AU2013203288B2 - Immunoglobulin frameworks which demonstrate enhanced stability in the intracellular environment and methods of identifying same - Google Patents

Abstract

Compositions are provided, which can be used as frameworks for the creation of very stable and soluble single-chain Fv antibody fragments. These frameworks have been 5 selected for intracellular performance and are thus ideally suited for the creation of scFv antibody fragments or scFv antibody libraries for applications where stability and solubility are limiting factors for the performance of antibody fragments, such as in the reducing environment of a cell. Such frameworks can also be used to identify highly conserved residues and consensus sequences which demonstrate enhanced solubility and 10 stability.

Description

Australian Patents Act 1990- Regulation 3.2 ORIGINAL COMPLETE SPECIFICATION STANDARD PATENT Invention Title "Immunoglobulin frameworks which demonstrate enhanced stability in the intracellular environment and methods of identifying same" The following statement is a full description of this invention, including the best method of performing it known to me/us: -1, mrm~ruJful nor Immunoglobulin frameworks which demonstrate enhanced stability in the intracellular environment and methods of identifying same This application is a divisional of Australian Patent Application No. 2012211497. , the entire contents of which is 5 incorporated herein by reference. FIELD OF THE INVENTION The invention relates to protein chemistry, molecular biology, and immunology. 10 Background of the Related Art Antibodies can recognize and target almost any molecule with high -specificity and affinity. This characteristic has been exploited to turn these natural 15 proteins into powerful tools for diagnostic and therapeu tic applications. Advances in recombinant DNA technology have facilitated the manipulation, cloning, and expressi on of antibody genes in a wide variety of non-lymphoid cells (Skerra, 1988; Martineau, 1998; Verma, 1998). A 20 number of different antibody fragments have been con structed to best suit the various applications. The smal lest entity that retains the full antigen-binding capaci ty of the whole parental immunoglobulin is the single chain Fv fragment (scFv) (Bird, 1988). This antibody 25 fragment comprises the variable regions of the heavy and the light chains linked by a flexible peptide-linker, which allows the expression of the protein from a single gene. Antibody fragments have several important 30 advantages in comparison to the entire immunoglobulin mo lecule. Due to their smaller size, the expression is fa cilitated and the yield is enhanced in a variety of ex pression host cells, such as E. coli cells (Pltckthui, 1996). Moreover, antibody fragments allow improved tumour 35 penetration in in vivo applications (Yokota, 1992) and they can be linked covalently to various effector molecu les for therapeutic approaches.

2 Naturally occurring antibodies, which are se creted by plasma cells, have evolved to function in an extracellular, oxidizing environment. To obtain their functional, folded structure, they generally require the 5 formation of disulfide-bridges within the separate doma ins, which are crucial for the stability of the immuno globulin fold. In contrast to full-length antibodies, scFv or Fab antibody fragments can, in principle, be functionally expressed in a reducing environment inside 10 any cell and directed to any compartment to target intra cellular -proteins and thus evoke specific biological ef fects (Biocca, 1991). Indeed, some intracellular single chain antibody fragments, which are called intrabodies, have been applied successfully to modulate the function 15 of intracellular target proteins in different biological systems. Thus, resistance against viral infections has been demonstrated in plant biotechnology (Tavladoraki, 1993; Benvenuto, 1995),binding of intrabodies to HIV -pro teins has been shown(Rondon, 1997), and binding to onco 20 gene products (Biocca, 1993; Cochet, 1998; Lener, 2000) has been described. Moreover, intracellular antibodies promise to be a valuable tool in characterizing the func tion of a vast number of genes now identified through the sequencing of the human genome (Richardson, 1995; Mara 25 sco, 1997). For example, they can be used in a functional genomics approach to block or modulate the activity of newly identified proteins, thereby contributing to the understanding of their functions. Finally, intrabodies have potential diagnostic and therapeutic applications, 30 for example in gene therapy settings. Despite these great prospects, the generation of functional intrabodies is still limited by their in stability and insolubility or propensity to aggregate. The reducing environment of the cytoplasm prevents the 35 formation of the conserved intrachain disulfide bridges, thus rendering a high percentage of antibody fragments unstable and, as a consequence, non-functional inside the cell (Biocca, 1995; Proba, 1997). Stability and solubili ty of antibody fragments therefore represents a major obstacle for the application of intrabodies as potential modulators of protein function in vivo. So far, no pre 5 dictions can be made about the sequence requirements that render an antibody fragment functional in an intracellu lar environment. There is, therefore, a need for antibody fragments which perform well in a broad range of diffe 10 rent cell types and can thus be used as frameworks for diverse binding specificities. Such frameworks can be used to construct libraries for intracellular screening or can serve as an acceptor for the binding portions of an existing antibody. 15 Besides being uniquely suited for intracellu lar applications, such antibody fragments or whole anti bodies based on very stable variable domain frameworks also have a distinct advantage over other antibodies in numerous extracellular and in vitro applications. When 20 -such frameworks are produced -in an oxidizing environment, their disulfide-bridges can be formed, further enhancing their stability and making them highly resistant towards aggregation and protease degradation. The in vivo half life (and thus the resistance towards aggregation and de 25 gradation by serum proteases) is, besides affinity and specificity, the single-most important factor for the success of antibodies in therapeutic or diagnostic appli cations (Willuda, 1999). The half-life of antibody frag ments can further be increased through the covalent at 30 tachment of polymer molecules such as poly-ethylene gly col (PEG) (Weir, 2002). Stable molecules of this type re present a significant advance in the use of antibodies, especially, but not exclusively, when the Fc functionali ty is not desired. 35 The great practical importance of antibody fragment libraries has motivated research in this area. Winter (EP 0368684) has provided the initial cloning and 4 expression of antibody variable region.genes. Starting from these genes he has created large antibody libraries having high diversity in both the complementary determi ning regions (CDRs) as well as in the framework regions. 5 Winter does not disclose, however, the usefulness of dif ferent frameworks for library construction. The teaching of PlQckthun (EP 0859841), on the other hand, has tried to improve the library design by limiting the frameworks to -a defined number of syn 1o thetic consensus sequences. Protein engineering efforts involving introduction of a large amount of rationally designed mutations have previously suggested mutations towards the respective consensus sequence as a suitable means for the improvement of the stability of isolated 15 variable immunoglobulin domains (Ohage 1999; Ohage 1999 and US 5,854, 027, hereby incorporated by reference). Plfickthun (EP 0859841) discloses methods for the further optimization of binding affinities based on these consensus sequences. The Pluckthun patent also 20 acknowledges the ongoing increase in knowledge concerning antibodies and accordingly aims at including such future findings in the library design. However, no possible further improvements of the synthetic consensus frame works are suggested. .25 The teachings of Winter, Pluckthun and others (e.g. Soderlind, WO 0175091 ) have thus tried to create large antibody libraries with a focus on high diversity in the CDRs for selection and application of the selected scFvs under oxidizing conditions. All of these libraries 30 are, however, not optimized for intracellular applica tions and thus not useful for selection and applications in a reducing environment, or other conditions which set special requirements on stability and solubility of the expressed antibody fragment. 35 The qualities required for antibody fragments to perform well in a reducing environment, e.g. the cytoplasm of prokaryotic and eukaryotic cells, are not 5 clear. The application of intracellular antibodies or "intrabodies" is theref ore currently limited by their Un predictable behavior under reducing conditions, which can affect their stability and solubility properties (Biocca, 5 1995; Wbrn, 2000). Present patent applications (EP1040201, EP1166121 and W00200729) and publications (Visintin, 1999) concerning intracellular screening for intrabodies focus on the screening technology but do not disclose specific antibody sequences which are functional 10 in eukaryotic cells, in particular in yeast, and, thus, useful for library construction in this context. Visintin and Tse have independently described the isolation of a so-called intracellular consensus se quence (ICS) (Visintin, 2002; Tse, 2002). This sequence 15 was derived from a number of sequences that had been iso lated from an antigen-antibody-interaction screen in yeast. The input into the intracellular screen was, however, heavily biased due to prior phage-display selec tion. Thus, all but one of the input-sequences belonged 20 to the VH 3 subgroup in the case of Visintin et al. The published consensus sequence ICS is fully identical to the consensus sequence for the human VE 3 subgroup de scribed by Knappik (2000) and EP0859841. 60 of the 62 amino acids of the ICS are also identical to the general 25 human VH-domain consensus sequence which was proposed by Steipe as a basis for the construction of variable doma ins with enhanced stability (United States Patent No. 6,262,238, hereby incorporated by reference) . These works were, in turn, based on earlier sequence collections 30 (i.e., Kabat, 1991 and definitions of variable domain subgroups and structural determinants (Tomlinson, 1992; Williams, 1996; Chothia, 1989 and Chothia, 1987). Howe ver, because the input to the intrabody selection was so heavily biased (i.e., in the case of Visintin et al. all 35 but one of the VH domains was VH3), the isolation of VH3 sequences from intracellular screening is not particular ly surprising. Due to the heavy bias of their input 6 library, the work of Tse et al. and Visintin et al. does not provide a thorough evaluation of the human variable domain repertoire as would be provided by an unbiased in quiry and as is required to identify the useful intrabody 5 frameworks present in the human repertoire. We have previously described a system, which allows for the selection of stable and soluble intrabo dies in yeast, independent of their antigen-binding spe cificity (Auf der Maur (2001), W0014B017). This approach 10 allows efficient screening of scFv libraries and the iso lation of specific frameworks,. which are stable and so luble in the reducing' environment of the yeast cell. The objective remains to actually isolate framework sequences and use the patterns in a first step to predict what se 15 quence types would be most stable in the reducing envi ronment and in a second step identify by analysis, recom bination and further in vivo and in vitro experiments the optimal sequence. 20 Brief Summary of the Invention The present invention fills a missing link in the field of antibody generation. It provides antibody variable domain framework sequences with superior charac 25 teristics regarding stability and solubility. These are crucial features for many relevant applications, such as in diagnostics, therapy or research. These frameworks can be used for grafting of existing binding-specificities or for the generation of antibody libraries with high sta 30 bility and solubility. ScFv libraries were used for the isolation of frameworks which are stable and soluble in the reducing environment of the yeast cell. The performance of the isolated frameworks has subsequently been characterized 35 in human cell lines and in in vitro experiments.. The de scribed frameworks can directly serve as acceptor backbo nes for existing binding specificities or to construct 7 CDR libraries by randomization of one or more of the hy pervariable loops for use in reducing or otherwise chal lenging environments. The isolated variable domain se quences have further been analyzed by alignment to iden 5 tify preferred sequence families. From those preferred variable domain sequence families, optimal sequences were chosen based on a structural analysis which excludes se quences containing framework residues which disturb the immunoglobulin fold. The identified variable domain se 10 quence candidates were subsequently recombined in all possible variations and the optimal combinations of va riable domains of the light and heavy chain were selected by analysis.of their performance in yeast, mammalian cells and in vitro. 15 These optimized scFvs and their constituting variable domain frameworks, as well as other antibody fragments or whole antibodies derived thereof, are ideal as, for example, acceptor backbones for existing binding specificities or for the construction of CDR libraries by 20 randomization of one or more of the hypervariable loops for use in reducing or otherwise challenging environ ments. Antibodies suitable for intracellular applica tions are by definition more stable and soluble. Accor dingly, their use will also be advantageous in applica 25 tions outside the intracellular environment. The invention provides compositions conpri sing frameworks of antibody variable domains and single chain Fv antibody (ScFv) fragments which can be incorpo rated into various antibody fragments or whole antibo 30 dies. Classes of antibody variable domains fragments are provided which are the most stable and soluble and thus best suited for intracellular applications. Specific fra mework sequences of antibody variable domains and scFv antibody fragments which show the highest performance in 35 intracellular assays are also provided. The invention al so provides specific framework sequences of antibody va riable domains and synthetic combinations of variable do- 8 mains of the light and heavy chain in scFv fragments which are, for example, optimal for intracellular appli cations and show an optimal performance in vitro regar ding stability and solubility. 5 The invention provides single-chain framework reagents that have the general structures:

