WO2005033130A2

WO2005033130A2 - Mutated ig binding domains of protein l

Info

Publication number: WO2005033130A2
Application number: PCT/GB2004/004174
Authority: WO
Inventors: Carolyn Enever; Ian Tomlinson
Original assignee: Domantis Limited
Priority date: 2003-10-03
Filing date: 2004-10-01
Publication date: 2005-04-14
Also published as: GB0323238D0; WO2005033130A3

Abstract

Isolated synthetic Ig binding domain of Protein L having enhanced binding affinity for an antibody VK domain, said binding affinity being enhanced over that of the wild type 131 binding domain of Protein L, said isolated Ig binding domain exhibiting at least one mutation of at least one amino acid residue corresponding to residues 25, 28, 31, 33-41, 45, 49, 50, 52-59, 61, 62, 76 and 78 of seq id no 1.

Description

TITLE: SYNTHETIC LG BINDING DOMAINS OF PROTEIN L

BACKGROUND OF THE INVENTION

The discovery of protein L in 1988 (Bjόrck, L. 1988 J Immunol 140 (4), 1194- 7) complemented the other widely used immunoglobulin binding reagents, protein A (Forsgren A, et. al,. 1966 J Immuno 97 (6), 822-7,) and protein G (Bjorck, L et. al, 1984 J Immunol 1984 133 (2) 969-74) for the purification, detection and immobilisation of antibodies. Protein L binds to antibodies from a wide range of species including approximately 50% of human and 75% of mouse antibodies through the VK region (Graille et. al, 2002 J Biol Chem 277 (49) 47500-6). Protein L has been reported to bind to human, rabbit, porcine, mouse and rat immunoglobulins. The location of this unique binding site in the framework region of the light chain of antibodies allows protein L to bind an alternative subset of antibodies compared to protein A and G and also to bind the range of antibody fragments such as scFv, Fab and single domains, F_v, disulphide bonded F_v, a Fab fragment and a F(ab )₂ fragment used in antibody engineering, if they have the correct K framework. Protein L has been found to bind to VK of subgroups I, III and IV (Nilson et al 1992 J Biol Chem 267 (4) 2234-9).

Nine species of B subunits have been found in various forms of Protein L Ig binding subunits. 5 so called B subunits were derived from strain 312 of Peptostreptococcus magnus and 4 were derived from strain 3316 of Peptostreptococcus magnus. The 312 strain protein L has five homologous antibody binding B subunits of 72-76 amino acids each with each successive subunit having a homology of 70-80% (Kastern W., Sjorb ng J., Bjorck L. (1992) J. Biol. Chem. 267: 12820-12825). The 3316 strain protein L has four homologous antibody binding subunits of 71-75 amino acids each with homology of 70-97% (Murphy et. al, 1994 Mol Microbiol 1994 72 (6) 911-20). WO 93/22438 reveals the deposit of ATCC 53516 comprising cDNA encoding the 3316 strain subunit C1. There is also a C^* sequence known in the art which is a derivative of the C1-C4 subunit sequences (Graille et. al, 2001 Structure 9 (8) 679-87) which has 66—97% to the C1-C4 domains. Each protein L Ig binding subunit has a similar secondary and tertiary structure consisting of a globular domain of a four stranded β sheet spanned by a central a helix preceded by a disordered N-terminus (Wikstrom M., Sjorbring J., Kastern W., Bjorck L., Drakenberg T., Forsen S (1993) Biochemistry 32: 3381-3386.). It was determined that the disordered N terminus was not involved in the Ig binding but that 61 amino acid residues, corresponding to residues 94-155 (that are numbered using the Wikstrom notation), were (Wikstrom et al 1995). Those residues 94-155 correspond to residues 20-81 of seq id no 1. Sequence id no 1 provides the amino acid sequence of residues 74-155 of Wikstrom sequence derived for B1 of Protein L of strain 312 of Peptostreptococcus magnus.

Commercially, protein L is produced in a recombinant tetrameric form. Due to the avidity effect of having four Ig-binding domains, the affinity of the tetrameric form of protein L is around 1.5 nM (Akerstrom et. al, 1989 J Biol Chem 264 (33) 19740-6) in comparison to 150 nM as determined by (Beckingham et. al, 1999 Biochem J. 340 (1 ) 193-9) and 160 nM as determined by (Kastern et al 1992 J Biol Chem 267 (18) 12820-5) for an individual Ig-binding domain using competition ELISA. Although tetrameric protein L is of sufficient affinity for purification, a higher affinity protein would prove useful in assays such as Enzyme-Linked Immuno-Adsorbant assays (ELISAs), Radioactive Immuno Assays (RIAs) and Western Blots. The use of a higher affinity reagent to either immobilise antibodies or antibody fragments on solid media or as a detection reagent (through conjugation to enzymes such as horseradish peroxidase) would allow detection of antigen-antibody binding events of low affinity. It would also allow detection where either the antigen or antibody is present at a low concentration due to poor expression, scarcity of reagent or miniaturisation. In this context, for a low affinity application, an affinity providing a binding signal that is increased over that of commercially available protein L would be desirable. Preferably the signal will be increased by at least two-fold. Suitably this can be higher. Any improvement up to 10 fold, 20 fold, 50 fold and higher is useful. Any improvement up to 100 fold is envisaged as forming a suitable improvement in binding signal over that of commercially available protein L. Suitably low affinity implies greater than one micromolar affinity. In this context low concentration implies a concentration below 1 mM, suitably below 1 μM, more preferably below 1 nM.

NMR spectroscopy has been used to analyse the effects of VK binding on individual amino acids in the B1 domain of protein L (pL domain) (Wikstrom et. al, 1995). This data suggested that the amino acids involved in binding were concentrated in the second β-strand, the α-helix and the loop connecting the α-helix with the third β-sheet.

The interaction between the V_u region of immunoglobulin and the B1 domain of protein L was analysed using heteronuclear NMR spectroscopy (Wikstrom et. al, 1995) by looking at the changes in chemical shift of individual amino acids upon complex formation. The residues divided into two groups in terms of their response to the addition of Ig; two thirds of the amino acids in the 61 amino acid domain remained unaffected by the addition of Ig, the other third showed changes upon binding. The resonances of 16 amino acids broadened beyond detection and a further 5 showed significant shift changes, thus implicating 21 residues as potentially being involved in binding.

The twenty-one amino acids that exhibited changes in chemical shift were A99, 1102*, Q109*, T110*. A111*. E112*, F113*. K114*, G115, A124, Y127*. A128, D129\ T130^*, L131*. K132^*, K133*. N135^*, G136\ N150^* and K152. (Using the numerical notation of Wikstrom). These amino acids correspond to residues A25, I28, Q35, T36, A37, E38, F39, K40, G41 , A50, Y53, A54, D55, T56, L57, K58, K59, N61 , G62, N76 and K78 respectively of seq id no 1. The asterisk indicates those amino acids whose resonances were broadened beyond detection. They are located mainly in the second β-strand, the C- terminal part of the α-helix and the loop connecting the α-helix with the third β- strand. According to Wikstrom (1995 reference quoted above) the changes observed in these amino acids could be due to their direct involvement in binding or alternatively conformational changes that take place upon binding. However, as the majority of the backbone amide shifts in the B1 domain show no change upon binding, large-scale conformational changes seem unlikely, indicating that the regions identified are part of the binding site.

Studies further assessing impact of mutations of protein L have been carried out by the group of David Baker. They mutated protein L outside the postulated Ig binding area of (Wikstrom et al 1995 J Mol Biol 250, 128-33). They presumably undertook mutations there as they felt the binding site was not to be touched if binding were to be retained. These mutants however were not produced as such with a view to testing their binding affinity, but for assessing the protein folding structure per se. Any impact such mutations may have had on binding affinity has not to our knowledge been assessed.

A small number of mutants with known identity have also been generated with one or two mutations brought about at amino acid residues within the postulated Ig binding domain as specified by Wikstrom (1995). None of these mutations have led to enhanced affinity of the B1 domain for VK domains either.

Phenylalanine 39 (F39) was replaced by tryptophan (F39W) by Beckingham (J.A. Beckingham , S.P. Bottomley, R.J. Hinton, B.J. Sutton and M.G. Gore (1997) Biochem. Soc. Trans. 25: 38S). This resulted in decreased affinity for human IgG. F39 (of the Beckingham notation), is identical to that of sequence id no 1.

Also tyrosine 53 (Y53) was mutated by Beckingham (Beckingham et al 1999). Y53 of the Beckingham notation, is identical to that of seq id no 1 and equates to tyrosine 127 of the Wikstrom notation) The Beckingham mutation was a substitution mutation of tyrosine by the structurally similar phenyl alanine. This (Y53F) mutation produced a dramatic reduction in affinity in the wild type protein L domain. Also Beckingham (Beckingham J.A., Housden N.G., Muir N.M., S.P. Bottomley and M.G. Gore (2001) 353, 395-401) used TNM

(tetranitromethane) as a chemical modification agent of protein L of strain 3316 of Peptostreptococcus magnus. TNM is known to modify predominantly tyrosine and cysteine fairly specifically at pH 7.5-8.0. As cysteine is absent in the B1 subunit, predominantly tyrosine residues would be expected to be targeted, thus allowing targeting of tyrosine 51 , 53 and 64 (again using the sequence id no 1 notation for residue numbers here). Tyrosine 51 and 53 are situated on opposite sides of the helix and tyrosine 64 is located on β strand 3.

Beckingham et al. (2001 ) generated a Y64W mutant and a Y53F, Y64W double mutant by means of site specific mutagenesis using PCR and subjected these mutants to TNM as well. Binding affinities of the resulting mutants were determined by ELISA against goat anti-(human Fc) specific IgG.

It appeared that tyrosine 64 was not available for TNM mutation. Kd determination of the Y64W mutant revealed little effect on the binding interaction with K chain. The Kd determination revealed a small increase in off-rate i.e. a small decrease in binding affinity.

The nitration of tyrosine 51 on the other hand led to a relatively small decrease in affinity for IgG. It is indicated by Beckingham et al. (2001 ) that 4 possible explanations for the behaviour of the tyrosine nitration mutants could be provided. Firstly, it could be because tyrosine 53 has a direct role in binding of the VK domain. Secondly it could be because a pKa alteration occurs due to the phenol group. Thirdly the bulky nitrate group might sterically hinder approach of the VK chain and fourthly secondary structural changes leading to a loss of binding cannot be excluded. The decrease in affinity upon mutation of Y53F however, indicates according to Beckingham that the Y53 residue is involved in stability of the complex of the B domain and the VK domain. It could indicate involvement of a hydrogen bond in the interaction of tyrosine with the Vκdomain. The experiments seem to suggest to Beckingham et al a two step procedure involving an initial encounter complex, followed by conformational change. Beckingham et al. however provide no further indication of any other residues besides Y53 that could be involved in those postulated processes.

Subsequently the co-crystal structure of a protein L domain (C* derived from strain 3316 of Peptostreptococcus magnus) in complex with Fab was solved (Graille et al 2001). Each protein L domain was found to bind two Fab molecules, the two Fab molecules being bound by two different interfaces at the V framework regions of the light chain. The residues apparently involved, seem well conserved in the different Peptostreptococcus magnus strains. The C* subunit appears to have similar affinity for VK domains (130 nM) as B1 exhibits. Using heteronuclear NMR it was ascertained most positions identified as being involved for B1 in the interaction with the V region of the light chain are also implicated in the first interface of C*. These positions are located on the β 2 strand and the o helix. There is, however, a discrepancy in the loop between the α helix and the β 3 strand, which could however be explained by mobility change upon complexation. Thus, according to Graille, it is unclear on the basis of this data whether there actually is any difference at those locations between the various species of subunits B and C*. Thus it remains unclear on the basis of the Graille publication, whether residues in that loop are involved or not in determining binding affinity either on their own or in combination with other residues located elsewhere in the binding domain Wikstrom postulated.

Graille revealed the existence of two binding sites within the Protein L Ig binding subunit. According to Graille none of the residues involved in forming hydrogen bonds in binding site 1 (T36, E38, K40, Y53 of seq id no 1) appeared to be involved in forming hydrogen bonds in binding site 2. Those latter residues were predominantly located on β strand 3 and the α helix (D55, T65, A66, D67, L68, G71 of seq id no 1 ). Six hydrogen bonds appear to be associated with the first site and six hydrogen bonds and 2 salt bridges appear to be involved in the second site.

Graille et al (2001 ) concluded that disruption of the hydrogen bond between the side chain of tyrosine 53 would occur upon phenylalanine mutation. Graille postulates this destroys binding at binding site 1 and that this brings about decreased affinity of the Y53F mutant. Graille et al. also created a D55A, Y64W mutant with mutations in binding site 2 to see whether that would have any effect on binding. This mutation however, seems to have no impact on the binding affinity. According to Graille et al. this indicates dominance of the first site over the second, with the dissociation constant apparently being at least one order of magnitude larger for binding site 1.

In the crystallised complex of Graille et al 12 amino acids in the protein L domain were apparently involved in the interaction, 7 of these being classed as core residues. These core residues are residues 35-40 and 53 (using the amino acid residue notation of seq id no 1). These core residues are also notably strictly conserved between 8 of the 10 known domains (see Figure 1 ). Graille indicates, based on the known sequence data of those various protein L domains, that the change of E38 to T38 that occurs in subunit B5 in comparison could weaken the interaction, but not disrupt it and likewise for the change of T36 to N36 in subunit C1 to the sequence of B1. Thus Graille implies that changes at these residues would not be expected to enhance affinity, but rather may bring about the opposite.

Also Graille indicates that the residue changes outside the structural core in subunit C2 and subunit C3 versus subunit C^* could be responsible for reduced affinity. Thus Graille implies residues outside the structural core impact on the affinity i.e., residues outside 35-40 and 53 influence the affinity. Specifically residues 49 and 52 are mentioned as potentially decreasing affinity, when those salt bridge forming residues of subunit C^* are replaced by Lys and Ala. Those particular substitutions allegedly disrupt the salt bridge and accumulate positive charges. Graille also mentions that protein L domains should retain all their hydrogen bonding interactions to retain binding. The data produced by Graille may go some way to explaining the previous NMR findings of Wikstrom regarding the relevance of residues in the loop connecting the a helix with the third β sheet. The NMR data suggested that residues in this area (D55-K59 and G62) were involved in the interaction and also as mentioned above were considered relevant to antibody Fab binding by Wikstrom. However, a comparison of the sequence data of the various protein L domains sequenced to date (see figure 1 ), which as Graille revealed do not exhibit any great variation in affinity, revealed only that some amino acids within that range were conserved, thus perhaps indicating a lack of relevance of some of residues D55-K59 and G62 on binding affinity.

