdis" or "Kd," as used herein, is intended to refer to the dissociation rate of a particular antibody-antigen interaction. The term "KD", as used herein, is intended to refer to the dissociation constant, which is obtained from the ratio of Kd to Ka (i.e. Kd/Ka) and is expressed as a molar concentration (M). K _D values for antibodies can be determined using methods well established in the art. A method for determining the K _D of an antibody is by using surface plasmon resonance, or using a biosensor system such as a Biacore ^® system.

The terms "monoclonal antibody" or "monoclonal antibody composition" as used herein refer to a preparation of antibody molecules of single molecular composition. A monoclonal antibody composition displays a single binding specificity and affinity for a particular epitope.

Antibody libraries can be derived from immunoglobulins, or fragments thereof, that are biased towards certain specificities present in immunized animals or naturally immunized, or infected, humans. Alternatively, antibody libraries can be derived from nal ^' ve immunoglobulins, or fragments thereof, i.e. immunoglobulins that are not biased towards specificities found in the immune system. Such libraries are referred to as "unbiased" libraries. In preferred embodiments the present invention provides unbiased antibody libraries, i.e. the libraries are not pre-exposed to the antigen of interest. Due to the absence of any bias, such libraries comprise antibodies binding to any potential target antigen of interest Typically, immune antibody libraries are constructed with VH and VL gene pools that are cloned from source B cells by PCR-based (or related) cloning techniques. In the same way it is also possible to generate unbiased, na ^' ive antibody libraries. Unbiased, nal ^' ve antibody libraries can however also be generated in a synthetic way in which the entire library is constructed entirely in vitro. Recombinant DNA technology is employed and may be used to mimic the natural biases and redundancies of the natural antibody repertoire. Such antibody libraries are referred to as "synthetic" antibody libraries. The term "fully synthetic" library refers to antibody libraries which are completely, i.e. fully, de novo constructed by DNA synthesis, e.g. by total gene synthesis, PCR-based methods, or related DNA technologies. In such libraries the entire DNA is constructed de novo, i.e. the part encoding the CDRs, as well as the parts (e.g. the framework regions) encoding the parts surrounding the CDRs of the antibodies of the library. The terms "synthetic" and "fully synthetic" therefore refer to the de novo origin of the DNA. In contrast, in a "semi-synthetic" antibody library only parts of the antibodies of the library are constructed de novo, whereas other parts, e.g. certain CDR regions, are derived from natural sources, (numerous reviews on this matter exist, see e.g. Sidhu et al. ; Nat Chem Biol (2006), 2, 682-8). In certain aspects, the present invention provides a synthetic rodent antibody library. In preferred aspect, the present invention provides a fully synthetic rodent antibody library.

The term "library" refers to a collection of disctinct molecules comprising typically more than 10 ³ , more than 1 0 ⁴ , more than 10 ⁵ , more than 10 ⁶ , more than 10 ⁷ , more than 10 ⁸ , more than 1 0 ⁹ or even more than 10 ¹⁰ members. A library in the context of the present invention is a mixture of heterogeneous polypeptides or nucleic acids. The library is composed of members, each of which have a single polypeptide or nucleic acid sequence. To this extent, library is synonymous with repertoire. Sequence differences between library members are responsible for the diversity present in the library. The library may take the form of a simple mixture of polypeptides or nucleic acids, or may be in the form of organisms or cells, for example bacteria, viruses, animal or plant cells, transformed with a library of nucleic acids. Preferably, each individual organism or cell contains only one or a limited number of library members. Advantageously, the nucleic acids are incorporated into expression vectors, in order to allow expression of the polypeptides encoded by the nucleic acids. In a certain aspect, a library may take the form of a population of host organisms, each organism containing one or more copies of an expression vector containing a single member of the library in nucleic acid form which can be expressed to produce its corresponding polypeptide member. Thus, the population of host organisms has the potential to encode a large repertoire of genetically diverse polypeptide variants. The term "collection" is essentially used interchangeably with the term "library".

The term "germline" refers to the nucleotide sequences of the antibody genes and gene segments as they are passed from parents to offspring via the germ cells. The germline sequence is distinguished from the nucleotide sequences encoding antibodies in mature B cells which have been altered by recombination and hypermutation events during the course of B cell maturation.

The term "nucleic acid" is used herein interchangeably with the term "polynucleotide" and refers to deoxyribonucleotides or ribonucleotides and polymers thereof in either single- or double- stranded form. The term encompasses nucleic acids containing known nucleotide analogs or modified backbone residues or linkages, which are synthetic, naturally occurring, and non-naturally occurring, which have similar binding properties as the reference nucleic acid, and which are metabolized in a manner similar to the reference nucleotides. Examples of such analogs include, without limitation, phosphorothioates, phosphoramidates, methyl phosphonates, chiral-methyl phosphonates, 2-O-methyl ribonucleotides, peptide-nucleic acids (PNAs).

Unless otherwise indicated, a particular nucleic acid sequence also implicitly encompasses conservatively modified variants thereof (e.g., degenerate codon substitutions) and complementary sequences, as well as the sequence explicitly indicated. Specifically, as detailed below, degenerate codon substitutions may be achieved by generating sequences in which the third position of one or more selected (or all) codons is substituted with mixed-base and/or deoxyinosine residues (Batzer et al., Nucleic Acid Res. 19:5081 , 1991 ; Ohtsuka et al., J. Biol. Chem. 260:2605-2608, 1985; and Rossolini et al., Mol. Cell. Probes 8:91 -98, 1994).

The term "operably linked" refers to a functional relationship between two or more polynucleotide (e.g., DNA) segments. Typically, it refers to the functional relationship of a transcriptional regulatory sequence to a transcribed sequence. For example, a promoter or enhancer sequence is operably linked to a coding sequence if it stimulates or modulates the transcription of the coding sequence in an appropriate host cell or other expression system. Generally, promoter transcriptional regulatory sequences that are operably linked to a transcribed sequence are physically contiguous to the transcribed sequence, i.e., they are cis-acting. However, some transcriptional regulatory sequences, such as enhancers, need not be physically contiguous or located in close proximity to the coding sequences whose transcription they enhance.

As used herein, the term, "optimized" means that a nucleotide sequence has been altered to encode an amino acid sequence using codons that are preferred in the production cell or organism, generally a eukaryotic cell, for example, a cell of Pichia, a Chinese Hamster Ovary cell (CHO) or a human cell. The optimized nucleotide sequence is engineered to retain completely or as much as possible the amino acid sequence originally encoded by the starting nucleotide sequence, which is also known as the "parental" sequence. The optimized sequences herein have been engineered to have codons that are preferred in mammalian cells. However, optimized expression of these sequences in other eukaryotic cells or prokaryotic cells is also envisioned herein. The amino acid sequences encoded by optimized nucleotide sequences are also referred to as optimized.

The term "amino acid" refers to naturally occurring and synthetic amino acids, as well as amino acid analogs and amino acid mimetics that function in a manner similar to the naturally occurring amino acids. Naturally occurring amino acids are those encoded by the genetic code, as well as those amino acids that are later modified, e.g., hydroxyproline, y-carboxyglutamate, and O- phosphoserine. Amino acid analogs refer to compounds that have the same basic chemical structure as a naturally occurring amino acid, i.e., an alpha carbon that is bound to a hydrogen, a carboxyl group, an amino group, and an R group, e.g., homoserine, norleucine, methionine sulfoxide, methionine methyl sulfonium. Such analogs have modified R groups (e.g., norleucine) or modified peptide backbones, but retain the same basic chemical structure as a naturally occurring amino acid. Amino acid mimetics refers to chemical compounds that have a structure that is different from the general chemical structure of an amino acid, but that functions in a manner similar to a naturally occurring amino acid.

The terms "polypeptide" and "protein" are used interchangeably herein to refer to a polymer of amino acid residues. The terms apply to amino acid polymers in which one or more amino acid residue is an artificial chemical mimetic of a corresponding naturally occurring amino acid, as well as to naturally occurring amino acid polymers and non-naturally occurring amino acid polymer. Unless otherwise indicated, a particular polypeptide sequence also implicitly encompasses conservatively modified variants thereof.

The term "conservatively modified variant" applies to both amino acid and nucleic acid sequences. With respect to particular nucleic acid sequences, conservatively modified variants refers to those nucleic acids which encode identical or essentially identical amino acid sequences, or where the nucleic acid does not encode an amino acid sequence, to essentially identical sequences. Because of the degeneracy of the genetic code, a large number of functionally identical nucleic acids encode any given protein. For instance, the codons GCA, GCC, GCG and GCU all encode the amino acid alanine. Thus, at every position where an alanine is specified by a codon, the codon can be altered to any of the corresponding codons described without altering the encoded polypeptide. Such nucleic acid variations are "silent variations," which are one species of conservatively modified variations. Every nucleic acid sequence herein which encodes a polypeptide also describes every possible silent variation of the nucleic acid. One of skill will recognize that each codon in a nucleic acid (except AUG, which is ordinarily the only codon for methionine, and TGG, which is ordinarily the only codon for tryptophan) can be modified to yield a functionally identical molecule. Accordingly, each silent variation of a nucleic acid that encodes a polypeptide is implicit in each described sequence.

For polypeptide sequences, "conservatively modified variants" include individual substitutions, deletions or additions to a polypeptide sequence which result in the substitution of an amino acid with a chemically similar amino acid. Conservative substitution tables providing functionally similar amino acids are well known in the art. Such conservatively modified variants are in addition to and do not exclude polymorphic variants, interspecies homologs, and alleles of the invention. The following eight groups contain amino acids that are conservative substitutions for one another: 1 ) Alanine (A), Glycine (G); 2) Aspartic acid (D), Glutamic acid (E); 3) Asparagine (N), Glutamine (Q); 4) Arginine (R), Lysine (K); 5) Isoleucine (I), Leucine (L), Methionine (M), Valine (V); 6) Phenylalanine (F), Tyrosine (Y), Tryptophan (W); 7) Serine (S), Threonine (T); and 8) Cysteine (C), Methionine (M) (see, e.g., Creighton, Proteins (1984)). In some embodiments, the term "conservative sequence modifications" are used to refer to amino acid modifications that do not significantly affect or alter the binding characteristics of the antibody containing the amino acid sequence. The terms "identical" or percent "identity," in the context of two or more nucleic acids or polypeptide sequences, refer to two or more sequences or subsequences that are the same. Two sequences are "substantially identical" if two sequences have a specified percentage of amino acid residues or nucleotides that are the same (i.e., 60% identity, optionally 65%, 70%, 75%, 80%, 85%, 90%, 95%, or 99% identity over a specified region, or, when not specified, over the entire sequence), when compared and aligned for maximum correspondence over a comparison window, or designated region as measured using one of the following sequence comparison algorithms or by manual alignment and visual inspection. Optionally, the identity exists over a region that is at least about 50 nucleotides (or 10 amino acids) in length, or more preferably over a region that is 100 to 500 or 1000 or more nucleotides (or 20, 50, 200 or more amino acids) in length.

For sequence comparison, typically one sequence acts as a reference sequence, to which test sequences are compared. When using a sequence comparison algorithm, test and reference sequences are entered into a computer, subsequence coordinates are designated, if necessary, and sequence algorithm program parameters are designated. Default program parameters can be used, or alternative parameters can be designated. The sequence comparison algorithm then calculates the percent sequence identities for the test sequences relative to the reference sequence, based on the program parameters.

A "comparison window", as used herein, includes reference to a segment of any one of the number of contiguous positions selected from the group consisting of from 20 to 600, usually about 50 to about 200, more usually about 100 to about 150 in which a sequence may be compared to a reference sequence of the same number of contiguous positions after the two sequences are optimally aligned. Methods of alignment of sequences for comparison are well known in the art. Optimal alignment of sequences for comparison can be conducted, e.g., by the local homology algorithm of Smith and Waterman (1 970) Adv. Appl. Math. 2:482c, by the homology alignment algorithm of Needleman and Wunsch, J. Mol. Biol. 48:443, 1970, by the search for similarity method of Pearson and Lipman, Proc. Nat'l. Acad. Sci. USA 85:2444, 1988, by computerized implementations of these algorithms (GAP, BESTFIT, FASTA, and TFASTA in the Wisconsin Genetics Software Package, Genetics Computer Group, 575 Science Dr., Madison, Wl), or by manual alignment and visual inspection (see, e.g., Brent et al., Current Protocols in Molecular Biology, John Wiley & Sons, Inc. (ringbou ed., 2003)). Two examples of algorithms that are suitable for determining percent sequence identity and sequence similarity are the BLAST and BLAST 2.0 algorithms, which are described in Altschul et al., Nuc. Acids Res. 25:3389-3402, 1977; and Altschul et al., J. Mol. Biol. 215:403-410, 1990, respectively. Software for performing BLAST analyses is publicly available through the National Center for Biotechnology Information. This algorithm involves first identifying high scoring sequence pairs (HSPs) by identifying short words of length W in the query sequence, which either match or satisfy some positive-valued threshold score T when aligned with a word of the same length in a database sequence. T is referred to as the neighborhood word score threshold (Altschul et al., supra). These initial neighborhood word hits act as seeds for initiating searches to find longer HSPs containing them. The word hits are extended in both directions along each sequence for as far as the cumulative alignment score can be increased. Cumulative scores are calculated using, for nucleotide sequences, the parameters M (reward score for a pair of matching residues; always > 0) and N (penalty score for mismatching residues; always < 0). For amino acid sequences, a scoring matrix is used to calculate the cumulative score. Extension of the word hits in each direction are halted when: the cumulative alignment score falls off by the quantity X from its maximum achieved value; the cumulative score goes to zero or below, due to the accumulation of one or more negative-scoring residue alignments; or the end of either sequence is reached. The BLAST algorithm parameters W, T, and X determine the sensitivity and speed of the alignment. The BLASTN program (for nucleotide sequences) uses as defaults a wordlength (W) of 1 1 , an expectation (E) or 10, M=5, N=-4 and a comparison of both strands. For amino acid sequences, the BLASTP program uses as defaults a wordlength of 3, and expectation (E) of 10, and the BLOSUM62 scoring matrix (see Henikoff and Henikoff, Proc. Natl. Acad. Sci. USA 89:10915, 1989) alignments (B) of 50, expectation (E) of 10, M=5, N=-4, and a comparison of both strands.

The BLAST algorithm also performs a statistical analysis of the similarity between two sequences (see, e.g., Karlin and Altschul, Proc. Natl. Acad. Sci. USA 90:5873-5787, 1993). One measure of similarity provided by the BLAST algorithm is the smallest sum probability (P(N)), which provides an indication of the probability by which a match between two nucleotide or amino acid sequences would occur by chance. For example, a nucleic acid is considered similar to a reference sequence if the smallest sum probability in a comparison of the test nucleic acid to the reference nucleic acid is less than about 0.2, more preferably less than about 0.01 , and most preferably less than about 0.001 .

The percent identity between two amino acid sequences can also be determined using the algorithm of E. Meyers and W. Miller (Comput. Appl. Biosci., 4:1 1 -17, 1988) which has been incorporated into the ALIGN program (version 2.0), using a PAM120 weight residue table, a gap length penalty of 12 and a gap penalty of 4. In addition, the percent identity between two amino acid sequences can be determined using the Needleman and Wunsch (J. Mol, Biol. 48:444-453, 1970) algorithm which has been incorporated into the GAP program in the GCG software package (available at www.gcg.com), using either a Blossom 62 matrix or a PAM250 matrix, and a gap weight of 16, 14, 12, 10, 8, 6, or 4 and a length weight of 1 , 2, 3, 4, 5, or 6.

Other than percentage of sequence identity noted above, another indication that two nucleic acid sequences or polypeptides are substantially identical is that the polypeptide encoded by the first nucleic acid is immunologically cross reactive with the antibodies raised against the polypeptide encoded by the second nucleic acid, as described below. Thus, a polypeptide is typically substantially identical to a second polypeptide, for example, where the two peptides differ only by conservative substitutions. Another indication that two nucleic acid sequences are substantially identical is that the two molecules or their complements hybridize to each other under stringent conditions, as described below. Yet another indication that two nucleic acid sequences are substantially identical is that the same primers can be used to amplify the sequence.

The term "recombinant antibody", as used herein, includes all antibodies that are prepared, expressed, created or isolated by recombinant means, such as antibodies isolated from an animal

(e.g., a mouse) that is transgenic or transchromosomal for rodent immunoglobulin genes or a hybridoma prepared therefrom, antibodies isolated from a host cell transformed to express rodent human antibody, e.g., from a transfectoma, antibodies isolated from a recombinant, combinatorial rodent antibody library, and antibodies prepared, expressed, created or isolated by any other means that involve splicing of all or a portion of a rodent immunoglobulin gene, sequences to other DNA sequences. Such recombinant rodent antibodies have variable regions in which the framework and

CDR regions are derived from rodent germline immunoglobulin sequences. In certain embodiments, however, such recombinant rodent antibodies can be subjected to in vitro mutagenesis (or, when an animal transgenic for rodent Ig sequences is used, in vivo somatic mutagenesis) and thus the amino acid sequences of the VH and VL regions of the recombinant antibodies are sequences that, while derived from and related to rodent germline VH and VL sequences, may not naturally exist within the rodent antibody germline repertoire in vivo.

The term "recombinant host cell" (or simply "host cell") refers to a cell into which a recombinant expression vector has been introduced. It should be understood that such terms are intended to refer not only to the particular subject cell but to the progeny of such a cell. Because certain modifications may occur in succeeding generations due to either mutation or environmental influences, such progeny may not, in fact, be identical to the parent cell, but are still included within the scope of the term "host cell" as used herein.

The term "vector" is intended to refer to a polynucleotide molecule capable of transporting another polynucleotide to which it has been linked. One type of vector is a "plasmid", which refers to a circular double stranded DNA loop into which additional DNA segments may be ligated. Another type of vector is a viral vector, wherein additional DNA segments may be ligated into the viral genome. Certain vectors are capable of autonomous replication in a host cell into which they are introduced (e.g., bacterial vectors having a bacterial origin of replication and episomal mammalian vectors). Other vectors (e.g., non-episomal mammalian vectors) can be integrated into the genome of a host cell upon introduction into the host cell, and thereby are replicated along with the host genome. Moreover, certain vectors are capable of directing the expression of genes to which they are operatively linked. Such vectors are referred to herein as "recombinant expression vectors" (or simply, "expression vectors"). In general, expression vectors of utility in recombinant DNA techniques are often in the form of plasmids. In the present specification, "plasmid" and "vector" may be used interchangeably as the plasmid is the most commonly used form of vector. However, the invention is intended to include such other forms of expression vectors, such as viral vectors (e.g., replication defective retroviruses, adenoviruses and adeno-associated viruses), which serve equivalent functions. Restriction sites that are "unique" are restriction sites that exist or appear only once on a given nucleic acid molecule. Typically such a nucleic acid molecule is a vector which encodes the library members of the present invention.

The term "mVL" is abbreviated for variable domain of the murine light immunoglobulin chain. The term "mVH" is abbreviated for variable domain of the murine heavy immunoglobulin chain. The term "rVL" is abbreviated for variable domain of the rat light immunoglobulin chain. The term "rVH" is abbreviated for variable domain of the rat heavy immunoglobulin chain. The term "IMGT" is abbreviated for ImMunoGeneTics, an on-line database, (Giudicelli, et al. (1997), Nucleic Acids Res 25, 206-21 1 )).