NH

2 -VL-linker-VH-COOH or

NH

2 -VH-linker-VL-COOH. 10 In another' embodiment of the invention the single-chain framework may be fused to a second protein moiety to yield a fusion construct of the general struc ture:

NH

2 -VL-linker-VH-second protein-COOH 15 NH 2 -second protein-VL-linker-VH-COOH. The orientation of the VH and VL regions in these fusion constructs may be reversed. In another embodiment -of the invention the 20 variable domains may be incorporated into a Fab fragment, which may additionally be. fused to -a second protein moie ty to yield fusion constructs of the general- structure:

NH

2 -VE-CH-second protein-COOH and NH 2

-VL-CL

COOH 25 The second protein may be fused to either N or C-terminus of either the heavy or the light chain. In a preferred embodiment, the second protein of the single-chain or Fab framework fusion construct is a protein which provides a read-out for intracellular as 30 says, either directly or via transcriptional activation. Another object of the invention is to provide framework classes of antibody variable domains and se quences of variable domains and scFvs which are suitable for grafting the hypervariable loops from existing anti 35 bodies, for example, in order to obtain antibodies which are functional in a reducing or otherwise challenging en vironment.

9 Another object of the invention is to provide framework classes of antibody variable domains and se quences of variable domains and scFvs which, for example, through randomization of one or more of the hypervariabie 5 loops of such frameworks, are suitable for the creation of libraries for use in a reducing or otherwise challeng ing environment. Another object of the invention is the use of the disclosed sequences in the identification of con 10 served residues and consensus sequences. The antibodies or antibody fragments result ing from the use. of the disclosed frameworks can be used as reagents in target validation and in therapy, preven tion and diagnosis of human, animal and plant diseases. 15 The antibodies can be used in the form -of protein or DNA encoding such a protein and are not limited to intracel lular applications. Brief Description of -the Drawings 20 Figure 1 shows the result of a typical "qual ity control" screen in yeast assayed by activation of lacZ expression (see, for example, Example 1) . The se lected, positive clones (black) were identified in sev 25 eral different screens and the corresponding sequences of the positive clones can be found in Tables 1 and 2. The selected sequences are compared to the positive control-, the very stable lambda-graft (dark grey) . Figure 2 shows the performance of the frame 30 works isolated from a typical "quality control" screen in yeast (black) in the human cell line Hela, assayed by the activation of luciferase expression in comparison to the very stable lambda-graft (dark grey) . The positive con trol Gal4-VP16 (white) gives the maximally possible level 35 of transcriptional activation in the system. Luciferase activity has been corrected for transfection efficiency.

10 Figure 3 shows the in vivo performance of the superior framework combinations assayed in yeast by the activation of lacZ expression. The framework sequences (black) are compared to the positive control (the very 5 stable lambda-graf t (dark grey)). The numbering of the frameworks is as described in Table 5. Figure 4 shows the in vivo performance of the superior framework combinations assayed in the human cell line Hela by the activation of luciferase expression and 10 illustrated in comparison to the very stable lambda-graft (dark grey) . The positive control, Gal4-VP16 (white) gives the maximal possible level of transcriptional acti vation in the system. Luciferase activity has been cor rected for transfection efficiency. 15 Figure 5 shows the in vivo performance of the superior framework combinations assayed .by the amount of soluble protein produced in the cytoplasm of yeast strain S. cerevisiae JPY9. Figure 6 shows the expression behavior of se 20 lected framework combinations in the periplasm of E.coli. The arrow indicates the location of the band correspond ing to the scFv frameworks. Figure 7 shows the in vivo performance of se lected superior framework combinations assayed in three 25 human cell lines (Hela, (black), Saos-2 (dark grey) and HEK 293 (white) )~, by the activation of luciferase expres sion and illustrated in comparison to the very stable lambda-graft. The positive control Gal4-VP16 gives the maximal possible level of transcriptional activation in 30 the system. Luciferase activity has been corrected for transfection efficiency. Figure 8 represents the resistance towards aggregation at 37 0 C of selected framework combinations as quantified by the amount of monomeric protein present be 35 fore and after incubation as indicated in PBS-buffer. Panel A is representative for frameworks 2.4 and 5.2 and panel B for frameworks 4.4, 6.4 and 7.3.

11 Figure 9 represents the resistance towards protease degradation aggregation in human serum at 37 0 C of selected framework combinations, quantified by the amount of soluble full-length protein present before and 5 after prolonged incubation. Figure 10 shows the in vivo performance of two selected binders on the novel framework 7.3 in the Fab-context, assayed in yeast interaction assay by the activation of lacZ expression. Expression of the Fab 10 chains is from a bi-directional galactose-inducible pro moter, on either an ars/den.or a 2 micron vectors. Ex pression from the Fab vector yields the antibody light chain and a VH-CH1-Gal4-AD fusion protein. Binders are directed against human Polo-like kinase1 (hPLKI1). Binding 15 to the target is compared with the unspecific binding to an unrelated antigen and the binding of the un-randomized framework 7.3. Note that the corresponding scFv that have been included for reference are expressed from an actin promoter (2 micron). 20 Figure 11 shows the in vivo performance of the scFv frameworks in the Fab-context assayed by the amount of soluble protein produced in the cytoplasm of the yeast strain JPY9. Expression of the Gal4-AD-scFv fu sion (actin/2 micron) is compared with the expression of 25 the corresponding Fab-construct, and with the parent fra mework 7.3 as a Fab, both from two different vectors (Gal-inducible, ars/cen and 2 micron) . Expression from the Fab vector yields the antibody light chain and a VH CH1-Gal4-AD fusion protein, which is detected in this 30 blot. Table 1 shows an alignment of all VH-domain framework sequences selected from various "quality con trol" screens in yeast. Table 2 shows an alignment of all VL-domain 35 framework sequences selected from various "quality con trol" screens in yeast.

12 Table 3 shows an alignment of randomly picked sequences from the library. Table 4 shows a statistical analysis of the sub-class frequency for VH- and VL-domains in the se 5 quences isolated with the 'quality control" system. Only those sequences were considered which were subsequently found to be positive in the quantitative yeast assay. The selected sequences are compared with the unselected li brary as determined from a limited number of random se 10 quences (Table 3). Table 5 shows the sequences used for further recombination and evaluation of the best combinations in scFvs and their respective abbreviations (abb.), sources and sub-family. 15 Detailed Description of the Invention Unless otherwise defined, all technical and scientific terms used herein have the same meaning as .20 commonly understood by one of ordinary skill in the art to which this invention belongs. Although methods and ma terials similar or equivalent to those described herein can be used in the practice or testing of the invention, suitable methods and materials are described below. All 25 publications, patent applications, patents, and other re ferences mentioned herein are incorporated by reference in their entirety. In the case of conflict, the present specification, including definitions, will control. In addition, the materials, methods, and examples are illu 30 strative only and not intended to be limiting. As used herein, "identity" refers to the se quence similarity between two polypeptides, molecules or between two nucleic acids. When a position in both of the two compared sequences is occupied by the same base or 35 amino acid monomer subunit (for instance, if a position in each of the two DNA molecules is occupied by adenine, or a position in each of two polypeptides is occupied by 13 a lysine), then the respective molecules are hamologous at that position. The "percentage identity" between two sequences is a function of the number of matching positi ons shared by the two sequences divided by the number of 5 positions compared x 100. For instance, if 6 of 10 of the positions in two sequences are matched, then the two se quences have 60% identity. By way of example, the DNA se quences CTGACT and CAGGTT share 50% homology (3 of the 6 total positions are matched) . Generally, a comparison is 10 made when two sequences are aligned to give maximum homo logy. Such alignment can be provided using, for instan ce, the method of Needleman et al., J. Mol Biol. 48: 443-453 (1970), implemented conveniently by computer pro grams such as the Align program (DNAstar, Inc.). 15 "Similar" sequences are those which, when aligned, share identical and similar amino acid residues, where similar residues are conservative substitutions for, or "allowed point mutations" of, corresponding amino acid residues in an aligned reference sequence. In this 20 regard, a "conservative substitution" of a residue in a reference sequence is a substitution by a residue that is physically or functionally similar to the corresponding reference residue, e.g., that has a similar size, shape, electric charge, chemical properties, including the abi 25 lity to form covalent or hydrogen bonds, or the like. Thus, a "conservative substitution modified" sequence is one that differs from a reference sequence or a wild-type sequence in that one or more conservative substitutions or allowed point mutations are present. The "percentage 30 positive" between two sequences is a function of the num ber of positions that contain matching residues or con servative substitutions shared by the two sequences divi ded by the number of positions compared x 100. For in stance, if 6 of 10 of the positions in two sequences are 35 matched and 2 of 10 positions contain conservative sub stitutions, then the two sequences have 80% positive ho mology.