In a later publication of Graille et al 2002 further elaboration on how the Wikstrom and Graille structures compare is provided. In these studies the D55A mutant is used in complex with a murine κ9 light chain, which is related to human κ4. This mutation ensures binding via binding site 2 is abolished, but retains binding via binding site 1.

Graille et al. (2002) addressed the fact here that their previous paper revealed binding to κ1 via two interfaces, which differed from the DSSA mutant κ4 binding complex in the actual amino acids involved on VK, however mostly these amino acids are of a similar nature. Notwithstanding the fact, that the V regions that were complexed with the protein L forms shared only 50% sequence identity among the contact residues, the interface was nevertheless largely conserved. A conserved core of residues from both partners apparently ensures the ability of protein L to maintain its interaction. Recognition is apparently ensured by only the backbone of the K chains and with the side chain of tyrosine 53 being responsible for β zipper contact.

Graille (2002) revealed that the two structures superimposed well. Basically the β 2 strand of protein L forms a β zipper structure antiparallel with the V ? strand of the K chain, via 3 hydrogen bonds between main chain atoms from both partners (protein L and the K chain). In the mouse κ9 construct with D55A 9 residues of protein L are involved, whereas for human κ1 with protein L, 11 residues are involved. Four hydrogen bonds out of 6 or 7 were common, with three mediating the β zipper and one being formed by the tyrosine side chain. Graille reveals that thus these residues are important for the recognition of a large population of VK light chains in a sequence independent manner. He also reveals how the hydrogen bonds from residues 40 and 38 are absent in one of the structures i.e. the κ9 binding structure in the case of mutant D55. He postulates that apparently this loss is compensated for by the N terminal part of the β2 strand. Graille also reveals in that structure that the E49 residue is not required to contact the light chain to ensure binding occurs.

Graille 2002 also provides a structural correlation between binding site 1 and 2 of protein L. The amino acids that are involved in binding site 1 , i.e. Phe39, Glu49, Tyr 53, Ala 37 and Arg 52 correspond to those involved in binding site 2, which are Tyr 64, Asp 55, Tyr 51 , Ala 66 and Arg 52 respectively. Binding site 1 has 4 hydrogen bonds in the β zipper region and binding site 2 has 2.

In summary, from the prior art it is thus clear, that a number of mutations of residues within the IgG binding domain have provided information about amino acid residues that are apparently implicated in binding. Also a number of residues have been mutated in Ig binding subunits Bή and C*. These mutations have either had no impact on binding affinity or have reduced or abolished binding affinity for IgG (or for a VK light chain fragment). There is considerable information available in the art as to the constitution and structure of the IgG binding domains of protein L, yet no information is available about the specific nature of any mutants or derivatives of such subunits exhibiting increased affinity towards a VK domain. More importantly none are described specifically exhibiting increased affinity towards multiple subclasses of VK domains in comparison to the affinity exhibited by the wild type B1 subunit. There is however, a need for such compounds to assist in purification of, isolation of, or screening for components consisting of or comprising VK domains of various subclasses. SUMMARY OF THE INVENTION

The instant invention for the first time provides specific examples of derivatives of subunits of protein L, (i.e. isolated synthetic Ig binding domains) having enhanced affinity over that of the wild type B1 binding domain for binding of an antibody VK variable region. The instant invention covers such derivatives with one or more mutations vis a vis the amino acid sequence of native B subunits. Specifically the derivative may comprise 2, 3, 4 or more mutations of residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 60, 62, 76 and 78 of seq id no 1. Numerous examples of isolated synthetic Ig binding domains with significantly improved affinity are now provided. As was apparent from the prior art, mutants of the B1 or C* domain had been made previously, which however either exhibited no change in affinity or exhibited decreased or abolished affinity for Ig binding, in particular for Ig binding of the VK domain. None of the prior art documents indicated which mutation(s), if any, of the residues located within residues 21-81 of sequence id no. 1 could significantly enhance affinity. On the contrary, the only specific teaching of examples of mutants provided within this area exhibited a decrease in affinity or little or no effect on affinity.

The data available to the skilled person on the nature of the binding site(s) was sometimes contradictory and insufficiently complete to allow prediction with any reasonable expectation of success with regard to which residues of known protein L Ig binding subunits could be mutated, without decreasing or abolishing Ig binding affinity, which residues could not be mutated because they decreased or abolished binding affinity and which mutations could result in enhanced affinity as opposed to decreased or abolished affinity.

The invention thus is directed at an isolated synthetic Ig binding domain of protein L having enhanced binding affinity for an antibody VK domain, said binding affinity being enhanced over that of the wild type B1 binding domain of protein L, wherein mutation has occurred of at least one of the amino acids corresponding to residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 60, 62 76 and 78 of seq id no 1. Sequence id no 1 amino acid residues 21-81 correspond to amino acid residues with sequence numbers 95-155 of the Wikstrom representation of B1. This is the sequence segment postulated by Wikstrom (1995) to be the Ig binding domain of subunit B1.

A suitable embodiment of the domain according to the invention will have binding affinity for Ig exceeding that of wild type B1 , wherein wild type B1 comprises residues 21-81 of the amino acid sequence of seq id no 1 as Ig binding domain. This sequence was derived for wild type B1 of Peptostreptococcus magnus strain 312 and corresponds to the sequence data provided by Wikstrom et al (1995) for subunit B1 of Peptostreptococcus magnus. The full length wild type B1 subunit is known to also have an additional 20 amino acids preceding the sequence of residues 21-81 of sequence id no 1 as N-terminal sequence, however that N-terminal sequence has been revealed in the prior art not to have been considered part of the Ig binding domain of B1. The isolated synthetic Ig binding domain according to the invention will thus also exhibit increased binding over such a full length B1 sequence.

As is apparent from the prior art, there are a number of homologous subunits that have been isolated from Peptostreptococcus magnus, which exhibit a large degree of homology to the amino acid sequence of wild type B1. Specifically the subunits exhibit a large degree of homology between the residues corresponding to the sequence section of residues 21-81 of seq id no. 1. Any of these other subunits may serve as starting point to produce an isolated synthetic Ig binding domain of protein L according to the invention. From figure 1 the alignment of these known sequences is apparent, thus revealing to the skilled person which amino acids of the respective Ig binding subunits of protein L correspond to the residues mentioned in the specification for subunit B1 in sequence id no 1. It is thus simple for the skilled person to ascertain, when looking at a derivative of any of the sequences B1, B2, B3, B4, B5, C1, C2, C3, C4 or C^* or of any other protein L )g binding subunit sequences, (in particular any derived from protein L of Peptostreptococci or Streptococci) which amino acid residues correspond to those indicated as mutatable according to the invention in sequence id no 1 , and which thus can suitably be mutated to provide an embodiment of isolated synthetic Ig binding domain according to the invention. The invention does not cover any sequences which are identical to naturally occurring known protein L Ig binding subunits or domains. The invention thus does not cover any of the sequences as such of figure 1.

Preferably an isolated synthetic Ig binding domain of protein L according to any of the embodiments of the invention mentioned above will exhibit at least 60% identity, with amino acid residues corresponding to the location and identity of amino acid residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 in seq id no 1. Other suitable embodiments exhibit at least 70% identity. They exhibit 1-8 mutations within the aforementioned group of residues. They may exhibit at least 79% identity, at least 83% identity, at least 87% identity or at least 92% identity whilst always containing at least one mutation when compared to those residues of sequence id no 1. "Homology searches can be performed using the BLAST algorithm contained in the Wisconsin Sequence Analysis Package (Genetics Computer Group, Unive.- sity Research Park, 575 Science Drive, Madison, Wl 5371 1 )".. Thus, a derivative with 1 to 10 mutations within the aforementioned residues is comprised within the scope of the invention. Also envisaged are embodiments exhibiting 1-5 mutations within the aforementioned group of residues. Also embodiments with 1-4 or 1-3 mutations are covered. Thus a derivative selected from a group of mutants consisting of an amino acid sequence with 1 , 2, 3, 4, 5, 6, 7, 8, 9 or 10 mutations in the aforementioned group of residues form suitable embodiments.

The isolated synthetic Ig binding domain according to the invention in any of the embodiments as disclosed, will suitably have multiple mutations compared to the wild type sequence from which it is derived i.e. its starting point. A suitable embodiment will have multiple mutations when its amino acid sequence is aligned with sequence 21-81 of seq id no 1 and is compared at positions corresponding to positions 25, 28, 31 , 33-41 , 45, 50, 52-59, 61 , 62, 76 and 78 of sequence id no. 1. The sequences encoding isolated synthetic Ig binding domains according to any of the embodiments of the invention described may have double, triple or quadruple mutations vis a vis any of the native protein L sequences that form the starting point within the residues corresponding to those positions 25, 28, 31 , 33-41 , 45, 50, 52-59, 61, 62, 76 and 78 of sequence id no 1. Isolated synthetic Ig binding subunits according to the invention with 2, 3, 4 or more mutations, at positions of the native sequence of the starting point sequence corresponding to positions 25, 28, 33-41, 45, 50, 52-59, 61 , 62, 71 and 78 of the protein B1 sequence are suitable embodiments of the instant invention. Suitably an isolated synthetic Ig binding domain according to the invention in any of the preceding embodiments comprises 1 , 2, 3, 4, 5 or 6 amino acid residues that are mutated in comparison to the amino acid residues 25, 28, 31 , 33-41, 45, 49, 50, 52-59, 61 , 62, 76 and 78 in seq id no 1. These will exhibit percentages of identity to those residues of seq id no 1 of 96.3%, 92.6%, 88.9%, 85.2%, 81.5% and 77.8% respectively. An isolated synthetic Ig binding domain according to the invention will usually exhibit less than 15 amino acid mutations in comparison to the residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76, and 78 of sequence id no 1. An embodiment with less than 15 mutations will exhibit over 48, 1% identity to those residues of sequence id no 1.

Either in combination with one or all of the characteristics provided for in the preceding embodiments or on its own, an isolated synthetic Ig binding domain of protein L according to the invention will have enhanced binding affinity for an antibody V/c domain, said binding affinity being enhanced over that of the wild type B1 binding domain of Protein L, said isolated Ig binding domain exhibiting at least one mutation of at least one amino acid corresponding to amino acid residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 in seq id no 1 , said isolated synthetic Ig binding domain further exhibiting at least 50% identity, preferably at least 59% identity, preferably at least 62% identity, more preferably at least 67% identity, most preferably at least 73% identity and most preferably more than 89% identity when compared to and aligned with the amino acid sequence 21-81 in seq id no 1. A suitable embodiment will comprise 1-30 mutations when aligned and compared to the amino acid sequence 21-81 in seq id no 1. Another suitable embodiment will comprise 1-24 mutations when aligned and compared to the amino acid sequence 21-81 in seq id no 1. An embodiment with 1-19 mutations when aligned and compared to the amino acid sequence falls within the scope of the invention as does an embodiment with 1-15 mutations, as does an embodiment with 1-6 mutations.

Suitably the Ig binding domain according to the invention exhibiting increased binding affinity as described in any of the embodiments of the invention will either consist of or comprise an amino acid sequence corresponding to residues of seq id no 1 mutated at least at any of the locations corresponding to 25, 28, 31 , 33-41 , 45, 49, 50, 52-54, 61 , 62, 76 and 80 of seq id no 1, said domain not having or comprising any Ig binding domain sequence depicted in figure 1.

The isolated synthetic Ig binding domain of protein L according to any of the embodiments of the invention can suitably be derived from the amino acid sequence of any Ig binding subunit of protein L of Peptostreptococci or Streptococci such as from Peptostreptococcus magnus. Suitably such an isolated synthetic Ig binding domain of protein L can thus be derived from domains of subunits B1 , B2, B3, B4, B5, C1, C2, C3, C4 or C^*. The B subunits may be derived from Peptostreptococcus magnus strain 312 and the C subunits may be derived from the Peptostreptococcus magnus strain 3316. The sequence data for these sequences from which the isolated synthetic Ig binding domain may be derived by way of mutation are presented in figure 1. The starting point of such a mutant may be the amino acid sequence or the corresponding encoding DNA sequence encoding a native subunit or Ig binding domain comprised within such subunit selected from any of the group consisting of subunits B1 to B4 and C1 to C*. A preference exists for those mutants derived from sequences of B1-B4, with an even larger preference for those derived from B1-B3. All the B domains appear to have equal affinity for κ1 light chains. Note the Ig binding domains within the various subunits correspond to those residues corresponding to residues 21-81 of sequence id no 1 , as is apparent from figure 1. Where the term starting point or derived from is employed, this does not necessarily imply that the isolated synthetic Ig binding domain according to the invention has to be made by mutating the amino acid sequence or corresponding DNA encoding sequence encoding a native protein L Ig binding subunit or Ig binding fragment thereof. The sequence encoding the Ig binding domain according to the invention may readily be made de novo in manners well known to the skilled person. The sequence may be made by using the nucleic acid sequence or amino acid sequence encoding a subunit whose sequence the isolated synthetic Ig binding domain exhibits the closest identity to when considering the residues corresponding to residues 21-81 of seq id no 1 and/or when considering residues 25,28,31 ,33-41 ,45,49,50,52-59,61 ,62,76 and 78 of seq id no 1. However it may also be possible to use a sequence as starting point in the mutation process that is further removed in identity from the native protein sequence that ultimately exhibits the closest identity to the sequence of the Ig binding domain according to the invention. By way of example by physically starting with a nucleic acid sequence encoding an amino acid sequence identical to that of B5, introduction of 8 mutations could create a B5 mutant equal in identity to B1 , by introducing another 3 mutations one could produce a mutant differing at only two positions from B2. In such a case even though the starting point of the mutations was a sequence identical to B5, the starting point as meant here or the sequence from which the sequence according to the invention is considered to be derived would be B2 and the number of mutations would be 2. Whereas the starting point of the mutations was the sequence of B5 into which 10 mutations has actually been introduced. Where the invention is thus claimed to be derived from a particular sequence it is meant to exhibit the closest identity at amino acid sequence level to that sequence and not necessarily to have been physically derived from that sequence. Suitably an embodiment according to the invention will exhibit no more than 15 mutations when comparing the residues corresponding to residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 of sequence id no 1 and also in comparison to aligned corresponding sequences of B2, B3, B4, B5, C1 , C2, C3, C4, and C^* of figure 1. The isolated synthetic Ig binding domain of protein L according to the invention will preferably exhibit increased affinity for the binding of a VK domain, suitably of sub classes K I, II, III and IV. The increase will be apparent versus that of one or more native Ig binding domains such as any of protein L subunits derived from Peptostreptococcus magnus, for example any of domains B1-B5 and C1-C*, in particular versus that of B1. It is shown that in particular binding of sub class VK I and sub class K III can be enhanced in the isolated synthetic Ig binding domains according to the invention, most particularly that of sub class I is improved. In one embodiment the VK domain comprises a VK germline gene segment wherein the germline gene segment is selected from DPK4, DPK8, DPK9 or DPK22 (see WO99/20749).