The term "subject" includes human and non-human animals. Non-human animals include all vertebrates, e.g., mammals and non-mammals, such as non-human primates, sheep, dog, cow, chickens, amphibians, and reptiles. Except when noted, the terms "patient" or "subject" are used herein interchangeably.

The term "treating" includes the administration of compositions or antibodies to prevent or delay the onset of the symptoms, complications, or biochemical indicia of a disease, alleviating the symptoms or arresting or inhibiting further development of the disease, condition, or disorder. Treatment may be prophylactic (to prevent or delay the onset of the disease, or to prevent the manifestation of clinical or subclinical symptoms thereof) or therapeutic suppression or alleviation of symptoms after the manifestation of the disease.

The term "position-dependent amino-acid usage" refers to the likelihood of occurrence of a particular amino acid sequence at a given position in a polypeptide. In the present invention, the position-dependent amino acid usage was determined for the re-arranged amino acid sequences classified by the individual germline gene. This enables the individual, precise design of the CDRs within its natural germline context.

In certain aspects the present invention provides a synthetic rodent antibody library which comprises members of at least two germline families. In other aspects said library comprises members of at least three, at least four or at least five germline families. In preferred aspects said synthetic rodent antibody library is a murine rodent library. In other aspects said synthetic rodent antibody library is a rat rodent library

In certain aspects the present invention provides a synthetic murine antibody library which comprises more than 50 % of the natural mouse VH repertoire . In other aspects said library comprises more than 60 %, more than 70 %, more than 75 %, more than 80%, more than 85 %, more than 90 % or more than 95 % of the natural mouse VH repertoire.

In certains aspects the present invention provides a synthetic murine antibody library which comprises members of at least two VH germline families. In other aspects said library comprises members of at least three, at least four or at least five VH germline families. In certain aspects said VH germline families are selected from VH1 , VH2, VH3, VH5 and VH9.

In certain aspects the present invention provides a synthetic murine antibody library which comprises more than 50 % of the natural mouse VL repertoire . In other aspects said library comprises more than 60 %, more than 70 %, more than 75 %, more than 80%, more than 85 %, more than 90 % or more than 95 % of the natural mouse VL repertoire.

In certains aspects the present invention provides a synthetic murine antibody library which comprises members of at least two VL germline families. In other aspects said library comprises members of at least three VL germline families. In certain aspects said VL germline families are VL- kappa germline families selected from VL1 , VL3, VL7, VL8 and VL19.

In certain aspects the present invention provides a synthetic murine antibody library which comprises at least one, at least two, at least three, at least four or at least five of the following VH germline genes: IGHV1 -72 ^* 01 (VH1 ), IGHV2-2 ^* 01 (VH2), IGHV3-6 ^* 01 (VH3), IGHV5-9 ^* 04 (VH5), and IGHV9-3 ^* 01 (VH9). In other aspects the present invention provides a synthetic murine antibody library which comprises the following VH germline genes: IGHV1 -72 ^* 01 (VH1 ), IGHV2-2 ^* 01 (VH2), and IGHV5-9 ^* 04 (VH5),

In certain embodiments the present invention provides a synthetic murine antibody library which comprises a variant VH1 germline gene IGHV1 -72 ^* 01 which contains a K77R and/or a T78V mutation. In certain aspects said variant VH1 germline gene IGHV1 -72 ^* 01 contains K77R mutation. In other aspects said variant contains a T78V mutation. In preferred aspects said variant contains a K77R and a T78V mutation.

In certain aspects the present invention provides a synthetic murine antibody library which comprises at least one, at least two, at least three, at least four, at least five or at least six of the following VL-kappa germline genes: IGKV1 -1 17 ^* 01 (Vk1 ), IGKV3-12 ^* 01 (Vk3), IGHV3-4 ^* 01 (Vk3), IGHV7-33 ^* 01 (Vk7), IGHV8 ^* 21 ^* 01 (Vk8) and GHV19-93 ^* 01 (Vk19). In other aspects the present invention provides a synthetic murine antibody library which comprises at least two of the following VL-kappa germline genes: IGKV1 -1 17 ^* 01 (Vk1 ), IGKV3-12 ^* 01 (Vk3), and IGHV3-4 ^* 01 (Vk3).

In certain aspects the present invention provides a synthetic murine antibody library which comprises the VH germline genes IGHV1 -72 ^* 01 (VH1 ), IGHV2-2 ^* 01 (VH2), and IGHV5-9 ^* 04 (VH5) and at least two of the following VL-kappa germline genes: IGKV1 -1 17 ^* 01 (Vk1 ), IGKV3-12 ^* 01 (Vk3), and IGHV3- 4 ^* 01 (Vk3).

In certain aspects of the present invention the germline genes require modification in order to include restriction sites and to achieve modularity of the library. Therefore, in certain aspects the present invention provides a synthetic murine antibody library comprising one or more of the following modified germline genes: IGHV1 -72 ^* 01 with a Q1 E mutation; IGHV1 -72 ^* 01 with a Q1 E, a K77R and a A78V mutation; IGHV2-2 ^* 01 with a Q1 E mutation; IGHV3-6 ^* 01 with a D1 E and a T26S mutation; IGHV5-9 ^* 04 with a K3Q mutation; IGHV9-3 ^* 01 with a Q1 E, a I2V and a K53E mutation and IGKV1 - 1 17 ^* 01 with a V2I mutation.

In certain embodiments the present invention provides a synthetic murine antibody library which comprises H-CDR3S which cover more than 50 % of the naturally occuring H-CDR3 lengths of the murine H-CDR3 repertoire. In other aspects said library comprises more than 60 %, more than 70 %, more than 75 %, more than 80%, more than 85 %, more than 90 % or more than 95 % of the naturally occuring H-CDR3 lengths of the murine H-CDR3 repertoire.

In certain aspects the present invention provides a synthetic murine antibody library which comprises H-CDR3s of the length of 7-12 amino acids. In other aspects the present invention provides a synthetic murine antibody library which comprises H-CDR3s of the length of 7 amino acids and/or 8 amino acids and/or 9 amino acids and/or 10 amino acids and/or 1 1 amino acids and/or 12 amino acids.

In certain aspects the present invention provides a synthetic murine antibody library which comprises a H-CDR3 design as shown in Figure 5 or Figure 6. In certain aspects the present invention provides a synthetic murine antibody library which comprises a H-CDR3 design as shown in Figure 6.

In certain aspects the present invention provides a synthetic murine antibody library in which the H-CDR3 region has a diversity of at least 1.0 ^* 10 ^s . In other aspects the present invention provides a synthetic murine antibody library which the H-CDR3 region has a diversity of at least 1 .0 ^* 10 ⁹ , of at least 1 .0 ^* 10 ¹⁰ , of at least 1 .0 ^* 10 ^{1 1} , of at least 1 .0 ^* 10 ¹² or of at least 1.0 ^* 10 ¹³ .

In certain aspects the present invention provides a synthetic murine antibody library which comprises L-CDR3s which cover more than 80 % of the naturally occuring L-CDR3 lengths of the murine L-CDR3 repertoire. In other aspects said library comprises more than 85 %, more than 90 % or more than 95 % of the naturally occuring L-CDR3 lengths of the murine L-CDR3 repertoire.

In certain aspects the present invention provides a synthetic murine antibody library which comprises a L-CDR3 of the length of 8 amino acids. In certain aspects the present invention provides a synthetic murine antibody library in which the L-CDR3 of essentially all members of the library is of the length of 8 amino acids.

In certain aspects the present invention provides a synthetic murine antibody library which comprises a L-CDR3 design as shown in Figure 7 or Figure 8. In certain aspects the present invention provides a synthetic murine antibody library which comprises a L-CDR3 design as shown in Figure 8.

In certain aspects the present invention provides a synthetic murine antibody library in which the L-CDR3 region has a diversity of at least 1 .0 ^* 10 ⁴ . In other aspects the present invention provides a synthetic murine antibody library which the L-CDR3 region has a diversity of at least 1 .0 ^* 10 ⁵ , of at least 1 .0 ^* 10 ⁶ or of at least 1.0 ^* 10 ⁷ . In certain aspects the present invention provides a synthetic murine antibody library in which at least one of amino acid residues 52, 52a, 54, 56 and 58 (Kabat numbering) of the H-CDR2 is diversified. In other aspects the present invention provides a synthetic murine antibody library in which at least two, at least three, at least four or at least five of amino acid residues 52, 52a, 54, 56 and 58 (Kabat numbering) of the H-CDR2 are diversified.

In certain aspects the present invention provides a synthetic murine antibody library which comprises a H-CDR2 design as shown in Figure 10 or Figure 1 1 . In certain aspects the present invention provides a synthetic murine antibody library which comprises a H-CDR2 design as shown in Figure 1 1 .

In certain aspects the present invention provides a synthetic murine antibody library in which the H-CDR2 region has a diversity of at least 1.0 ^* 10 ⁵ . In other aspects the present invention provides a synthetic murine antibody library which the H-CDR2 region has a diversity of at least 1 .0 ^* 10 ⁶ , of at least 1 .0 ^* 10 ⁷ , of at least 1 .0 ^* 10 ⁸ , of at least 1 .0 ^* 10 ⁹ or of at least 1.0 ^* 10 ¹⁰

In certain aspects the present invention provides a synthetic murine antibody library with a VH1 germline gene which comprises a diversity of at least 1 .0 ^* 10 ⁵ in the H-CDR2 region. In other aspects the present invention provides a synthetic murine antibody library with a VH1 germline which comprises a diversity of at least 1 .0 ^* 1 0 ⁶ , of at least 1 .0 ^* 10 ⁷ , of at least 1 .0 ^* 10 ⁸ or of at least 1 .0 ^* 10 ⁹ in the H-CDR2 region.

In certain aspects the present invention provides a synthetic murine antibody library with a VH2 germline gene which comprises a diversity of at least 1 .0 ^* 10 ³ in the H-CDR2 region. In other aspects the present invention provides a synthetic murine antibody library with a VH2 germline gene which comprises a diversity of at least 1 .0 ^* 10 ⁴ , of at least 1 .0 ^* 10 ⁵ , of at least 1 .0 ^* 10 ⁶ or of about 1 .0 ^* 10 ⁷ in the H-CDR2 region.

In certain aspects the present invention provides a synthetic murine antibody library with a VH5 germline gene which comprises a diversity of at least 1 .0 ^* 10 ⁵ in the H-CDR2 region. In other aspects the present invention provides a synthetic murine antibody library with a VH5 germline which comprises a diversity of at least 1 .0 ^* 1 0 ⁶ , of at least 1 .0 ^* 10 ⁷ , of at least 1 .0 ^* 10 ⁸ or of at least 1 .0 ^* 10 ⁹ in the H-CDR2 region. In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are efficiently displayed. In other aspects the present invention provides a synthetic murine antibody library wherein all or substantially all VH-VL combinations, with the exception of all VLB7-combinations are efficiently displayed. In yet other aspects the present invention provides a synthetic murine antibody library comprising at least two, at least three, at least four or at least five VH germline genes and at least two, at least three, at least four or at least five VL germline genes, wherein each of the VH-VL combinations comprised in said library is efficiently displayed. Efficiency of display can be measured by sandwich phage ELISA as described in the present invention.

In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are well expressed in E. coli in Fab format. In other aspects the present invention provides a synthetic murine antibody library wherein all or substantially all VH-VL combinations, with the exception of all VLB7-combinations are well expressed in E. coli in Fab format. In yet other aspects the present invention provides a synthetic murine antibody library comprising at least two, at least three, at least four or at least five VH germline genes and at least two, at least three, at least four or at least five VL germline genes, wherein each of the VH-VL combinations comprised in said library is well expressed in E. coli in Fab format. Expression in Fab format in E. coli can be quantified. In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are expressed at levels of more than 5 mg/L in a bacterial culture. In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are expressed at levels of more than 10 mg/L in a bacterial culture. In certain aspects the present invention provides a synthetic murine antibody library wherein most VH-VL combinations are expressed at levels of more than 15 mg/L in a bacterial culture. In certain aspects the present invention provides a synthetic murine antibody library wherein at least two, at least three, at least four or at least five VH-VL combinations are expressed at levels of more than 20 mg/L in a bacterial culture.

In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are well expressed in a mammalian system in IgG format. In other aspects the present invention provides a synthetic murine antibody library wherein all or substantially all VH-VL combinations, with the exception of all VLB7-combinations are well expressed in a mammalian system in IgG format. In yet other aspects the present invention provides a synthetic murine antibody library comprising at least two, at least three, at least four or at least five VH germline genes and at least two, at least three, at least four or at least five VL germline genes, wherein each of the VH-VL combinations comprised in said library is well expressed in a mammalian system in IgG format. Expression in a mammalian system in IgG format can be quantified. In certain aspects said mammalian system is a mammalian suspension culture. In other aspects said mammalian system is a mammalian adherent cell culture. In certain aspects said IgG format is an lgG1 format. In other aspects said IgG format is an lgG2a format. In certain aspects said mammalian system comprises HKB1 1 cells. In other aspects said mammalian system comprises PERC.6 cells. In yet other aspects said mammalian system comprises CHO cells. In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are expressed at levels of more than 5 mg/L in a mammalian system in IgG format. In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations are expressed at levels of more than 10 mg/L in a mammalian system in IgG format. In certain aspects the present invention provides a synthetic murine antibody library wherein most VH-VL combinations are expressed at levels of more than 15 mg/L in a mammalian system in IgG format. In certain aspects the present invention provides a synthetic murine antibody library wherein at least two, at least three, at least four or at least five VH-VL combinations are expressed at levels of more than 20 mg/L in a mammalian system in IgG format.

In certain aspects the present invention provides a synthetic murine antibody library wherein all or substantially all of the members of said library are stable in isoproanol at a concentration of 30 % (v/v). It is an assumption that all CDR derivatives of a stable framework will behave like the tested mastergenes.

In certain aspects the present invention provides a synthetic murine antibody library wherein all or substantially all VH-VL combinations are thermally stable. Thermal stability can be measured as described in the present application. In certain aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations have a T _m of more than 62°C. In other aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations have a T _m of more than 64 °C. In yet other aspects the present invention provides a synthetic murine antibody library wherein essentially all VH-VL combinations have a T _m of more than 66°C. In yet other aspects the present invention provides a synthetic murine antibody library wherein most VH-VL combinations have a T ^" ™ of more than 68°C. In yet other aspects the present invention provides a synthetic murine antibody library wherein many VH-VL combinations have a T _m of more than 70 °C. In certain aspects the present invention provides a synthetic murine antibody library wherein all or substantially all VH1 -combinations have a T _m of more than 70 °C. In other aspects the present invention provides a synthetic murine antibody library wherein all or substantially all VH1_RV-combinations have a T _m of more than 70 °C. In yet other aspects the present invention provides a synthetic murine antibody library wherein all VLB1 -combinations have a T _m of more than 68°C. In yet other aspects the present invention provides a synthetic murine antibody library wherein all VLB3-combinations have a T _m of more than 68°C. In yet other aspects the present invention provides a synthetic murine antibody library wherein all VLB7-combinations have a T _m of more than 68°C. In yet other aspects the present invention provides a synthetic murine antibody library wherein all VLC3-combinations have a T _m of more than 68 °C.

In certain aspects the present invention provides rodent antibodies which possess favourable biophysical properties. Such antibodies are devoid of short comings of antibodies derived from synthetic libraries which have not undergone in-vivo maturation. Such favourable and desired biophysical properties include higher stability, higher expression levels and a low tendency for aggregation.

In certain aspects the present invention provides an antibody isolated from a rodent library contemplated by the present invention. In certain aspects said antibody may be a modified or a variant antibody of an antibody isolated from a rodent library contemplated by the present invention. . In other aspects said modified or variant antibody may be a humanized antibody.

Each compositions and attendant aspects, and each method and attendant aspects, which are described above can be combined with another in a manner consistent with the teachings contained herein. According to the embodiments of the inventions, all methods and the steps in each method can be applied in any order and repeated as many times in a manner consistent with the teachings contained herein. The invention is further described by the following examples, which do not limit the invention in any manner. For example, certain design features described below may be unique to the "HuCAL MOUSE" or "HuCAL RAT" design and not required in an alternative rodent library contemplated by the present invention.

Examples

Example 1 : Generation of a murine antibody library (HuCAL MOUSE)

Example 1 .1 : Selection of VL- and VH-qermline sequences for generation of the murine master genes

Public available data were analyzed for usage of murine VH germline families (see Kaushik et al., Res. Immunol. (1996) 147, 9-26). The following table summarizes the occurrence of the different VH germlines:

Table 1 :

% in Kabat DB Name (Kaushik % in Kaushik et al. % in Kaushik et al.

VH (var. strains) et a I. (1996)) (BALB/c) (var. strains)

VH1 33 J558 30 45

VH2 24 Q52 14 10

VH3 24 30-60 8 17

VH5 16 7183 1 1 1 1

VH9 10 VGA 9 10

SUM 87 72 93

Our aim was to generate a library that comprises as many germline families as possible. A library containing one member of each of the listed germline-families as mastergene would include more than 85 % of the mouse VH repertoire, and exactly this was the goal of our task. Germline-families often comprise more than one germline gene. For the listed families the number of germline gene ranges from 7 (VH2) to 134 (VH1 ) - the exact number depends on the mouse strain. Since most animal studies focus on BALB/c- and C57BL-6 mouse strains, only germline genes favoured by these strains were considered as basis for master-genes of the murine library. See Table 2. A reasonable intersection of VH-germline genes of both mouse strains was found and used exclusively.

Table 2:

# of germline-genes in family

VH (var. strains) (BALB/c & C57BL-6)

VH1 134 35

VH2 7 2

VH3 5 4

VH5 20 9

VH9 8 7

Among the murine light chains, kappa light chains constitute the majority of all light chains, accounting to about 90% of serum immunoglobulins isolated from mice (Frank and Gutman, 1988, Mo I Immunol 25, 953-960). Therefore only VL-kappa was considered in the murine library. Since BALB/c mice and C57BL-6 mice do not share the same VL-germline-usage, the development of VL- mastergenes for BALB/c and C57BL-6 was done separately. Table 3 shows the germline families that occur most frequently.

Table 3:

BALB/c C57BL-6

VL-kappa -family Usage in% VL-kappa -family Usage in%

1 40 3 49 We aimed to prepare a library that contains one member of each of the listed germline families. Such a library will include >85% of the VL-sequences of BALB/c and C57BL-6 mice. Although germline family kappa-3 is used by both strains, two separate master genes are required, because different germline-genes, with in total 5 nucleotide differences, are preferred. Thus, in total 6 VL- mastergenes are required. Certain alterations of these germline genes are required to form master genes which maintain the modularity of the library.

Example 1 .2: Selection of master genes

Two approaches were taken for the selection of the most frequently used germline genes, which are to be used as master genes:

a) Identification of the most prominent germline gene within the respective germline families b) Identification of the germline gene, which is most akin to the consensus of all germline- genes within one family.

Ideally, both approaches would result in the same germline-gene.