14 "VH domain" refers to the variable part of the heavy chain of an immunoglobulin molecule. "VL domain" refers to the variable part of the light chain of an immunoglobulin molecule. 5 VH or VL "subtype" refers to the subtype de fined by the respective consensus sequence as defined in Knappik (2000). The term "subfamily" or " subclass" is used as synonym for "subtype". The term "subtype" as used herein refers to sequences sharing a high degree of iden 10 tity and similarity with the respective consensus se quence representing their subtype. Whether a certain. va riable domain sequence belongs to a "subtype" is deter mined by alignment of the sequence with either all known human germline segments of the respective domain, or the 15 defined consensus sequences and subsequent identification of the greatest homology. Methods for' determining homolo gies and grouping of sequences by using search matrices, such as BLOSTM (Henikoff 1992) are well known to the per son skilled in the art. 20 "Amino acid consensus sequence" as used her ein refers to an amino acid sequence, which can be gene rated using a matrix of at least two or preferably more aligned amino acid sequences, and allowing for gaps in the alignment, it is possible to determine the most fre 25 quent amino acid residue at each position. The consensus sequence is that sequence which comprises the amino acids which are most frequently represented at each position. In the event that two or more amino acids are equally re presented at a single position, the consensus sequence 30 includes both or all of those amino acids. The amino acid sequence of a protein can be analyzed at various levels. For example, conservation or variability could be exhibited at the single residue level, multiple residue level, multiple residue with gaps 35 etc. Residues could exhibit conservation of the identical residue or could be conserved at the class level. Exam ples of amino acid classes include polar but uncharged R 15 groups (Serine, Threonine, Asparagine and Glutamine); positively charged R groups (Lysine, Arginine, and His tidine) ; negatively charged R -groups (Glutamic acid and Aspartic acid); hydrophobic R groups (Alanine, Isoleu 5 cine, Leucine, Methionine, Phenylalanine, Tryptophan, Va line and Tyrosine); and special amino acids (Cysteine, Glycine and Proline) . Other classes are known to one of skill in the art and may be defined using structural de terminations or other data to assess substitutability. 10 In that sense a substitutable amino acid could refer to any amino acid which could be substituted and maintain functional conservation at that position. "Polynucleotide consensus sequence" as used herein refers to a nucleotide sequence, which can be ge 15 nerated using a matrix of at least two or preferably more aligned nucleic acid sequences, and allowing for gaps in the alignment, it is possible to determine the most fre quent nucleotide at each position. The consensus sequence is that sequence which comprises the nucleotides which 20 are most frequently represented at each position. In the event that two or more nucleotides are equally represen ted at a single position, the consensus sequence includes both or all of those nucleotides. "Structural sub-element" as used herein re 25 fers to stretches of amino acid residues within a protein or polypeptide that correspond to a defined structural or functional part of the molecule. These can be loops (i.e. CDR loops of an antibody) or any other secondary or func tional structure within the protein or polypeptide (i.e., 30 domains, a-helices, S-sheets, framework regions of anti bodies;j"tc.) . A structural sub-element can be identified using known structures of similar or homologous polypep tides, or by using the above mentioned matrices of aligned amino acid sequences. Here the variability at 35 each position is the basis for determining stretches of amino acid residues which belong to a structural sub element (e.g. hypervariable regions of an antibody).

16 "Sub-sequence" as used herein refers to a ge netic module which encodes at least one structural sub element. It is not necessarily identical to a structural sub-element. 5 "Antibody CDR" as used herein refers to the complementarity determining regions of the antibody which consist of the antigen binding loops as defined by Kabat et al. (1991). Each of the two variable domains of an an tibody Fv fragment contain, for example, three CDRs. 10 "Antibody" as used herein is a synonym for immunoglobulin". Antibodies according to the present in vention may be whole immunoglobulins or fragments the reof, comprising at least one variable domain of an imnu noglobulin, such as single variable domains, Fv (Skerra, 15 1988), scFv (Bird, 1988; Huston, 1988), Fab , (Fab'.)2 or other fragments well known to a person skilled in the art. "Antibody framework" as- used herein refers to the part of the variable domain, either VL or VH, which 20 serves as a scaffold for the antigen binding loops of this variable domain (Kabat et al., 1991). -Rationally.engineered scFv fragments have de monstrated a clear correlation between the thermodynamic stability of a scFv fragment and its in vivo performance 25 (W6rn, 2000; Auf der Maur, 2001). Using a recently deve loped system named "Quality Control" (Auf der Maur, 2001), specific antibody variable domain framework se quences which are suitable for intracellular applications have been isolated (Table 1 and 2), characterized (Fig. 1 30 and 2) and further improved (Fig. 3 to 9 and Table 3). As observed in our previous experiments, well performing frameworks selected in the intracellular assay show a high in vitro stability as demonstrated by their resi stance to aggregation and protease degradation at 37 0 C 35 (Fig. 8 and 9). Moreover, a pattern emerged which allows a selection of frameworks for intracellular applications on a more general basis, depending on their framework 17 subfamily (Table. 4) . Specific antibody variable domain sequences useful for intracellular applications are disclosed here, as well as the general pattern. This al lows, on the one hand, the use of these sequences as fra 5 mework donors in grafting experiments to obtain functio nal intrabodies which retain the binding specificity of the loop donor. Additionally, antibody libraries can be constructed using the disclosed sequences as frameworks. Such libraries are suitable for intracellular selection 10 systems under reducing conditions, such as those in pro karyotic and eukaryotic cells. Additionally, the disclo sed sequences may be used to identify, for example, con served sequences or residues or motifs. The grafting of structural sub-elements, for example, those of the bin 15 ding loops of an antibody (e.g. Jung, 1997), as well as the making of libraries of antibodies or fragments the reof (e.g. Vaughan, 1996; Knappik, 2000) has been descri bed in detail and is well known to a person skilled in the art. 20 Because intracellular applications expose the antibody fragments to very unfavorable conditions (i.e. increased temperatures, reducing environment), the .se quences disclosed in the present invention have acquired features that make them resistant to the most adverse 25 conditions. Therefore, when compared to "average" sequen ces, the disclosed sequences are of outstanding stability and solubility as is demonstrated by their resistance to wards aggregation and protease degradation (Fig.8 and 9) These features, together with their excellent expression 30 yield make the disclosed antibody framework sequences uniquely suitable not only for intracellular use, but especially for all therapeutic and diagnostic applica tions where long half-life, robustness, and ease of pro duction are of great concern. 35 The present invention enables the design of polypeptide sequences comprising at least the variable part of an antibody that are useful for applications in a 18 reducing or otherwise challenging environment. In a first embodiment, the invention provides a collection of anti body framework sequences useful for intracellular appli cations (Table 1 and 2). In a first step, a library of 5 diverse sequences is screened independent of binding af finity using the Quality control system in yeast. The isolated sequences can be evaluated for their intracellu lar performance in yeast and in mammalian cells (Fig. 1 and 2). 10 In one embodiment of the invention, the col lection of isolated sequences is analyzed by alignment to identify the antibody variable domain sub-classes and consensus sequences that are suitable for intracellular applications. 15 In a further preferred embodiment of the in vention, the collection of antibody framework sequences described above is further analyzed by alignment to each other and grouping into sub-families. All frameworks be longing to one sub-type are compared regarding their in 20 tracellular performance in yeast and in mammalian .cells (Fig. 1 and 2, as an example) and regarding the occur rence of negative, neutral or positive exchanges -in their amino-acid sequence relative to the respective sub-type consensus. A person skilled in the art can distinguish 25 between positive, neutral and negative changes based on the structural environment of the particular exchanged residue in the immunoglobulin domain. Subsequently, fra mework sequences of variable antibody domains are chosen which show the best intracellular performance and which 30 are devoid of negative exchanges compared to their re spective sub-type consensus. Preferably, sequences are selected which further contain amino-acid exchanges which are considered positive. In a further preferred embodiment, the selec 35 ted antibody variable domains of the heavy and the light chain are subsequently recombined in all possible combi nations into scFv fragments, in order to identify the 19 combinations with the highest stability and solubility. To this end the novel, recombined scFv fragments are eva luated for their performance under reducing conditions in intracellular interaction assays in yeast (Fig. 3) and in 5 mammalian cell lines (Fig. 4 and 7) and for soluble in tracellular expression in yeast (Fig. 5) . Promising com binations are further evaluated for their behavior under oxidizing conditions by analyzing the periplasmic expres sion yield in E.coli (Fig. 6), the resistance to aggrega 10 tion at elevated temperatures (Fig. 8) and the resistance to aggregation and protease degradation upon prolonged incubation in human serum at 37 0 C (Fig. 9). These data are used to identify the scrv framework best suitable for any specific application, either intracellular, or under 15 oxidizing conditions. The selected and optimized framework sequen ces disclosed herein have a significant advantage not on ly in intracellular applications, but in all applications which can profit from increased stability and/or solubi 20 lity of the scFv. Examples are the long-term storage at high concentrations required for diagnostic applications, and prolonged functional half-life in serum at 37 0 C (as required, for example, in therapeutic applications). According to one aspect of the present inven 25 tion, there is provided an intrabody framework comprising a single-chain framework having the general structure:

