AU8238291A - Binding domains - Google Patents

Description

BINDING DOMAINS

The present invention relates to molecules comprising binding domains and methods for their preparation and use. In particular the present invention relates to molecules comprising domains which are synthetic analogues of a natural single variable domain of a member of an immunoglobulin family or superfamily. The present invention also relates to methods for designing said molecules comprising domains, the molecules so designed, and kits and methods for use of said molecules in therapy and diagnosis.

Antibodies and other members of the immunoglobulin superfamily, such as T cell receptors, have the ability to recognise molecules eg. antigens, specifically and bind them with high affinity. In naturally occurring antibodies, the binding site for antigen is formed by the juxtaposition of variable (V) domains from both heavy (H) and light (L) chains. Within each of these chains are three stretches of amino acids, the complementarity determining regions (CDRs), which comprise the residues which interact with antigens. The three CDRs alternate with four framework regions (FR). Winter et al. demonstrated that a single V domain was able to bind antigen with high affinity and specificity [Ward et al. Nature 341, 544-546 (1989)]. It was proposed that these single domain antibodies (VH) would have advantages for

ET several applications owing to their small size (relative molecular mass (rmm)=13,000) compared to whole antibodies (rmm=160,000) and other antibody fragments.

However, VH domains have unique disadvantages that disappointingly limit their utility. The difficulties encountered reflect at least two properties of VH domains which may be related. They are expressed in low quantities when cloned in bacteria (ca. 200μg/ml of culture supernatant compared to lOmg/ml for Fv fragments) and during purification of VH domains, substantial amounts of material are lost. Concentration of the VH single domain, for example using ultrafiltration and purification on chromatography columns, often leads to poor recoveries. This probably reflects non-specific binding to surfaces. This has been observed directly by Ward et al. (1989) supra., who isolated a high proportion of single-domain antibodies exhibiting significant (and non-specific) binding to plastic. In vivo, non-specific binding in tissues would lead to poor performance in applications such as tumour imaging or cancer therapy studies.

These difficulties are not generally experienced with whole antibodies or with fragments of antibodies such as Fv or Fab fragments. Therefore, the problems appear to be a characteristic of antibody fragments containing unpaired single domains.

Thus, the present invention seeks to ameliorate any of the above or other problems associated with single variable domain binding members, whilst retaining antigen binding for their full commercial exploitation.

The present applicant has realised that the most likely cause of the unfavourable properties of single domain antibodies is the exposure to aqueous solvent, of the hydrophobic face of a single variable domain eg. the VH single domain. In native antibodies, this face interacts with the adjacent hydrophobic face of the VL domain and is buried within the antibody molecule. Exposure of the face would lead to strong interactions with surfaces, for example chromatography matrices, from which material could not be recovered. Additionally, the present applicant believes that exposure of the face to aqueous solvent may lead to decreased stability of the single variable domain eg. the VH single domain, leading to turn to unfolding and loss of binding activity during purification procedures. Most importantly, this hydrophobic face is a potential source of non-specific binding, which considerably limits the utility of these single variable domain molecules in vivo and in vitro.

Although the VH single domain is quite small, its activity depends on a complex tertiary structure comprising interactions between disparate parts of the molecule [Chothia, C. et al. J. Mol. Biol. 186, 651-663 (1985)]. In several cases, there are side-chains in the framework which have important interactions with those in the antigen combining site. For example, in the anti- progesterone antibody DB3, tryptophan 47 (framework) has

S been shown to contact progesterone (Arevalo, J.H. Taussig, M. and Wilson, I. personal communication), a -hydroxy vitamin K in the -hydroxy vitamin K-bindi antibody IgGl NEW [Twining, S.S. and Atassi, M.Z. J Biol. Chem. 253, 5259 (1978)]. In some antibodies interactions of residue 71 (framework) with CDR residue have been shown to be important for the maintenance o antigen binding [Chothia, C. et al. Nature 342, 877-88 (1989)]. It will be apparent to those skilled in th art, that any modifications to the basic antibody o variable domain structure will most likely hav pleiotropic consequences on the tertiary structure of th antibody or variable domain and render them unable t bind antigen with the desired specificity and affinity. This is exemplified in recent experiments with anti lysozyme single-domain antibodies. A single amino aci change, substituting Asn 35 with His, was found t improve the expression level in E.coli by about 1000 fold. Critically however, the altered molecule was foun to bind lysozyme only very weakly (E.S.Ward, L.Reich ann, G.P.Winter, personal communication). Because the interactions responsible for assembly of the active structure, particularly of VH domains, are not completely understood, any amino acid changes made to the basic structure must be selected extremely carefully. Selection of any changes is further complicated by the fact that the residues at the interface of the VH single domain with the VL domain, come from the CDRs as well as

^S the framework. The framework residues (for example, residues 37,39,45,47,91,93 and 103) at the interface of the anti-lysozyme antibody D 1.3 [Amit, A.G. et al, Science 233 747-753 (1986), Chothia, C. et al., Science 233, 755-758 (1986)] are highly conserved in antibodies of all species. They are invariant or, rarely, have substitutions which are usually conservative, i.e. replacement of the amino acid with one having similar chemical properties. For example, aliphatic hydrophobic residues are usually substituted by similar aliphatic residues. Thus, if substitutions are made in these framework residues, there is a strong risk of disrupting the structure of the molecule and hence binding of antigen by the antibody or single variable domain eg. VH. The CDR residues vary between antibodies and determine the specificity and affinity of binding to antigen. Since the applicant wishes to retain the capacity to bind a variety of antigens with high affinity and specificity, the ability to vary these CDR residues needs to be retained.