Example 1 .2.1 : Identification of the most prominent qermline-qene within the respective qermline- families

The germline gene used most abundantly was identified by comparison of the germline genes to re-arranged sequences. About 9,400 re-arranged murine VH and 2,500 Vic-sequences were extracted from a database flatfile, which was downloaded on from IgBLAST on the NCBI-Server (url: http://www.ncbi.nlm.nih.gov/igblast/). Sequence-fragments and sequences of bad quality were excluded. The remaining sequences were classified into germline families by application of filters for unique, germline specific amino acid pattern. This raw version of family-sorting was corrected by scoring the re-arranged antibodies against germline genes. Here the position-dependent amino-acid usage of each re-arranged sequence was compared to the amino-acid usage of each germline gene of various strains. By doing so, each re-arranged sequence got assigned to a nearest germline gene. If wrong assigned re-arranged sequences were identified, they were sorted to the correct germline family, leading to distribution shown in Table 4 (VH) and 5 (VL).

Table 4:

VH: Germline- # re-arranged

family sequences % distribution

VH1 5980 70.9

VH2 597 7.1

VH3 246 2.9

VH5 762 9.0

VH9 117 1.4

others 730 8.7

SUM 8432 100

Table 5:

VL:

Germline- # re-arranged

% distribution

family sequences

Vk1 543 21 .6

Vk3 232 9.2

Vk7 14 0.6

Vk8 187 7.4

Vk19 44 1 .8

others 1494 59.4

SUM 2514 100

The position-dependent amino-acid usage of the re-arranged sequence was compared with the amino-acid usage of the BLAB/c and C57BL-6-specific-germline-genes to identify the most frequent used germline genes. Each re-arranged sequence got assigned a nearest BLAB/C/C57BL-6- germline gene. If two or more germline genes were identified as nearest germline gene for a rearranged sequence, all germline-genes were considered as nearest germline genes. For calculation of the preferably used germline-gene within a family the usage of each germline-gene was summed up. The results for the preferred germline genes within a family were surprisingly unambiguous and clear. Table 6 shows an example of the results for the usage of the four germline genes within the VH3-family:

Table 6: absolute relative

germline-gene usage usage [%]

D 13203I IGHV3-3*03| BALB/c 0 0.0

AC073589| IGHV3-3 ^* 01 |C57BL/6 5 1.9

AJ223544|IGHV3-6 ^* 03| BALB/c 14 5.4

AC073590| IGHV3-6 ^* 01 |C57BL/6 242 92.7

SUM 261 100

# re-arranged sequences 246

For the example shown the sum of the absolute usage of germline-genes (261 ) differs from the number of re-arranged sequences (246), because 1 5 of the re-arranged sequences had been assigned to two or more germline-genes.

Table 7 shows the germline genes that are the most prominent in the different VH-families. Table 8 shows the most prominent germline genes for the different VL-families. Different germline- genes were identified for Vk3 of BALB/c and C56BL-6 as most prominent ones.

Table 7:

% of prominent germline-

Germline- Name of prominent # re-arranged gene

family germline-gene sequences in-arranged sequences

VH1 AC163348| IGHV1 -72*01 5980 49.7

VH2 AC090887| IGHV2-2*01 597 87.0

VH3 AC073590| IGHV3-6 ^* 01 246 92.7

VH5 AF120472| IGHV5-9 ^* 04 762 49.8

VH9 AC073563_| IGHV9-3 ^* 01 1 17 63.1

Table 8:

% of prominent

Germline- Name of prominent # re-arranged germline-gene

family strain germline-gene sequences in-arranged sequences

Vk1 BALB/c IGKV1 -1 17 ^* 01 543 90%

Vk3 BALB/c IGKV3-12 ^* 01 232 24%

Vk3 C57BL-6 IGHV3-4 ^* 01 232 75%

Vk7*) BALB/c IGHV7-33*01 246 100%

Vk8 C57BL-6 IGHV8*21 *01 187 37% Vk19*) C57BL-6 IGHV19-93 ^* 01 44 100%

SUM 1484

^* ) Germline-family has one master-gene only

Example 1 .2.2: Identification of the qermline-qene, which is most akin to the consensus of all qermline-qenes within one family

Consensus sequences of the germline-genes of the BALB/c C57BL-6-germline intersection were prepared. The comparison of the consensus sequences to the germline genes was done analogously as described herein above: the position-dependent amino-acid usage of the consensus sequences was compared to the amino-acid usage of the BLAB/c and C57BL-6-specific-germline- genes. By doing so, the consensus sequences got assigned a nearest BLAB/c and C57BL-6-germline gene. The result is shown in Table 9.

Table 9:

Name of

Germline- Name of prominent Germline- prominent

family germline-gene family strain germline-gene

VH1 AC163348|IGHV1 -72 ^* 01 Vk1 BALB/c IGKV1 -1 17 ^* 01

VH2 AC090887|IGHV2-2 ^* 01 Vk3 BALB/c IGKV3-12 ^* 01

VH3 AC073590|IGHV3-6 ^* 01 Vk3 C57BL-6 IGHV3-4 ^* 01

VH5 AF120472|IGHV5-9 ^* 04 Vk7 BALB/c IGHV7-33 ^* 01

VH9 AC073563_|IGHV9-3 ^* 01 Vk8 C57BL-6 IGHV8*28*01

Vk19 C57BL-6 IGHV19-93 ^* 01

The validity of this approach was confirmed by BLAST analyses of the consensus-sequences against all germline genes. Except for Vk8, the identification of the most prominent germline gene within the respective germline families, and the germline gene within the highest similarity to the consensus or BLAST search, came up with the same set of germline genes. For Vk8 IGKV8-28 ^* 01 was chosen as master gene, since is shows less deviations to the consensus-sequence than IGKV8- 21 ^* 01 . The usage of IGKV8-28 ^* 01 for rearranged sequences in germline-family Vk8 is 1 1 %. For mVH1 a second variant was included with K77R and T78V mutations, because the formation of a stabilising salt bridge between R77 and D100 might add extra high stability to this VH1 - variant (Kaufmann et al., J.Mol.Biol, (2002) 318, 135-47). See Figure 1 .

The mastergenes eventually selected (see herein below) required slight modifications in order to achieve modularity of the library. The modifications made to generate restriction are summarized in Tables 10 (VH) and 1 1 (VL).

Table 10:

Name of nearest Deviations to nearest

MG

germline-gene germline-gene

mVH1 AC163348| IGHV1 -72 ^* 01 Q1 E

Q1 E, K77R, A78V

mVH1_RV AC163348| IGHV1 -72 ^* 01

(stabilizing salt bridge)

mVH2 AC090887| IGHV2-2 ^* 01 Q1 E

mVH3 AC073590| IGHV3-6 ^* 01 D1 E, T26S (BspEI-site)

mVH5 AF120472|IGHV5-9*04 K3Q (Mfel-site)

mVH9 AC073563_| IGHV9-3 ^* 01 Q1 E, I2V, K53E (Xhol-site)

Table 1 1 :

Name of Deviations to

nearest nearest

MG strain germline-gene germline-gene

Vk1 = B1 BALB/c IGKV1 -1 17 ^* 01 V2I (EcoRV-site)

Vk3 = B3 BALB/c IGKV3-12*01 -

Vk7 = B7 BALB/c IGHV7-33*01 -

Vk3 = C3 C57BL-6 IGHV3-4 ^* 01 -

Vk8 = C8 C57BL-6 IGHV8 ^* 28 ^* 01 -

Vk19 = Cl9 C57BL-6 IGHV19-93 ^* 01 -

Also, to create full-length master genes from the germline genes, antibody framework 4 had to be added to the sequences. IMGT (http://irnqt.cines.fr/) provides access to murine J-regions, which form antibody framework 4 in antibody-development. Figure 2 shows the framework regions for the heavy chains, Figure 3 for the light chains. J-region JH2 was selected for the VH-mastergene, since the associated 3' end of HCDR3 (FDY) is preferred in HCDR3 and JH2 provides al required RE-sites (Styl and Blpl). J-regions JK1 (=JK2 in FW4-sequence) was selected for complementation of the VL-mastergene, covering 85% of the mouse JK-usage. The last amino acid of framework 4 was changed from Ala to Thr to provide a BsiWI-site.

Example 1 .3: Design of murine CDRs

The sites of the closest contact between antibody and antigen are the complementary determining regions (CDR) of the antibody. H-CDR3 and L-CDR3 are playing major roles in antigen- binding, thus variability was mainly introduced into these two CDRs. Additional limited variability was introduced at antigen-contact-positions of the H-CDR2. Germline sequences were used for all other CDRs. For improvement of affinities of selected murine binders to obtain affinities comparable to the human lead candidate additional maturation cassettes were synthesized for H-CDR2.

Cysteine residues were generally avoided in the design of the CDRs because cysteine residues can be engaged in the formation of disulfide bonds covalently linking a binder with the panning target or in formation of antibody homodimers. Such binders are unlikely to be target- selective.

Re-arranged antibody-sequences were compiled and analysed for design of murine CDRs (see Example 1 .1 ). Tables 4 (VH) and 5 (VL) show the data-sets used.

Example 1 .3.1 : Design of H-CDR3

The statistical analysis of the H-CDR3 was done for all re-arranged sequences irrespective of the germline family, since this CDR is encoded germline-independent by the D- and J-segment. Although tyrosine, glycine and serine predominate in both, human and mouse H-CDR3 sequences, amino acid utilization differs among sequences of the same length between mouse and human variable heavy chains. These differences in overall amino acid frequencies and in the local distribution of amino acid residues within the H-CDR3-loops contribute to dramatic differences in the diversity and predicted structures of the H-CDR3 repertoires of mouse and human. (Zemlin et al., 2003, J Mo! Biol 334, 733-749). Based on the analysis of mouse rearranged sequences some positions were identified, which deviated extremely from the composition of human H-CDR3 design. Especially Kabat position 95, 96 and 102 showed another amino acid usage than the human H-CDR3 (see Table 12). Specially designed murine H-CDR3s are therefore required.

Table 12:

Mouse Human

Illustration reduced to amino acids with an appearance of >=5%.

In comparison to human antibodies, the length-range of H-CDR3 found in murine antibodies is restricted. While human H-CDR3 covers lengths from 2 to 28 amino acids, murine H-CDR3 uses only a set of 3 to 16 amino acids, as was found in analysis of data from Ig-BLAST. For results see Table 13. Similar results were described by Zemlin et al. (Zemlin et al., 2003, J Mol Biol 334, 733-749). TRIM ^® technology (see WO 93/21203) was used to design mouse H-CDR3s covering a length of 7 - 12 amino acids. This range covers about 80 % of all murine H-CDR3S.

Table 13: HCDR3 length

(HuCAL-numbering) # of sequences %

1_aa 0 0

2_aa 1 0

3_aa 72 1

4_aa 1 13 1

5_aa 326 4

6 aa 341 4

7_aa 446 6

8_aa 645 8

9_aa 1705 21

10_aa 131 1 17

1 1_aa 1054 13

12 aa 1 189 15

13_aa 355 4

14_aa 187 2

15_aa 120 2

16_aa 46 1

17_aa 19 0

18 aa 9 0

19 aa 2 0

20 aa 0 0

21_aa 0 0

total # of sequences 7941 100

The position-dependent relative amino acid distribution was determined for every single length, resulting in the natural amino acid distribution. The single amino acid distributions were aligned by grouping for the D-segment, leading to a right-centred alignment. Results are shown in Figure 4. To summarize all lengths in one design, the weighted average of the amino acids was calculated for each position. See Figure 5.

This calculated amino acid distribution was slightly modified to finalize the design of the H- CDR3 in the library:

^* Unification of the amino acids distribution to limit the set of different TRIM-mixes

^* Realization of amino acids with an appearance of >= 10%

^* Supplementation of other amino acids (except cysteine) by 19 amino acid-mix, which was added to each position

^* Complete avoidance of cysteine

^* Examination for and reduction of certain restriction sites to achieve modularity

^* Lower usage of tryptophan, which is to be used only in the 19aa-mix ^* Avoidance of similar amino acids at one position (e.g. Glu OR Asp)

^* Avoidance of glycosylation-sites (NxS, NxT)

After considering and implementing the required modifications the H-CDR3 design shown in Figure 6 was used.

The diversity of the H-CDR3 of the length of 12 amino acids is the product of the diversities at the single positions and is about 2.5 ^* 10 ¹³ . The overall diversity is the sum of the diversities of the distinct lengths, 7-12 amino acids, and is about 2.6 ^* 10 ¹³ .

Example 1 .3.2: Design of L-CDR3

The statistical analysis of the L-CDR3 was done for all re-arranged sequences irrespective of the germline family, since this CDR is partly encoded germline-independent by the J-segment. Based on analysis of mouse rearranged VL-sequences, it was found, that a L-CDR3-length of 8 amino acids occurs in about 85% of the sequences (see Table 14). Thus the length of the L-CDR3 in the library was set to a length of 8 amino acids.

Table 14:

Length #sequences %

7aa 173 9.6

8aa 1571 87.6

9aa 50 2.8

SUM 1794 100.0

The position dependent amino acid distribution of the 1571 re-arranged V-kappa sequences with the length of 8 amino acids was calculated. Results are shown in Figure 7.

Reason for the variability at framework position 1 00 is probably the usage of different JK- segments. This calculated amino acid distribution was slightly modified to finalize the design of the L- CDR3 in the library: ^* Unification of the amino acids distribution to limit the set of different TRIM-mixes

^* Realization of amino acids with an appearance of >= 10%

^* Supplementation of other amino acids (except cysteine) by 19 amino acid-mix, which was added to each position

^* Complete avoidance of cysteine

^* Examination for and reduction of certain restriction sites to achieve modularity

^* Lower usage of tryptophan, which is to be used only in the 19 amino acid-mix (especially positions 91 and 96)

^* Avoidance of similar amino acids at one position (e.g. Glu OR Asp)

^* Avoidance of glycosylation-sites (NxS, NxT, present at -5% in nat. a distribution)

^* Fix position 90 to 100% Q and position 95 to 100% P

After considering and implementing the required modifications the L-CDR3 design shown in Figure 8 was used.

The diversity of the L-CDR3 of the length of 8 amino acids is the product of the diversities at the single positions and is about 4.7 ^* 10 ⁷ , which is equal to the overall diversity.

Example 1 .3.3: Design of H-CDR2

It had been described that positions 52, 52a, 54, 56, 58 (Kabat-numbering) of H-CDR2 are most responsible for antigen-contact in H-CDR2 region. See (http://www.bioc.uzh.ch/antibodv/Structures/AgContact/index. html. Variation of these positions therefore should enable the selection of antibodies with antigen-tailored binding pocket comprising H- CDR2.

We investigated the amino acid distribution for distinct germ-line sequence families of rearranged antibodies represented by the three selected mVH master-genes (mVH1 _RV, mVH2 and mVH5). A list of residues which were found frequently at particular positions of H-CDR2, was compiled. In order to preserve prevalence of the germline sequence, the alternative residues present in the rearranged antibodies but not in germ-line sequences were intentionally underrepresented in the design. The variation of the positions was introduced into the selected master-genes by wobbles. See Figure 9.

H-CDR2 maturation cassettes were designed to improve the affinities of selected murine binders.

The position-dependent amino acid distribution of the H-CDR2 was investigated for distinct germ-line sequence families of rearranged antibodies represented by the three selected mVH master- genes (mVH1_RV, mVH2 and mVH5). Based on analysis of the rearranged mouse sequences, the amino acid distribution shown in Figure 10 was deduced by calculation of the position-dependent amino acid distribution.

This calculated amino acid distribution was slightly modified to finalize the design of the H- CDR2 in the library:

^* Unification of the amino acids distribution to limit the set of different TRIM-mixes

Realization of amino acids with an appearance of >= 10%

^* Supplementation of other amino acids (except cysteine) by 19 amino acid mix, which was added to some positions

^* Complete avoidance of cysteine

^* Avoidance of glycosylation-sites (NxS, NxT)

^* Examination for and reduction of certain restriction sites to achieve modularity

^* Enhance variability of VH2-HCDR2-cassette by addition of 19 amino acid mix at position 53 and 54.

^* Fix all position from position 62 onwards to an amino acid to provide a constant region for the

PCR-backward-primer

After considering and implementing the required modifications the H-CDR2 design shown in Figure 1 1 was used.

The diversity of the H-CDR2-cassettes is the product of the diversities at the single positions and is about 1 .1 ^* 10 ⁹ for mVH1 -H-CDR2, 9.9 ^* 10 ⁶ for mVH2-H-CDR2 and 3.6 ^* 1 0 ⁹ for mVH5-H-CDR2. Example 1 .4: Cloning, expression and purification of the murine library

Bacterial strains and vectors used for cloning

Molecular cloning was carried out using the E. coli strains XL1 -Blue (Stratagene) or Top10F+ (Invitrogen). Master genes were synthesized by GeneArt AG (Regensburg, Germany). The products from gene synthesis were cloned into a pGA-vector, and then re-cloned into tricistronic CysDisplay pMorph30-vector as described below. Phage display vector pMorph30 was derived from plasmid pMorph23 (Rothe et al., 2008, J Mol Biol 376, 1 182-1200) basically as follows: hexa-His-tag attached at the C-terminus of VH was replaced for the amino acid sequence HKHKHK, and a chi site in gi ll was eliminated. Restriction endonucleases were from New England Biolabs or Fermentas, and T4 DNA ligase was from Invitrogen. Reaction conditions were carried out according to the manufacturer's recommendations.

Accession numbers of original master-genes

IMGT Mus musculus sub-database- / NCBI-accession numbers of germ-line sequences which were used for construction of the master-genes:

mVH1 : IGHV1 -72 ^* 01 / AC163348, mVH2: IGHV2-2 ^* 01 / AC090887, mVH3: IGHV3-6 ^* 01 / AC073590, mVH5: IGHV5-9 ^* 04 / AF120472, mVH9: IGHV9-3 ^* 01 / AC073563, mVLB1 : IGKV1 -1 17 ^* 01 / D00081 , mVLB3: IGKV3-12 ^* 01 / K02159, mVLB7: IGKV7-33 ^* 01 / AF044198, mVLC3: IGKV3-4 ^* 01 / Y15968, mVLC8: IGKV8-28 ^* 01 / AJ235947, mVLC19: IGKV19-93 ^* 01 / AJ235935, J K2: IGKJ2 ^* 01 / V00777, CK: D IGKC ^* 01 / V00807, JH2: IGHJ2 ^* 01 / V00770, CH1 + hinge: IGHG1 ^* 02 / L35252

Considerations for the synthesis of master-genes

Genes corresponding to the selected representative germ-line sequences were optimized on DNA level by applying following criteria: (i) rare codon usage has to be optimal for E. coli while human/murine rare codons were avoided whenever possible; (ii) undesirable DNA RNA features such as internal TATA-boxes, chi-sites, ribosomal entry sites, AT-rich or GC-rich sequence stretches, sequences preventing mRNA-transport into cytoplasm (I NS or CRS elements), repeat sequences, RNA secondary structures, and cryptic splice donor and acceptor sites were avoided; (iii) introduction of restriction endonuclease (RE) sites conflicting with the HuCAL cloning-modularity were avoided; and (iv) restriction endonuclease sites required for the modular cloning were introduced. Glycosylation in variable domains is undesirable mainly because the carbohydrate-attachment is often found to affect binding of antigen due to steric hindrance, and is also responsible for heterogeneity of proteins expressed in eukaryotic hosts. Hence we ascertained that the designed VUVH-master-genes did not encode for protein with potential sites for N-linked glycosylation.