NH

2 -VL-linker--VH-COOH; or

NH

2 -VHlinker-VL-COOH wherein the -VH framework is of subtype la, lb 30 or 3. In another embodiment, the orientation of the VH and VL regions is reversed in the single chain frame work described above. According to one aspect of the present inven 35 tion, there is provided an intrabody framework comprising a single-chain framework having the general structure: NH2-VL-linker-VH-COOH; or 20

NH

2 -VH-linker-VL-COOH wherein the VH framework is of subtype la, lb or 3 and the VL framework is of subtype X1, X3 or X1. In another embodiment, the invention provides 5 a single-chain framework fused to a second protein moiety to yield a fusion construct of the general structure:

NH

2 -VL-linker-VH-second protein-COOH; or

NH

2 -second protein-VL-linker-VH-COOH wherein the VH framework is of subtype la, lb 10 or 3 and the VL framework is of subtype X1, X3 or x1. In another embodiment, the orientation of the VH and VL regions in these fusion constructs may be re versed. In another embodiment, the variable domains 15 may be incorporated into a Fab fragment which may addi tionally be fused to a second protein moiety to yield fu sion constructs of the general structure: NH2-VH-CH-second protein-COOH and NH 2

-VL-CL

COOH 20 The second protein may be fused to either N or C-terminus of either the heavy or the light chain. As disclosed herein, there is a very strong preference in intracellular applications for VH framework of the subtype 3, but also for la and 1b. Regarding the 25 light chain variable domain (VL), there is a clear pref erence by numbers for frameworks of the kappa 1 type; but lambda 1 and 3 are also enriched. These framework subty pes, i..e. VH la, lb and 3 combined with a kappa 1, lambda lor 3 VL domain are therefore best suited for intracellu 30 lar use and other applications which require he folding properties of the scFv. Therefore, in order to reduce the amount of molecules which are not functional in the re ducing environment, libraries for intracellular screening systems should preferentially be constructed from a mix 35 ture of these framework subtypes.

21 In a preferred embodiment, the VH domain of the antibody fragments of the invention is of the sub group la, lb or 3. In a preferred embodiment, the VL domain of 5 the antibody fragments of the invention is of the sub group kappa, lambda 1 or 3. In a preferred embodiment, antibody fragments used as frameworks are selected from the group consisting of: 1.1, 2.1, 3.1, 4.1, 5.1, 1.2, 2.2, 3.2, 4.2, 5.2, 10 1.3, 2.3, 3.3, 4.3, -5.3, 7.3, 1.4, 2.4, 3.4, 4.4, 5.4, and 6.4 as described in Table 5. In one embodiment of the invention, at least two and preferably more frameworks are identified and then analyzed. A database of the protein sequences may be 15 established where the protein sequences are aligned with each other. The alignment can then be used to define; for example, residues, sub-elements, sub-sequence or sub groups of framework sequences which show a high degree of' similarity in both the sequence and, if that information 20 is available, in the structural arrangement. The length of the sub-elements is preferably, but not exclusively ranging between 1 amino acid (such as one residue in the active site of an enzyme or a structu re-determining residue) and 150 amino acids (for -example, 25 whole protein domains). Most preferably, the length ran ges between 3 and .25 amino acids, such as most commonly found in CDR loops of antibodies. In another embodiment, consensus nucleic acid sequences, which are predicted from the analysis are syn 30 thesized. This can be achieved by any one of several me thods well known to the practitioner skilled in the art, for example, by total gene synthesis or by PCR-based ap proaches. In another embodiment, the nucleic acid se 35 quences are cloned into a vector. The vector could be a sequencing vector, an expression vector or a display (e.g. phage display) vector, all which are well known to 22 those of skill in the art. A vector could comprise one nucleic acid sequence, or two or more nucleic sequences, either in different or the same operon. In the last case, they could either be cloned separately or as contiguous 5 sequences. In one embodiment, the- polypeptides have an amino acid pattern characteristic of a particular spe cies. This can for example be achieved by deducing the consensus sequences from a collection of homologous pro 10 teins of just one species, most preferably from a collec tion of human proteins. A further embodiment of the present invention relates to fusion proteins by providing for a DNA se quence which encodes both the polypeptide, as described 15 above, as well as an additional moiety. In further embodiments, the invention provi des for nucleic acid sequences, vectors containing the nucleic acid sequences, host cells containing the vec tors, and polypeptides obtainable according to the me 20 thods described herein. In a further embodiment, the invention provi des for synthesizing or otherwise placing restriction si tes at the end of the nucleic acid sequences of the in vention allowing them to be cloned into suitable vectors. 25 In a further preferred embodiment, the inven tion provides for vector systems being compatible with the nucleic acid sequences encoding the polypeptides. The vectors comprise restriction sites, which would be, for example, unique within the vector system and essen 30 tially unique with respect to the restriction sites in corporated into the nucleic acid sequences encoding the polypeptides, except for example the restriction sites necessary for cloning the nucleic acid sequences into the vector. 35 In another embodiment, the invention provides for a kit, comprising one or more of the list of nucleic acid sequences, recombinant vectors, polypeptides, and 23 vectors according to the methods described above, and, for example, suitable host cells for producing the poly peptides. All of the above embodiments of the present 5 invention can be effected using standard techniques of molecular biology known to one skilled in the art. In another embodiment, the nucleic acid se quence is any sequence capable of encoding the polypepti des of the invention. 10 In another embodiment, the .inventive nucleic acids are used in gene therapy. In another embodiment, the single chain fra mework is .a variant of any one of sequences. 1.1, 2.1, 3.1, 4.1, 5.1, 1.2, 2.2, 3.2, 4.2, 5.2, 1.3, 2.3, 3.3, 15 4.3, 5.3, 7.3, 1.4, 2.4, 3.4, 4.4, 5.4, 6.4 (Table 5), where "variant" as used herein refers to a sequence that exhibits 90% or greater identity, while maintaining en hanced stability. In another embodiment, the single chain fra 20 mework is a derivative of any one of sequences 1.1, 2.1, 3.1, 4.1, 5.1, 1.2, 2.2, 3.2, 4.2, 5.2, 1.3, 2.3, 3.3, 4.3, 5.3, 7.3, 1.4, 2.4, 3.4, 4.4, 5.4, 6.4 (Table 5) where "derivative" as used herein refers to a sequence that maintains only those amino acids that are critical 25 to the function and stability of the molecule. Isolated neutral or positive exchanges in the framework as descri bed in example 3, are not considered to be relevant change to the antibody frameworks of the present inventi on. 30 In a preferred embodiment of the invention, the single chain framework is fused to a second protein, wherein that protein provides a read-out for intracellu lar assays. The read-out can be either direct, for ex ample in the form of a fusion to a detectable protein, 35 e.g. GFP (green fluorescent protein), enhanced blue fluo rescent protein, enhanced yellow fluorescent, protein en hanced cyan fluorescent protein which can be observed by 24 fluorescence, or other fusion partners with different de tection methods. Alternatively, a read-out can be achie ved through transcriptional activation of a reporter ge ne, where the fusion partner in the scFv-fusion protein 5 is either a transcriptional activator, such as the Gal4 activation domain, or a DNA-binding protein, such as the LexA- or Gal4 DNA-binding domain, which activates the transcription of a reporter gene of an enzyme, such as 0 galctosidase, luciferase, a-galactosidase, Q 10 glucuronidase, chloramphenicol acetyl transferase and others, which in turn provide a read-out. Fusion pro teins, which provide a read out are well known to one of skill in the art. Another embodiment of the invention is an an 15 tibody comprising a framework described herein. Another embodiment of the invention is the use of the antibody of the instant invention. A further preferred embodiment of the inven tion is the use of the described framework classes of an 20 tibody variable domains and sequences of variable domains and scFvs for grafting of hypervariable loops from exist ing antibodies, in order to obtain antibodies which are functional in a reducing or otherwise challenging envi ronment. 25 Another further preferred embodiment of the invention is the use of the described framework classes of antibody variable domains and sequences of variable. domains and scFvs, for example through randomization of one or more of the hypervariable loops of such frame 30 works, for the creation of libraries for applications in a reducing or otherwise challenging environment. As would be apparent to one of ordinary skill in the art, the inventive molecules described herein may be used in diagnostic and therapeutic applications, tar 35 get validation and gene therapy.

25 The invention may be illustrated by the fol lowing examples, which are not intended to limit the scope of the invention in any way.

26 References Agatep, R., Kirkpatrick, D. L., Parchaliuk, R. A., Woods and Gietz, R. D. (1998). "Transformation of 5 'Saccharomyces cerevisiae by lithium acetate/single stranded carrier DNA/polyethylene glycol protocol. " Tech nical Tips Online (http://tto. trends. com). Auf der Maur, A., Escher, D. and Barberis, A. (2001). "Antigen-independent selection of stable intra 10 cellular single-chain antibodies." FEBS Lett 508: 407 412. Benvenuto, E. and Tavladoraki, P. (1995). "IMtmotherapy of plant viral diseases." Trends icrobiol 3(7): 272-5. 15 Biocca, S. and. Cattaneo, A. (1995). "Intra cellular immunization: antibody targeting to subcellular compartments." Trends Cell Biol 5: 248-252. Biocca, S., Di Luzio, A.., Werge, T. and Cat taneo, A. (1991). "Intracellular immunization: expression 20 of antibody domains in the cytoplasm and in the nucleus of mammalian cells." Cytotechnology 5 Suppl 1: 49-50. Biocca, S., Pierandrei-Amaldi, P. and Catta neo, A. (1993). "Intracellular expression of anti-p21ras single chain Fv fragments inhibits meiotic maturation of 25 xenopus oocytes. " Biochem Biophys Res Commun 197 (2) : 422 7. Biocca, S., Ruberti, F., Tafani, M., Pierandrei-Amaldi, P. and Cattaneo, A. (1995). "Redox state of single chain Fv fragments targeted to the 30 endoplasmic reticulum, cytosol and mitochondria. Bio/Technology 13 (10): 1110-5. Bird, R. E., Hardman, K. D., Jacobson, J. W., Johnson, S., Kaufman, B. M., Lee, S. M., Lee, T., Pope, S. H., Riordan, G. S. and Whitlow, M. (1988). "Single 35 chain antigen-binding proteins.," Science 242(4877): 423 6.