The instant invention is directed not only at binding affinity enhancement for binding to K chains of human origin, though that is preferred, but also at binding to antibody of animal origins such as murine. It is one of the advantages of protein L that it recognises a large number of K light chains of diverse origin and subclasses and is not restricted to merely one specific type of light chain. It is this characteristic that the isolated synthetic Ig binding domain of protein L preferably also retains. Suitably therefore the isolated synthetic Ig binding domain will recognise a multitude of K sub classes and more suitably thus also a multitude of K light chain germline gene segment. Of particular interest are isolated synthetic Ig binding domains according to the invention which exhibit increased affinity towards germline gene seqment DPK9. Also of interest are isolated synthetic Ig binding domains according to the invention which exhibit enhanced binding affinity to K chains of human or non-human animal origin e.g. murine or other rodent.

A preferred embodiment of an isolated synthetic Ig binding domain of protein L according to the invention in any of the other embodiments described in the preceding section will further retain at least the potential to form at least 4 hydrogen bonds with a VK domain. Preferably the Ig binding domain will be able to form more than 4 hydrogen bonds with one (or more) VK domain(s) of choice. For example, an embodiment that is capable of formation of 5, 6 or even 7 hydrogen bonds with one and the same VK domain is envisaged. Suitable embodiments of the invention comprise any or all of the characteristics of the embodiments of the invention described in the preceding sections with 4, 5, 6 or 7 hydrogen bonds being present, when the isolated synthetic Ig binding domain according to the invention binds to a VK domain. Specifically, such is envisaged when binding of the isolated synthetic Ig binding domain in any of the preceding embodiments of the invention is possible to a K light chain of any of the sub classes K I, II, III or IV, more specifically to any of the germline gene segments representative of any of the aforementioned sub classes. Suitably this may be ascertained if binding occurs to any of the following germline gene segments DPK1 , DPK4, DPK8, DPK9 or DPK22. Suitable alternative gene segments are available to the skilled person (see references cited elsewhere in this description, which are incorporated by reference).

In any of the embodiments of the invention the isolated synthetic Ig binding domain according to the invention may be mutated vis a vis a native sequence of an Ig binding domain of protein L, specifically of protein L of Peptostreptococcus magnus, e.g. strains 312 or 3316 by means of substitution mutation and/or chemical modification. Preferably substitution mutation occurs. Any modification exhibited by the isolated Ig binding domain according to the invention in any of the embodiments described which, in comparison to the amino acid sequence corresponding to a native sequence of a Ig binding domain of protein L, results in the formation of an additional hydrogen bond between the Ig binding domain according to the invention and a Vκdomain, which hydrogen bond would not be present when the native sequence from which the domain according to the invention was derived and/or exhibits the closest identity to, binds to the same VK domain, is also covered by the invention. Most particularly embodiments wherein that native sequence corresponds to B1 , B2, B3, B4, B5, C1 , C2, C3, C4 or C*, more specifically wherein that native sequence comprises a sequence presented in Figure 1 are of interest. A modification in such an embodiment according to the invention will preferably be brought about on at least one of the amino acid residues corresponding to residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 of sequence id no 1.

The affinity of the isolated synthetic Ig binding domain of protein L is preferably increased more than two times over that of wild type B1 , preferably increased more than 4 times over that of wild type B1. Even more preferably the affinity will be increased at least eight times over that of the wild type B1. Suitably wild type B1 will comprise the amino acid sequence provided in seq id no 1. Preferably it will consist of the amino acid sequence presented by Wikstrom et al as involved in binding IgGs. In a suitable embodiment of the isolated synthetic Ig binding domain of protein L according to the invention, the affinity will increase such that KD of binding to the VK domain Vκl is lower than 100nM, preferably this will even be lower than 50nM. In a more preferred embodiment the increase will be such that the K_D is preferably below 5nM for binding of the isolated synthetic Ig binding domain according to the invention to VK domain Vκl, most preferably below 1.5nM. The KD can be determined in a number of methods of which Surface Plasmon Resonance measurement (as is detailed in the examples) is a suitable example. Suitably the BIA core process can be used for such measurement. Alternative processes for determining influence on binding affinity are available to the skilled person. Examples of such methods are stop-flow fluorimetry or ELISA. Details of how to use these processes have been detailed for example in a number of the references cited in the introduction to this specification. By way of example Wikstrom, Beckingham and Graille have all used such processes and provided the appropriate references in the papers quoted elsewhere in this description. The increased affinity can be determined by Surface Plasmon Resonance by using immobilised target binding partner of choice and determination of a) increased binding which is expressed as resonance units and/or b) reduced off-rate and/or c) K_D. The increased affinity, by which ever means measured, can be determined as either increased binding or as expressed by a reduced off-rate. As mentioned before, there are a number of processes available to the skilled person to determine the off-rates in complex formation competitive assays. In view of the equivalence in the nature of the binding sites for VK light chains exhibited by the 10 known sequences encoding Ig binding sub units of protein L and the knowledge of the sequence details of these proteins and the knowledge of the instant invention as presented herein, the skilled person can easily extrapolate the residue numbering used here (based on seq id 1 , which is the sequence derived from Peptostreptococcus magnus sub unit B1 of strain 312) to determine corresponding mutants based on other sub units of protein L that exhibit Ig binding affinity. Preferred points of mutation are those situated at positions corresponding to the amino acid residues T36, A37, E38, K40, T47, E49, A52, D55, T56, K58 and K59 of seq id no 1 in an isolated synthetic Ig binding domain according to the invention. In particular mutations occurring at one or more positions corresponding to T36, E38, A52 and T56 are suitable embodiments of the invention. In a preferred embodiment the isolated synthetic Ig binding domain of protein L according to the invention will exhibit substitution of a bulkier amino acid residue at least at one of those corresponding positions. Mutations may occur either by means of chemical modification of the amino acid or by replacing the amino acid residue by a different amino acid residue, more preferably of a bulkier nature. Specifically a number of substitution mutation possibilities are presented here for an Ig binding domain according to the invention. Embodiments of isolated synthetic Ig binding domains of the invention may further comprise amino acid residues corresponding to positions of other native Ig binding protein L subunits such as any of protein L subunits of Peptostreptococcus magnus and/or subunits B1 , B2-B5, C1-C4 and C^*. The mutations are those of the type corresponding to T36I, T36Q, T36W, T36E, T36H, T36N, T36S, T36V, A37E, A37V, E38K, E38G, E38R, E38L, E38P, E38N, E38S, E38T, E38V, E38A, E38Q, K40Q, K40I, K40R, T47A, T47I, T47M, T47V, T47S, T47R, T47L, E49K, A52R, A52W, A52Y, A52R, A52G, A52K, A52L, A52Q, A52T, A52V, D55G, D55N, T56I, T56L, T56A, T56N, T56V, T56S, T56H, K58M, K58R, K59A, K59I, K59Q and K59R of seq id no 1. In the case of a mutation occurring at a position corresponding to T47 suitably that mutation is a substitution mutation which introduces an amino acid L, A, R, M, I, V, S or L more suitably A, M, V, L or I. A bulkier amino acid than T may be introduced. Also I or A may alternatively be introduced with good effect. Particularly a mutation at position T47, was considered to be responsible for enhancing expression of the resultant sub unit or Ig binding domain in E-coli or yeast. The invention further covers any isolated synthetic Ig binding domain of protein L according to any of the embodiments described herein which exhibit enhanced expression in E-coli or yeast over that of wild type B1. The invention also covers any of the embodiments of isolated synthetic Ig binding domain of protein L described herein wherein the solubility is enhanced in respect to that of the wild type B1 subunit in the corresponding solute under corresponding conditions. Specifically an isolated Ig binding domain of protein I according to the invention comprising a sequence with a mutation at a position corresponding to that of position T47 of sequence id no 1 exhibits solubility that is enhanced in respect to that of the wild type B1 subunit in the corresponding solute under corresponding conditions.

Another preferred isolated synthetic Ig binding subunit according to the invention comprises or consists of a sequence with a mutation at a position corresponding to T56 of seq id no 1. Suitably such a mutation is a substitution mutation, wherein an amino acid such as A, N, S, V, L, H or I, suitably a bulkier amino acid than T such as I. Most suitably V is introduced as substituent amino acid.

Isolated synthetic Ig binding domain of protein L according to any of the preceding embodiments of the invention, comprising or consisting of an amino acid sequence wherein a mutation occurs at an amino acid position corresponding to T36 of sequence id no 1 in B1 of protein L form embodiments of the invention. The mutation may be a substitution mutation introducing an amino acid such as N, Q, W, E or I. Suitably a bulkier amino acid such as I, Q or E is introduced at a position in the sequence of the isolated synthetic Ig binding domain according to the invention at a position corresponding to that of T36 of sequence id no 1. A particularly suitable embodiment of an isolated synthetic Ig binding domain of protein L according to any of the preceding embodiments comprises or consists of an amino acid sequence, wherein a substitution mutation of a T to I occurs.

It was found that an isolated synthetic Ig binding domain of protein L according to any of the preceding embodiments, comprising or consisting of an amino acid sequence wherein a mutation occurs at an amino acid position corresponding to A52 of the sequence id no 1 in B1 of protein L also forms a specific embodiment of the invention of interest. The mutation may be a substitution mutation wherein the amino acid is substituted by Y, R or W, most suitably R or W.

Another embodiment of an isolated synthetic Ig binding domain of protein L in accordance with any of the preceding embodiments, is provided comprising or consisting of an amino acid sequence wherein a mutation occurs at an amino acid position corresponding to E38 of the sequence id no 1 in B1 of protein L. The mutation may be a substitution mutation of E38 wherein the substitution occurs by V, A, T, L, G, Q, K, more preferably K, G or T, most preferably K or T.

As revealed earlier multiple mutations in the amino acid sequence of an isolated synthetic Ig binding domain according to the invention are envisaged, by way of example such may be provided when the sequence of the Ig binding domain according to the invention is aligned to sequence id no 1 and compared at positions corresponding at least to one or more of amino acid positions corresponding to T36, E38, T47 and T56 of the sequence id no 1. Suitably such an embodiment of the invention may also include a mutation at a position corresponding to A52 or at one or more of positions corresponding to T36, E38, A52 and T56 of sequence id no 1.

Further examples of multiple mutations in the amino acid sequence of an isolated synthetic Ig binding domain according to the invention that form suitable embodiments are provided, when the sequence of the Ig binding domain according to the invention is aligned to sequence id no 1 and compared, revealing a combination of mutations that are present at least at amino acid positions corresponding to positions T36 and T56 of sequence id no 1. Specific examples of such embodiments suitably comprise mutations corresponding to T36I or T36Q and T56I or T56V of sequence id no 1. Quadruple combination mutations such as (T36I E38K A52 R T56I), (T36I E38K A52R T56V), (T36Q E38R A52R T56I), (T36Q E38L A52R T56V), specifically (T36I E38K A52 R T56I) are preferred embodiments of such multiply mutated sequences comprised within the sequence of an isolated synthetic Ig binding domain according to the invention. Alternatively or also, the mutations may at least occur at a combination of amino acid positions corresponding to T36 and T56 with E38 in sequence id no 1. A particularly suitable embodiment of this type comprises a sequence comprising the following mutations at positions corresponding to those of (T36I E38K T56I) of sequence id no 1. Alternatively or also, the mutations may at least occur at a combination of amino acid positions corresponding to T36 and E38 in sequence id no 1. Suitably the mutation at a position corresponding to T36 of sequence id no 1 is I or Q, preferably I. Suitably the mutation at a position corresponding to E38 of sequence id no 1 is K or T, preferably K. Alternatively or also the mutations may at least occur at a combination of amino acid positions corresponding to A52 and T56 in sequence id no 1. Suitably the mutation at a position corresponding to A52 of sequence id no 1 is R or W, preferably R. Suitably the mutation at a position corresponding to T56 of sequence id no 1 is I or V or L.

Particularly interesting embodiments of an isolated synthetic Ig binding domain of protein L according to the invention comprise a sequence wherein amino acids corresponding to positions (T36, E38, T47, A52, T56) of sequence id no 1 are present as follows: (T36I, E38T, T47S, A52L, T56L), (T36E, E38T, T47V, A52Y, T56I), (T36H, E38Y, T47V, A52Y, T56V) or (T36Q, E38R, T47L, A52R, T56V), preferably (T36I, E38K, T47S, A52R, T56I) (T6I, E38K, T47V, A52R, T56I) or (T36I, E38K, T47L, A52R, T56I). Additionally any of the preceding embodiments of an isolated synthetic Ig binding domain of protein L according to the invention are also provided, wherein a mutation occurs in the amino acid sequence of the Ig binding domain according to the invention at a position corresponding to Y53F of the seq id no 1, with the proviso said mutation allows hydrogen bond formation of the type created by Y53 when B1 of wild type protein L binds Ig i.e. a stabilising hydrogen bond having the same function as that of Y53. As was apparent from the prior art, mutation of an amino acid corresponding to this position that abolishes the hydrogen bond of the side chain of tyrosine abolishes binding affinity of the domain to Ig. Therefore, if an isolated Ig binding domain according to the invention is to comprise a mutation at a position corresponding to Y53 of sequence id no 1 it should retain a form of hydrogen bonding at that position. Preferably the embodiments of the isolated synthetic Ig binding domains according to the invention are also provided, wherein no mutation occurs in the amino acid sequence of the Ig binding domain according to the invention at a position corresponding to Y53 of the seq id no 1 in B1. In other words in an embodiment of the invention the isolated synthetic Ig binding domain according to the invention in any embodiment described elsewhere herein will contain Y at a position corresponding to Y53 of sequence id no 1.