Constant domains in Fab format

The constant domain of the kappa light chain is encoded by the CK gene (IGKC ^* 01 / V00807, see above) . The N-terminal alanine residue was replaced by threonine which allowed the introduction of an BsiWI site at this position on the DNA level. A cysteine residue located in the C- terminus of wild-type CK was replaces by alanine.

The constant domain of the heavy chain in Fab format is encoded by CH1 gene-fragment. We have chosen gene IGHG1 ^* 02 because it represents consensus of CH1 -sequences of the murine lgG1 type deposited in the IMGT database. The only mismatch was a Glu-residue in a germ-line sequence-fragment TWPSE, and therefore the Glu-residue was changed to a Gin. Addition of one alanine residue in front of the N-terminal sequence KTTPPS allowed the introduction of a Blpl site. The first four residues of the hinge (sequence VPRD) followed by residues Glu-Phe (necessary for the introduction of an EcoRI site) were added at the C-terminus of the CH1 domain.

Gene assembly in tricistronic phage display plasmid pMorph31

The designed genes encoding for murine constant or variable domains were generated by PCR-assembly of a series of oligonucleotides, and cloned into a cloning pGA vector at GeneArt AG. It was convenient and faster to replace human VL, VH, Ck and CH1 genes for murine equivalents in two different interim vector molecules. Two derivatives of the pMorph30 plasmid (see above) denominated as "plasmid L" and "plasmid H", were constructed with large fragments of dummy DNA in order to facilitate cloning as follows: in "plasmid L", dummy DNA (about 3 kb) was inserted in place of light chain-coding region, whereas in "plasmid H", dummy DNA (about 2 kb) was inserted in place of heavy chain-coding region. Dummy DNA in "plasmid L" was then replaced for ITICK and mVL in two steps: ITICK gene (see above) was inserted by using BsiWI / Sphl , and afterwards mVL kappa master- genes (see above) were inserted by using EcoRV / BsiWI. Dummy DNA in "plasmid H" was replaced for mCH1 and mVH also in two steps: mCH 1 gene (see above) was inserted by using Styl (or Eco1 30l) / EcoRI, and then mVH master-genes (see above) were inserted by using Mfel / Styl. DNA fragments encoding master-genes mVH 1 , mVHI RV, mVH2, mVH5 or mVH9 (all linked to mCH 1 - region) were combined (cross-cloned) by using Sphl / Hindi 11 with master-genes mVLB1 , mVLB3, mVLB7, mVLC3, mVLC8, and mVLC19 in pMorph30/mCK plasmid resulting in 36 mVL-mVH combinations. After the cross-cloning, the display plasmid was denominated pMorph31 indicating the fact that the plasmid encodes besides of gpl l l also murine variable and constant domains, and is hence suitable for display of murine Fabs.

Re-cloning of VL-VH master gene combinations into bicistronic pMorphx9_Fab_FH vector for expression in E. coli

In order to analyze soluble expression of VL-VH framework combinations in Fab format, we have re-cloned by using Xbal and EcoRI all 36 mVL-mVH master-gene combinations (as Fab- encoding inserts) into pMorphx9_Fab_FH vector (Rauchenberger et al. , 2003, J. Biol. Chem. 278, 38194-38205) where FH indicates that the Flag- and hexa-His-tags were attached C-terminally to the Fab.

Conversion of VL-VH master gene combinations to murine lgG2a format

In order to analyze expression of master-gene combinations in full-length murine lgG2a kappa format in mammalian cells, we have subcloned the 36 mVL-mVH master gene combinations into pMorph4_h/m_lgG vector which is a derivative of pMorph2_h_lgG vector (SteidI et al. 2008. Mol Immunol 46(1 ):1 35-44). In the pMorph4_h/m_lgG vector, the light and heavy chains were encoded in one plasmid; however, each chain was translated together with its N-terminal leader sequence from a separate transcript (transcription was driven by two CMV-promoter sites located in the plasmid). Conversion of VL-VH master gene combinations to murine lgG1 format

In order to analyze expression of master-gene combinations in full-length murine lgG1 kappa format in mammalian cells, we subcloned the 36 mVL-mVH master gene combinations into pMorph4_h/m_lgG vector, which is a derivative of pMorph2_h_lgG vector (SteidI et al. 2008. Mol Immunol 46(1 ):135-44). In the pMorph4_h/m_lgG vector, the light and heavy chains are encoded on one plasmid; however, each chain is translated together with its N-terminal leader sequence from a separate transcript (transcription was driven by two CMV-promoter sites located in the plasmid).

Expression and purification of Fab

Expression in E. coli ^ - cells from the pMx9_Fab_FH plasmid was carried out in 500 ml of 2xYT-medium supplemented with chloramphenicol (34 μg/ml) and glucose (0.1 %). After induction with IPTG (0.75 mM), the cells were grown at 30°C for 20 h. Cell pellets were lysed during incubation with lysis buffer (25 mM Tris/HCI, , pH 7.4, 0.5 M NaCI, 10 mM imidazol, 0.2 % lysozyme, 2 mM MgCI ₂ , 20 U/ml benzonase and EDTA-free protease inhibitor cocktail from Roche) for 30 min at RT. Fab-purification from cleared cell lyzates (centrifugation and filtration, 0.2μηι) was carried out by 2- step purification using Akta xpress for automated processing. First step was IMAC chromatography with Ni ²⁺ ions, where in case of stability evaluation an additional wash step with 30% isopropanol in PBS was carried out. After elution with high concentration of imidazol the eluted peak was reloaded on a gel filtration column, and proteins were separated by size exclusion using PBS pH 7.4. Finally, elution fractions which correspond to expected size were pooled.

Expression and purification of IgG

Eukaryotic HKB1 1 (Cho et al. ; J Biomed Science (2002) 9, 631 -8) cells were transiently transfected with the pMorph4_h/m_lgG plasmid encoding the heavy and light chains. Cell culture supernatant was harvested from 3 to 7 days post transfection. After adjusting the pH of the supernatant to 8.0, addition of NaCI (2 M final concentration) and sterile filtration (0.45 μπ ), the solution was subjected to standard protein A affinity chromatography (MabSelect SURE, GE Healthcare). If not stated otherwise, buffer exchange was performed to 1 x Dulbcecco ^' s PBS (pH 7.2, Invitrogen) and samples were sterile filtered (0.2 μιη). Purity of IgG was analysed under denaturing conditions by SDS-PAGE or by using Agilent BioAnalyzer and in native state by HP-SEC.

Phage preparation

Phage amplification and purification was carried out essentially as described previously (Krebs et al., 2001 , J Immunol Methods 254, 67-84) with a following modification. Phagemids (VL-VH framework combinations encoded in display plasmid) were propagated in E. coll Top10F+ cells in 2xYT/Cam/Tet/glucose (1 %) medium. After helper phage infection (hyperphage from Progen, Cat. No.: PRHYPE, multiplicity of infection: 40), centrifugation and resuspension of the cell pellet in 2xYT/Cam/Kan/Tet/I PTG (0.25 mM), phages were produced during 20 h-incubation at 22 °C.

Phage ELISA

Black Maxisorp microtiter plates (Nunc) were coated with anti-pi ll antibody (MoBiTec, Cat. No.: PSKAN3) for determination of relative pi ll-level ( ^« phage concentration), or with anti mouse F(ab ^' ) ₂ antibody (Jackson ImmunoResearch Lab.; Cat. No. : 1 12-006-072) for determination of relative Fab- level in phage samples. Although the exact epitope of the anti mouse F(ab ^' ) ₂ antibody is not known (it is presumed to be located in the constant domain), we reasoned that binding of the antibody to murine Fab is not likely to be strongly affected by sequence-variation in Fv-domain, and thus the antibody should be suitable for general assessment of murine Fab-display on tip of filamentous phage. TBS pH 7.4 containing Tween 20 (0.05%) and chemi-blocker (from Chemicon; diluted 1 :2) was used for blocking of non-specific binding sites in the microtiter plate and on phages. After incubation with phages, wells were washed with TBS-T, and incubated sequentially with biotinylated anti-bacteriophage antibody (Sigma Cat. No. B2661 ) and avidin-HRP (BD Pharmingen; Cat. No.: 554058). "Quanta Blu" substrate for HRP was applied, and fluorescence signals were recorded on Tecan-Spectrafluor instrument (excitation: 320 nm, emission: 430 nm). The relative pi ll- and Fab- levels in phage sample were determined by using reference phage (displaying a previously selected binder converted into Fab with murine constant domains) and the standard 2-state model applied for non-linear regression. The relative display rate was defined as a ratio of the Fab- and pi l l-levels.

Example 1 .5: Characterization of the murine antibody library by way of exemplary antibody 4D5 Good presentation on phages and high expression in cells can be achieved only for well folded and stable Fab-molecules. Thermodynamic stability of Fab-molecule depends on intrinsic stability of the individual domains (heavy and light chains) as well as on non-covalent interaction between the two domains (Ewert et al., 2003, J Mol Biol 325, 531 -553). Tight interaction between Fd (heavy chain) and CK (light chain) are of particular importance because in the CysDisplay ^® system the two domains cannot be covalently linked via disulfide bond since cysteine residue in the C-termini of constant domains was eliminated. Therefore, instead of testing VL- and VH-frameworks individually, we analysed VL-VH combinations which could be efficiently folded and expressed in E. coli (Fab format) and in mammalian cells (IgG format).

As a test case we used the HCDR3 sequence of antibody 4D5 (WGGDGFYAMDY). This sequence had already been used previously for the assessment of human master-genes prior the construction of the original HuCAL library (Knappik et al., 2000, J Mol Biol 296, 57-86)

Analysis of display efficiency

Efficient display of library members on phage particles is a prerequisite for successful phage display selections. We used sandwich phage ELISA for quantification of display of the 36 murine VL- VH combinations in Fab-format. Our data indicate that most of the tested mVL-mVH combinations, with the exception of all mVLB7-combinations, were displayed on phages. There was no significant difference in display between Fab-molecules containing mVH1 - or mVH1_RV. Although statistically insignificant, there is a trend indicating that the VH5-combinations are slightly better displayed than all other murine Fabs Results are depicted in Figure 12.

Soluble Fab expression

Next, we investigated whether the favorable folding and expression characteristics of tested murine Fab in bacterial host can also be observed also in soluble format. In this format both heavy and light chains are directed into the periplasm of E. coli where they achieve their final fold and associate with each other to form a stable Fab-fragment. As shown in Figure 13 all tested murine VL-VH combinations, with the exception of all mVLB7- combinations, were well expressed in E. coli and the yield of purified Fab exceeded 5 mg/L bacterial culture. The VH5-combinations (particularly, VLC3-, VLC8-, VLC19-VH5) seem to be slightly better expressed than all other murine Fabs. There was no significant difference in expression between Fab-molecules containing mVH1 - or mVH1_RV. Fab-expression in soluble- and and in display-forms correlates remarkably well (see Figure 14). Furthermore, characterization of isolated material revealed that 98-100% of purified Fab was at monomeric form. This indicates usefulness for the development of murine surrogate antibodies.

Soluble IgG-expression

The vast majority of antibody-based therapeutic biologicals currently on the market are in IgG- format for a variety of reasons: (i) the half-life of IgG molecules in the human body is very high (about 3 weeks) due to the interaction of the IgG with the neonatal receptor (FcRn); (ii) IgG molecules are highly soluble, thermodynamically stable and relatively resistant to proteases in blood; and (iii) IgG possess ADCC (antibody-dependent cell-mediated cytotoxicity) and/or CDC (complement-dependent cytotoxicity) activity, which are required for elimination of tumor cells.

Since the expression of a particular VL/VH-combination in Fab-format not necessarily correlates with the expression of the same VL/VH-combination in IgG-format, we opted to characterize the expression of all mutual combinations of synthesized murine VL- and VH- mastergenes also in IgG format.

Two distinct mouse IgG formats, lgG1 and lgG2a, were chosen for the expression test. The murine lgG2a isotype was found in about 27% of all monoclonal IgG isolated from Balb/c mouse (Rousseaux and Bazin, 1979, Vet Immunol Immunopathol. 1, 61 -78), and we tested all VL-VH master-gene combinations in this format. In addition, few selected VL/VH-combinations were tested also in murine lgG1 format which was found in about 40% of all monoclonal IgG isolated from Balb/c mouse (Rousseaux and Bazin, 1979, Vet Immunol Immunopathol. 1, 61 -78).

All murine VL- and VH-mastergenes were re-cloned into mammalian expression vectors, and the resulting plasmids were used for transient transfection of a mammalian suspension cell line as described herein above. Data are shown in Figures 15 and 16, which indicate that the expression level of almost all mVL-mVH combinations tested, in both murine lgG1 and lgG2a formats, largely exceed a value of 10 mg/L. Similar to the observations made for the Fab format, there was no difference in expression between IgG-molecules containing mVH1 - or mVH1_RV. Very low expression level of mVLB7-combinations parallels the very low prokaryotic expression of the mVLB7- combinations observed in the Fab format. Altogether, against the odds, we were able to soluble express almost all of the mVL-mVH combinations tested.

All murine VL and VH-mastergene combinations were purified in a 2-step purification process using Akta express. Expressed Fab-fragments are bound to chelated Ni2+ and washed either with 15 column volumes of 20 mM Na-Phosphatbuffer/500 mM NaCI, pH 7.4, 20 mM Imidazole or, in a separate purification, using 20 column volumes of 30% lsopropanol/20mM Na-Phosphatbuffer pH 7.4. Second step of purification (SEC) was identical for both procedures (with or without isopropanol treatment) and is decribed herein above. After purification both protein preparations were compared regarding yield, concentration, purity, ratio of VH and VL and thermal stability. A wash step with 30% isopropanol while the Fab is still bound to the IMAC column was introduced in order to receive Fab fragments that are of higher purity. However for some Fab fragments this treatment was seen to be too harsh because the interface - the hydrophobic interaction between heavy and light chain - was not strong enough to endure this procedure. The heavy and light chains of such Fab fragments were eluted at different stages and therefore the ratio between the two chains did deviate from 1 :1 .

Thermal stability

Protein expression level in cellular host depends on many parameters. Some of them are related to intrinsic protein folding; however, many are dependent on the metabolic status of the host, such as activity of translational and translocation machineries, or levels of secreted proteases and ubiquitin. These latter protein-extrinsic properties are difficult to control, and vary often substantially in experiments performed in different days. Due to the variations observed in the expression tests described above we characterized the intrinsic protein properties by using protein samples purified during the expression tests. Practically, measurement of thermal stability can be performed in a small volume in a standard thermo cycler for real-time PCR, and is therefore suitable for relative comparison of intrinsic stability among a large group of tested proteins. On the other hand, thermal denaturation of proteins, in contrary to chemical denaturation by chaotropic agents, is typically irreversible, and cannot be used directly for determination of absolute thermodynamical parameters, such as AG of the protein transition between its two conformational states.

Results are shown in Figure 18. The most thermally stable murine VL-VH combinations were are the combinations VLB3-VH1 RV and VLB3-VH1 RV. Of importance, even the thermal stability of the tested mVLB7-combinations was reasonably high. This was unexpected considering the comparably low expression levels detected for all mVLB7-combinations in Fab and IgG-formats.

Selection of Fv-frameworks for library construction

During the assessment of the 36 different frameworks (combinations of 6 VL- with 6 VH master-genes) described above, we observed that basically all Fv-frameworks were expressed in Fab-format E. col '/ ^' and in IgG-format in mammalian cells with appreciable yields exceeding usually 10 mg/L. The only exceptions were all mVLB7-combinations and several of the tested mVLC19- combinations (see above). In our selection of the most favorable Fv-frameworks we strongly emphasized thermodynamical stability of Fv-molecules. In this regard, the mVLB3-mVH1 RV and mVLC3-mVH1_RV with mid-point of thermal unfolding (Tm) at about 76°C were clear choice (see Figure 18). Furthermore, mVH1 , mVLB3 and mVLC3 germ-line sequences are frequently found in the IMGT database (see Table 15). Two other selected Fv-frameworks comprised mVLB3 and mVLC3 in combination with mVH5. mVH5 is highly used in murine heavy chains (Table 15), mVH5- combinations were well expressed in Fab format in E. coli (see Figures 12 and 13), and mVLB3- mVH5 and mVLC3-mVH5 sequences were found in the Kabat database of re-arranged murine antibodies (Table 16).

Another favorable Fv-framework comprising distinct VH was mVLB1 -mVH2 due to relatively high abundance of this Fv-framework combination in the Kabat database (Table 16), and due to high usage of mVLBl in murine light chains (Table 15). Diversified LCDR3- and HCDR3-library cassettes were therefore subcloned into the selected Fv-frameworks resulting in five murine sublibraries (see below). Table 15 indicates the usage of VK and VH germline families for the mouse strains BALB/c and C57BL-6.

Table 15:

Table 16 shows the result of the statistical analysis of the combined VL-VH sequence representation in the Kabat database.

Table 16:

Statistical analysis of∞mbined VL-VH sequence representation in the Kabat database

To determine whether those VH-VL combinations that resulted in well expressing antibodies in the library of the present invention also exist in naturally selected antibodies, the representation of those antibodies was analysed in the Kabat database. The combined VH-VL framework sequences of the antibodies where blasted against all frameworks of the complete Kabat antibody sequences, comprising a sequence for both VH and VL.

In the Kabat database 1501 complete mouse antibody entries with complete sequence information (heavy and light chain) are present. To cover all possible heavy and light chain combinations of the BALB/c mouse strain, 101 (heavy chains) x 23 (light chains) = 2323 complete antibodies would be needed in the sample set. For C57BL6 the number would be even higher: 195 (heavy chains) x 74 (light chains) = 14430 complete antibodies. Taking into account the small sample set and the fact, that in the Kabat database these two mouse strains are not the only ones which are represented it is rather unlikely, that one of the VH-VL chain combinations of the mouse library of the present invention exists in this dataset. Any number of matches larger than one was therefore taken as an indication, that this combination might in fact be a naturally preferred one. The best represented chain combinations where chosen for final library design.

Example 1 .6: Diversification of the murine library

Introduction of DNA-stuffers for insertion of diversified CDR3 -cassettes into the display plasmid pMorph31

DNA regions encoding for L- or H-CDR3 are short (<40 bp). In our modular system, both original CDR3 encoded in the master-gene and diversified CDR3-cassette are flanked by a pair of unique restriction sites, which allow for the introduction of diversified CDR3-cassette in place of the original CDR3. Inefficient plasmid digest at these restriction sites can result in a substantial background comprising vector with original CDR3 because it is not technically feasible to efficiently separate by gel electrophoresis linearized plasmids (-5500 bp) in which CDR-encoding region is present or absent. In order to prevent the vector background regardless of the digest-efficacy, we decided to introduce a DNA-stuffer (fragments of dummy DNA of -500 bp in size) in place of CDR3- encoding region. DNA stuffers were introduced into the plasmid pMorph31 at the CDR3-flanking RE- sites resulting in plasmids pMorph31_mVLy/MBP-mVH//AP. In particular, AP-stuffer derived from alkaline phosphatase gene was inserted in place of H-CDR3 by using BssHII / Styl, and MBP-stuffer derived from maltose binding protein gene was inserted in place of L-CDR3 by using Bpil / BsiWI. Noteworthy, due to the digest mode of Bpil (type II RE), we had to prepare two variants of the MBP- stuffer (one for mVLB1 and another one for mVLB3/mVLC3) which differ in the proximity of Bpil-site depending on the master-gene sequence.