27 Chothia, C. and Lesk, A. M. (1987). "Canoni cal structures for the hypervariable regions of immuno globulins." J. Mol. Biol. 196(4): 901-17. Chothia, C., Lesk, A. M., Tramontano, A., Le 5 vitt, M., Smith-Gill, S. J., Air, G., Sheriff, S., Pad lan, E. A., Davies, D., Tulip, W. R., Colman, W. R., Spi nelli, S., Alzari, P. [. and Poljak, R. J. (1989). "Con formations of immunoglobulin hypervariable regions." J.Mol. Biol. 342: 877-883. 10 Johnson, G., Kabat, E. A. and Wu, T. T. (1996). kabat database of sequences of proteins of immu nological interest. WEIR'S Handbook of experimantal Immu nology I Immunochemistry and Molecular Immunology. Cam bridge, MA, Blackwell Science Inv: 6.1-6.21. 15 Kabat, E. A., Wu, T. T., Perry, H. M., Got tesman, K. S. and Foeller, C. (1991). -Sequences of pro teins of immunological interest. NIH Publication 91-3242, Wasinghton, DC, US Department of health and human servi ces. 20 Knappik, A., Ge, L., Honegger, A.,' Pack, P., Fischer, M., Wellnhofer, G., Hoess, A., Wolle, J., Pluckthun, A. and Virnekas, B. (2000). "Fully synthetic human combinatorial antibody .libraries (HuCAL) based on modular consensus frameworks and CDRs randomized with 25 trinucleotides. " JT Mol Biol 296 (1) : 57-86. Cochet, 0., Kenigsberg, M., Delumeau, I., Vi rone-Oddos, A., Multon, M. C., Fridman, W. H., Schweig hoffer, F., Teillaud, J. L. and Tocque, B. (1998). "In tracellular expression of an antibody fragment 30 neutralizing p21 ras promotes tumor regression. " Cancer Res 58(6): 1170-6. Corpet, F. (1988). "Multalin. Multiple se quence alignment with hierarchical clustering. " Nucleic Acids Res 16(22): 10881-10890. 35 Cox, J. P., Tomlinson, I. M. and Winter, G. (1994) . "A directory of human germ-line V kappa segments 28 reveals a strong bias in. their usage. " Eur J Immunol 24: 827-836. de Haard, H. J., van Neer, N., Reurs, A., Hufton, S. E., Roovers, R. C., Henderik, P., de Bruine, 5 A. P., Arends, J.-W. and Hoogenboom, H. R. (1999). "a large non-immunized human Fab fragment phage library that permits rapid isolation and kinetic analysis of high af finity antibodies." J. Biol. Chem. 274: 18218-18230. Escher, D., Bodmer-Glavas, M., Barberis, A. 10 and Schaffner, W. (2000). "Conservation of glutamine-rich transactivation function between yeast and humans." Mol Cell Biol 20(8): 2774-82. Escher, D. and Schaffner, W. (1996). "Impro ved "activator trap" method for the isolation of tran 15 scriptional activation domains from random DNA frag ments. " Biotechniques 21(5): 848-54. Krebber, A., Bornhauser, S., Burmester, J., Honegger, A., Willuda, J., Bosshard, H. R. and Plnckthun, A. (1997). "Reliable cloning of functional variable doma 20 ins from hybridomas and spleen cell repertoires employing a reengineered phage display system. " J Immunol Methods 201: 35-55. Lener, M., Horn, I. R., Cardinale, A., Messi na, S., Nielsen, U. B., Rybak, S. X., Hoogenboom, H. R., 25 Cattaneo, A. and Biocca, S. -(2000). "Diverting a protein from its cellular location by intracellular antibodies. The case of p21Ras." Eur J Biochem 267 (4): 1196-205. Marasco, W. A. (1997). "Intrabodies: turning the humoral immune system outside in for intracellular 30 immunization." Gene Ther 4(1): 11-5. Martineau, P., Jones, P. and Winter, G. (1998). "Expression of an antibody fragment at high levels in the bacterial cytoplasm." J. Mol. Biol. 280 (1): 117-27. 35 Ohage, E. and Steipe, B. (1999). "Intrabody construction and expression. I. The critical role of VL domain stability." J IMol Biol 291(5): 1119-28.

29 Ohage, E. C., Wirtz, P., Barnikow, J. and Steipe, B. (1999). "Intrabody construction and expressi on. II. A synthetic catalytic Fv fragment." J Mol Biol 291(5): 1129-34. 5 Plickthun, A., Krebber, A., Krebber, C., Horn, U., Kn~pfer, U., Wenderoth, R., Nieba, L., Proba, K. and Riesenberg, D. (1996). Producing antibodies in Escherichia coli: from PCR to' fermentation. Antibody En gineering, a practical approach. Oxford, Oxford Universi 10 ty Press. Proba, K., Honegger, A. and Pluckthun, A. (1997). "A natural antibody missing a cysteine in VH: consequences for thermodynamic stability and folding. " J. Rol. Biol. 265(2): 161-72. 15 Richardson, J. H. and Marasco, W. A. (1995). "Intracellular antibodies: development and therapeutic potential. " Trends Biotechnol 13(8) : 306-10. Rondon, I. J. and Marasco, W. A. (1997). "In tracellular antibodies (intrabodies) for gene therapy of 20 infectious diseases. " Annu Rev Microbiol 51: 257-83. Skerra, A. and Plnckthun, A. (1988). "Assem bly of a functional immunoglobulin Fv fragment in Esche richia coli. " Science 240 (4855): 1038-41. Tavladoraki, P., Benvenuto, E., Trinca, S., 25 De Martinis, D., Cattaneo, A. and Galeffi, P. (1993). "Transgenic plants expressing a functional single-chain Fv antibody are specifically protected from virus at tack." Nature 366(6454): 469-72. Tomlinson, I. M., Walter, G., Marks, J. D., 30 Llewelyn, M. B. and Winter, G. (1992). "The repertoire of human germline VH sequences revealy about 50 groups of VH segments with different hypervariable loops." J Mol Biol 227: 776-798. Tse, E., Lobato, M. N., Forster, A., Tanaka, 35 T. , Chung, G. and Rabbitts, T. (2002). "Intracellular an tibody capture technology: application to selection of 30 intracellular antibodies recognising the BCR-ABL oncoge nic protein. " J. Mol Biol 317 ( (1)): 85-94. Verma, R., Boleti, E. and George, A. J. (1998). "Antibody engineering: comparison of bacterial, 5. yeast, insect and mammalian expression systems." J .ImmU nol Methods 21-6(1-2): 165-81. Visintin, M., Settanni, G., Maritan, A., Gra zios.i, S., Marks, J. D. and Cattaneo, A. (2002). "The in tracellular antibody capture technology (IACT) : towards a 10 consensus sequence for intracellular antibodies." J. Mol. Biol. 317((1)): 73-83. Visintin, ilM., Tse, E., Axelson, H., Rabbitts, T. H. and Cattaneo, A. (1999). "Selection of antibodies for intracellular function using a .two-hybrid in vivo sy 15 stem." Proc Natl Adad Sci U S A 96 (21):.11723-8. Welschhof, M., Ternesss, P., Kolbinger, F., Zewe, M., Dibel, S., Dbrsam, H., Hain,. C., Finger, M., Jung, M., Moldenhauer, G., Hayashi, N., Little, M. and .Opelz, G.. (1995)'. "Amino acid sequence based PCR primers 20 for amplification of rearranged human heavy and light chain immunoglobulin variable region genes." J Immunol Methods 179: 203-214. Williams, S. C., Frippiat,. J. P'., Tomlinson, I. N., Ignatovic, 0., Lefranc, M. P. and Winter, G. 25 (1996). "Sequence and evolution of the human germline V lambda repertoire." J Mol Biol 264: 220-232. Wbrn, A., Auf der Maur, A., Escher, D., Ho negger, A., Barberis, A. and Plickthun, A. (2000). "Cor relation between in vitro stability and in vivo perfor 30 mance of anti-GCN4 intrabodies as cytoplasmic inhibi tors." J. Biol. Chem. 275(4): 2795-803. Yokota, T., Milenic, D. E., Whitlow, M. and Schlom, J. (1992). "Rapid tumor penetration of a single chain Fv and comparison with other immunoglobulin forms." 35 Cancer Res 52(12) : 3402-8. Auf der Maur, A., Zahnd, C., Fischei, F., Spinelli, S., Honegger, A., Cambillau, C., Escher, D. , 31 Plfickthun, A. and. Barberis, A. (2002). ,,Direct -in vivo screening of intrabody libraries constructed on a highly stable single-chain f ramework." J Bio2 Chem 277 (47): 45075-45085. 5 Gietz, R.D. and Sugino, A. (1988). "New yeast-Escherichia coli shuttle vectors constructed with in vitro mutagenized yeast genes lacking six-base pair restriction sites" Gene 74: 527-534. Willuda, J., Honegger, A.; Waibel, P., Schu 10 biger, A. Stahel, R., Zangmeister-Wittke, U. and Pluckthun, A. (1999). " High thermal stability is essen tial for tumor targeting of antibody fragments: engi neering of a humanized anti-epithelial glycoprotein-2 (epithelial cell adhesion molecule) single-chain Fv frag 15 ment" Cancer Research 59: 5758-5767. Wbrn, A. and Pl ckthun, A. (1999). "Different equilibrium stability behavior of scFv fragments: identi fication, classification, and improvement by protein en gineering" Biochemistry 38: 8739-8750. 20 The invention is further illustrated in the following non-limiting examples. 25 Example 1 Selection of intrabody frameworks through screening of a human library in the "quality control" system in yeast 30 Screening with the "quality control" system for stable frameworks was essentially performed as de scribed in detail by Auf der Maur (WO0148017, Auf der Maur 2001, each hereby incorporated by reference). The plasmids for expression of the scFv 35 fusion constructs for screening in yeast were derived from pESBA-Act (Warn, 2000). It contains the yeast TRP1 gene for transformation selection in S. cerevisiae and 32 the. 2 micron origin of replication to ensure high copy numbers. Moreover it has a constitutive actin promoter for strong expression and the GAL11 transcriptional ter mination sequence, separated by a multiple cloning site. 5 For handling in bacterial systems, it also has a bacte rial origin of replication and the amp resistance gene. The Gal4 activation domain (AD amino acids 768-881) was originally amplified by PCR using pGAD424 (Clontech) as template with primers including the SV40 T 10 antigen nuclear localization signal N-terminal to the' Gal4-AD. The DNA-fragments encoding amino .acids 263-352 of GalliP were amplified by PCR and cloned in frame, N terminal to the SV40-NLS-Gal4-AD-construct. The human scFv- library, amplified from human spleen-cell cDNA as 15 described elsewhere (Welschhof, 1995; Krebber, 1997; de Haard, 1999), was cloned in frame, N-terminal to this fu sion construct via SfiI-sites, and in the orientation V linker-VH where the linker has the sequence (GGGS) 4 . Ex pression thus yields a fusion protein of the general 20 structure scFv-Galllp-SV40 NLS-Gal4AD. Screening was carried out in the yeast strain S. cerevisiae YDE172 (MATa ura3-52 leu2A1 trpld63 his3A200 lys2A 385 gal4A 11) (Auf der Maur, 2001), which was derived from the strain JPY9 (Escher, 2000) by inte 25 grating the .divergently oriented LacZ and HIS3 reporter genes under the control of the natural UASG from Gall GALIO regulatory sequences into the his3A200 locus. Transcriptional activation of the reporter system is me diated by the Gal4-AD moiety of the scFv-fusion con 30 struct, following the specific interaction of its Gal11P moiety with the Gal4-DNA-binding-domain (DBD, amino acids 1-100) . The Gal4-DBD is provided by expression from a second plasmid, pMP83. It contains the yeast LEU2 gene for transformation selection in S. cerevisiae and the ARS 35 CEN origin of replication. Moreover, it has a constitu tive actin promoter for strong expression and the GAL11 transcriptional termination sequence. For handling in 33 bacterial systems, it also has a bacterial origin of rep lication and the amp resistance gene. For screening, the yeast strain S. cerevisiae YDE172 was co-transformed with a scFv-library as fusion 5 construct on the pESBA-Act2 vector while the pMP83-vector provided the Gal4-DBD. A standard lithium acetate trans formation protocol was used (Agatep, 1998). Following transformation, the cells were plated on drop-out plates (-Trp/-Leu/-His) containing 80 mX 3-aminotriazole. Colo 10 nies were picked after 3 days incubation at 30*C and re streaked on drop-out plates (-Trp/-Leu/-His) containing 80 mM 3-aminotriazole. -Those that re-grew were tested for LacZ-expression by development of blue color in a filter assay on plates containing the substrate X-Gal. Positive 15 clones were taken for further analysis involving isola tion of the scFv-carrying plasmid from yeast, transforma tion into E.coli DH5a, isolation of plasmid from single colonies of E.coli and re-transformati-on into freshly prepared yeast strain S. cerevisiae YDE172 for the assay 20 as described below. All methods were performed according to standard procedures, well known. to.a person of ordi nary skill in the art. In addition, a modified screening procedure was used were the scFv was directly fused to both a DNA 25 binding domain (LexA amino acids 1-202) and an activation domain (Gal4, amino acids 768-881) to yield a fusion con struct of the following structure: scFv-LexA-NLS-Gal4AD. The plasmids for expression of the scFv-fusion constructs for screening in yeast were derived from pESBA-Act2. It 30 contains the yeast TRP1 gene for transformation selection in S. cerevisiae and the 2 micron origin of replication to ensure high copy numbers. Moreover, it has a constitu tive actin promoter (for strong expression) and the GAL1 transcriptional termination sequence separated by a mul 35 tiple cloning site. For handling in bacterial systems, it also has a bacterial origin of replication and the amp resistance gene.