Also any of the preceding embodiments of an isolated synthetic Ig binding domain of protein L according to the invention are also provided, wherein no mutation occurs in the amino acid sequence of the Ig binding domain according to the invention at a position corresponding to Q35 of seq id no 1. In other words in an embodiment of the invention, the isolated synthetic Ig binding domain according to the invention in any embodiment described elsewhere herein will contain Q at a position corresponding to Q35 of sequence id no 1.

Also any of the preceding embodiments of an isolated synthetic Ig binding domain of protein L according to the invention are also provided, wherein no mutation occurs in the amino acid sequence of the Ig binding domain according to the invention at a position corresponding to F39 of the seq id no 1. In other words in an embodiment of the invention the isolated synthetic Ig binding domain according to the invention in any embodiment described elsewhere herein, will contain F at a position corresponding to F39 of sequence id no 1 , alternatively the substitution of F by W at that position is also envisaged as a suitable embodiment.

Finally an isolated synthetic Ig binding domain of protein L according to any of the preceding embodiments whose sequence when aligned and compared to sequence id no 1 , additionally comprises a mutation in one or more positions corresponding to positions A37, K40, E49 or D55 of seq id no 1 is also specifically provided.

An embodiment of the isolated synthetic Ig binding domain according to the invention may additionally comprise an N-terminal sequence of at least 1-25 amino acids preceding its actual Ig binding sequence. A suitable embodiment of that N terminal preceding sequence is provided in seq id no 2 which reveals the sequence from amino acid 6 that is present as N terminal section to wild type subunit B1 from Protein L of Peptostreptococcus magnus strain 312. Suitably any natively occurring N terminal sequence of a protein L subunit, that precedes the sequence corresponding to residues 21-81 of sequence id no 1 in the native subunit may be present as N terminal sequence preceding the Ig binding section of an isolated synthetic Ig binding domain according to the invention. Suitably the native N terminal sequence will correspond to that present in the native subunit to which the Ig binding section of the isolated synthetic Ig binding domain of the invention exhibits closest identity. Suitably the N terminal sequence is selected from any of Protein L Ig binding subunits derived from Peptostreptococcus magnus, such as B1 , B2, B3, B4, B5, C1 , C2, C3, C4 or C*. The identity of such N terminal sequences is available from the prior art and also from Figure 1.

The isolated Ig binding domain according to the invention may, in one embodiment of the invention, have the same length as a full length native Ig binding subunit of protein L. The length of the isolated Ig binding domain according to the invention may however alternatively be shorter. The length may for example be the same as that of the Ig binding domain of a native Ig binding subunit of protein L. The length may thus be that of an amino acid sequence which corresponds to the length of that section of a native Ig binding subunit of protein L corresponding to residues 21-81 of sequence id no 1 , which corresponds to the Ig binding Wikstrom fragment. Thus the length of the Ig binding domain according to the invention may be the length of a native protein L Ig binding subunit minus the N-terminal residues of the subunit, that are not involved in Ig binding, preceding residue 20 of the fragment corresponding to residues 21-81 of sequence id no 1. The isolated synthetic Ig binding domain according to the invention may however, also have a length corresponding to a truncated version of the fragment of amino acids 21-81 of sequence id no 1 of protein L subunit B1 or a truncated version of the corresponding Wikstrom Ig binding sequence derived from any of the other protein L subunits. The isolated synthetic Ig binding domain of protein L according to any of the above embodiments of the invention may thus suitably have a length of at least 25 amino acids, suitably at least 30 amino acids, suitably at least 40 amino acids. The isolated synthetic Ig binding domain according to the invention may suitably comprise or consist of amino acid residues corresponding to amino residues 35-62 of seq id no 1 as minimum structure with at least one mutation at any of the amino acid residues corresponding to 35-41 , 45, 49, 50, 52-59, 61 , 62 of seq id no 1. The isolated synthetic Ig binding domain of protein L in an embodiment of the invention will preferably be shorter than 81 amino acids, suitably shorter than 62 or 61 amino acids. Preferably the sequence of an isolated Ig binding subunit according to the invention will further exhibit identity of at least 60% between amino acid residues 35-41 , 45, 49, 50, 52-59, 61 , 62 of seq id. no. 1 when aligned with a segment of corresponding length of a native subunit sequence of protein L. Most preferably, such percentage of identity is calculated excluding amino acid residues corresponding to 35-41 , 45, 49, 50, 52-59, 61 , 62 of sequence id no 1 , such percentage may however also be calculated including those residues too. The percentage of identity may readily be higher and even may be 100% when compared to a native protein L subunit, most preferably such identity is ascertained by comparison to B1. A suitable embodiment of the invention in addition exhibits at least 50% identity, suitably at least 80% identity, in comparison to the amino acid residues 21-81 of sequence id no 1 , when the amino acid sequence of the isolated synthetic Ig binding domain of the invention is aligned to those residues 21-81 , whereby percentage identity is calculated for the length of the sequences exhibiting overlap (as is common for sequence alignment calculation). If by way of example the Ig binding domain according to the invention corresponds to section 31-70 of the sequence id. no 1 , then 50% identity means 20 of the 40 residues must be identical.

If by way of example the Ig binding domain comprises 100 amino acid residues, only the number of residues overlapping with a section or the full length of seq id. no. 1 for residues 21-81 are used in the calculation.

A suitable embodiment according to the invention, besides exhibiting at least one mutation and an identity of at least 60% at amino acid residues corresponding to 35-41 , 45, 47, 50, 52-59, 61 , 62 of seq id no 1 also exhibits identity of at least 60%, maybe even 100% in the sequence corresponding to the remaining sections of amino acid residues 21-81 of sequence id. no. 1.

Suitably the length of an isolated synthetic Ig binding domain of protein L according to the invention has a length of at least 30 amino acids, suitably at least 40 amino acids.

Usually it will have a length shorter than 81 amino acids in length. It may have a length shorter than 61 amino acids in length.

Suitably an embodiment exhibiting 0-24 mutations in the Ig binding subunit outside the residues corresponding to 35-41 , 45, 49, 50, 52-59, 61 , 62 of sequence id no 1 falls within the scope of the invention, as such will exhibit at least 60% identity with subunit B1 outside the residues corresponding to 35- 41 , 45, 49, 50, 52-59, 61 , 62 of sequence id no 1.

The isolated synthetic Ig binding domain of protein L according to the invention may also be provided with a linker sequence, for example, when the domain is to be used in multimeric form. Suitably the length of that linker will correspond to the length of linker present in native protein L. Preferably, the length of linker will be that present in the subunit of the corresponding protein L from which the isolated synthetic subunit has been derived or exhibits closest identity to at amino acid level. In the case of closest identity being wild type B1 for example the length of that linker will be 14 amino acids in length. For B2, B3 and B4 this will be 10, for B5 11. Quite suitably the linker may have the same identity as any of the naturally occurring linkers of the known variants of protein L.

As is described in the prior art multiple Ig binding subunits are present in native protein L, and this multiplicity of subunits enhances the affinity for binding Ig tremendously. There is a so called avidity effect brought about by the combination of subunits into one compound. The isolated synthetic Ig binding domain of the instant invention may thus suitably also be used in combination with at least one other Ig binding subunit or Ig binding domain. Note an Ig binding subunit is the full length sequence that corresponds to amino acids 1-81 of sequence id no 1 in native Ig binding subunits from protein L. The isolated synthetic Ig binding domain according to the invention may thus form part of a polypeptide, consisting of multiple Ig binding subunits. This polypeptide according to the invention will preferably exhibit enhanced Ig binding affinity over that of a single Ig binding component comprised therein. It will preferably exhibit enhanced Ig binding affinity over that of a wild type B1 subunit. The polypeptide according to the invention may comprise one or more synthetic Ig binding domains according to any of the embodiments of the invention described above. The polypeptide according to the invention may further comprise at least one native Ig binding subunit or at least one Ig binding domain, wherein such an Ig binding domain corresponds to amino acid residues 21-81 of sequence id no 1 of any native protein L Ig binding subunit. Thus a polypeptide according to the invention may comprise a mixture of synthetic and native Ig binding domains and/or subunits or may consist solely of a multiplicity of sequences corresponding to those of isolated synthetic Ig binding subunits according to the invention. Suitably the polypeptide will comprise as Ig binding subunit components 1-5 domains in total, wherein at least one will be an isolated synthetic Ig binding domain in accordance with any of the embodiments described in the preceding sections. A suitable embodiment will thus be formed by a polypeptide comprising as Ig binding components one sequence fragment corresponding to an isolated synthetic Ig binding domain according to any of the embodiments of the invention described herein and further comprising 1 , 2, 3 or 4 additional Ig binding domains. Those additional Ig binding domains may be native Ig binding subunits and/or have native Ig binding domains. The polypeptide according to the invention may however also comprise more than one sequence corresponding to that of an isolated synthetic Ig binding domain according to the invention. Suitably in any embodiment of the polypeptide according to the invention the total of Ig binding domains does not exceed 5. An effective embodiment consists of a polypeptide comprising either 4 or 5 Ig binding domains, as these are comparable to the known native structures of protein L. The polypeptide according to the invention may comprise sequences corresponding to those of 2-5 isolated synthetic Ig binding domains according to the invention, suitably 3, 4 or 5, preferably 4 or 5.

The native Ig binding domains are suitably selected from protein L subunits. Suitably the native Ig binding subunits are those with an amino acid sequence identical to those occurring in nature in protein L. The Ig binding subunits of protein L found in Peptostreptococcus magnus are preferred embodiments of such native Ig binding subunits. As described above such subunits are well known in the art, as are their sequences. Figure 1 provides sequence details of suitable embodiments of Ig binding domains of those subunits. Suitable embodiments of so called native subunits are subunits B1 , B2, B3, B4, B5, C1 , C2, C3, C4 and C^*. These are suitably derived from Peptostreptococcus magnus strains 316 and 3312 (see references cited in the preceding text). In particular a polypeptide according to the invention will comprise as additional Ig binding domain at least one Ig binding domain present in subunits selected from B2, B3 and B4. It may comprise a combination of any of Ig binding domains present in B2, B3 or B4. It may also comprise a combination consisting of the three Ig binding domains present in subunits B2, B3 and B4 linked to one or more sequences corresponding to that of an isolated synthetic Ig binding domain according to the invention. A preferred embodiment will comprise as Ig binding domain one sequence fragment corresponding to an isolated synthetic Ig binding domain according to the invention linked to 3 or 4 native Ig domains, said native Ig domains being selected so that the polypeptide according to the invention exhibits at least 3 B domains or at least 3 C domains. Suitably the 3 native domains will be different native domains. Clearly such an embodiment may suitably comprise as the 3 native domains, domains different to the domain to which the isolated synthetic Ig binding domain exhibits closest identity, e.g. an isolated synthetic Ig binding domain according to any embodiment of the invention exhibiting closest identity to the Ig binding domain of B4 will be linked to Ig binding domains of subunits B1 , B2 and B3, in any order or as a polypeptide comprising in N-C terminal order domains B1 linked to domain B2, with domain B2 linked to domain B3, with domain B3 linked to the isolated synthetic Ig binding domain according to any of the embodiments of the invention. Also a polypeptide according to the invention may comprise as Ig binding domain an isolated synthetic Ig binding domain according to the invention exhibiting closest identity to B1 , which is linked to the Ig binding domains of subunits B2, B3 and B4. Here also this embodiment may be such that the order proceeding from N terminus to C terminus is B1 , B2, B3 and B4. It is however not necessary to maintain the domains in the same order within the polypeptide according to the invention as occurs in native protein L. It is also possible that a polypeptide according to the invention links an isolated domain according to the invention exhibiting closest identity to B1 to Ig binding domains of subunits B1 , B2 and B3, in any particular order, or in the order presented here when proceeding from the N terminus to C terminus of the polypeptide sequence. Wherein the preceding section native Ig binding domains are mentioned, the corresponding Ig binding subunit may also be applied.

The polypeptide according to the invention may comprise one or more sequences corresponding to those of an isolated synthetic Ig binding domain according to the invention. A polypeptide according to the invention may be produced by separately preparing various segments and subsequently joining these or by preparing the polypeptide as one single unit, or by a combination of these processes. The Ig binding domains may be preformed as individual domains or as multimers and subsequently joined to form the polypeptide.

By way of example using recombinant nucleic acid technology, well known in the art, the polypeptide can be produced either as one complete Ig binding subunit or domain from one nucleic acid strand or as separate units or domains, whose various amino acid sequences, i.e. component polypeptide segments may be chemically joined. It is also possible to join all amino acids chemically to form the polypeptide of choice de novo. All such technologies are available to the skilled person and the selected method will in general depend on the economic considerations and availability of reagents. The skilled person is well equipped to decide which technology to use.

By way of example, DNA encoding protein L can be isolated from the chromosomal DNA from Peptostreptococcus magnus 312 based on the nucleic acid information derivable from the amino acid sequence details of figure 1 or derivable from the prior art. That encoding chromosomal DNA or cDNA derived therefrom may be used as a template and any defined fragment of nucleic acid desired can then be amplified from that template with the aid of e.g. PCR (Polymerase Chain Reaction), a well known and ubiquitously used technology. One can also use PCR to introduce the required nucleic acid mutations at the desired locations, by the corresponding use of primers in site directed mutagenesis. This also is standard methodology for anybody routinely carrying out generation of mutated DNA sequences. (See for example PCR Technology Ed: PCR Technology Principles and Applications for DNA Amplification and Kunkel T.A. (1985) [Rapid and efficient site specific mutagenesis without phenotype selection Proc. Nat. Acad Sci USA 82, 488- 492]). Thus by the correct choice of primers the various fragments of the protein L subunits may be generated, with or without linker sequences and the desired mutations of any subunit of choice may be engineered. All of this is routine procedure for a person skilled in the art of recombinant DNA technology. One can routinely attach various nucleic acid fragments generated in this manner by use of DNA ligase and express this ligated sequence using an expression vector in a host cell of choice (Sambrook, J.E. Fritsch and T. Maniatis 1989 Molecular cloning, A laboratory Manual 2^nd ed. Cold Spring Harbor laboratories, cold Spring Harbor, New York, USA.). A suitable host cell may be an E co/ cell or a yeast cell, such as a Saccharomyces cerevisiae, Hansenula or Pichia host cell. Numerous expression vectors optimised for the host cell of choice may be routinely used. It is also possible using phage display technology to generate multiple different mutations of native binding domain subunit sequences or full length protein L subunit sequences, with mutations targeted to the locations corresponding to one or more positions 25, 28, 31 , 33-41, 45, 49, 50, 52-59, 61 , 62, 76 and 78 of sequence id no 1 in nucleic acid sequences encoding one or more native Ig binding subunits or Ig binding domains and to screen these for binding to a VK domain of choice in order to generate Ig binding domains or polypeptides according to the invention. A suitable reference providing details required of phage display technology is Antibody Phage Display: Methods and Protocols (Methods in Molecular Biology) Philippa O'Brien, Robert Aitken.