Construction of HCDR2 mini-library cassette by using primers with wobbled positions

The intended variability (five amino acid sequence variants per master-gene) was introduced into the selected master-genes in 10 cycles of PCR with Platinum Pfx DNA polymerase (Invitrogen), and a set of forward and reverse primers. Forward primers containing wobbled positions with defined nucleotide ratios were used as follows:

GGTCGCGGCCTCGAGTGGATTGGCCGCATTG(70%)/A(15%)/T(15%)ATCCGAA CAG(80%)/A(20 %) CGG CG G C ACC AAATATAACG AA for mVH1_RV,

GGCAAAGGTCTCGAGTGGCTGGGCGTGATTTGGA(80°/o)/G(20%)GCGG(70%)/A (30%)CGGCA(7 0%)/G(30%)GCACCGATTATAACGCCGCC for mVH2, and

GAAAAACGCCTCGAGTGGGTGGCGACCATTAG(70%)/A(30%)CGGTGGCA(70%)/G( 30%)GTGGC AA(60%)/G(40%)CACCTATTATCCAGATAGC for mVH5.

The H-CDR2-regions were completed by constant flanking regions, representing a part of antibody framework 2 and framework 3. Thus it was possible to PCR-amplify and clone the cassettes at a later stage. Since the flanking regions of the VH-mastergenes are different, separate oligos had to be synthesised for the different mastergenes:

Sequence analysis of about 30 randomly picked clones per framework indicated that the desired variants (five per framework) were represented in the HCDR2 with a frequency similar to our design. Only about 10% of randomly picked clones carried unintended nucleotide deletions mainly in the positions which were wobbled in the PCR-primers. In order to ensure low abundance of frame- shifted clones in the library we decided to dilute the plasmid pool encoding variations in HCDR2 with plasmid encoding wild-type VH-framework. Hence in the final library the minor variants in HCDR2 are about 10-fold less abundant than in our initially HCDR2-design.

After this quality control, the PCR products were digested with Xhol / BssHII, and cloned into the respective pMorph31 _mVLy7MBP-mVH/7AP plasmids (see the paragraphs on the introduction of the DNA stuffers above). Plasmid pool containing the introduced variety in mastergenes was mixed 1 :10 with plasmid encoding the original master-gene, and used for introduction of the LCDR3-library cassette.

Synthesis of trinucleotide-containing oligonucleotides and CDR3 library cassette preparation

Trinucleotide-containing oligonucleotides were synthesized as described previously (Knappik et al., 2000). Diversified trinucleotide-containing ("TRIM") oligonucleotide was mixed with "overlapping" oligonucleotide corresponding to the complementary strand overlapping with the trinucleotide-containing oligonucleotide in nondiversified positions. Both oligonucleotides (at concentrations 0.1 -0.2 μΜ) were assembled and amplified in 10 PCR cycles by using end primers (at concentration 1 μΜ) and Platinum Pfx DNA polymerase (Invitrogen). The forward and reverse end primers annealed to the 5' and 3' ends of the CDR cassette, respectively, and facilitated introduction of restriction sites necessary for cassette cloning. Sequences of all oligonucleotides used for preparation of CDR3 cassettes are listed in Tables 17 (L-CDR3) and 28 (H-CDR3). The amplified CDR3 cassettes were purified on agarose gels and isolated as described elsewhere (Rothe et al., 2008, J Mol Biol 376, 1 182-1200). Noteworthy, four LCDR3 cassettes (one specific for mVKB3 and mVKC3, others for mN/κΒΙ , VKB7 and VKC8) had to be synthesized for each selected VL-framework differing in the proximity of the Bpil site (type I I RE) which was necessary for the LCDR3-subcloning.

Table 17:

mVkC8 Lib66 GTGCAGGCCGAAGACCTGGCCGTGTATTACTGC

Overlapping oligos

mouse Lib71 CGGCCACGTACGTTTCAGTTCCAGTTTGGTGCCCGCACCAAAGG

Vk rev primer

mouse Lib69 CGGCCACGTACGTTTCAGTTC

Table 18:

Sub-cloning of CDR3 library cassettes into master-genes in phagemid format

The amplified LCDR3-library cassettes were inserted in place of the MBP DNA stuffer in pMorph31_mVL/-mVH//AP plasmids with diversified HCDR2 by using Bpil / BsiWI. We used typically ~2 μg vector for ligation with 2-molar excess of the LCDR3-cassette. The vector-insert mixture was pre-incubated 10 minutes at 56 °C prior addition of T4 DNA ligase and buffer, and after the overnight incubation at 16°C, the ligation mixture was incubated 10 minutes at 65 °C in order to inactivate the ligase. Ligated DNA was then purified by 2-butanol/glycogen precipitation, and electroporated into Topl OF' competent cells (Invitrogen). After transformation and outgrowth for 1 h at 37°C in SOB medium, the number of transformed cells was determined by titration on LB/cam/glucose (1 %) agar plates. Amplification of the transformed clones was performed in liquid medium (LB/cam/glucose (1 %)) overnight at 22°C and then at 30°C until OD ₆₀ onm 2.0 was reached. An aliquot of the liquid culture was used for glycerol frozen stock, and the rest was used for DNA maxiprep by using Qiagen kit.

The highly variable L-CDR3-cassette was completed by constant flanking regions, representing a part of antibody framework 3 and the terminal VL-antibody framework 4. Thus it was possible to PCR-amplify and clone the cassette at a later stage. Since the flanking regions of the VL- mastergenes were different, separate oligos had to be synthesised for the different mastergenes: mVkB3=mVkC3

5 ^' GAAGACGCCGCCACCTATTATTGC-T8-C/\G- 79-770-77 1-T12-CCG-T13-ACCTTTGGCGG TGGCACCA mVkB1

5 ^' GAAGACCTGGGCGTGTATTATTGC- T8-CAG- T9-T10-T1 1-T12-CCG-T13-ACCTTTGGCGG TGGCACCA mVkB7

5 ^' GAAGACCTGACCCATTATTACTGC- T8-CAG-T9-T10-T1 1 -T12-CCG-T13-ACCTTTGGCGG TGGCACCA mVkC8

5 ^' GAAGACCTGGCCGTGTATTACTGC- TS-C/IG- 79-7 0-77 7-772-CCG-773-ACCTTTGGCGG TGGCACCA

wherein T8 - T13 are TRIM-mixes.

For H-CDR3 construction, a separate CDR cassette was generated for each H-CDR3 length. The composition of the separate cassettes is shown in Figure 19.

The highly variable H-CDR3-cassette was completed by constant flanking regions, representing a part of antibody framework 3 and the terminal VH-antibody framework 4. Thus it was possible to PCR-amplify and clone the cassettes at a later stage. Since the flanking frameworks of the different VH-mastergenes are identical, one oligo could be used for all mastergenes:

5 ^' GTGGAAGACTGCGCGCGG-T1 -T1 -(T2)-(T3)-(T3)-(T3)-(T3)-T4-T5-T6-W- T7-TGGGGCCAAGGCGTCT, wherein T1 - T7 are TRIM-mixes and W are wobble as described herein above.

Amplified diversified HCDR3 library cassettes of various lengths (7, 8, 9, 1 0, 1 1 , and 12 residues) were mixed together at molar ratios reflecting the planned HCDR3 length distribution, and subcloned into the pMorph31_mVLy-mVH/7AP plasmids with diversified HCDR2 and LCDR3 by using BssHII / Styl. Typically, 15-20 μg vector were ligated with 2-molar excess of the HCDR3-cassette mixture. The ligation-, transformation-, and clone propagation- conditions used for subcloning of the HCDR3 cassette-mixture did not vary from the conditions used for subcloning of the LCDR3 cassettes. VL- and VH-sequence analysis of -70 clones and restriction analysis of 24 randomly picked clones and of pool DNA-maxiprep from each sublibrary were performed in order to assess the library quality.

The theoretical diversity of 2.4- 10 ⁸ individual clones per sublibrary was readily covered after the transformation of plasmid pool into E. coli (Table 19).

Table 19:

We obtained about 2.7- 10 ⁹ individual clones per sublibrary, or 1 .3- 10 ¹⁰ individual clones for the entire HuCAL MOUSE library comprising all five diversified murine Fv-frameworks (Figures 19-24). The achieved diversity of the HuCAL Mouse antibody library is 2 orders of magnitude larger than B cell population of an immunologically mature mouse (Rajewsky et al., 1987, Science 238, 1088- 1094), and the quality of the library was very high - about 87% of analyzed sequences were in-frame and did not contain undesirable amino acid mutations (see Table 19). Sequence analysis of randomly picked clones indicated that amino acid distribution in LCDR3 and HCDR3 did not vary substantially from our design (Figures 20-25), and also the actual HCDR3 length-distribution was well in accordance with the design (Figure 26).

Sub-cloning of LCDR3 library cassette into auxiliary cloning vector for affinity maturation

The amplified LCDR3-library cassettes were inserted in place of a DNA stuffer in pGA1 8 cloning vector by using Bpi l / BsiWI. This vector does not possess any promoter and replicates in bacteria with a high copy number. The ligation-, transformation-, and clone propagation- conditions used for subcloning of the LCDR3 cassettes into the plasmid pMorph31 _mVLy-mVH/7AP or into the cloning vector pGA18 were basically the same.

The diversity of the LCDR3 maturation cassettes is shown in Table 20

Table 20:

Amplification of HCDR2 maturation cassettes

The insert comprising HCDR2 maturation-cassettes for mVH was prepared in two steps. In the first step, diversified trinucleotide-containing ("TRI M") oligonucleotide was mixed with "overlapping" oligonucleotide corresponding to the complementary strand overlapping with the trinucleotide- containing oligonucleotide in nondiversified positions. Both oligonucleotides (at concentrations 0.1 -0.2 μΜ) were assembled and amplified in 10 PCR cycles by using forward and reverse "mat" primers (at concentration 1 μΜ) and Platinum Pfx DNA polymerase (Invitrogen) resulting in PCR product A/. In parallel, the framework 3 of mVH/ ^' was amplified from pMorph31_mVLy7MBP-mVH/7AP template (see above) in 10 PCR cycles by using forward and reverse "FW3" primers (at concentration 1 μΜ) resulting in PCR product B/ ^' . In a next step, the corresponding PCR products A/ and B/ ^' (each at concentration -20 nM) were assembled and amplified in 10 PCR cycles by using corresponding forward "mat" and reverse "FW3" primers resulting in PCR products encoding diversified HCDR2/FW3/ ^' and flanked by Xhol and BssHII RE-sites and 5' and 3', respectively. Sequences of all oligonucleotides used for preparation of maturation HCDR2 cassette are listed in Table 21 .

Table 21 :

ratl_ib90 rat VH1_HCDR2mat rev CCACGGTAAAGGTCGCTTTG

ratLib91 rat VH1 FW3 for CAAAGCGACCTTTACCGTGG

ratl_ib92 rat VH1 FW3-AP rev CGACGCCCAGCGCGCAGTAATACACC ratLib93 rat VH5_HCDR2mat for ACCAAAGGTCTCGAGTGG

ratLib94 rat VH5 HCDR2mat_overlap CACGGCTAATGGTAAAACGGCCTTTCACGCT ratLib95 rat VH5_HCDR2mat rev CACGGCTAATGGTAAAACGG

ratLib96 rat VH5 FW3 for CCGTTTTACCATTAGCCGTG

ratl_ib97 rat VH5 FW3-AP rev CGACGCCCAGCGCGCAATAATAGGTC

The H-CDR2-cassettes were completed by constant flanking regions, representing a part of antibody framework 2 and framework 3. Thus it was possible to PCR-amplify and clone the cassettes at a later stage. Since the flanking regions of the VH-mastergenes are different, separate oligos had to be synthesised for the different mastergenes. mVH1_RV_HCDR2mat

5' GGTCGCGGCCTCGAGTGGATTGGC-MMT1 -ATT-MMT2-CCG-MMT1 -MMT3-MMT4-MM T1 -MMT5-MMT1 -TAT-AAC-W1 -AAATTCAAAAGCCGGGTGACC

mVH2_HCDR2mat

5' GGCAAAGGTCTCGAGTGGCTGGGC-W2-ATT-MMT6-MMT7-MMT8-GGC-MMT9-ACC- MMT1 -TAT-AAC-MMT10-GCCTTTATTAGCCGCCTGAGCATC

mVH5_HCDR2mat

5' GAAAAACGCCTCGAGTGGGTGGCG-MMT1 -ATT-MMT1 1 -MMT12-MMT13-W3-MMT12- MMT14-MMT15-TATTATW4GATAGCGTGAAAGGCCGCTTTACC,

wherein MMT1 - MMT15 are TRIM-mixes and W are wobbles as described herein above

The diversity of the HCDR2 maturation cassette is shown in Table 22:

Table 22: HCDR2 maturation cassette

theoretical percentage of

framework library size

diversity correct clones [%]

VH1 RV 1 .1 E+09 4.0E+09 03 ⁿ o

Mouse VH2 9.9E+06 2.3E+09 86%

VH5 3.6E+09 6.0E+09 4°

Rat VH5 1 .2E+08 4.0E+09 88°o

Sub-cloning of HCDR2 maturation cassettes into auxiliary cloning vector

First, we sub-cloned the three selected murine VH-master-genes mVH1 RV, mVH2 or mVH5 (fragments encoding residues 1 - 94, i.e. ORF starting from the N-terminus of FW1 till very C- terminus of FW3) from pMorph31 phage-display vector (see above) into an auxiliary cloning vector pGA1 8 by using Mfel / BssHI I . In next step, the master-gene region between Xho I and BssH I I RE- sites was replaced by a DNA stuffer (about 600 bp). We inserted the DNA stuffer in order to facilitate sub-cloning of the amplified HCDR2 maturation diversified cassette into the cloning vectors pGA1 8_mVH; by using Xhol / BssH I I in a subsequent step. Then, we sub-cloned the amplified diversified HCDR2 (see the chapter above) into the cloning vector pGA18_mVH/ ^' in place of the DNA stuffer by using Xhol / BssH II . Typically, 1 5-20 μg vector were ligated with 2-molar excess of the HCDR2-cassette. The ligation-, transformation-, and clone propagation- conditions used for subcloning of the HCDR2 cassette did not vary from the conditions used for subcloning of the LCDR3 cassettes.

Summary

The mouse antibody library generated in the presented study comprises diversified LCDR3 and HCDR3 regions. CD R3 library cassettes were generated from mixed trinucleotides (TRIM) which encode only desirable residues in the CD R3-cassettes preventing introduction of cysteine- and stop- codons. Since it has been recognized that antigen-antibody interaction involves also framework residues and therefore it might be beneficial to screen for binders from a conformation-space broader than one offered by a variety of distinct sequences in CDRs only, we selected five different murine Fv-frameworks. Hence five distinct libraries with diversified LCDR3 and HCDR3 were generated in parallel. Sequencing of unselected clones indicated that overall incidence of undesirable DNA mutations, deletions and insertions introduced by PCR amplification of the LCDR3 and HCDR3 cassettes was low (-10%). In order to comply with the essential requirements of the HuCAL ^® cloning- modularity, following mutations were introduced into germ-line sequences: V2I in mVLB1 , T25S in mVH3, K46E in mVH9 and K66R/A67V in mVH1_RV. The amino acid sequence corresponding to positions 1 - 4 in all mVHs was changed into EVQL.

An overview of the all synthesized master genes encoding variable domain is presented in Figures 27-31 . Kabat indicates the position of amino acid residue in VL- or VH-sequences. mVL denotes murine variable domain in the light chain (exclusively of the type kappa), and mVH denotes murine variable domain in the heavy chain. Murine VK germ-line genes encode mVL starting from position 1 (at the N-terminus of framework 1 ) to 95 (at the C-terminus of L-CDR3), and murine JK germ-line genes encode for the very C-terminal residue in L-CDR3 (constant position 96) and whole framework 4 in mVL (positions 97 - 1 08). We have chosen murine J K2, since Tyr is the most frequent residue in position 96 of rearranged antibodies found in the IMGT database.

Murine VH germ-line genes encode mVH starting from position 1 (at the N-terminus of framework 1 ) to 94 (at the very C-terminus of framework 3). HCDR3 sequence is encoded by the D germ-line genes. We reasoned that biochemical and biophysical properties of the VH-domain depend also on its HCDR3 sequence, and therefore we decided to introduce one particular HCDR3 sequence into all mVH master-genes for the purpose of the master-gene comparison. The HCDR3 sequence (WGGDGFYAMDY) was derived from antibody 4D5, and had been used previously for the assessment of human master-genes prior the construction of the original HuCAL library (Knappik et al., 2000, J Mol Biol 296, 57-86). Murine JH germ-line genes encode for a C-terminal part of HCDR3 and whole framework 4 (positions 1 03-1 13). We have chosen murine JH2 (sequence WGQGTTLTVSS) for the design of framework 4 because JH2 was frequently found in rearranged antibodies deposited in the IMGT database, and allowed us to introduce two restriction endonuclease sites (Styl and Blpl) indispensable for the HuCAL cloning modularity.

Example 1 .7: Test panning with the murine library Murine M-CSF (macrophage colony stimulation factor) was used as antigen for test panning. Here we handled each sub-library separatly and in addition used a pooled full library consisting of all sub-libraries.

Panning mM-CSF was diluted in PBS to coat with the standard concentration of 50 μg/ml. For each sublibrary 2 wells of a microtiter plate were coated with antigen using 300 ul antigen solution per well. The plate was stored O/N at 4°C. Then the antigen solution was removed from the coated wells by rapidly inverting the plate over a plastic tray. The coated wells were washed twice with 400 μΙ PBS and blocked with for 2 h at RT on a microtiter plate shaker.

Meanwhile the phage blocking mixtures were incubated in 2 ml reaction tubes for 2 h at RT shaking gently. After the blocking procedure the wells were washed 2x with 400 μΙ PBS and the 300 μΙ of the pre-blocked phage mix transferred into each blocked well. It was incubated for 2 h at RT on a microtiter plate shaker. After that the phage solution from the antigen coated wells were removed by rapidly inverting the plate over a plastic tray and plates were washed with the following washing conditions (Table 23):

Table 23:

1 st round 2nd round 3rd round

3x PBST quick 1 x PBST quick 10x PBST quick

2x PBST for 5 min 4x PBST for 5 min 5x PBST for 5 min

3x PBS quick 1 x PBS quick 10x PBS quick

2x PBS for 5 min 4x PBS for 5 min 5x PBS for 5 min

All washing steps were done at RT. After the washing steps all traces of the wash solution were removed by carefully tapping the microtiter plate on a new stack of paper towels.

For the elution of specifically bound phage, we added 300 μΙ 20 mM DTT in 10 mM Tris/HCI, pH8.0, to each selection well and incubated at RT for 10 min without shaking. The DTT eluates of each selection were collected. E. coli TG1 with an OD600nm of 0.6-0.8 was added to the phage eluates of each selection and were incubated in an incubator without shaking. After infection bacteria were plated out evenly on two large LB/Cm/Glu agar plates for each selection and incubated O/N at 37°C.