34 Screening was carried out in the yeast strain S. cerevisiae ImmunaLHB (MATa ura3-52 leu2A1 trp1d63 his3A200 lys2A 385) which was derived from the strain JPY5 by integrating the divergently oriented LacZ and 5 HIS3 reporter genes under the control of a bi-directional promoter with six LexA-binding sites (integrating re porter plasmid pDE200, Escher 2000) into the his3A200 lo cus and by integrating the LEU2 reporter gene under the control of a promoter with eight LexA-binding sites (de i rived from EGY48) into the leu2A1 locus. Transcriptional activation of the reporter system is mediated by the Gal4-AD moiety of the scFv-fusion construct. Screening was carried out essentially as described above using drop-'&ut medium (-Trp/-Leu/-His) and 3-aminotriazole con 15 centrations up to 40 mm. Example 2 Evaluation of in vivo performance a) in yeast 20 For quantitative analysis of -the performance of the selected frameworks in yeast (Fig.1 and 3), S. cerevisiae -strain Immuna LHB was transformed with the isolated scFvs as LexA-Gal4-AD-fusion constructs on the pESBA-Act2 vector by following a standard lithium acetate 25 transformation protocol (Agatep, 1998). Following trans formation, the cells were plated on drop-out plates ( Trp). 2 ml overnight-cultures in drop-out medium (-Trp) were inoculated in duplicates from streaks containing several colonies and grown at 30 0 C. Cultures were di 30 luted in 1 ml drop-out medium (-Trp) to an optical den sity at 600 nm (OD600) of 0.7. They were then grown at 30 OC for 2h. For the assay, 100 pl cell culture were taken, mixed with 900 pl buffer, 45 pl Chloroform and 30 sl 0.1% SDS, vortexed and incubated at room temperature 35 for 5 minutes. The color development was initiated by the addition of 0.2 ml ONPG (4 mg/ml) and stopped with 0.5 ml Na 2