The subsequent isolation and purification of the expression product is also a matter of routine. The cells can be lysed for example or the cells may secrete the expressed product. The expressed product may be purified using any art recognised protein purification technique e.g. ion exchange chromatography, gel filtration or affinity chromatography using an immunoglobulin as ligand or using any other VK domain comprising compound known to bind protein L or a subunit of protein L as ligand. The methods may be used as such or in combination in a conventional manner.

The sequence of the resulting cloned nucleic acid can also be routinely sequenced to confirm the identity of the sequence. EP B 0 662 086 for example shows details of cloning and expressing and isolating the native protein L subunit regions B1-B4 using recombinant DNA technology. An analogous method can be used to obtain the native subunit fragments desired to generate an embodiment of the invention.

After isolation, one can use any known method of purification of the expression product of the bacterial host e.g. using Ni NTA agarose. The expression product can also be subjected to binding affinity testing with any compound comprising or consisting of a VK domain e.g. by Surface Plasmon Resonance. The equivalent test can be carried out using a native B1 subunit or native protein L comprising the B1 subunit as control, to confirm that the isolated synthetic Ig binding domain or polypeptide comprising such indeed exhibit increased binding affinity. The protein can for example be isolated using SDS PAGE gel electrophoresis (e.g. 10% acrylamide concentration) and be transferred to do either a Western blot or a Slot Dot Blot using radioactively labelled proteins and a VK domain comprising binding partner. The Examples reveal such a process. (The affinity could also be measured using Fluoresence quench spectroscopy, Beckingham et. al, 1999 or Competition ELISA, Kastern et. al, 1992).

When a polypeptide according to the invention is prepared the skilled person will appreciate that maintaining a certain amount of diversity in the various monomers will assist in the preparation of the polypeptide. In particular such is preferred if the polypeptide is being prepared from one nucleic acid sequence encoding all the Ig binding domains. The skilled person may in such an event use the same sequences for the various component domains, however preferably the domains will in that event be separated by a linker sequence to prevent hairpin formation and recombination events. As indicated above the isolated synthetic Ig binding domains according to the invention may themselves be provided with N terminal linker sequences and, it is envisaged that where a polypeptide according to the invention is formed from one nucleic acid sequence it will comprise linker sequences between the various Ig binding subunits. These linker sequences may be produced separately from the nucleic acid encoding the Ig binding domains and attached preceding expression of the nucleic acid or may form an integral part of the nucleic acid sequence encoding one or more Ig binding domains according to the invention or of fragments used to construct the Ig binding domain(s) according to the invention. They may simply also be introduced via PCR. The linker sequences may also be made de novo in a polypeptide production process which proceeds by way of chemically linking amino acids. Thus a recombinant nucleic acid encoding the various Ig . binding domain components may comprise linker sequences linking the domains and/or preceding the N terminal domain within the polypeptide according to the invention. In an embodiment of the invention, a polypeptide as described in any of the preceding embodiments may consist of or comprise a glutathione S transferase sequence linked to the sequence encoding or forming the Ig binding segment of the polypeptide. A suitable embodiment of such a polypeptide according to the invention may thus by way of example comprise a dimer of isolated synthetic Ig binding domain according to the invention linked to the GST fragment.

As described in the prior art protein L and Ig binding domains and subunits of protein L have been described and found useful for various applications. In the references cited in the introduction to the art numerous applications are described. The isolated synthetic Ig binding domains according to the invention and the polypeptide according to the invention may also be used for those applications. They are considered incorporated herein by reference. Specifically the invention provides for use of the isolated synthetic Ig binding domains according to the invention and the polypeptides according to the invention in screening for components consisting of or comprising a VK domain. The invention also provides for use of the isolated synthetic Ig binding domains according to the invention and the polypeptides according to the invention in isolation of components consisting of or comprising a VK domain. The invention further provides for use of the isolated synthetic Ig binding domains according to the invention and the polypeptides according to the invention in purification of components consisting of or comprising a VK domain. The invention provides for use of the isolated synthetic Ig binding domains according to the invention and the polypeptides according to the invention in immobilisation of components consisting of or comprising a VK domain. All of there procedures may be carried out in a manner analogous to those routinely used in the art for screening, purifying, isolating or immobilising antibodies or antibody fragments using a proteinaceous binding ligand. Specifically analogues to those using protein L as binding ligand. In the case of immobilisation this suitably occurs on a solid support. A solid support may be CNBr activated sepharose, agarose, plastic surfaces, polyacrylamide etc as commonly used in the biotechnology industry for immobilising proteins. The isolated synthetic Ig binding domain or polypeptide may suitably be labelled where appropriate. For example labelling may occur with biotin, alkaline phosphatase, radioactive isotopes, fluorescein etc that are used to routinely label proteins. All of the above uses can be carried out using techniques known in the art for procedures using binding affinity of a target compound for a binding ligand thereof. Particularly suited are those already available in the field of immunology, using antibodies or antibody fragments as a binding partner. A suitable example of a useful application is of course use in binding assays. Examples of binding assays of interest are ELISA, RIA or Western blot. Other binding assays will also be apparent to the skilled person and are considered to fall within the scope of the invention.

By way of example arrays of components consisting of or comprising a VK domain can simply be screened using an isolated synthetic Ig binding domain according to the invention or using a polypeptide according to the invention in a manner known per se for screening using a binding partner capable of binding the desired target.

Basically any procedure allowing and requiring binding to a compound comprising or consisting of a VK domain can be carried out with the isolated synthetic Ig binding domain according to the invention or the polypeptide according to the invention. Specifically, advantageously, it is now possible to use the isolated synthetic Ig binding domain according to the invention or the polypeptide according to the invention in situations where the VK domain comprising component has low affinity for binding to native protein L or a native Ig binding subunit of protein L. In the introduction low affinity has been further defined. Specifically it is now possible to advantageously use the isolated synthetic Ig binding domain according to the invention or the polypeptide according to the invention in situations where the VK domain comprising component is present in low concentrations, for example in concentrations too low to successfully use native protein L or a native Ig binding subunit of protein L. In the introduction low concentration has been further defined.

The compounds according to the invention may be used in the analysis, screening, isolation or preparation of antibodies and in general for diagnostic and biological research. In particular applications requiring binding of multiple classes of antibodies such as screening, purification, isolation or immobilisation are envisaged as being suitable applications of the compounds according to the invention. Specifically advantageous is the application of the compounds according to the invention in situations where Fc binding proteins or fragments will not achieve the desired objective as the target to be bound lacks an Fc component yet has a VK domain component. One could also use Protein L as such as an alternative embodiment to use for such an application. Suitably an embodiment of the subject invention comprises a complex of an isolated synthetic Ig binding domain or polypeptide according to the invention bound to an antibody fragment lacking a Fc component. By way of example such antibody fragment may be scFv. Preferably the antibody fragment has a lambda light chain. Preferably the antibody fragment precludes any possibility of interaction between the fusion partners of the complex. Thus the domains or polypeptides according to the invention are suitable to be effective fusion partners of therapeutic antibody fragments. In this manner a broad spectrum of antibody dependent effector functions is recruited through the creation of a scFv-B cell superantigen fusion protein. The identity of superantigen is provided in WO99/20749. This fusion protein is able to recruit any Ig bearing a VK domain of the K I, II, III and IV subclasses towards the target bound by the scFv. Thus the effector functions can be recruited by a scFv without the need for the addition of the Fc region. Preferred methods of linking include the use of polypeptide linkers, as described, for example, in connection with scFv molecules (Bird et al., (1988) Science 242:423-426). Discussion of suitable linkers is provided in Bird et al. Science 242, 423-426; Hudson et al , Journal Immunol Methods 231 (1999) 177-189; Hudson et al, Proc Nat Acad Sci USA 85, 5879-5883. Linkers are preferably flexible, allowing the two single domains to interact. One linker example is a (Gly Ser)_n linker, where n=1 to 8, eg, 2, 3, 4, 5, 6 or 7. The linkers used in diabodies, which are less flexible, may also be employed (Holliger ef al., (1993) PNAS (USA) 90:6444-6448).

An additional aspect of the invention covers a complex of an isolated synthetic Ig binding domain according to the invention or a polypeptide according to the invention to a VK domain comprising component.

In any of the embodiments of the invention mentioning a VK domain, may suitably be a VK I, II, III or IV subclass.

The invention naturally also covers nucleic acid sequences encoding the isolated synthetic Ig binding domains according to the invention or polypeptides according to the invention and the corresponding amino acid sequences, as well as a host cell comprising such nucleic acid and/or expressing such nucleic acid sequence. The invention also covers a method of generating a polypeptide or isolated synthetic Ig binding domain according to the invention using recombinant DNA technology. Such a method may also further comprise a combination of well known and obvious techniques for obtaining the required protein product. The invention also covers kits comprising at least one isolated synthetic Ig binding domain or polypeptide according to the invention in addition to instructions for carrying out an assay in the form of a kit. Such a kit may comprise additional reagents required for carrying out an assay requiring binding of the Ig binding domain or polypeptide to a compound comprising or consisting of a VK domain. Suitably the Ig binding domain or polypeptide according to the invention are present in immobilised form on a carrier in the kit or else the kit further comprises a carrier for immobilisation of protein as additional reagent.

FIGURE DESCRIPTION: Figure 1 provides a comparison and alignment of Ig binding sections of native protein L subunits, basically an alignment of the various Wikstrom Ig binding segments, which run from residues 21-81. The subunits have been aligned versus B1 of Peptostreptococcus magnus 312. The subunits B2, B3, B4 and B5 were also derived from that strain. The subunits C1 , C2, C3, C4 and C8 were derived from strain Peptostreptococcus magnus 3316. The numbering provided corresponds to the numbering provided by Wikstrom (Wikstrom et al 1995, cited elsewhere in the specification). The sequence fragments preceding residue 20 provide examples of N terminal fragments not involved in Ig binding that occur in native protein L subunits, which may be used as linker sequences or parts of linker sequences in the polypeptides according to the invention or may be comprised either N terminally or C terminally to the Ig binding fragment of an isolated synthetic Ig binding domain according to the invention.

Figure 2

This figure reveals a selection of amino acid sequences, known from the art that represent various frameworks of K sub classes. The residues involved in the protein L-V kappa interaction in the two different co-crystal structures are compared. Those residues that are involved in the interaction in both co- crystal structures are coloured in red, those residues that are only involved in the first co-crystal are coloured in blue and those residues that are only involved in the second co-crystal structure are coloured in green.

Figure 3 This figure reveals the percentage of mutations based on degree of identity between the native subunits of Protein L in relation to subunit B1. This is useful to ascertain from which native subunit an isolated synthetic Ig binding domain according to the invention is derived. The B1 refers to the fragment corresponding to the Wikstrom fragment, residues 21-81 of sequence id no. 1. The 23 refers to the residues that may be mutated, the total of residues 35-41 , 45, 49-50, 52-59, 61 , 62 of Sequence id no. 1. The table reveals how many residues in the relevant section of each subunit differ vis a vis sequence id no. 1 in the native subunits.

Figure 4

Curves used to calculated the on-rate and off-rate of the IKRI domain when bound to the Vκl scfV (DPK9 framework). Higher concentrations of the IKRI domain were used to calculate the off-rate as the effect of re-binding was minimised at higher concentrations and lower concentrations were used to calculated the on-rate as at the higher concentrations, the on-rate was so rapid that is was impossible to separate the association phase from the increase in resonance units due to the introductions of a different buffer.

Figure 5

Biotinylated B1 and IKRI were compared as detection reagents in ELISA. A dilution series of B1 and IKRI-biotin in triplicate was carried out across a plate containing the Vκl scFv supernatant immobilised on protein A. Binding of the biotinylated reagents was carried out using streptavidin-HRP. As can be seen from the figure, IKRI-biotin gave an increased signal compared to B1-biotin at all concentrations tested.

Figure 6 We created a model system to determine whether an immunoglobulin domain of protein L fused to a single chain Fv (or other antibody fragment) was capable of recruiting effector functions to an antigen bound target. The antigen bound target consisted of Red Blood Cells coated with the hapten fluorescein, and the protein L-scFv fusion consisted of the IKRI (B1 mutant) domain fused to an anti-fluorescein scFv (with a λ light chain), E2. The scFv should bind to the antigen coated cell and the IKRI domain can bind all immunoglobulins with a kappa light chain and therefore should be able to recruit both complement and Fc receptor mediated effector function through bound immunoglobulin.

Figure 7 Results of agglutination assay to determine whether the fusion protein E2- IKRI could cause agglutination of fluorescein coated red blood cells in the presence (or absence) of immunoglobulin (mouse lgMκ and human lgG1κ). The results indicate that although agglutination occurs in the absence of immunoglobulin at high fusion protein concentrations, agglutination occurs at much lower fusion protein concentrations in the presence of immunoglobulin, particularly the pentameric IgM. (++++ corresponds to strong agglutination i.e. formation of single pellet with no cells remaining in suspension, +++-+ corresponds to decreasing amounds of agglutination i.e. a less defined pellet and increasing numbers of cells remaining in suspension and - corresponds to no visible agglutination i.e. all cells remain in suspension).

Figure 8

Assay investigating the ability of the fusion protein (E2-IKRI) to stimulate superoxide burst via the Fcγ Rl receptor on U937 cells. The ability of the fusion protein to stimulate a superoxide burst through the formation of immunoglobulin complexes on the surface of fluorescein coated Red Blood Cells was compared to that of a positive control system where an immunoglobulin bound an antigen coated surface directly. Both systems were found to stimulate bursts of comparable magnitude indicating that use of the fusion protein to recruit immunoglobulin to an antigen coated surface was as effective as an immunoglobulin bound directly to the cell surface.

Two different lgG1 kappas (IgG 1k and Gam-1) were compared to an lgG1 lambda (IgGII) in their ability to stimulate the superoxide burst. Both lgG1 kappas produced a significant burst (IgGIK also stimulated a burst in the absence of fusion protein probably due to aggregation of the IgG in the sample) whereas the lgG1 lambda did not stimulate any significant burst indicating that the burst was dependent on fusion protein binding (the IKRI mutant only binds to lambda light chains).

Figure 9 This assay shows the effect of fusion protein and lgG1κ concentration on the superoxide burst from U937 cells and indicates that the burst is dependent on the presence of both the fusion protein and the lgG1.