For the following panning rounds bacterial suspensions of each pool were collected and used to propagate phages for an additional panning round as described above.

After each round of panning against mM-CSF the phage titer was determined. The expected range goes from 1 x10 ¹⁰ -1 x10 ¹² phage/mL for the input and 10 ⁴ -10 ⁶ phage/mL for the output. Table 24 shows the input and the output after each round of panning and all values are in the expected range.

Table 24:

1st round 2nd round 3rd round

Pool INPUT OUTPUT INPUT OUTPUT INPUT OUTPUT

VH1 VkB3 1x10 ¹² 2.6x 10 ⁵ 3.4 x 10 ¹⁰ 8x 10 ⁴ 2 x 10 ¹⁰ 2.6 x 10 ⁶

VH2 VkB1 1x10 ¹² 8 x 10 ⁴ 4.6 x 10 ⁹ 4.1 x 10 ⁴ 3.6 x 1 0 ⁹ 3.2 x 10 ⁶

VH5 VkB3 1x10 ¹² 3.7 x 10 ⁴ 2.6x 10 ¹⁰ 5.2 x 10 ⁴ 4x 10 ⁹ 1 .8 x 10 ⁵

VH1 VkC3 1x10 ¹² 1 .6 x 10 ⁵ 2 x 10 ¹⁰ 1 .8 x 10 ⁴ 5.8 x 1 0 ⁹ 4 x 10 ⁵

VH5 VkC3 1x10 ¹² 8 x 10 ⁴ 2.8 x 10 ¹⁰ 1 .2 x 10 ³ 1.8 x 1 0 ¹⁰ 1 .2 x 10 ⁶

Full library 1x10 ¹² 2.4 x 10 ⁵ 2.4 x 10 ¹⁰ 2.88 x 10 ³ 1.4 x 1 0 ¹⁰ 2 x 10 ⁶

Sub-cloning of Selected Pools

DNA was prepared and with digested with EcoH\, Xba\ and Bmti. Nucleic acids were separated on a preparative 1 % DNA agarose gel andthe band referring to the Fab encoding DNA (~ 1470 bp) was extracted, ligated into pMorphx9_FH vector and electroprated into competent TG1 F- cells. Single colonies were picked and incubated in round bottom 96-well microtiter plates in 100 μΙ 2xYT/Cm/1 %Glu medium O/N at 30°C. The next morning, 100 μΙ sterile 30% glycerol in 2xYT to each well of the O/N master plate were added and stored at -80 °C.

Micro-expression of Fab Protein For the preparation of expression plates 5 μΙ per well was replicated from the master plate to the corresponding well of 96-well microtiter plate containing 100 μΙ 2xYT/Cm/0.1 % Glu medium per well. Thereafter the microtiter plates were shaken at 30 °C and 400 rpm until the cultures are slightly turbid (~ 2-4 h) with an OD600nm of ~ 0.5 following induction by addition of 20 μΙ 2xYT/Cm containing 3 mM IPTG. Each plate wase sealed with gas-permeable tape and covered with a lid and incubated O/N at 30°C shaking at 400 rpm.

Primary ELISA Screening

384-well Maxisorp microtiter plates were coated with 20 μΙ antigen solution (mM-CSF) per well at a concentration of 1 pg/ml diluted in PBS and incubated at 4°C O/N. Controls were applied as decribed below. The antigen solution as removed from the coated ELISA plate by rapidly inverting them over the sink and washing the wells once with TBST. ELISA plate was blocked with 100 μΙ 5% MTBST for 2 h at RT, shaking gently. To each well of the expression plate, 40 μΙ BEL buffer containing 2.5 mg/ml Lysozyme was added and shaked for 1 h at 22°C at 400 rpm. After lysis of bacteria 40 μΙ 12.5% MTBS was added per well and incubated for an additional 30 min at 400 rpm and 22°C for blocking. The blocked ELISA screening plate was rinsed once with TBST and tapped it on a stack of paper towels. Afterwards 20 μΙ of the blocked BEL extracts was transferred from the expression plate to the corresponding well of the blocked ELISA plate. Likewise 20μΙ of the expression positive control extract was transferred to well of the ELISA plate. The ELISA plate was incubated for 1 .5h at RT on a microtiter plate shakter, shaking gently. The ELISA plates was wahed 5x quickly with TBST. Then 20 μΙ goat anti-mouse F(ab')2 antibody-AP conjugate diluted 1 :5000 in TBST was added and incubated 1 h at RT. After 1 h incubation the ELISA plates were 5x washed quickly with TBST and 20 μΙ AttoPhosTM substrate was added to each well (diluted 1 :5 in TBST). Fluorescence readings at excitation of 440±25 nM and emission of 550±35 nM was taken.

Controls:

well E12 => positive control for expression and extraction

well F12 => background control of goat anti-mouse F(ab)2 antibody-AP conjugate

well G12 => background control of the AttoPhosTM substrate

well H12 => background of ELISA plate For selecting the best binders an ELISA screening was performed after three rounds of solid- phage panning. 184 for each sublibrary and 2x184 for the full library of the transformed clones were picked and transferred in 96-well microtiter plate for the masterplates. Then a periplasmatic expression was induced and an ELISA screening was performed with mM-CSF as coating antigen. Table 25 shows as example of the fluorescence ELISA readings of one plate of the full library screening. Only the hits with intensity signals of 5-10 (in green) and over 10 (in blue) fold over the background were selected. Hits with intensity signals of 2-5 fold over the background are in yellow.

Table 25:

Sequencing of ELISA Positive Fabs

The heavy and the light chain of primary hits were sequenced. Tables 26 (VH1 VkC3), 27 (VH5VkC3), 28(full library), 29 (VH1 VkB3), 30 (VH2VkB1 ) and 31 (VH5VkB3) show CDR regions of some unique binders. The binders isolated from the same sublibraries comprise the same CDRI and CDR2 regions.

Table 26 (selected binders from the sublibrary VH1 VkC3):

Heavy chain

Clone HCDR1 HCDR2 HCDR3

mM-CSF1.3.7C4 YTFTSYWMH RIDPNSGGTKYNEKFKS APYYAFAY

mM-CSF1.3.7E5 YTFTSYWMH RIDPNSGGTKYNEKFKS SDYRGAFAY mM-CSF1.3.7G2 YTFTSYWMH RIDPNSGGTKYNEKFKS SYYYAFDV

mM-CSF1.3.7G7 YTFTSYWMH RIDPNSGGTKYNEKFKS DAGVDDFDV mM-CSF1.3.8A3 YTFTSYWMH RIDPNSGGTKYNEKFKS SRAYAGYVFDY mM-CSF1.3.8B9 YTFTSYWMH RIDPNSGGTKYNEKFKS SYSYYEAFDY mM-CSF1.3.7G5 YTFTSYWMH RIDPNSGGTKYNEKFKS SYTSRFDY

Light chain

Clone LCDR1 LCDR2 LCDR3

mM-CSF1.3.7C4 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQYNSYPL mM-CSF1.3.7E5 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQYWSRPY mM-CSF1.3.7G2 KASQSVDYDGDSYMN YAASNLESGIPARFSGS SQYWSYPF mM-CSF1.3.7G7 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQSSHYPL mM-CSF1.3.8A3 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQSNAYPR mM-CSF1.3.8B9 KASQSVDYDGDSYMN YAASNLESGIPARFSGS SQKIQRPY mM-CSF1.3.7G5 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQDYEKPY

Table 27 (selected binders from the sublibrary VH5VkC3):

Heavy chain

Clone HCDR1 HCDR2 HCDR3

mM-CSF1.3.9C8 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG RWYYAFDY mM-CSF1.3.10G5 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG RYYHYFSDFDY mM-CSF1.3.10H5 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG WIYYAFDY

Light chain

Clone LCDR1 LCDR2 LCDR3

mM-CSF1.3.9C8 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QLGQEKPY mM-CSF1.3.10G5 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQGNHFPA mM-CSF1.3.10H5 KASQSVDYDGDSYMN YAASNLESGIPARFSGS QQGTSKPY

Table 28 (selected binders from the full library):

Heavy chain

Clone Framework HCDR1 HCDR2 HCDR3

mM-CSF1 .3.12D6 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS SDGSTYAFAY mM-CSF1 .3.13A8 VH1 YTFTSYWMH RIYPNSGGTKYNEKFKS SSSYTYAFDY mM-CSF1 .3.13D4 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS SDSYTYAFAY mM-CSF1 .3.13G1 1 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS SYYYAFDY mM-CSF1 .3.14C2 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS TYYYAFDY mM-CSF1 .3.14D1 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS TYKYAFDY mM-CSF1 .3.13E10 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS STAYGYRYYFDY mM-CSF1 .3.14G10 VH1 YTFTSYWMH RIDPNSGGTKYNEKFKS SYKYAFDY Light chain

Clone Framework LCDR1 LCDR2 LCDR3 mM-CSF1.3.12D6 VkB3 R AS KS VSTSG YS YM H YLASNLESGVPARFSGS QQYSEYPF mM-CSF1.3.13A8 VkB3 R AS KSVSTSG YS YM H YLASNLESGVPARFSGS QQYSSYPF mM-CSF1.3.13D4 VkB3 R ASKS VSTSG YSYMH YLASNLESGVPARFSGS FQYYERPH mM-CSF1.3.13E10 VkB3 R ASKS VSTSG YSYMH YLASNLESGVPARFSGS EQIYSFPL mM-CSF1.3.13G1 1 VkB3 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS QQYYSYPF mM-CSF1.3.14C2 VkB3 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS QQYHKYPF mM-CSF1.3.14D1 VkB3 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS QQYYEKPY mM-CSF1.3.14G10 VkB3 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS QQYNHVPF

Table 29 (selected binders from the sublibrary VH1 VkB3):

Heavy chain

Clone HCDR1 HCDR2 HCDR3 mM-CSF1.3.15B8 YTFTSYWMH RIDPNSGGTKYNEKFKS SYRSYMDY mM-CSF1.3.15G8 YTFTSYWMH RIDPNSGGTKYNEKFKS SYRSYFDY mM-CSF1.3.15A10 YTFTSYWMH RIDPNSGGTKYNEKFKS SYRSYFDV

Light chain

Clone LCDR1 LCDR2 LCDR3 mM-CSF1.3.15B8 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS QQHWEYPH mM-CSF1.3.15G8 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS QQYWQYPF mM-CSF1.3.15A10 R AS KSVSTSG YSYMH YLASNLESGVPARFSGS SQYWSYPF

Table 30 (selected binders from the sublibrary VH2VkB1 ):

Heavy chain

Clone HCDR1 HCDR2 HCDR3 mM-CSF1.3.17F6 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG SYYGSKYPFDY mM-CSF1.3.18D8 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG SYYGIFDY mM-CSF1.3.18H4 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG WIYYAFDY

Light chain

Clone LCDR1 LCDR2 LCDR3 mM-CSF1 .3.17F6 RSSQSIVHSNGNTYLE YKVSNRFSGVPDRFSGS AQRSHYPL mM-CSF1.3.18D8 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQEEHVPL mM-CSF1.3.18H4 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQYSSYPF

Table 31 (selected binders from the sublibrary VH5VkB3):

Heavy chain

Clone HCDR1 HCDR2 HCDR3

mM-CSF1.3. .19C9 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG MYYSYGMTFDY mM-CSF1.3. .20C3 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG MYYASGMAFDY mM-CSF1.3. .20D7 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG YRYGSGYYFDY mM-CSF1.3. .20F1 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG YTYSMGYYFDY mM-CSF1.3. .20F9 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG RYYDNLTFAV mM-CSF1.3. .20A6 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG SGYGYYFAY mM-CSF1.3. .20A12 FTFSSYTMSWVRQTPEKRLE TISGGGGNTYYPDSVKG TSSSFWYAFDY

Light chain

Clone LCDR1 LCDR2 LCDR3

mM-CSF1.3. .19C9 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQDDHYPY mM-CSF1.3. .20C3 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQDNEYPY mM-CSF1.3. .20D7 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQGSHYPR mM-CSF1.3. .20F1 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQGSHYPR mM-CSF1.3. .20F9 RASKSVSTSGYSYMH YLASNLESGVPARFSGS QQEEEWPD mM-CSF1.3. .20A6 KASQSVDYDGDSYMN YAASN LESGIPARFSGS SQRSHYPQ mM-CSF1.3. .20A12 KASQSVDYDGDSYMN YAASN LESGIPARFSGS QQLYSYPK

A diversity of different frameworks combinations was isolated from the panning with the full library, with the most abundant binders comprising a VH1 VkB3 combination. Figure 32 shows an overview of the combinations identified.

Expression of Fab Antibody Fragment

First, a preculture (10 ml) of fresh transformed binders in E. coli TG1 F- in expression vector pMORPHx9_FH was inoculated in 2xYT/Cm/1 % Glu medium and incubated for 3-5 h at 30°C . After that, 500 ml 2xTY/Cm/0.1 %Glu medium with the whole volume of preculture was inoculated and incubated at 30°C in a shaker at 200 rpm until an OD600nm of 0.5 was reached. I PTG (1 M) was added to a final concentration of 0.75 mM for the induction of expression (375 μΙ/500ηιΙ culture). The Fabs were expressed at 30 °C O/N shaking at 200 rpm. The next day, the bacteria were spun down at -4000 g for 30 min at 4°C and the pellets were frozen at 20 °C for at least O/N.

Purification using IMAC and SEC

Bacterial pellets were resuspended in Lysis Buffer comprising 25 mM Tris pH8.0, 0.5 M NaCI, 0.1 % Lysozyme, 2 mM MgCI ₂ and 10 U/ml Benzonase.. After incubation the supernatant was filtered through a 0.2 μιη filter. IMAC Purification of Fab with His6-Tag using Aekta Express was applied for fully automated purification at room temperature 20 °C: filtered samples were loaded on an IMAC column (HiTrap 1 ml Chelating HP), washed with 20 mM Na-Phosphatbuffer pH7.4, 500 mM NaC, 20 mM Imidazole, and eluted with 250 mM lmidazol/20 mM Na-Phosphatbuffer/500 mM NaCI, pH7.4. The eluted peaks were apllied on a gel filtration column (HiLoad 16/60 Superdex 75) and the purified protein were eluted into a deepwell plate in PBS. Figure 33 shows an overview of the obtained expression yields of mM-CSF specific Fab fragments.

Affinity constants determination by ELISA

Wells of a microtiter plate were coated with 1 9/ιτιΙ mM-CSF in PBS, 50 μΙ_/ννβΙΙ, o/n at 4°C. After washing the wells were blocked with 350-400 μΙ_ 5% MTBST for 2h at RT. In the meantime dilutions of each purified Fab for test were prepared : 1000 / 100 / 10 / 1 / 0.1 / 0.01 nM concentration (in duplicates). The blocked ELISA microtiter plates was washed once with TBST. After that 50 μΐ of the prepared dilutions of purified Fabs were transferred in the corresponding wells of the ELISA plates.

NK1 : Coating with selection antigen, -(-blocking, +PBS, -secondary conjugated antibody +substrate NK2: Coating with PBS, +blocking, +PBS, +PBS, +secondary conjugated antibody, +substrate NK3: Coating with selection antigen, -(-blocking, +PBS, -ι-secondary conjugated antibody, +substrate The ELISA plates were incubated with protein dilutions for 2 h at RT on a microtiter plate shaker, washed 3x quickly with TBST and 100 μΙ_ goat anti-mouse Fab antibody-AP conjugate (diluted 1 :5000 in MTBST) was added and incubated for 1 h at RT shaking gently. The plate was washed 3x quickly with TBST and add 100μΙ_ AttoPhos substrate (1 :5) was added to each well and measurement of Fluorescence (ELISA Reader) was taken. Data were analysed using a 4-parameter fit. Figure 34 shows the results for some mM-CSF specific Fab fragments.Over 30 % of the Fabs isolated have an EC50 value in the range of 1 -10 nM, 17% of the Fabs in the range of 100-1000nM, 15% in the range of 10-100nM and 7% in the picomolar range.

Fab conversion and expression and purification of a mlgG2

One of the Fab antibody fragments which bound to murine M-CSF and which showed binding activity in an ELISA was converted into into murine lgG2a format by cloning of the respective VH and VL sequences into an appropriate eukaryotic expression vector harboring the respective constant light chain and murine gamma 2a chain. This expression vector was used for transfection of HEK- 293 cells. lgG2a antibodies were subsequently purified from HEK293 cell culture supernatants by Protein A affinity chromatography on an Akta system.

Affinity determination of a murine lgG2a antibody by BiaCore

Binding of mouse anti-m M-CSF antibody as mlgG2a sub-type was measured towards immobilized mM-CSF. For this HBS-EP as running buffer at 30 μΙ/min was used at 20 °C and antibody was diluted in running buffer to following concentrations: 10, 5, 2.5, 1 , 0.5 nM. Following cycle set-up was chosen: association for 180 sec, dissociation of 900 sec, regeneration of 30 sec using 10 mM glycine pH 2.0 at 30 μΙ/min and a final stabilization of 60 sec. All concentrations were measured in duplicates and reference subtracted (blank channel 1 ). Sensograms were fitted using a "bivalent ligand" binding model and constants are shown in table 32.

Table 32:

ka SE kd SE Rmax KD SE Chi ²

Ligand tc

(1/Ms) (ka) (1/s) (kd) (RU) (M) (tc) (RU ² )

2.83 1.3 < < 1 .41 1.5 mM-CSF local 4.5

E+05 E+03 E-06 E-08 E-12 E+09 E+07 Example 1 .8: Test panning with the murine library

Murine GM-CSF (granulocyte macrophage colony stimulation factor) was used as antigen for an another test panning. Panning, selection and characterization of the binder was essentially identical to Example 1 .7 herein above, except that GM-CSF was used instead of M-CSF.

One of the Fab antibody fragments which bound to murine GM-CSF and which exhibited activity in an FDCP-1 proliferation assay was converted into into murine lgG2a format by cloning of the respective VH and VL sequences into an appropriate eukaryotic expression vector harboring the respective constant light chain and murine gamma 2a chain. This expression vector was used for transfection of HKB1 1 cells. lgG2a antibodies were subsequently purified from HKB1 1 cell culture supernatants by Protein A affinity chromatography. The Fab fragment and the murine lgG2a antibody were then comparatively tested in a FDCP-1 proliferation assay. Briefly murine cell line FDCP-1 proliferates in the presence of murine GM-CSF. Different dilutions of the anti-murine-GM-CSF lgG2a or Fab were added to wells containing FDCP-1 cells and murine GM-CSF. After and incubation for 72h (37°C; 5% C02) cell viability was measured by adding XTT reagent (Roche) according to the manufacturer ^' s recommendation..

Results are shown in Figure 46. Both binders were active in the assay with comparable efficacy, thereby demonstrating that functional binders can be isolated from the library, which also preserve their activity after conversion into full length immunoglobulin format.