CO

3 (1 M) . The activity was calculated by taking into 35 account the OD600 of the assay culture, as well as the incubation time of the color development and the culture volume used Clones that were at least equal to or better 5 than the positive control (the very stable lambda-graft described before (W5rn, 2000; Auf der Maur, 2001)) were sequenced to identify the framework subtype (framework subtype definitions according to Tomlinson, (1992), Cox, (1994) and Williams, (1996)). Sequencing revealed a 10 striking preference for certain framework subtypes. For the heavy chain variable domain (VH), framework subtypes 2 and 6 were never found and 4 was markedly reduced among the positive clones. Corrected for the performance of the isolated sequences in-the yeast intracellular assay, 15 there is a very strong preference for VH framework of the subtype 3, but also for la and lb in intracellular appli cations. Regarding the light chain variable domain (VL); there is a clear preference for frameworks of the kappa 1, lambda 1 and lambda 3 sub-types (Table 4) . 20 These framework subtypes, i.e. VH la, lb and 3 combined with a kappa 1, lambda 1 and lambda 3 VL do main are therefore best suited for intracellular use and other applications with stringent requirements concerning the folding properties of the scFv. Libraries for intra 25 cellular screening systems should, for example, preferen tially be constructed from a mixture of these framework subtypes only, to reduce the amount molecules which are not functional in the reducing environment. 30 b) in mammalian cells Hela cell line was used for quantitative analysis of the performance of the selected frameworks in human cells (Fig. 2, 4 and 7). The luciferase reporter gene was provided from a co-transfected pGL3 (Promega) 35 reporter plasmid containing the luciferase under the con trol of the natural Ga14 UAS. The mammalian expression vectors used for transient transfection contains the Ga14' 36 (1-147) fused on the C-terminus to the VP16-AD under the control of a CMV promoter. The isolated scFvs were cloned in frame, C-terminal to a Gal4(1-147)-VP16-fusion to yield a Gal4 (1-147) -VP16-scFv- fusion protein upon expres 5 sion. Cells were cultured in DMEM supplemented with 2.5% FCS and 2 m[ 1-glutamine. Transient transfections were carried out according to the Polyfect-protocol (Qiagen) in 60 mm tissue culture plates using 0.01-0.1 pg of the vector containing the scFv-construct, 0.5 pg of a CMV 10 promoter-driven Gal4 (1-147) -VP16-scFv expression plasmid and 0.5 pg of a LacZ expression vector as reference for transfection efficiency. Cells were harvested 24-48 hours after transfection, resupended in 1000 pl buffer and lysed by three freeze-thaw-cycles. The cell lysate was 15 centrifuged and the supernatant assayed for luciferase. activity using luciferase assay solution (Promega) and for LacZ activity according to the standard protocol. The obtained luciferase activity was corrected with the LacZ activity to account for the variation in transfec 20 .tion efficiency. Example 3 Multiple alignment and analysis of the se quence comparison 25 To elucidate the general pattern of framework sequences suitable for intracellular applications, all positive clones (i.e. those that grow under selective conditions in the quality control system) were isolated and the part coding f:or the scFvs was sequenced. Subse 30 quently, the scFv sequences were divided in their light and heavy-chain component to allow alignment of the re spective domains (Tables 1 and 2) according to the struc tural adjusted numbering scheme of immunoglobulin domains by Honegger (2001). 35 To allow evaluation of the obtained data, an alignment representing the unselected library was genera ted (Table 3). In order to obtain unselected sequences, 37 the library was transformed in E.coli cells which do not express the scFv-genes and clones were picked at random for plasmid isolation and sequencing of the scFv sequence. The library covers the human antibody reper 5 toire as expected and thus has no bias towards specific subgroups, other than expected by the expression pattern generally found in humans. The VH and VL sequences were grouped accor ding to their subgroup. Changes to the subgroup-specific 10 consensus sequence were highlighted. A person skilled in the art can distinguish between positive, neutral and.ne gative changes based on the structural environment of the particular exchanged residue (e.g. Honegger, 2001). An exchange of a residue belonging to a particular group of 15 amino acids to a residue of the same group is in general validated as a neutral exchange. An exchange of a residue belonging to the group of hydrophobic amino acid pointing into the hydrophobic core of the protein to one amino acid of the group of polar but uncharged or positively or 20 negatively charged amino acids would be highly unfavorab le because unsatisfied hydrogen donor/acceptor sites disturb tight packing of the hydrophobic core. Such a change is therefore considered negative. An exchange of a residue belonging to the group of polar but uncharged re 25 sidues at the surface of the immunoglobulin domain to an amino acid of the group of positively or negatively char ged residues is highly favorable as the solubility of the protein is increased. Such a change is therefore valida ted positively, whereas the exchange from a polar to a 30 hydrophobic residue is highly unfavorable as the solubi lity of the protein is decreased and is therefore valida ted negatively. At positions with a conserved positive phi-angle, an exchange of any amino acid to glycine is validated positively whereas an exchange of gylcine to 35 any amino acid is validated negatively because glycine is the only amino acid which is able to form a positive phi angle. The loss of a conserved salt bridge between posi- 38 tions 45 - 53, 45 - 100, 77 - 100 and 108 - 137 because of an exchange from an amino acid of the group of positi vely or negatively charged residues to an uncharged amino acid results in a decreased thermodynamic stability and 5 is therefore considered negative. Finally, we chose 7 VL domains and 4 VH doma ins that were preferentially selected during the quality control (i.e. showing the least negative and most positi ve. exchanges from the consensus sequence and cover the 10 subgroups) and that each show high in vivo performance in yeast. The sequences are summarized in Table 5 and inclu de two VK1 (k I 27 (1.x) and k III 25(2.x)), two VK3 (k IV 103 (3..x) and k 1V135 (5.x)), one VX1 (k IV 107 (4.x)), two VX3 (a33 (7.x) and a43 (6.x)), one VH1b (a33 15 .(x..3)) and three VH3 (a fw1O (x.2), - a43 (x.4) and a44 (x.1)). These VL and VH- domains were shuffled giving 22 novel combinations in' the scFv format (1.1, 2.1, 3.1, 4.1, 5.1, 1.2, 2.2, 3..2, 4.2, 5.2, 1.3, 2.3, 3.3, 4.3, 5.3, 7.3, 1.4, 2.4, 3.4, 4.4, 5.4,.6.4.). 20 Example 4 Evaluation of in vivo performance of shuffled domains a)- Performance in an intracellular assay in 25 yeast and mammalian cells The 22 combinations were tested for their in vivo performance in yeast and mammalian cells as descri bed in example 2 (Fig. 3 and 4). 30 b) Expression of soluble protein under re ducing conditions in yeast To compare the yields of soluble protein upon expression under reducing conditions, the selected frame works were expressed as a fusion to Gal4 AD in the 35 cytoplasm of yeast S. cerevisiae. The fusion constructs on the pESBA-Act2 vector had the general structure Gal4 AD-scFv. They were transformed as described above into 39 the yeast S. cerevisiae strain JPY9 and plated on -Tzp, drop-out plates. 5 ml overnight--cultures in drop-out medium ( Trp) were inoculated from streaks containing several co 5 lonies and grown at 30 0 C. Cultures were diluted in 50 ml drop-out medium (-Trp) to an optical density at 600 nm (OD600) of 0.5. They were grown at 30 0 C for 5h. For the native cell extract, 2.5 ml cell culture normalized to an ODGOO of 3 were harvested by centrifugation, frozen 10 in liquid nitrogen and subsequently resuspended in 75pl Y-PER (Pierce) containing protease inhibitor (PMSF). The resuspended cell pellet was vortexed shortly and incuba ted (slightly shaking) at 20 0 C for 20 min. Insoluble and aggregated material was pelleted at maximal speed in an 15 eppendorf centrifuge at 4 0 C for 10 min. The supernatant was mixed with loading dye, -heated to 100 0 C for 5 min. and separated on a 12% SDS-PAGE. The soluble Gal4 AD-scFv fusion constructs were visualized by western blotting via detection of the Ga14-moiety with an anti-Gal4AD monoclo 20 nal mouse antibody (Santa Cruz Biotechnology) as a prima ry antibody and an anti--mouse-peroxidase conjugate (Sig ma) as secondary antibody and using a chemoluminescent substrate (Pierce) (Fig. 5) . SDS-PAGE and western blot ting procedures are well known to a person of ordinary 25 skill in the art. c) Expression behavior in the periplasm of E.coli For evaluation of periplasmic expression be 30 havior in E. coli .(Fig. 6), isolated scFvs-frameworks we re cloned in a bacterial vector harbouring the cam resi stance gene (catR) and the lacI repressor gene (Krebber, 1997) , with a N-terminal pelB-leader sequence and a C terminal his-tag under the control of the lac promo 35 ter/operator. Competent E.coli JM83 were transformed with these plasmids. 50 ml dYT-medium containing 35 mg/l chlo ramphenicol in shaking flasks was inoculated 1:40 with an 40 over-night culture and incubated at 30 0 C. Cells were in duced at an OD600 of 0.8 with 1mM IPTG and harvested af ter 3 hours of induction by centrifugation. The. pellet was resuspended in 50 mM Tris, pH 7.5, 500 mM NaCl and 5 normalized to an OD600 of 10. Samples of each scFv frag ments were analyzed either directly (total extract) or after sonification followed by centrifugation (soluble fraction) by SDS-PAGE. The amount of soluble protein was estimated from the Coomassie-stained gel. 10 Example 5 Detailed evaluation of 5 combinations with superior properties for extracellular use Five .combinations were chosen as examples 15 which show good performance both in yeast and mammalian intracellular assays, -yield .soluble protein during ex pression in yeast and E.coli, and cover the subgroups which were preferentially selected during the quality control (2.4, 4.4, 5.2, 6.4 and 7.3, see Table 5 for de 20 tails). We analysed these combinations in greater detail to further evaluate their use under reducing, as well as oxidizing conditions. a) Performance in an intracellular assay in 25 different mammalian cells The quantitative analysis of the performance of the five combinations in human cells was carried out using Hela. cells and in addition using the human osteo sarcoma cell line Saos-2 and the human embryonal kidney 30 cell line HEK293 as performed in Example 2 (Fig. 7). b) Performance in vitro Expression and purification For evaluation of the in vitro performance, 35 the five superior combinations were expressed in the periplasm of E.coli (Fig. 6). The amount of 0.1 1 dYT medium containing 35 mg/1 chloramphenicol in shaking 41 flasks was inoculated 1:40 with an over-night culture and incubated at 300C. Cells were induced at an OD550 of 1.5 with 1mM IPTG and harvested after 2 hours of induction- by centrifugation. For purification of the scFvs, the cell 5 pellet was resuspended and lysed by sonication. Following centrifugation in SS34 at 20krpm, 4 0 C for 30 minutes, the supernatent was applied to a Ni-MC-affinity column (Hi TrapT" Chelating HP, 1ml, Amersham Pharmacia) at pH 7.5 and eluted with 200 mM imidazol using an Akta Basic sys 10 tem from Amersham Pharmacia. The purity of the .scFv frag ments was greater than 98% -as determined by SDS-PAGE (data not shown).. The concentration of the purified pro tein was determined using the calculated extinction coef ficient at 280 nm. The yield of .soluble purified protein 15 was normalized to a culture volume of 1 1 with an OD600 of 10 and varied from 8 to over 55 mg. Resistance to aggregation Resistance towards aggregation has been shown 20 to correlate with thermodynamic stability (Wbrn, 1999) in vitro and the efficiency of tumor localization in a xeno graf ted tumor model in mice (Willuda, 1999). In order to test for the stability, resistance to aggregation and re versibility of unfolding, 200 sl samples of the purified 25 proteins at concentrations of 6 pM in 50 mM Tris, pH 7.5, 100 mM NaCl were either kept 4 days at 4 0 C or 4 days at 37 0 C or 3 days -at 4 0 C followed by an incubation of 15 or 60 minutes at 1000C, slow cooling down to room tempera ture and an overnight incubation at 40C. The oligomeric 30 state of each sample was subsequently analyzed on a gel filtration column equilibrated with 50 mM Tris, pH 7.5, 100 mM NaCl to estimate the amount of aggregated versus monomeric material (Fig. 8). The proteins were injected on a Superdex-75 column (Amersham Pharmacia) in a volume 35 of 100 pl and a flow-rate of 1 ml/min on a Akta Basic system (Amersham Pharmacia)..