Figure 10 Figure (a) shows an example of the resetting of Red Blood Cells round U937 cells (reproduced from Holliger et. al, 1997)

Figure (b) Ability of the fusion protein to stimulated resetting of fluorescein coated Red Blood cells through lgG1κ and the Fcγ RIIA receptor on K562 cells. Rosetting was determined to have taken place if 5 or more Red Blood Cells were clustered round a single K562 cell. The results indicate that the fusion protein is capable of stimulating resetting through the Fcγ RIIA receptor in the presence of lgG1 , which binds the receptor but no in the presence of lgG2 or lgG4, which do not bind the receptor. The results also indicate that the fusion protein is as effective at stimulating resetting of fluorescein coated Red Blood Cells in the presence of lgG1 as a positive control system using NIP coated Red Blood Cells and an anti-NIP lgG1.

The following section reveals some examples of the various embodiments of the invention.

EXAMPLES

1) Selecting and screening of mutants and assessment of binding affinity

Summary

Parallel selections were carried out against an anti-BSA scFv (13CG2) using phage display. Almost 400 clones from these selections were screened against the Vκl scFv and approximately 50 clones were identified with a potential improvement in affinity over the wild-type B1 domain including one clone with substantial improvement. These 50 clones were sequenced and the pattern of mutations across the 50 clones analysed; from the analysis five amino acid hotspots emerged whose mutational frequency was much greater than the other ten positions in the original library, and within these five positions, some amino acids were present at a much higher frequency than others and whilst the mutants that exhibited the strongest binding contained a number of these favoured mutations, no single mutant contained them all. Therefore a series of mutants were created with different combinations of the most favoured mutations. The off-rates of these mutants were analysed with the mutant IKRI having the most improved affinity for the scFv over the wild- type B1 domain. The affinity of this mutant for the scFv (13CG2) was then measured in detail.

Detail

Two types of selection were carried out in parallel using phage from the library, one using streptavidin coated beads to immobilise the biotinylated antigen and capture any binding phage and the other using a streptavidin coated BIAcore chip on which to immobilise the antigen and capture the phage. In both types of selection, the phage were selected against biotinylated 13CG2 (an anti-BSA single chain Fv (scFv) with a K light chain of the κ1 subclass - DPK9 (Tomlinson and Winter, unpublished). Three rounds of phage selection were carried out against the scFv immobilised on the streptavidin coated beads and a single round of selection was carried out against the scFv immobilised on the streptavidin coated BIAcore chip.

Subsequently Surface Plasmon Resonance was used to screen 192 mutants from each round of selection. This was chosen as many of the antibodies used as detection and immobilisation reagents in ELISA bind protein L to a greater or lesser extent result in high background. As SPR does not require reagents it was found to be a more reliable and reproducible method of measuring binding. The binding was assessed either as increase in resonance units corresponding to amount bound above background or as a reduced off rate. On this basis, 47 clones showed improved binding and one clone G5 (T36I, E38K, T56I) showed a significant improvement in binding.

The pattern of mutations in the 47 clones was analysed and it was noticeable that some positions had a much higher mutation rate than others in the group of selected clones and that some amino acids were much more favoured than others at these positions. It was also noticed that the most improved clone (G5) had the highest number of these favoured mutations but no single clone had them all.

To create a clone with the optimal combination of mutations, eight clones were created with different combinations of mutations that had led to the greatest improvements in affinity across the five positions that were most commonly mutated in the first group of selected clones, if two amino acids were commonly selected at a single position, both were included. Of all the combinations tested, IKRI appeared to have the slowest off-rate. The affinity of this clone in comparison to the wild-type B1 clone was then analysed in more detail. IKRI is mutant (T36I, E38K, A52R, TS6I).

Before the affinity of the B1 and IKRI domains were compared, the multimerisation state of the two domains was analysed using gel filtration chromatography to ensure that any increase in affinity was not due to the formation of dimers or other higher multimers. A sample of both domains was passed down an analytical scale HR-75 column (Pharmacia), and both were eluted in a single peak corresponding to a molecular weight of around 10 kD (results not shown) indicating that both domains were monomeric.

The affinity of B1 and IKRI for a Vκ1 scFv was measured using Surface Plasmon Resonance. To ensure that the measurements were accurate, i.e. to reduce the effects of rebinding on the off rate, the streptavidin chip was coated with a lower amount of Vκ1 scFv (approximately 600 resonance units) and high flow rate was used (30μl/min). A dilution series of twelve different concentrations was used ranging from 23nM - 55μM in PBS. The off-rates were calculated at the higher concentrations where the chip was most likely to be saturated and the effects of re-binding would be minimised although in general the off-rates were fairly consistent over the entire concentration range. The portion of the curve that was fitted in the calculation of both the on and the off-rates was immediately after the injection start and stop points, once any effects of the change in buffer had passed. The on-rates were calculated from the lower concentrations as at the higher concentrations, binding reached equilibrium very rapidly while the increase in resonance units due to a change of buffer were still affecting the measurements whereas at the lower concentrations, equilibrium was reached more slowly allowing time for the measurement of the on-rate. From averaging (n >3) of the most accurate measurements (see figure 4 and table below) the affinities were calculated as follows:

B1 kd = 0.06105 s- ka = 3.95 e⁵ M^"1 S^"1 KD = 154nM

IKRI kd = 0.0084475 s -1 ka = 5.02 e .5° n M/ι-^"1¹ S o-^"1¹ KD = 16.8nM Thus, the affinity of the B1 domain mutant, IKRI is approximately 9 times higher than that of the wild type B1 domain. The improvement in affinity is due almost entirely to a reduction in off-rate rather than any significant change in the on-rate. This is not unexpected as the selection was designed to select mutants with off-rate improvements rather than an improvement in the on-rate. The affinity of B1 for Vκ1 scFv is virtually identical to that calculated using other methods such as competition ELISA (Kd = 160nM) (Kastern et. al, 1992) and also similar to that calculated for the C^* domain from strain 3312 (Kd = 130nM) using stopped-flow fluorimetry (Beckingham et. al, 1999), indicating that Surface Plasmon Resonance under the experimental conditions used here is a valid method for affinity determination of the interactions of pL with VK domains.

Use of IKRI as a reagent for improved detection in ELISA To determine whether the use of IKRI as compared to the wild-type B1 domain could lead to increased sensitivity of detection of antibody-antigen binding reactions, both domains were biotinylated and used as detection reagents in ELISA to detect the binding of the Vκ1 (13CG2) scFv to protein A. Although the number of lysine residues (and correspondingly primary amine groups) has increased from seven to eight between the wild-type B1 domain and the IKRI mutant, it is unlikely that any difference in signal is due to increased biotinylation of the domain. If biotinylation did take place at K38 then the presence of the biotin group within the binding site would most probably disrupt binding reducing the affinity of the interaction and therefore the signal from that domain.

The measurements were carried out in triplicate and results can be seen in Figure 5. As can be seen from the figure, the use of IKRI-biotin resulted in:

1 ) an increase in signal at any given concentration of the biotinylated B domains;

2) a 10-fold reduction in the amount of biotinylated reagent required compared to wild-type to elicit the same signal and;

3) an increase in the maximum signal at the maximum pL concentration compared to the wild-type.

These improvements will allow the use of much smaller amounts of detection material to elicit the same signal and potentially allowing the detection of lower affinity antigen-antibody interactions. We have therefore created a detection reagent based on the B1 domain of protein L with higher affinity for Vκl light chains and therefore improved sensitivity of detection in binding assays.

3) Binding to different K frameworks To deflate any argument that the selection of the B1 domain mutant IKRI has simply created a B domain whose affinity for Vκ1 scFv has been improved but at the expense of binding to other sub-types or other frameworks (the framework used in this case was DPκ9) or that the affinity improvement shown may be specific only to this sub-type and framework, we investigated whether increased binding affinity extended to the other K sub-types and frameworks. B1 -biotin and IKRI-biotin were compared at equal concentration for their ability to detect different scFvs. The frameworks tested were four Vκl frameworks (DPK 1 , DPK4, DPK 8 and DPK 9) and one VKIII framework (DPK 22), which all had heavy chains of the V_HIII class and were therefore able to bind Protein A.

DPK 4 and DPK 9 share the same sequence in the protein L binding region and gave a strong signal in ELISA. The other four frameworks gave much weaker signals, which could be due to poor expression or an intrinsically lower affinity for all domains of protein L. However all of the frameworks tested gave a significantly stronger signal when detected with IKRI-biotin as compared to B1 -biotin indicating that the development of IKRI has created a mutant with improved affinity for a range of K sub-types and frameworks.

4) Superantigens as fusion partners for therapeutic antibody fragments Many therapeutic monoclonal antibodies mediate their effects through the effector functions including antibody dependent cell mediated cytotoxcity (ADCC), phagocytosis and complement activation. ADCC and phagocytosis are activated through binding of the Fc region of the antibody to Fc_γR receptors on immune effector cells such as natural killer cells, macrophages, neutrophils and B cells and complement dependent cytotoxicity is activated through the interaction of the Fc region with complement proteins such as C1q, C3 and C4.

Monoclonal antibodies developed through immunisation are full-size and contain an Fc region capable of mediating some if not all of these functions depending on the isotype. However, recombinant antibody fragments such as scFv and Fab do not contain an Fc region and are unable to recruit these effector functions. Depending on their mode of action in vivo, the ability of an antibody to recruit effector functions may be crucial to its therapeutic effect, particularly if the antibody is directed towards a tumour. The ability to recruit effector functions can be restored to antibody fragments for example by converting them into a full monoclonal antibody. Alternative methods have also been developed to overcome this problem using a range of recombinant antibody formats including bispecific antibody fragments such as diabodies where one half of the diabody is directed against a therapeutic target and the other directs the fragment towards the Fcγ receptor, or antibody fragment - fusion proteins such as antibody fragments fused to toxins, growth factors or T-cell superantigens. The use of a T-cell superantigen, staphylococcal enterotoxin A (SEA) to direct cytotoxic T-cell dependent cell lysis towards a range of carcinomas through the use of a fusion between SEA and various carcinoma reactive antibodies is well documented and has resulted in several phase I clinical trials demonstrating the potential of this method in antibody mediated therapy.

Here, we present a novel approach to simultaneously recruit a broad spectrum of antibody dependent effector functions through the creation of a scFv - B-cell superantigen fusion protein. This fusion protein is able to recruit any immunoglobulin bearing a light chain of the κ l, II, III and IV subclasses towards the target bound by the scFv. In this way, the effector functions can be recruited by a scFv without the need for the addition of the Fc region (figure 6).

To investigate whether the B domains of protein L could be used as effective fusion partners to therapeutic antibody fragments we fused the high affinity B1 mutant IKRI to the C terminus of an anti-FITC scFv (E2). E2 was chosen on the basis of its antigen, FITC, which can easily be coated onto Red-Blood Cells as it is commercially available in a N-hydroxy-succinimide form. E2 also has an extremely high affinity for FITC (K_d = 0.2nM) and the scFv has a lambda light chain, which should preclude intramolecular interaction of the light chain with the protein L fusion partner that might occur with a scFv with a K light chain.

The gene for scFv E2 was isolated by phage display and was therefore cloned in the display vector pCANTAB-6 (a phagemid vector) as a fusion to the gene III protein. The gene III protein was to be replaced with the B1 domain mutant IKRI using the Not I site in the multiple cloning site and the Eco Rl site C- terminal to the gene III protein. A glycine/serine linker was introduced between the two proteins to allow flexibility of movement and correct folding. Furthermore, two stop codons were introduced at the C-terminus of IKRI to replace those removed through the use of the Notl/Eco Rl restriction sites including the amber stop codon preceding gene III in the original construct. Thus, all expressed protein should consist of the full-length fusion protein and there should be no premature termination after the scFv in TG1 (or other suppressor strains).

The fusion protein was expressed in TG1 and purified from the supernatant after overnight expression using Protein A sepharose.

To determine whether the two halves of the fusion protein could simultaneously bind FITC and antibody, an ELISA was carried out with using supernatant from an overnight induction of the fusion protein. The plate was coated overnight with 13CG2 (anti-BSA), a single chain Fv known to bind protein L followed by blocking, incubation with the supernatant and detection with FITC-HRP. This resulted in a strong signal (results not shown) indicating that both halves of the fusion protein could bind their respective targets simultaneously.

We first investigated whether the fusion protein E2-IKRI could cause agglutination of FITC coated Red Blood Cells in the absence of any immunoglobulin and in the presence of mouse monoclonal lgMκ and human polyclonal lgG1κ (Sigma). The results of this assay are shown in (figure 7). A dilution series of the fusion protein was incubated with Red Blood Cells coated with FITC or uncoated Red Blood Cells either in the absence of immunoglobulin or in the presence of human lgG4κ or mouse lgMκ. The fusion protein appeared to be highly effective at agglutinating the Red Blood Cells in the presence of IgM, slightly less effective in the presence of IgG and only effective in the absence of immunoglobulin at high concentrations. One would expect agglutination to be most effective in presence of IgM due to the pentameric nature of the immunoglobulin providing multiple sites for cross- linking of the FITC coated Red Blood Cells through the E2-IKRI fusion protein. lgG4 is monomeric in comparison, however each IgG molecule has two protein L binding sites, one on each light chain, which is less effective than the 10 binding sites on an IgM pentamer but still able to promote agglutination. Finally agglutination of the Red Blood Cells in the absence of any immunoglobulin is probably due to low levels of multimeric fusion protein formed through the dimerisation of the single chain Fv to form diabodies.

This assay provided the first evidence of the ability of the pL domain in E2- IKRI to recruit immunoglobulin whilst the scFv portion of the fusion protein remained bound to its cognate antigen, in this example FITC.

The ability of the fusion protein to cause agglutination of FITC coated Red Blood Cells in the presence of immunoglobulin demonstrated the ability of the fusion protein to simultaneously bind antigen and immunoglobulin. However in order to function as a therapeutic reagent, the fusion protein must have the ability not only to recruit immunoglobulin, but also to recruit Fc effector functions such as those of the classical complement pathway. Antibody dependent complement activation occurs through the formation of IgG or IgM antibody complexes on cell surface antigen (in this case complexing occurs through the fusion protein), which can then bind C1q. Binding of C1q to immunoglobulin activates C1r which then cleaves and activates the serine protease C1s leading to the activation of the complement dependent cascade of activities including the activation of inflammatory mediators, opsonisation of pathogens, the removal of immune complexes and the formation of membrane attack complexes leading to the lysis of pathogens and certain cells.