Example 2: Generation of a rat antibody library (HuCAL RAT)

Example 2.1 : Design of the rat antibody master genes

Since no data were published about the usage of rat VH-or VL germline-families, all public available rat antibody sequences (498 sequences in total) were subjected to analysis. Sequences were classified into germline families by comparison of the re-arranged sequences to the germline genes. In total 498 re-arranged rat VH sequences were compared to 1 19 rat germline genes extracted from IMGT. The position-dependent amino-acid usage of each re-arranged sequence was compared to the amino-acid usage of each germline gene of the strain Rattus norvegicus. By doing so, each re-arranged sequence got assigned a nearest germline gene, leading to following distribution (Table 33):

Table 33:

VH1 48

VH2 25

VH3

VH4

\ V/UnoK 41 o

VH6 3

VH7 4

VH8

VH9

VH1 0 1

VH1 1 2

SUM 498

Further analysis for the identification of the VH mastergenes within the germline genes was done exclusively for the VH germline-families which include an reasonable amount of re-arranged sequences of VH1 and VH5.

Two approaches were taken for selection of the most frequent used germline genes, which are to be used as master genes:

a) Identification of the most prominent germline gene within the respective germline families b) Identification of the germline gene, which is most akin to the consensus of all germline- genes within one family.

Ideally, both approaches would result in the same germline gene.

Example 2.1 .1 : Identification of the most prominent VH germline gene within the respective germline families

The position-dependent amino acid present in each re-arranged sequence was compared with the amino acid present in the rat germline genes in order to identify the most frequent used germline genes. Each of the re-arranged sequences got assigned a nearest rat germline-gene. If two or more germline genes were identified as nearest germline gene for a re-arranged sequence, all germline- genes were considered as nearest germline genes. For calculation of the preferably used germline gene within a family the usage of each germline gene was summed up.

The results for the preferred germline genes within a family were not as clear as for the selection of the mouse germline-gene. The following germline genes are the two most prominent ones in the different VH-families (Table 34):

Table 34:

Name of prominent # re-arranged % of prominent germline-gene

family germline-gene sequences in-arranged sequences

VH1 AABR03048898|IGHV1 S15 ^* 01 48 10

VH1 >AABR03050300|IGHV1 S23"01 48 10

VH5 AABR03048730|IGHV5S45 ^* 01 415 16

VH5 AABR03048499|IGHV5S23*01 415 14

Example 2.1 .2: Identification of the germline gene, which is most similar to the consensus of all germline genes within one family

The consensus germline sequences of 25 single VH1 and 21 single VH5 germline genes were prepared.

Consensus_of all VH1 germline-genes:

QVQLQQSGAELVKPGASVKISCKASGYTFTSYYMH..WVKQRPGQGLEWIGYIYPGS..G GTNYNEKF KG K ATLT V D KSS ST AYM Q LS S LTS E D S A V YYC AR

Consensus_of all VH5 germline-genes:

EVQLVESGGGLVQPGRSLKLSCAASGFTFSNYGMA..WVRQAPTKGLEWVASISTDG..S STYYRDSV KGRFTISRDNAKSTLYLQMDSLRSEDTATYYCAR These consensus sequences were compared against each individual germline gene of the VH1 family and the VH5 family, respectively, and the individual germline genes were evaluated in respect of identity to the consensus sequence (Table 35).

Table 35:

Germline- Name of prominent # deviations to

family germline-gene consensus [aa]

VPR >AABR03050300|IGHV1 S23 ^* 01 1 1 (of 98 aa)

VH5 AABR03048730| IGHV5S45 ^* 01 2(of 98 aa)

This result was confirmed by BLAST analyses of the consensus sequences against all germline genes.

Example 2.1 .3: Compilation of a rat VL mastergene

Kappa light chains constitute a majority of rat light chains, being found in 90% of pooled light chains isolated from laboratory rats (Frank and Gutman, 1988, Mol Immunol 25, 953-960). Since no VL germline sequences, and only a limited set of re-arranged sequences, were available a different approach was taken for the development of a rat VL mastergene. The 44 sequences available were grouped according to the length of the LCDR1 , followed by formation of consensus sequences.

The consensus sequences of a first group (LCDR1 -length = 17aa, number of sequences = 1 1 ) and of a second group (LCDR1 -length = 1 1 aa, number of sequences = 33) were used as rat VL mastergenes in the library. See Figure 35.

Example 2.1 .4: Summary

No modifications of the germline genes (for VH) or the consensus-sequence (for VL) were required for modularity reasons, since all required restriction sites were already present in the selected genes. Table 36 summarizes the nearest germline genes for the mastergenes selected:

Table 36: Name of nearest

MG germline-gene

rVH1 IGHV1 S23 ^* 01

rVH5 IGHV5S45 ^* 01

rVL1 consensus-sequencel

rVL2 consensus-sequence2

For completion of the rVH germline genes to full length mastergenes, antibody framework 4 was added to the sequences. IMGT provides access to rat J-regions, which form antibody framework 4 in antibody-development. J-region JH2 (sequence WGQGVMVTVSS) was selected for complementation of the VH-mastergene, since it was frequently found in rearranged antibodies deposited in the IMGT database. In addition JH2 provides al required RE-sites (Styl and Blpl). See Figure 36.

For completion of the rVL to a full length mastergene, J-regions JK2-3 was selected, since Tyr was found most frequently in the set of re-arranged sequences, which were used for analysis. The last amino acids of framework 4 (RT) were added to include a BsiWI-site. See Figure 27.

Example 2.2: Design of rat CDRs

Design of L-CDR3 and H-CDR3 library cassettes

Design of H-CDR2 maturation cassette

Overview of the synthesized master genes encoding variable domain is presented in Figure 21. Kabat numbering indicates position of an amino acid residue in a VL- or VH-sequences. rVL denotes rat variable domain in the light chain (exclusively of the type kappa), and rVH denotes rat variable domain in the heavy chain. The two rat VK consensus sequences rVL1 and VL2 encode VL starting from position 1 (at the N-terminus of framework 1 ) to 95 (at the C-terminus of L-CDR3). Rat JK germ-line genes encode for the very C-terminal residue in L-CDR3 (constant position 96) and whole framework 4 in rVL (positions 97 - 108). We have chosen rat JK2 (Burstein et al., 1982, Proc Natl Acad Sci U. S A 79, 5993-5997), since Tyr is the most frequent residue in position 96 of rearranged antibodies found in the IMGT database. Rat VH germ-line genes encode rVH starting from position 1 (at the N-terminus of framework 1 ) to 94 (at the very C-terminus of framework 3). H-CDR3 sequence is encoded by the D germ-line genes. We reasoned that biochemical and biophysical properties of the VH-domain depend also on its HCDR3 sequence, and therefore for the purpose of the master-gene comparison we decided introducing a particular HCDR3 sequence derived from antibody 4D5 into the both rVH master-genes (see below). Rat JH germ-line genes encode for a C-terminal part of HCDR3 and whole framework 4 (positions 103-1 1 3). We have chosen rat JH2 (sequence WGQGVMVTVSS) for the design of framework 4 because JH2 was frequently found in rearranged antibodies deposited in the IMGT database, and it allowed us to introduce two restriction endonuclease sites (Styl and Blpl) indispensable for the HuCAL cloning modularity.

In order to comply with the essential requirements of the HuCAL ^® cloning-modularity, the amino acid sequence corresponding to positions 1 - 4 in both rVHs was fixed to EVQL.

Example 2.3: Cloning, expression and purification of the rat library

Bacterial strains and vectors used for cloning

As described above for the murine library, molecular cloning was carried out using the E. coli strains XL1 -Blue (Stratagene) or Top10F+ (Invitrogen). The products from gene synthesis (master- genes) were first cloned into a pGA-vector at GeneArt AG, and then re-cloned into tricistronic CysDisplay pMorph30-vector as described below. Phage display vector pMorph30 was derived from plasmid pMorph23 (Rothe et al., 2008, J Mol Biol 376, 1 182-1200) basically as follows: hexa-His-tag attached at the C-terminus of VH was replaced for AA-sequence HKHKHK, and a chi site in gill was eliminated. Restriction endonucleases were from New England Biolabs or Fermentas, and T4 DNA ligase was from Invitrogen. Reaction conditions were carried out according to the manufacturer's recommendations.

Gene construction IMGT Rattus norvegicus sub-database- / NCBI-accession numbers of germ-line sequences which were used for construction of the master-genes:

rVH1 : IGHV1 S23 ^* 01 / AABR03050300, rVH5: IGHV5S45 ^* 01 / AABR03048730, JK2: IGKJ2-3 ^* 01 / J00746, JH2: IGHJ2 ^* 01 / X56791 , CK: D IGKC ^* 01 / V01241 , CH1 + hinge: IGHG1 ^* 01 / AABR03048905.

Considerations for the synthesis of master-genes

As described above, genes corresponding to the selected representative germ-line sequences (accession numbers are given below) were optimized on DNA level by applying following criteria: (i) rare codon usage preferably is optimal for E. coli while human/murine rare codons should be avoided whenever possible; (ii) undesirable DNA/RNA features such as internal TATA-boxes, chi-sites, ribosomal entry sites, AT-rich or GC-rich sequence stretches, sequences preventing mRNA-transport into cytoplasm (INS or CRS elements), repeat sequences, RNA secondary structures, and cryptic splice donor and acceptor sites preferably are avoided; (iii) introduction of restriction endonuclease (RE) sites conflicting with the HuCAL ^® cloning-modularity were avoided; and (iv) restriction endonuclease sites required for the modular cloning were introduced. Glycosylation in variable domains is undesirable mainly because the carbohydrate-attachment is often found to affect binding of antigen due to steric hindrance, and is also responsible for heterogeneity of proteins expressed in eukaryotic hosts. Hence we ascertained that the designed VL/VH-master-genes did not encode for protein with potential sites for N-linked glycosylation.

Constant domains in Fab format

Constant domain of the kappa light chain is encoded by CK gene. N-terminal Ala-residue was replaced by Thr which allows introduction of the BsiWI site at this position on the DNA level. A cysteine residue which is located in the C-terminus of CK and is normally involved in formation of the disulfide bond linking the light and heavy chains in natural Fab- or IgG-molecules was replaces for Ala. Since the C-terminal "free" Cys-residue was removed, the light and heavy chains are no longer covalently linked together in our modified Fab format. This modification is the main feature of the CysDisplay ^® which allows for selective elution of bound phages (Rothe et al., 2008, J Mol Biol 376, 1 182-1200). Constant domain of the heavy chain in Fab format is encoded by CH1 gene-fragment. We have chosen germ-line gene IGHG1 ^* 01 which more resembles CH1 -sequence in a successful rat therapeutic anti-CD52 antibody CAMPATH-1 (Cheetham et al., 1998, J Mol Biol 284, 85-99 ). Addition of one Ala-residue in front of the N-terminal sequence ETTAPS allowed us introduction of the Blpl site. First three residues of the hinge (sequence PRD) followed by residues Glu-Phe needed for introduction of EcoRI site were added at the C-terminus of the CH1 domain.

Gene assembly in tricistronic phage display plasmid

The designed genes encoding for rat constant or variable domains were generated by PCR- assembly of a series of synthesized oligonucleotides, and cloned into an interim pGA vector at GeneArt AG. It was convenient and faster to replace human VL, VH, Ck and CH1 genes for rat equivalents in two different interim vector molecules. Derivatives of the pMorph30 plasmid (Rothe et al., 2008, J Mol Biol 376, 1 182-1200), designated as "plasmid L" and "plasmid H", were constructed with relatively large fragment of dummy DNA in order to facilitate cloning as follows: in "plasmid L", dummy DNA (~3 kb) was incorporated in place of light chain-coding region, whereas in "plasmid H", dummy DNA (about 2 kb) was present in place of heavy chain-coding region. Dummy DNA in "plasmid L" was replaced in two steps: rCic gene (see above) was inserted into the vector by using BsiWI / Sphl, and then rVL kappa master-genes (see above) were inserted into the vector by using EcoRV / BsiWI. Dummy DNA in "plasmid H" was also replaced in two steps: rCH1 gene (see above) was inserted into the vector by using Styl (or Eco130l) / EcoRI, and then rVH master-genes (see above) were inserted into the vector by using Mfel / Styl. Two DNA fragments encoding master-genes rVH1 or rVH5 (both linked to mCH1 -region) were excised from the "plasmid H", and inserted by using Sphl / Hindlll into a panel of "plasmid L" plasmids encoding a variety of rVL master-genes resulting in 4 rVL-rVH combinations. The tricistronic "plasmid L" encoding for gpl ll, rVL/rdcD also and rVH/rCH 1 was denominated as pMorph32 indicating the fact that the plasmid encodes rat variable and constant domains.

Re-cloning of rat VL-VH master gene combinations into bicistronic pMorphx9_Fab_FH vector for expression in E. coil

As described above, in order to analyze soluble expression of VL-VH framework combinations in Fab format, we have re-cloned by using Xbal and EcoRI all four VL-VH master-gene combinations (as Fab-encoding inserts) into pMorphx9_Fab_FH vector (Rauchenberger et al., 2003, J. Biol. Chem. 278, 38194-38205) where FH indicates that the Flag- and hexa-His-tags were attached C-terminally to the Fab.

Conversion of rVL-rVH master gene combinations to rat lgG2b and 2c formats

In order to analyze expression of master-gene combinations in full-length rat kappa lgG2b and 2c formats in mammalian cells, we have subcloned the four rVL-rVH master gene combinations into pMorph2_h/r_lgG vector which is a derivative of pMorph2_h_lgG vector (Steidl et al. 2008. Moi Immunol 46(1 ): 135-44). Transcription is driven in this plasmid by CMV-promoter.

Expression and purification of Fab

As described in the case of mouse, expression in E. coli TG1 - cells from the pMx9_Fab_FH plasmid was carried out in 500 ml of 2xYT-medium supplemented with chloramphenicol (34 μg/ml) and glucose (0.1 %). After induction with IPTG (0.75 mM), the cells were grown at 30 °C for 20 h. Cell pellets were lysed during incubation with lysis buffer (200 mM sodium phosphate, pH 7.4, 0.5 M NaCI, 10 mM imidazol, 0.2 % lysozyme, 2 mM MgCI ₂ , 20 U/ml benzonase and EDTA-free protease inhibitor cocktail from Roche) for 30 min at RT. Fab-purification from cleared cell lyzates was carried out by IMAC with Ni ²⁺ ions. After elution, imidazol-containing buffer was replaced for PBS on the PD10 desalting columns (Amersham Pharmacia Biotech).

Expression and purification of IgG

Also, as described above, eukaryotic HKB1 1 cells were transiently transfected with the pMorph4_h/m_lgG plasmid encoding the heavy and light chains. Cell culture supernatant was harvested from 3 to 7 days post transfection. After adjusting the pH of the supernatant to 8.0, addition of NaCI (2 M final concentration) and sterile filtration (0.45 μιη), the solution was subjected to standard protein A affinity chromatography (MabSelect SURE, GE Healthcare). If not stated otherwise, buffer exchange was performed to 1 x Dulbcecco ^' s PBS (pH 7.2, Invitrogen) and samples were sterile filtered (0.2 μητι). Purity of IgG was analysed under denaturing conditions by SDS-PAGE or by using Agilent BioAnalyzer and in native state by HP-SEC. Phage preparation

Phage amplification and purification was carried out essentially as described previously (Krebs et al., 2001 , J Immunol Methods 254, 67-84) with a following modification. Phagemids (VL-VH framework combinations encoded in display plasmid) were propagated in E. coll Top10F+ cells in 2xYT/Cam/Tet/glucose (1 %) medium. After helper phage infection (hyperphage from Progen, Cat. No.: PRHYPE, multiplicity of infection: 40), centrifugation and resuspension of the cell pellet in 2xYT/Cam/Kan/Tet/I PTG (0.25 mM), phages were produced during 20h incubation at 22 °C.

Phage display sandwich ELISA

Black Maxisorp microtiter plates (Nunc) were coated with anti-pi ll antibody (MoBiTec, Cat. No.: PS KAN 3) for determination of relative pill-level (=phage concentration), or with anti rat F(ab ^' ) ₂ antibody (Jackson ImmunoResearch Lab.; Cat. No. : 1 12-006-072) for determination of relative Fab- level in phage samples. Although the exact epitope of the anti rat F(ab ^' ) ₂ antibody is not known (it is presumed to be located in the constant domain), we reasoned that binding of the antibody to rat Fab is not likely to be strongly affected by sequence-variation in Fv-domain, and thus the antibody should be suitable for general assessment of rat Fab-display on tip of filamentous phage. TBS pH 7.4 with Tween 20 (0.05%) and chemi-blocker (from Chemicon; diluted 1 :2) was used for blocking of nonspecific binding sites in the plate and on phages. After incubation with phages, wells were washed with TBS-T, and incubated sequentially with biotinylated anti-bacteriophage antibody (Sigma Cat. No. B2661 ) and avidin-HRP (BD Pharmingen; Cat. No.: 554058). "Quanta Blu" substrate for HRP was applied, and fluorescence signals were recorded on Tecan-Spectrafluor instrument (excitation: 320 nm, emission: 430 nm). The relative pi l l- and Fab-levels in phage sample were determined by using reference phage (displaying a previously selected HuCAL Fab converted into rat format) and the standard 2-state model applied for non-linear regression. The relative display rate was defined as a ratio of the Fab- and pi l l-levels.

Example 2.4: Characterization of the rat antibody library by way of exemplary antibody 4D5

We reasoned that good presentation on phages and high expression in cells can be achieved only for well folded and stable Fab-molecules. It is known that thermodynamic stability of Fab- molecule depends on intrinsic stability of the individual domains (heavy and light chains) as well as on non-covalent interaction between the two domains (Ewert et al., 2003, J Mol Biol 325, 531 -553). Tight interaction between Fd (heavy chain) and CK (light chain) are of particular importance because in the CysDisplay ^® system the two domains cannot be covalently linked via disulfide bond since cysteine residue in the C-termini of constant domains was eliminated. Therefore, instead of testing VL- and VH-frameworks individually, we sought to identify VL-VH combinations which could be efficiently folded and expressed in E. coli (Fab format) and in mammalian cells (IgG format).

As a test case we used the HCDR3 sequence of antibody 4D5 (WGGDGFYAMDY). This sequence had already been used previously for the assessment of human master-genes prior the construction of the original HuCAL library (Knappik et al., 2000, J Mol Biol 296, 57-86).

Analysis of display efficiency

Since efficient display of library members on phage particles is a prerequisite for successful outcome from phage display selections, we decided to use sandwich phage ELISA quantification of display for four rat VL-VH combinations in Fab-format. Our data indicate that all four tested rVL-rVH framework combinations were clearly displayed on phages (Figure 38).

Soluble Fab expression

Also soluble expression of the four rat VL-VH combinations in Fab format was tested in E. coli. We found that the VL1 -combinations with VH 1 or VH5 were expressed surprisingly at 3-4-fold higher levels than the VL2-combinations (Figure 39). The expression level of VL1 -VH5 was remarkably high (about 15 mg/L), and led finally to our decision to choose this Fv as the framework for the construction of the rat library.

Soluble IgG-expression

Due to the fact that at present the most successful antibody format approved for human therapy is IgG we decided to characterize expression of the four rat VL-VH master-gene combinations also in IgG format. Two distinct rat IgG formats, lgG2b and 2c, were chosen for the expression test. Whereas rat gamma heavy chain γ1 and y2a share 94% sequence homology, and best resemble mouse γ1 , rat y2b is equivalent to the mouse Y2a/Y2b pair as regards both nucleotide sequence and antibody effector functions (strong complement activation and binding to FCYR) whilst rat Y2C resembles mouse γ3 (high sequence homology) and weakly activates complement (Bruggemann, 1988, Gene 74, 473-482) . Rat lgG2b binds poorly to proteins A and G under low (physiological) ionic strength. In contrary, rat lgG2c binds strongly to proteins A and G ; nevertheless is poorly soluble in solutions of low ionic strength (Rousseaux and Bazin, 1 979, Vet Immunol Immunopathol. 1, 61 -78).