42 Resistance to protease degradation To determine the stability of the isolated frameworks towards protease degradation, a parameter that is important for therapeutic applications, we incubated 5 the purified frameworks in human serum at 37 0 C (Fig. 9). Purified, his-tagged scFv-protein (see above) at a concentration of 50 pM was diluted tenfold into hu man serum to a final concentration of 5 pM in 90% serum. The samples were then either incubated at 370C for either 10 3 days or 1 day, or taken directly for loading. Before loading insoluble and aggregated material was pelleted at maximal speed in an eppendorf centrifuge at 4DC for 10 mnn. The supernatant was diluted six-fold with a loading dye to reduce the amount of serum loaded on the gel, hea 15 ted to 1000C for 5 min. and separated on a 12% SDS-PAGE. The soluble his-tagged scFv fragments were visualized by western blotting via detection of the his-tag with an an ti-his monoclonal mouse antibody (Qiagen) as primary and an anti-mouse-peroxidase conjugate (Sigma) as secondary 20 antibody and using a chemoluminescent substrate (Pierce). SDS-PAGE and western blotting procedures are well known to a person of ordinary skill in the art. Example 6 25 Selection of antigen binders through screen ing of a randomized CDR-library on the framework 7.3 in the interaction screening system in yeast Screening with the interaction system for an 30 tigen binders was essentially performed as described in detail before (Auf der Maur, 2002). The plasmids for expression of the scFv fusion constructs for screening in yeast were derived from pESBA-Act2. It contains the yeast TRP1 nutritional 35 marker and the 2 micron origin of replication. Moreover it has a constitutive actin promoter for strong expres sion and the GAL11 transcriptional termination sequence, 43 separated by a multiple cloning site. For handling in bacterial systems, it also has a bacterial origin of rep lication and the amp resistance- gene. The Gal4 activation domain (AD amino acids 5 768-881) was originally amplified by PCR using pGAD424 (Clontech) as template with primers including the SV40 T antigen nuclear localization signal N-terminal to the Gal4-AD. The scFv library was obtained by PCR amplification of the scFv-framework 7.3 using primers 10 randomizing 7 amino acids within the CDR3 of VH. The re sulting PCR-product was cloned in the framework 7.3, pre sent in the vector in the orientation VL-linker-V, as a C-terminal fusion to Gal4-AD. Expression thus. yields a fusion protein of the general structure Gal4-AD-scFv. 15 Screening was carried out in .the yeast strain S. cerevisiae Immuna LHB (MATc ura3-52 leu2A1 trpld63 his3A200 lys2A 385). It was derived from the strain JPY5 by integrating the divergently oriented LacZ and HIS3 re porter genes under the control of a bi-directional pro 20 moter with six LexA-binding sites (integrating reporter plasmid pDE200, Escher 2000) into the his3A200 locus and by integrating the LEU2 reporter gene under the control of a promoter with eight LexA-binding sites (derived from EGY48) into the leu2Al locus. 25 Transcriptional activation of the reporter system is mediated by the Gal4-AD moiety of the scFv fusion construct, following the specific interaction of its scFv moiety with the antigen-moiety of the bait fusion protein. The bait-fusion protein consists of the 30 kinase domain of the human polo-like kinase 1 (hPlkl-KD) fused C-terminal to the DNA-binding LexA protein. The ki nase domain (amino acid 2-332) was PCR amplified from a hPlkl cDNA using the upstream primer 5'-tgctctagaagt gctgcagtgactgcag-3' (Seq. Id.No. 12) and downstream pri 35 mer 5'-ggttgtcgacttacaggctgctgggagcaatcg-3' (Seq. Id. No.13). The resulting PCR product was cloned C-terminal of LexA via XbaI and Sa1 into the bait vector. The bait 44 vector contains the URA3 nutritional marker and an Ars Cen origin of replication. Expression of the bait-fusion protein is driven by a constitutively active actin promo ter. Transcription is terminated by the GAL1I termination 5 sequence. The bait vector also carries a bacterial origin of replication and the amp resistance gene for propagati on in bacterial systems. For screening the yeast strain S. cerevisiae Immuna LHB was co-transformed with a scFv-library as fu .10 sion to Gal4-AD on the pESBA-Act2 vector and the bait vector providing the LexA-hPLK1-KD fusion by following a standard lithium acetate transformation protocol (Agatep, 1998). Following transformation, the cells were plated on drop-out plates (-Trp/-Leu/-Ura). Colonies were picked 15 -after 3 to 5 days -incubation at 30 0 C and restreaked on drop-out plates (-Trp/-Leu/-Ura). Those that re-grew were tested for LacZ expression by development of blue color in a filter assay on plates containing the substrate X Gal. Positive clones were taken for further analysis in 20 volving isolation of the scFv-carrying plasmid from yeast, transformation into E.coli DH5a, isolation of plasmid from single colonies of E.coli, sequencing and re-transformation into freshly prepared yeast strain S. cerevisiae Immuna LHB for the assay as described below. 25 All methods were performed according to standard proce dures, well known to a person of ordinary skill in the art. Example 7 30 Evaluation of in vivo performance of Fab constructs derived from novel scFv frameworks To evaluate the beneficial effect of using stable variable domain frameworks on different antibody formats, Fab expression vector were constructed for use 35 in the yeast interaction screen. a) Fab constructs for intracellular screening in yeast 45 Two different expression vectors were con structed to allow different expression levels. The vec tors are based on either yEplac 112 (2 micron) or yCplac22 (ars/cen) backbones (Gietz, 1988). Both contain 5 the yeast -TRP1 nutritional marker, an inducible, bi directional Gall/Gal1O promoter, a bacterial origin of replication and the amp resistance gene for handling in bacterial systems. In one direction, tbe VH domain of the framework 7.3 was cloned N-terminal to the CH1-domain of 10 IgGi including the C-terminal cysteine, followed by a linker and the Gal4 activation domain (AD amino acids 768-881) including the SV40 T-antigen. On the other side, the VL domain of the framework 7.3 was cloned N-terminal to the CL (lambda) -domain including the C-terminal cys 15 teine. The terminators are Gall1 terminator on the side of the heavy chain and Cyclin 1 terminator .on the side of the light chain. b) Perf ormance in an intracellular assay in 20 yeast For quantitative analysis of the performance of the antigen binders in scFv and Fab format in yeast (Fig.1 and 3) , S. cerevisiae strain Immuna LHB was co transformed with the isolated scFvs as Gal4-AD-fusion 25 constructs on the pESBA-Act2 vector and the bait vector containing the LexA-hPLK1-KD fusion by following a stan dard lithium acetate transformation protocol (Agatep, 1998). Following transformation, the cells were plated on drop-out plates (-Trp, -Ura, Glc). 2 ml overnight 30 cultures in drop-out medium (-Trp, -- Ura, Glc) were inocu lated in duplicates from streaks containing several colo nies and grown at 30 0 C. Cultures were diluted in 1 ml drop-out medium (-Trp, -Ura, Gal) to an optical density at 600 nm (OD600) of 0.7. They were grown at 30 OC for 35 5h. The assay was carried out as described above. c) Expression of soluble protein under re ducing conditions in yeast - 46 To compare the yields of soluble protein upon expression under reducing conditions, the scFv and Fab constructs, together with the hPLK1-KD-bait vector, as described above were expressed in the cytoplasm of yeast S. cerevisiae. They were transformed as described above into the yeast strain YDE173 and plated on-Trp,-Ura, drop-out plates 5 containing glucose. 5 ml overnight-cultures in drop-out medium (- Trp, -Ura, Glc) were inoculated from streaks containing several colonies and grown at 30*C. Cultures were diluted in YPAG to an optical density at 600 nm (OD600) of 0.5. They were grown at 30 *C for 7.5h. For the native cell extract, 2.5 ml cell culture normalized to an OD600 of 3 were harvested by 10 centrifugation, frozen in liquid nitrogen and subsequently resuspended in 75pl Y-PER (Pierce). The resuspended cell pellet was vortexed shortly and incubated slightly shaking at 20*C for 20 min. Subsequently insoluble and aggregated material were pelleted at maxi mal speed in an eppendorf centrifuge at 4*C for 10 min. The supernatant was mixed with loading dye, heated to 100*C for 5 min and separated on a 12% SDS-PAGE. The soluble 5 Gal4-AD-scFv fusion and the heavy chain part of the Fab fused to the Gal4-AD were visualized by western blot- ting via detection of the Gal4-moiety with an anti-Gal4- AD monoclonal mouse antibody (Santa Cruz Biotechnology) as primary and an anti-mouse peroxidase conjugate (Sigma) as secondary antibody and using a chemoluminescent substrate (Pierce) (Fig. 11). SDS-PAGE and western blotting procedures are well known to D a person of ordinary skill in the art. Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" and "comprising", will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any integer or step or group of integers or steps. 5 The reference in this specification to any prior publication (or information derived from it), or to any matter which is known, is not, and should not be taken as acknowledgement or admission or any form of suggestion that that prior to publication (or information derived from it) or known matter forms part of the common general knowledge in the field of endeavour to which this specification relates. 0

Claims (14)

1. A single chain variable fragment (scFv fragment) comprising a VH and a VL domain, characterized in that the VL domain comprises the framework sequence of the sequence of Seq. Id, No. 4 or a variant or derivative thereof and the VI domain comprises the framework sequence of the sequence of Seq. Id. No. I or a variant or derivative thereof,

2. The scFv of claim 1, being stable under reducing conditions. 3, The seFv of claim 1, wherein the framework is fused to a second protein

4. The scFv of claim 3, wherein said second protein is selected from the group consisting of GFP, enhanced blue fluorescent protein, enhanced yellow fluorescent protein, enhanced cyan fluorescent protein, a transcriptional activator or a DNA-binding domain.

5. The scFv of claim 4, wherein said second protein is a transcriptional activator.

6. The scFv of claim 5, wherein said transcriptional activator is a Gal4 activation domain,

7. The scFv of claim 4, wherein said second protein is a DNA-binding domain.

8. The scFv of claim 7, wherein said DNA-binding domain is a LexA- or Gal4 binding domain.

9. Use of the scFv of any one of claims I to 8 in intracellular applications.

10. Use of the scFv of any one of claims I to 8 for the creation of libraries for applications in a reducing environment. 11L Use of the scFv of any one of claims 1 to 8 in diagnostic and therapeutic application, target validation and gene therapy.

12. A nucleic acid encoding the scFv of any one of claims I to 8.

13. A vector comprising the nucleic acid of claim 12.

14. A host cell comprising the nucleic of claim 12.

15. The use of the nucleic acid of claim 12 in gene therapy.

16. The single chain variable fragment of claim 1 or the use of claims 9, 10, 11 or 15. or the nucleic acid of claim 12, or the vector of claim 3, or the host cell of claim 14, substantially as hereinbefore described with reference to the Examples and the accompanying drawings.