To determine whether the fusion protein could activate the complement dependent cascade in the presence of Ig, we investigated the ability of the fusion protein to cause complement dependent lysis of FITC coated Red Blood Cells (RBCs). RBCs were coated with 200μg/ml FITC and incubated with a dilution series of fusion protein (0.019-117μg/ml) either the presence or absence of IgMκ (lμM), followed by the addition of 5% guinea pig complement in complement fixation diluent (CFD) and incubation at 37°C for 30 minutes. Haemolysis was measured by reading OD 405nm and compared to a positive control (FITC coated Red Blood Cells + MQ water). The results are presented in figure 7.

As can be seen, E2-IKRI is highly effective at promoting complement dependent lysis of Red Blood Cells, with lysis at the higher concentrations of fusion protein equivalent to that of the positive control. Maximal lysis is reached at approximately 10-40μg/ml and is observable down to 1 μg/ml, the loss of lysis with decreasing amounts of fusion protein demonstrating that complement-dependent cell lysis is completely dependent on the presence of fusion protein. Complement lysis assays were also carried out with the two halves of the fusion protein separately to determine whether both halves of the fusion protein were needed for this activity with the results confirming that both halves of the fusion protein are necessary to stimulate complement dependent lysis (see figure 7).

Unexpectedly, lysis does not appear to be dependent on the presence of lgMκ (mouse), with maximal lysis reaching higher levels in the absence of lgMκ than in the presence of lgMκ, however lysis appears to become dependent on the presence of lgMκ as the concentration of FITC on the surface of the Red Blood Cells decreases (results not shown). The presence of lysis in the absence of IgM is probably due to high levels of immunoglobulin in the guinea pig complement serum providing sufficient Ig for complement activation and as the concentration of FITC on the Red Blood Cells decreases, the immunoglobulin in the serum (particularly if serum activation is mostly taking place through IgG which is less effective at activating complement than IgM) no longer proves sufficient to stimulate lysis and becomes more reliant on the addition of IgM. An ELISA confirmed that both protein L binding IgG and IgM were present in the guinea pig complement serum (data not shown) explaining the activation of complement activity in the apparent absence of any immunoglobulin.

The demonstration of complement dependent lysis that is absolutely dependent on the presence of fusion protein confirms the ability of the fusion protein to recruit complement-mediated activities to an antigen present on the surface of a particular target cell.

We investigated whether the fusion protein could recruit effector functions through the Fcγ Rl receptor such as the respiratory burst mediated when aggregated antibodies bind to the Fcγ Rl receptors for example on monocytic cells like macrophages leading to the generation of a variety of toxic products, the most important being hydrogen peroxide (H₂O₂), the superoxide anion (O₂ ^' ) and nitric oxide (NO).

We investigated the ability of the fusion protein to complex lgG1κ on the surface of FITC coated Red Blood Cells. The complexed lgG1κ would then bind the Fcγ Rl receptor on monocyte like U937 cells (stimulated by IFN-γ for 72 hours) triggering a respiratory burst diagnosed by the production of the superoxide anion. Superoxide production is measured through its interaction with lucigenin (bis-Λ/-methylacridinium nitrate) resulting in chemiluminesence measured over a period of 30 minutes using a luminometer. U937 cells have both Fcγ Rl and Fcγ Rll on their surface but it has been shown that in this cell line, Fcγ Rll is effectively 'silent' and does not have significant interaction with IgG (Lund et. al, 1991 ). First, we simply investigated whether the fusion protein was capable of stimulating a superoxide burst. We compared the superoxide burst from U937 cells alone to U937 cells incubated with fusion protein (300 μg/ml), lgG1κ (l00μg/ml) and Red Blood Cells coated with 200μg/ml FITC and U937 cells incubated with individual assay components as illustrated in figure 8. The results of this assay indicated that the fusion protein was able to stimulate a superoxide burst in the presence of lgG1κ and FITC coated Red Blood Cells. None of the three components (lgG1 , FITC coated RBCs and E2-IKRI) stimulated a response alone and the burst also appeared to be dependent on the presence of a K chain in the light chain of the lgG1 as there was almost complete loss of response when the lgG1κ was replaced with the IgGlλ. The size of the burst produced through complexation of the lgG1κ on the surface of the Red Blood Cells through the fusion protein was compared to a control system in which the lgG1κ binds both it's cell-surface antigen and the Fcγ Rl receptor directly. The size of the two bursts was comparable indicating that the addition of the fusion protein had no negative effect on the ability of complexed lgG1 to stimulate monocytes through the Fcγ Rl receptor.

Once we had established that the fusion protein was capable of stimulating a superoxide burst we carried out a dilution series of the fusion protein with two different lgG1κ (Gam-1) (100μg/ml) and Wid (12.5μg/ml)) to look at the effect of decreasing fusion protein concentration on the size of the respiratory burst. For Gam-1 lgG1 the concentration range of fusion protein used was (0.4 - 100μg/ml) with the maximal response at 10μg/ml with the burst tailing off slightly above 10μg/ml and decreasing rapidly between 10 and 1 μg/ml with no response (above background) at concentrations below 1 μg/ml (see figure 9). For Wid lgG1 the concentration range of the fusion protein was lower (0.625 - 20μg/ml) with a maximal response at 20μg/ml with no tailing of the response at the higher concentrations, again the response decreased steadily to 1 μg/ml with no response below this concentration. The differences in concentration of the two IgGs did not appear to have much effect on the ability of the fusion protein to stimulate a superoxide burst indicating that neither of the IgG concentrations used is low enough to be limiting. Dilution series of the two IgGIs at a constant fusion protein concentration (10μg/ml) were carried out. The concentration range of Gam-1 used was (0.333-1 OOμg/ml). There was only a significant burst at concentrations above 10μg/ml, with the maximal response at 33μg/ml and a slight decrease in response at the highest concentration of 10Oμg/ml (figure 9). Again the concentration range of Wid IgG was slightly lower (1.56-50μg/ml) with some response even at the lowest concentration of 1.56μg/ml then a rapid increase in response up to the maximal response at 12.5μg/ml with a slight loss of activity to a highest concentration of 50μg/ml. As the serum concentration of lgG1κ is approximately 5.85 mg/ml (9 mg/ml lgG1 of which 65% will contain a K light chain), the concentrations needed to stimulate a respiratory burst would easily be present in vitro. The increased activity of Wid as compared to Gam- 1 may be due to differences in their K chain sequence (for instance one may contain a K chain of sub-class I whereas another may contain a K chain of sub-class III) with differences in affinity of the pL domain for these sequences leading to differences in activity of the IgGs in this system. Alternatively differences in the ability of the two IgGs to activate the cell via this receptor due to differences in their glycosylation state.

The reduction in the magnitude of the superoxide burst as both the concentration of the fusion protein (E2-IKRI) and the two IgGIs decrease demonstrate that the superoxide burst is dependent on both of these assay components. Unfortunately the magnitude of the response cannot be compared between the two IgGs as the experiments were run on different days with different batches of activated U937 cells, and there is wide variability in the ability of different batches of U937 cells to produce a superoxide burst.

The ability of the fusion protein to produce a superoxide burst comparable in magnitude to that of the positive control indicates that the fusion protein is able to efficiently recruit effector functions via the FcyRI receptor. Also, the fusion protein was able to activate the Fcγ Rl receptor through all IgG tested indicating that the fusion protein should be able to recruit effector functions through any VK antibody.

Finally, we investigated the ability of the fusion protein to recruit effector functions through the Fcγ Rll receptor using K562 leukocyte cells. In this case we were investigating the ability of the fusion protein to opsonise the FITC coated Red Blood Cells in the presence of lgG1. Opsonisation determined through a rosetting assay by counting the number of Red Blood Cells clustered round a single K562 cell in the presence of lgG1 and fusion protein. Rosetting was considered to have taken place if five or more RBCs were clustered round a single K562 cell.

We compared the ability of the fusion protein together with Wid lgG1κ to rosette FITC coated Red Blood Cells round K562 cells with the ability of an anti-NIP (5-iodo-4-hydroxy-3-nitrophenylacetyl) IgGI K to rosette NIP coated Red Blood Cells round K562 cells. Red Blood Cells coated with 100 μg/ml NIP or 100μg/ml FITC were incubated with either 20μg anti-NIP lgG1 or 20μg Wid lgG1 plus 2μg fusion protein (E2-IKRI) respectively. The rosetting reaction was set up by the addition of K562 cells to the sensitised Red Blood Cells and incubated for 15 minutes before staining of the cells with acridine orange and scoring of the rosetted cells under a fluorescence microscope. The results of this assay are shown in figure 10.

From the results it can be seen that the fusion protein is capable of stimulating rosetting, demonstrating that the fusion protein can recruit effector functions through the Fcγ Rll receptor as well as the Fcγ Rl receptor. The level of rosetting is equivalent in the FITC and NIP systems providing further evidence that the use of a fusion protein to recruit effector functions to an antigen coated cell through IgG is no less effective than using an IgG directly. The level of rosetting appears to be strictly dependent on the presence of fusion protein and IgG with no rosetting taking place in the absence of either. There is also a rapid decrease in the amount of rosetting with decreasing FITC concentration on the surface of the Red Blood Cells suggesting that if the antigen is not present at a high enough concentration then multimerisation of the IgGs cannot take place and the Fcγ R receptor will not be activated. It had previously been shown that lgG2 and lgG4 do not interact with the Fcγ RIIA receptor on K562 cells (Walker et. al, 1989) making it likely that the receptor is of the Fcγ RIIA (R131 ) form (Gessner et. al, 1988). These results were replicated in this assay as neither lgG2 nor lgG4 elicited rosetting (both had _κ light chains and lgG4 caused agglutination of the FITC coated RBCs in the presence of fusion protein indicating that it did bind the fusion protein) again confirming that rosetting was taking place through the Fcγ RIIA receptor.

Discussion

Therapeutic antibodies function via several different methods in vitro, they may act directly through binding to a target molecule by inducing apoptosis, inhibiting cell growth, mimicking or blocking a ligand or by interfering with a key function (Esteva and Hayes, 1988, Maloney, 1988) or the antibody itself may act as an effector through the activation of antibody-dependent cellular cytotoxicity (ADCC) or the complement dependent cascade (CDC) or it may involve effector elements such as cytotoxic drugs, enzymes, radioactive isotopes (Adair et. al, 1992, Peterson, 1998, Russell et. al, 1992). Several therapeutic antibodies against cancer are unconjugated and have been shown to elicit their activity through the effector functions. For instance Herceptin, a humanised monoclonal antibody for breast cancer targeted to the p185/Her2 protein (Carter, 1992) exhibited severely reduced efficacy in mice lacking Fcγ Rl and Fcγ Rill (Clynes et. al, 2000) suggesting that ADCC may be important in the mechanism of action of this drug. Rituxan, a chimeric anti-CD20 monoclonal antibody for non-Hodgkin's lymphoma (Leget, 1998) was also shown to be ineffective in Fcγ Rl/ Fcγ RIM deficient mice (Clynes et. al, 2000) and when an lgG4 form of Rituxan (lgG4 shows poor binding to Fcγ R and C1q) was tested in primates it showed no effect compared to the efficacious lgG1 form (Anderson, 1997). In cases where the monoclonal antibody utilises the Fcγ R bearing cells either through ADCC, to activate cytotoxic T-cells or simply as a cross-linking agent, improving the binding of the monoclonal antibody to the Fcγ R could improve the efficacy of the monoclonal antibody. The development of selection techniques such as phage (McCafferty et. al, 1990) and ribosome display (Hanes and Plϋckthun, 1997) has lead to the isolation of large numbers of recombinant antibody fragments such as scFv and Fab against therapeutic targets. These antibody fragments lack an Fc region and are therefore unable to recruit the effector functions that may be needed for their therapeutic effect. To overcome this problem, scFvs and Fabs are often converted into full antibodies, although alternatives such as the use of bi-specific diabodies have been developed. Bispecific diabodies have dual specificity with one half of the diabody directed against an antigen on the surface of the target cell and the other used to recruit either T-cell or Fcγ receptor functions. A combination of two or more bispecific diabodies may be used, for instance although either bispecific CD19 (B cell marker) x CD3 or CD19 x CD16 (Fcγ RIIIA) diabodies alone lead to partial tumour regression in SCID mice with an established Burkitt's lymphoma, the combination of the two diabodies together with CD28 co-stimulation resulted in complete elimination of tumours in 80% of animals (Kipriyanov et. al, 2002) demonstrating the advantage of simultaneously recruiting different populations of human effector cells. Here we investigated an alternative approach to recruiting a range of antibody dependent effector functions through scFv and Fabs by fusing a single antibody binding domain of the B-cell superantigen, protein L to a scFv against the hapten, FITC.

The potential of superantigen fusion proteins was first highlighted through the use of a T-cell superantigen (staphylococcal enterotoxin A (SEA))-monoclonal antibody fusion to redirect cytotoxic T-cell dependent cell-lysis to cancerous tissue over 10 years ago (Kalland et. al, 1991 ). Since this initial study, SEA fusions to different antibody fragments including scFv, Fab and diabodies have been used to treat a range of cancers including pancreatic, colorectal and B-cell malignancies (Gidlof et. al, 1997, Giantonio et. al, 1997, Nielsen et. al, 2000). Although T-cell mediated functions may be crucial in the destruction of tumours, Ig-mediated effector functions, which could be recruited through a B-cell superantigen also have an important role to play as discussed earlier. In terms of a B-cell superantigen, protein L is the optimal choice due to the location of its binding site in the variable region making binding independent of Fc type. Protein L also binds approximately half of the circulating immunoglobulins in humans and two-thirds in mice (Graille et. al, 2002) making it suitable for both treatment in humans and analysis in mouse models. Although Protein A and Protein G both bind immunoglobulin in the Fab region, the affinity of this interaction is much lower than that of protein L with Fab. The use of the B1 domain mutant with its higher affinity for light chain, may also improve the efficacy of the fusion protein further. The other common B-cell superantigens, protein A and protein G also have binding sites in the Fc region of the antibody at the CH₂/CH₃ hinge (Deisenhofer, 1981 , Sauer-Eriksson et. al, 1995), a position that could potentially interfere with Fc receptor interaction making them less suitable as fusion partners.