In practice, the rat VL- and VH-mastergenes were re-cloned into mammalian expression vectors, and the resulting plasmids were used for transient transfection of a mammalian suspension cell line as described in Materials and Methods. We have not observed dramatic differences in expression level of the four rat combinations in rat lgG2b format; however, VH5-combinations seemed to be expressed about 5-fold better than VH 1 -combinations in rat lgG2c format where VL2- VH5 was slightly better expressed than VL1 -VH5. Results for lgG2b are shown in Figure 40, results for lgG2c in Figure 41 . The expression levels of the tested rat VL-VH- combinations in Fab and IgG formats did not correlate.

All rat VL and VH-mastergene combinations were purified in a 2-step purification process using Akta express. Expressed Fab-fragments were bound to chelated Ni2+ and washed either with 1 5 column volumes of 20mM Na-Phosphatbuffer/500mM NaCI, pH 7.4, 20 mM Imidazole or, in a separate purification, using 20 column volumes of 30% lsopropanol/20mM Na-Phosphatbuffer pH 7.4. Second step of purification (SEC) was identical for both procedures (with or without isopropanol treatment) and is described above. After purification, both protein preparations were compared regarding yield, concentration, purity, ratio of VH and VL and thermal stability. A wash step with 30% isopropanol while the Fab is still bound to the IMAC column was introduced in order to receive Fab fragments that are of higher purity. However for some Fab fragments this treatment was seen to be too harsh because the interface - the hydrophobic interaction between heavy and light chain - was not strong enough to endure this procedure. The heavy and light chains of such Fab fragments were eluted at different stages and therefore the ratio between the two chains did deviate from 1 : 1 . Figure 42 shows an overview of expression yields of all tested Fab rat framework combinations after washing on IMAC with and without isopropanol in a color code for selection criteria. Thermal stability

As described for the murine library, protein stability and resistance to environmental stress (such as proteases, heat or chaotropic agents) are an intrinsic property of protein and are typically independent of the expression conditions. Tm values (all within the range 67-75°C) shown in Figure 43 indicate that thermal stability of four rat VL-VH variants tested in lgG2b and lgG2c formats did not vary; however, stability of the rat VL2-combinations in Fab-format (Tm about 48°C) was lower than the stability of the rat VL1 -combinations (Tm about 65 °C).

Selection of Fv-frameworks for library construction

Considering scarcity of available sequences of re-arranged rat antibodies in databases making solid statistical assessment of germ-line sequence-prevalence difficult, we opted to construct rat combinatorial phagemid library using a single robust Fv-framework. We selected the VL1 -VH5 combination because: (i) it had the highest display in the Fab-format on the tip of filamentous phage (see Figure 35); (ii) it was clearly best expressed in the soluble Fab-format in E. coli (about 17 mg/L); (iii) its expression level in tested IgG-formats was good (>10 mg/L) ; and (iv) its thermal stability was high (Tm at about 70°C). Diversified LCDR3- and HCDR3-library cassettes were cloned into this selected Fv-framework as described herein below.

Example 2.5: Diversification of the rat library

Construction of HCDR2 mini-library cassette by using primers with wobbled positions

It had been observed (see http://www.bioc.uzh.ch/antibody/Structures/AgContact/index.h tml) that the most important residues in HCDR2 involved in the antigen-contact are located in positions 52, 52a, 54, 56, and 58 (Kabat numbering). Variation at these positions allows natural in vivo selection of antibodies with antigen-tailored binding pocket comprising HCDR2. We investigated the amino acid distribution for germline sequence families of rearranged antibodies represented by our three selected rVH master-genes (rVHI RV, rVH2 and rVH5), compiled a list of residues which are frequently found in particular positions of HCDR2, and introduced these variations into the selected master-genes. In order to preserve prevalence of the germ-line sequence, the alternative residues found in the rearranged antibodies but not in germ-line sequences were intentionally underrepresented in the design (see Figure 44). Sequence analysis for about 30 randomly picked clones per framework indicated that desired variants (five per framework) were represented in the HCDR2 with a frequency similar to our design; however, about 10% of randomly picked clones carried unintended nucleotide deletions mainly in the positions which were wobbled in the PCR- primers (data not shown). In order to ensure low abundance of frame-shifted clones in the library we decided to dilute the plasmid pool encoding variations in HCDR2 with plasmid encoding wild-type VH- framework. Hence in the final library the minor variants in HCDR2 are about 10-fold less abundant than in our initially HCDR2-design.

The intended variability (five amino acid sequence variants in master-gene) was introduced by DNA amplification in 10 cycles of PCR with Platinum Pfx DNA polymerase (Invitrogen) and a set of forward and reverse primers. The variability in rVH5 was introduced by using forward primer containing wobbled positions with defined nucleotide ratio:

ACCAAAGGTCTCGAGTGGGTGGCGACCATTAG(70%)/C(30%)CTATGATGGCAGC AG(60% )/A(20%)/C(20%)CACCTATTATCGTGATAGC.

The resulting PCR product was digested with Xhol / BssHII, and cloned into the rVH5 master- gene in pMorph32 plasmid. Plasmid pool containing the introduced variety in the master-gene was diluted 1 :10 with plasmid encoding the original master-gene, and used for introduction of the LCDR3- library cassette.

Construction of LCDR3 library cassette

MBP-stuffer for rat VL1 is identical as mVLB1 , and LCDR3 library-cassettes for mVLB1 and rVL1 are identical. LCDR3 library cassette was constructed as described above in the case of mouse library cassette. The design itself was identical to the muine LCDR3, while the flanking regions differed, to reflect the natural framework of the LCDR3:

These flanking regions, represent a part of antibody framework 3 and the terminal VL-antibody framework 4. Thus it was possible to PCR-amplify and clone the cassette at a later stage. rVkl

5 ^' GAAGACGTGGCGACCTATTTTTGC-T8-CAG-T9-T10-T1 1 -T12-CCG-T13

ACCTTTGGTGCGGGCACCA

wherein T8 - T13 are TRIM-mixes.

Sequences of all oligonucleotides used for preparation of LCDR3 cassette are listed in Table

Table 37:

Construction of HCDR3 library cassette

HCDR3 library cassette also was constructed exactly as described above in the case of mouse library cassette. The design of the HCDR3-cassette is identical to the design developed for the murine HCDR3. Thus, all information about the HCDR3-TRIM and the PCR amplification-oligos can be found in the respective chapter for the murine library.

Example 2.6: Test panning with the rat library

The rat library consisting of one framework combination rVL1 rVH5 with diversifications in L- CDR3, H-CDR3 and H-CDR2 was used to generate antibodies specific for mouse GM-CSF. Phages from the rat library were subjected to three rounds of selection on murine GM-CSF (AbD Sereotec, Germany). GM-CSF was biotinylated according to manufacturer's instruction (Amersham, GE Healthcare). Biotinylated murine GM-CSF is hereinafter referred to as "Bio-GM- CSF". Phages were pre-blocked with 2x ChemiBLOCKER containing 0.1 % Tween for two hours and subsequently pre-adsorbed to uncoated NeutrAvidinTM plates for one additional hour at room temperature. This step is included to remove phages specific to the non-target NeutrAvidin. The pre- cleared rat library phage supernatant, obtained as described above, was incubated with 200 nM Bio- GM-CSF for 1 hour at room temperature on a rotator. Phages bound to Bio-GM-CSF were captured on Neutravidin coated wells (NeutrAvidin strips; Pierce) for 30 min and subsequent unbound phage were washed away as outlined in Table 38. Specific bound phages were eluted by adding 100 μΙ of 25 mM DTT in 10 mM Tris HCI, pH 8.0 to each well and incubating for 10-1 5 min at RT shaking. The DTT eluates of each selection were collected. E. coli TG1 with an OD600nm of 0.6-0.8 were added to the phage eluates of each selection and incubated for 30 min in a water bath. After infection bacteria were plated out evenly on two large LB/Cm/Glu agar plates for each selection and incubated over night at 37 °C. For the following panning rounds bacterial suspensions of each pool were collected and used to propagate phages for an additional panning round as described above. In total three subsequent rounds of phage selection were performed with increasing stringency of washing (see Table 38)

Table 38: Washing conditions applied in phage panning

Sub-cloning of selected phage, micro-expression of Fab fragments were performed as outlined for the mouse library herein above. Primary screening in ELISA was performed in solution with biotinylated antigen in capture mode. Essentially 20 μΙ anti-rat antibody (goat anti Rat IgG, F(ab ^' )2, Jackson ImmunoResearch Laboratories, Inc) was immobilized to maxisorb plates at a concentration of 1 .5 μ9/ιτιΙ in PBS incubated o/n at 4°C.

Blocking of rat BEL lysates was done in 3% BSA/TBST 30-60 min at 22 °C. Rat Fab fragments were captured from bacterial crude lysate (BEL) by incubation for 1 hour at RT. After washing 5 times with TBST 20 μΙ biotinylated mGM CSF (diluted to 0.75μg/ml in 0.5%BSA TBST) was added to the plates and incubated for 0,5h. After washing 5 times with TBST, the biotinylated mGM CSF captured via specific rat Fab was detected by adding 20μΙ avidin-HRP diluted 1 :5000 in 0.5%BSA TBST at RT for 1 h. After additional washing 5 times with TBST 20μΙ quanta blue substrate to the screening plates using a MultiDrop384. After 5-10min, fluorescence was measure with the GENios Pro (Tecan). Signals 5-10 fold over background were considered as mGM-CSF specific binding of Fab fragments.

Table 39 summarizes the screening result of the mGM-CSF panning. 262 clones were screened in ELISA on captured bio-mGM-CSF as described above. 52 clones show 5-fold clear signal over background were subjected to sequencing of VH to identify unique Fab fragements. Out of those 51 sequenced clones five unique Fab sequences were identified, which show sequence diversity in H-CDR3 sequence. The H-CDR1 and the H-CDR2 show no sequence variation and are identical to rat VH5 germline sequence.

Table 39: Overview Selection rat HuCAL library against mouse GM-CSF

Primary Unique

Sublibrary # Screened Hit rate # Sequenced

hits Sequence rVH5-Vkl 262 51 19% 51 5

These five specific Fab fragments were expressed and purified as described in mouse sections described herein above. Heavy chain CDR sequences are shown in Table 40. Expression and biophysical features are summarized in Table 41 . All five clones are well expressed in E.coli with expression yield in a range from 5-15 mg/L. On a size exclusion chromatography column all five Fab fragments showed a monomer elution profile, indicating that rat Fragment have no tendency towards misfolding or antibody aggregation. Thermal stability was assessed using the method described in section "Thermal stability" of the mouse library herein above. Except clone mGM-CSF_rat2 , all other four rat Fab fragments show an apparent temperature unfolding Tm in the range of 68,5 °C up to 72°C, which reflects the unfolding temperature of the library rVL1_VH5 framework.

Table 40: Heavy chain complementarity-determining region sequence of mouse GM-CSF specific antibodies

Cone H-CDR1 H-CDR2 H-CDR3

mGM-CSF_rat_1 FTFSNYGMAWVRQAPTKGLE TISYDGSSTYYRDSVKG DSWGYYRRKFDY mGM-CSF_rat_2 FTFSNYGMAWVRQAPTKGLE TISYDGSSTYYRDSVKG GYYTGSFD mGM-CSF_rat_3 FTFSNYGMAWVRQAPTKGLE TISYDGSSTYYRDSVKG NDYGGWRYKFDY mGM-CSF_rat_4 FTFSNYGMAWVRQAPTKGLE TISYDGSSTYYRDSVKG SDWGGMRYKMDY mGM-CSF rat 5 FTFSNYGMAWVRQAPTKGLE TISYDGSSTYYRDSVKG YGWYKGRYKFDY

Table 41 : Expression and thermal denaturation data mGM-CSF specific rat antibodies

Expression and Stability

Frame work Expression rate Monomer

Clone Tm [°C] combination mg/1 portion [%]

mGM-CSF_rat_l rVH5-Kl 5 100 70 mGM-CSF_rat_2 rVH5-Kl 7 99 55 mGM-CSF_rat_3 rVH5-Kl 15.5 100 72 mGM-CSF_rat_4 rVH5-Kl 12 100 71 .5 mGM-CSF rat 5 rVH5-Kl 9.5 100 68.5

The specificity of four of the rat Fab fragments for GM-CSF was demonstrated in a specificity ELISA (see Figure 41 ). All four rat antibodies gave a highly specific signal on directly immobilized GM-CSF on Maxisorp plates, but no signal with the unrelated antigen Estradiol-BSA.

This examples highlights the fact that both, the framework as well as the diversified CDRs contribute to the overall stability of an antibody molecule. By careful selection of highly stable antibody frameworks and sophisticated analysis and re-synthesis of CDR, antibody libraries can be synthetically build up which show superior behavior regarding expression and bio-physical characteristics and well defined binding specificities.

Certain embodiments of the present invention:

1 . A synthetic rodent antibody library.

2. The library of claim 1 , wherein said library comprises members of at least two VH germline families.

3. The library of claim 1 or 2, wherein said library comprises at least two VL germline families.

4. The library of any of the preceding claims, wherein said rodent is selected from mouse, rat, squirrel, chipmunk, gopher, porcupine, beaver, hamster, gerbil, guinea pig, degu, chinchilla, prairie dog, and groundhog.

5. The library of claim 4, wherein said rodent is selected from mouse and rat.

6. The library of claim 4 or 5, wherein said mouse is selected from the laboratory strains Balb/c and C57BL/6.

7. The library of any of the preceding claims, wherein said library comprises more than 50 % of the natural mouse VH repertoire.

8. The library of claim 7, wherein said library comprises members of at least two VH germline families selected from VH1 , VH2, VH3, VH5 and VH9.

9. The library of claim 8, wherein said library comprises members of at least two of the following VH germline genes: IGHV1 -72 ^* 01 (VH1 ), IGHV2-2 ^* 01 (VH2), IGHV3-6 ^* 01 (VH3), IGHV5-9 ^* 04 (VH5), and IGHV9-3 ^* 01 (VH9).

10. The library of claim 9, wherein said library comprises the following VH germline genes: IGHV1 - 72 ^* 01 (VH1 ), IGHV2-2 ^* 01 (VH2), and IGHV5-9 ^* 04 (VH5).

1 1. The library of any of the preceding claims, wherein said library comprises more than 50 % of the natural mouse VL repertoire. 12. The library of claim 1 1 , wherein said library comprises members of at least two VL germline families selected from VL1 , VL3, VL7, VL8 and VL19.

13. The library of claim 12, wherein said library comprises at least two of the following VL-kappa germline genes: IGKV1 -1 17 ^* 01 (Vk1 ), IGKV3-12 ^* 01 (Vk3), and IGHV3-4 ^* 01 (Vk3).

14. The library of any of the preceding claims, wherein said library comprises the VH germline genes IGHV1 -72 ^* 01 (VH1 ), IGHV2-2 ^* 01 (VH2), and IGHV5-9 ^* 04 (VH5) and at least two of the following VL- kappa germline genes: IGKV1 -1 17 ^* 01 (Vk1 ), IGKV3-12 ^* 01 (Vk3), and IGHV3-4 ^* 01 (Vk3).

15. The library of any of the preceding claims, wherein said library comprises H-CDR3s which cover more than 50 % of the naturally occurring H-CDR3 lengths of the murine H-CDR3 repertoire.

16. The library of any of the preceding claims, wherein said library comprises H-CDR3s of the length of 7-12 amino acids.

17. The library of any of the preceding claims, wherein said library has a H-CDR3 design as shown in Figure 5 or Figure 6.

18. The library of any of the preceding claims, wherein said library has a diversity of at least 1.0 ^* 10 ^s in the H-CDR3 region.

19. The library of any of the preceding claims, wherein said library comprises L-CDR3s which cover more than 80 % of the naturally occurring L-CDR3 lengths of the murine L-CDR3 repertoire.

20. The library of claim 19, where the L-CDR3 of essentially all members of the library is of the length of 8 amino acids.

21. The library of any of the preceding claims, wherein said library has a L-CDR3 design as shown in Figure 5 or Figure 6.

22. The library of any of the preceding claims, wherein said library has a diversity of at least 1.0 ^* 1 0 ⁴ in the L-CDR3 region.

23. The library of any of the preceding claims, wherein said library is a mouse library and at least one of amino acid residues 52, 52a, 54, 56 and 58 (Kabat numbering) of the H-CDR2 is diversified 24. The library of any of the preceding claims, wherein said library has a H-CDR2 design as shown in Figure 10 or Figure 1 1 .

25. The library of any of the preceding claims, wherein said library has a diversity of at least 1.0 ^* 10 ⁵ in the H-CDR2 region.

26. The library of any of the preceding claims, wherein said library is a murine library comprising a VH1 germline gene with a diversity of at least 1 .0 ^* 10 ⁵ in the H-CDR2 region.

27. The library of any of the preceding claims, wherein said library is a murine library comprising a VH2 germline gene with a diversity of at least 1 .0 ^* 10 ³ in the H-CDR2 region.

28. The library of any of the preceding claims, wherein said library is a murine library comprising a VH5 germline gene which comprises a diversity of at least 1 .0 ^* 10 ⁵ in the H-CDR2 region.

29. The library of any of the preceding claims, wherein essentially all VH-VL combinations of said library are efficiently displayed.

30. The library of any of the preceding claims, wherein essentially all VH-VL combinations are well expressed in E. coli in Fab format.

31. The library of any of the preceding claims, wherein essentially all VH-VL combinations are well expressed in a mammalian system in IgG format.

32. The library of any of the preceding claims, wherein all VH-VL combinations are thermally stable.

33. The library of any of the preceding claims, wherein the H-CDR3 regions of essentially all members of the library are flanked by unique restriction sites.

34. The library of any of the preceding claims, wherein the L-CDR3 regions of essentially all members of the library are flanked by unique restriction sites.

35. The library of any of the preceding claims, wherein the H-CDR2 regions of essentially all members of the library are flanked by unique restriction sites.

36. The library of any of the preceding claims, wherein said library is an unbiased library.

37. A collection of nucleic acid molecules encoding the library members of any of the preceding claims. 38. A vector encoding the nucleic acid molecules of claim 37.

39. A recombinant host cell comprising the nucleic acid molecules of claim 37 or the vector of claim 38.

40. A method to isolate an antibody specific for an antigen, said method comprising the steps of:

(a) contacting the library of any one of claims 1 -36 with an antigen;

(b) removing those members of the library which do not bind to the antigen; and

(c) recovering those members of the library bound to the antigen.

41. An antibody isolated from a library according to any one of claims 1 -36.

42. An antibody variant derived from the antibody according to claim 41 .

43. An antibody variant according to claim 42, wherein said variant is a humanized antibody.

It is to be understood that the description, specific examples and data, while indicating exemplary embodiments, are given by way of illustration and are not intended to limit the present invention. Various changes and modifications within the present invention will become apparent to the skilled artisan from the discussion, disclosure and data contained herein, and thus are considered part of the invention.