The B domain of protein L is able to bind to light chains of K classes I, III and IV and as such binding is independent of Fc receptor type and includes immunoglobulin from all classes which than can then recruit the full spectrum of effector functions. In the assays used here, the B1 domain was not only able to bind all the lgκ tested (both mouse and human) but was also to complex immunoglobulin on the antigen coated surface bound by the scFv. These immunoglobulin complexes were then able to activate complement- mediated cell lysis, superoxide burst through the Fcγ Rl receptor and rosetting through the Fcγ Rll receptor and presumably would be able to activate through the Fcγ Rill receptor found on Natural Killer cells resulting in ADCC, an important mechanism in the destruction of tumours. The use of the fusion protein has also proved to be an effective activator of the Fcγ R, as both the fusion protein and an lgG1κ in combination and an lgG1κ alone (in the positive control system) resulted in equivalent maximal activity in both the superoxide and rosetting assays, providing further evidence that the use of the fusion protein provides an effective alternative to bispecific diabodies in terms of effector recruitment. Not only does the system described herein provide an effective alternative to monoclonal antibodies, it also offers some advantages. The fusion protein can be expressed in a fully functional form in E.coli compared to monoclonal antibodies, which must be expressed in mammalian cell culture due to the need for glycosylation of the CH₂ domain. E.coli offers several advantages over mammalian culture including rapid growth, high expression levels and ease of transformation (Verma et. al, 1998) and the need for less sophisticated production facilities with no subsequent testing for any retroviral contamination, all leading to lower production costs in terms of a therapeutic product.

Utilising serum immunoglobulins circulating in vivo to mediate effector functions rather than a single monoclonal antibody expressed in mammalian cell culture should also enable recruitment of the full range of effector functions as the fusion protein has the ability to bind to all antibody classes (e.g. IgM, IgG, IgE and IgA). The glycosylation state of the immunoglobulin has been shown to have a profound effect on its ability to recruit complement and bind the Fc receptors (Wright and Morrison, 1997) and monoclonal antibodies expressed in cell culture have been shown to have altered N-linked glycosylation in several systems (Monica et. al, 1993, Borys et. al, 1993) potentially affecting the ability of the antibody to function in vivo. Finally, the large size of whole immunoglobulin (150 kD) can lead to poor tumour penetration. The fusion protein is much smaller (35 kD) and therefore would be expected to be able to penetrate the tumour much more efficiently. However this has to countered against the much shorter half life of smaller fragments leading to faster clearance of the circulation possibly limiting the amount of antibody that reaches the tumour in the first place (Baxter et. al, 1995, Viti et. al, 1999). However it is likely that the serum half-life of the scFv - protein L fusion protein would be significantly increased by virtue of binding to serum immunoglobulin as has been shown previously for an anti- immunoglobulin diabody (Holliger et. al, 1997).

It may be that the most efficient treatment of cancerous tissue would be through the recruitment of both T-cells and B-cells simultaneously (Holliger et. al, 1997). This could be achieved through the use of T-cell and B-cell superantigen fusions to the same or different scFvs or Fabs against cell- surface antigens on the same tissue administered simultaneously. Alternatively a diabody could be created using with one chain of the diabody fused to a B-cell superantigen and the other fused to a T-cell superantigen creating a single molecule capable of recruiting both T-cell and Fc and complement mediated cytotoxicity. A diabody would also have an advantage as comparison of the t_{1 2} for cell surface retention of a scFv versus a diabody showed a 30-fold improvement in tumour retention (Adams, 1998).

Conclusion

The initial results presented herein have proved promising with the fusion protein able to activate all the effector functions tested including complement mediated cell lysis, superoxide production activated through the Fcγ Rl and rosetting through the Fcγ Rll receptor, and is likely able to activate Fcγ Rill mediated functions as well. The use of the fusion protein of 35 kD to recruit effector functions as an alternative to a full length monoclonal antibody of 150 kD offers several advantages, most notably ease of expression in E.coli and increased tumour penetration.

Claims

1. Isolated synthetic Ig binding domain of Protein L having enhanced binding affinity for an antibody VK domain, said binding affinity being enhanced over that of the wild type B1 binding domain of Protein L, said isolated Ig binding domain exhibiting at least one mutation of at least one amino acid residue corresponding to residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 of seq id no 1.

2. Isolated synthetic Ig binding domain of protein L according to claim 1 , comprising multiple mutations of at least one of the amino acids of the synthetic binding domain corresponding to amino acid residues corresponding to residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 of seq id no 1 , wherein multiple suitably comprises mutations to between 2-6 amino acid residues.

3. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the Ig binding domain retains the potential to form at least 4 hydrogen bonds with the VK domain, suitably up to 7 hydrogen bonds.

4. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the binding affinity is increased at least 2 times over that of wild type B1.

5. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the binding affinity is increased at least 8 times over that of wild type B1.

6. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the binding affinity is increased such that K_D of binding to VKI is lower than 100nM.

7. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the binding affinity is increased such that KD of binding to V/cI is lower than 1 ,5 nM.

8. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the binding affinity is determined using surface plasmon resonance.

9. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims 1-8, wherein increased affinity is determined by surface plasmon resonance using immobilised target binding partner of choice, e.g. antibody fragments and determination of a) increased binding expressed as increased resonance units and/or b) reduced off-rate.

10. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the VK domain is provided by an antibody or antibody fragment, wherein the antibody fragment may suitably be selected from scFv, Fab and dAb, Fv, disulphide bonded Fv, a Fab' fragment and a F(ab')₂ fragment.

11. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the wild type B1 to which the binding affinity of the synthetic binding domain is compared has the amino acid sequence of seq id no 2.

12. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the amino acid sequence of the isolated synthetic Ig binding domain of protein L exhibits at least 60% identity at amino acid sequence positions corresponding to amino acid residues 25, 28, 31 , 33-41 , 45, 49, 50, 52- 59, 61 , 62, 76 and 78 of seq id no 1.

13. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the amino acid sequence of the isolated synthetic Ig binding domain of protein L exhibits at least 92% identity at amino acid positions corresponding to amino acid residues 25, 28, 31 , 33-41 , 45, 49, 50, 52-59, 61 , 62, 76 and 78 of seq id no 1.

14. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the isolated synthetic Ig binding domain of protein L exhibits at least 50% identity with amino acid residues 21-81 of seq id no 1.

15. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the isolated synthetic Ig binding domain of protein L exhibits at least 80% identity with amino acid residues 21-81 of seq id no 1.

16. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims with a length of at least 30 amino acids.

17. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims with a length of at least 40 amino acids.

18. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims with a length shorter than 81 amino acids.

19. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims shorter than 61 amino acids.

20. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, derived from the amino acid sequence of protein L of Peptostreptococcus magnus.

21. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, mutated vis a vis the amino acid sequence of a subunit of protein L of Peptostreptococcus magnus, said subunit being selected from the group consisting of subunits B1 , B2, B3, B4, B5, C1 , C2, C3, C4 and C* of strains 312 and 3316.

22. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the affinity is increased for binding of any of VK domain subclasses K I, II, III and IV.

23. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the affinity is increased for binding of Vκl.

24. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the affinity is increased towards a V domain comprising a V sequence of a VK germline gene segment selected from DPK1, DPK4, DPK8, DPK9 and DPK22, preferably DPK4 and DPK9.

25. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the affinity is increased towards the V domain comprising a V sequence of a VK germline gene segment DPK9.

26. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation is a substitution mutation or chemical modification, preferably a substitution mutation.

27. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at one or more amino acid positions corresponding to amino acid residues T36, A37, E38, K40, T47, E49, A52, D55, T56, K58 and K59 of sequence id no 1.

28. Isolated synthetic Ig binding domain of protein L according to the preceding claim, wherein the mutation occurs at one or more amino acid positions corresponding to T36, E38, A52 and T56.

29. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation is a substitution mutation corresponding to at least one of the mutations T36I, T36Q, T36W, T36E, T36H, T36N, T36S, T36V, A37E, A37V, E38K, E38G, E38R, E38L, E38P, E38N, E38S, E38T, E38V, E38A, E38Q, K40Q, K40I, K40R, T47A, T47I, T47M, T47V, T47S, T47R, T47L, E49K, A52R, A52W, A52Y, A52R, A52G, A52K, A52L, A52Q, A52T, A52V, D55G, D55N, T56I, T56L, T56A, T56N, T56V, T56S, T56H, K58M, K58R, K59A, K59I, K59Q and K59R wherein the residue numbers are amino acid positions of sequence id no 1.

30. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at an amino acid position corresponding to T47 of the sequence id no 1 in B1 of protein L.

31. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, the expression of which in E. coli or yeast is enhanced over that of the wild type B1.

32. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the solubility of the mutant is enhanced over that of the wild type B1.

33. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at an amino acid position corresponding to T56 of the sequence id no 1.

34 Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at an amino acid position corresponding to T36 of sequence id no 1.

35 Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at a position corresponding to A52 of sequence id no 1.

36. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at an amino acid position corresponding to E38 of sequence id no 1.

37. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutation occurs at least at one or more amino acid positions corresponding to T36, E38, T47 and T56 of sequence id no 1 , optionally including A52 or at one or more of positions corresponding to T36, E38, A52 and T56.

38. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutations are at least at a combination of amino acid positions corresponding to positions T36 and T56 of seq id no 1 , or at least at a combination of positions corresponding to positions T36, A52 and T56, most suitably (T36I E38K A52R T56I).

39. Isolated synthetic Ig binding domain of protein L according to the preceding claim, wherein the mutations are at least at a combination of amino acid positions corresponding to T36 and T56 with E38 in sequence id no 1.

40. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutations are at least at a combination of amino acid positions corresponding to T36 and E38 in sequence id no 1.

41. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the mutations are at least at a combination of amino acid residues corresponding to A52 and T56 in sequence id no 1.

42. Isolated synthetic Ig binding domain of protein L according to any of claims 1- 38, wherein amino acids corresponding to positions (T36I, E38F, T47S, A52L, T56L) of sequence id no 1 are present as follows: (I36, T38, S47, L52, L56), (T36E, E38T, T47V, A52Y, T56I), (T36H, E38Y, T47V, A52R, T56V) or (T36Q, E38R, T47L, A52R, T56V), (T36I, E38K, T47S, A52R , T56I), (T36I, E38K, T47S, A52R, T56I), (T36I, E38K, T47L, A52R, T56I).

43. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein a mutation occurs at a position corresponding to Y53F of seq id no 1 , with the proviso said mutation allows hydrogen bond formation of the type created by Y53 when B1 of wild type protein L binds Ig.

44. Isolated synthetic Ig binding domain of protein L according to any of claims 1-42, wherein the amino acid residue corresponding to Y53 of sequence id no 1 is Y.

45. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the amino acid residue corresponding to Q35 of sequence id no 1 is Q.

46. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, wherein the amino acid residue corresponding to F39 of sequence id no 1 is F or W.

47. Isolated synthetic Ig binding domain of protein L according to any of the preceding claims, additionally comprising a mutation in one or more positions corresponding to positions A37, K40, E49, D55 of seq id no 1.

48. Polypeptide comprising at least one isolated synthetic Ig binding domain of protein L according to any of the preceding claims.

49. Polypeptide according to claim 48 comprising multiple isolated synthetic Ig binding domains of protein L according to any of claims 1-47.

50. Polypeptide according to claim 48 or 49 further comprising at least one other B subunit of Protein L.

51. Polypeptide according to any of claims 48-50 wherein 1-4 Ig binding subunits of protein L are present, suitably those subunits are selected from B2,

B3 and B4 or combinations thereof.

52. Polypeptide according to any of claims 48-50 wherein 2-5 isolated synthetic Ig binding domains of protein L are present, suitably 4 or 5.

53. Polypeptide according to any of claims 48-52, wherein the isolated synthetic Ig binding domain of protein L is preceded by GST as a fusion peptide, suitably the polypeptide comprises a dimer of isolated synthetic Ig binding domains of protein L according to any of claims 1-47.

54. Use of an isolated synthetic Ig binding domain of protein L or polypeptide according to any of the preceding claims in any of the following procedures: screening for, isolation of, purification of or immobilisation on a solid support of components consisting of or comprising a VK domain.

55. Use according to claim 54 in a binding assay such as an ELISA or Western blot.

56. Use according to claim 54 in screening an array of components consisting of or comprising a VK domain, for example an array of antibody or antibody fragments.

57. Use of a mutant or polypeptide according to any of claims 54-56 in screening for, isolation of, purification of or immobilisation on a solid support of components comprising or consisting of a VK domain with a low affinity.

58. Use of a mutant or polypeptide according to any of claims 54-57 in screening for, isolation of, purification of or immobilisation on a solid support of components comprising or consisting of a VK domain from samples in which they are present in a low concentration.

59. Complex of either an isolated synthetic Ig binding domain of protein L according to any of claims 1-47 or of a polypeptide according to any of claims 48- 53 with a VK domain, suitably a VK I, II, III or IV.

60. A complex of an isolated synthetic Ig binding domain) according to any of claims 1-47 or a polypeptide according to any of claims 48-53 with an antibody fragment lacking an Fc component.

61. A complex according to claim 60, wherein the antibody fragment is a scFv, a Fab, a single domain, a Dab, a Fv, a disulphide bonded Fv, a Fab' or a F(ab')₂.

62. A complex according to any of claims 60-61 , wherein the antibody fragment contains no VK domain.

63. A complex according to any of claims 60 to 62, wherein the antibody fragment has a lambda light chain.

64. A pharmaceutical composition comprising a complex according to any of claims 60-63, capable of eliciting a broad spectrum of antibody dependent effector functions.

65. Use of a complex according to any of claims 60-63 as active ingredient in the preparation of a medicament for tumour therapy.

66. A recombinant or isolated nucleic acid sequence encoding an isolated synthetic Ig binding domain according to any of claims 1-47 or a polypeptide according to any of claims 48-53.

67. An expression vector comprising a recombinant or isolated nucleic acid sequence according to claim 66.

68. A host cell comprising a recombinant or isolated nucleic acid sequence according to claim 66 or an expression vector according to claim 67, said host cell preferably being capable of secretion.

69. A method of generating a polypeptide according to any of claims 48-53 or an isolated synthetic Ig binding domain according to any of claims 1-47 using recombinant DNA technology and protein purification in a manner known per se.

70. A kit comprising at least one isolated synthetic Ig binding domain according to any of claims 1-47 or a polypeptide according to any of claims 48-53 in addition to instructions for carrying out an assay requiring binding to a component comprising or consisting of a VK domain, optionally said kit comprising additional reagents required for carrying out the assay.

71. A kit according to claim 70 wherein the Ig binding domain or polypeptide is immobilised on a carrier or wherein a carrier for immobilisation of protein is separately present.