The present invention therefore provides a molecule comprising a binding domain having a polypeptide sequence which is an analogue of part or all of a naturally occurring molecule comprising a binding domain having specificity for a particular binding member, in which analogue one or more amino acids are altered to reduce the hydrophobicity of said analogue as compared to the natural binding substance.

SUBSTITUTE SHEET The analogue may have substantially the same bindin characteristics as the natural binding substance eg. wit respect to specificity, affinity, or avidity. In som cases these characteristics may be improved. Th alteration may be any amino acid alteration which reduce hydrophobicity of the polypeptide sequence eg. amino aci substitution, deletion or addition.

The molecule comprising a binding domain ma comprise an antibody or other receptor molecule and fragments and derivatives of antibodies and receptor molecules. In particular, the molecule may comprise a single variable domain of the type that is present in an antibody molecule. The alteration may be in a complementarity determining region and/or in a framework region. Preferably, the alteration is in a framework region. Where the alteration reducing hydrophobicity, is in a framework region, the complementarity determining regions may also be altered by way of amino acid substitution, deletion, addition or inversion to alter the specificity and or binding characteristics of the binding substance.

The molecule comprising a binding domain may comprise any one or more of the alterations described in examples 2 to 11. Thus the present invention provides a single chain variable domain which is a synthetic analogue of another single chain variable domain of a member of an immunoglobulin family or superfamily, and in which analogue, one or more interface amino acids residues of the domain is altered as compared to the said another domain, in which a said altered amino acid is substituted with a residue which occurs in an analogous position in a member of an immunoglobulin family or superfamily, so that the analogue is more hydrophilic than the said another domain.

The altered amino acid residue may be in a framework region. The altered amino acid residue may be in a complementarity determining region. The synthetic analogue may have essentially the same binding activity as the said another domain. The sequence of a complementarity determining region may be additionally altered by way of amino acid substitution, deletion, addition, or inversion, to alter the specificity and/or binding characteristics of the analogue as compared to the said another domain.

The single chain variable domain may be a synthetic analogue of a single variable immunoglobulin heavy chain domain. In which case, the one or more of the amino acid residues 37,39,45,47,91,93 and 103 may be altered. The amino acid alterations may comprise one or more of the following: i) substitution of valine 37 with glutamine or threonine; ii) substitution of glutamine 39 with glutamate; iii) substitution of leucine 45 with glutamine; iv) substitution of tryptophan 47 with aspartate or glycine; v) substitution of tyrosine 91 with threonine serine or methionine; vi) substitution of alanine 93 with serine o glutamate; vii) substitution of tryptophan 103 with glutamat or tyrosine or threonine; viii) substitution of valine 37, leucine 45, tryp tophan 47, alanine 93 and/or tryptophan 10 with any of asparagine, threonine or serine; ix) substitution of valine 37 with threonine an glutamine 39 with glutamate and tryptophan 4 with glycine; x) substitution of tyrosine 91 with serine o methionine and alanine 93 with glutamate and tryptophan 103 with threonine. The single chain variable domains according to the present invention may be coupled to a further molecular moiety. The further molecular moiety may be an enzymic-, fluorescent-, or radio-label, or a portion of an immunoglobulin.

The present invention also provides diagnostic kits which comprise a single chain variable domain as described above together with one or more ancillary reagents for carrying out the diagnostic test.

The present invention also provides therapeutic composition which comprises at least a single chain variable domain as described above. The composition may also comprise one or more excipients.

In an aspect of the invention there is provided single chain variable domains eg. VH single chain domain frameworks, with improved properties as indicated above and which allow the generation of specificities for various binding partners by substitution of different CDRs into the new framework. The applicant also provides that the framework residues of molecules comprising single domains previously isolated are substituted to make them more polar. The modified molecule should retain the ability to bind the desired antigen. Preferably, the changes would not render the interface immunogenic when administered to humans.

The present invention also provides a method for making a single chain variable domain which is a hydrophilic synthetic analogue of another single chain variable domain of a member of an immunoglobulin family or superfamily, which comprises:

(i) inspecting the interface regions of a said single chain variable domain to identify hydrophobic amino acid residues; and

(ii) producing a said analgue of said single chain variable domain in (i) in which one or more of said hydrophobic residues is substituted with a less hydrophobic residue which occurs in an analogous position in a member of an immunoglobulin family or superfamily.

The method may comprise:

(a) obtaining the nucleotide sequence encoding one or more of the identified hydrophobic amin acid residues;

(b) using site directed mutagenesis to alter th nucleotide sequence to introduce a triple coding for the substitute amino acid,

(c) using the altered nucleotide sequence in recombinant expression system to express th synthetic analogue.

In the method, more than one amino acid residue ma be substituted. The substitute amino acids may be derived from naturally monomeric members of the immunoglobulin superfamily. The naturally monomeric member may be Thy-1. The synthetic analogue may have essentially the same binding activity as the said another domain.

Where the molecule comprising a binding domain is an immunoglobulin, or a fragment or derivative of an immunoglobulin, an amino acid site suitable for alteration may be identified by: i) examining the molecule for hydrophobic amino acids expected to be on the surface of the binding substance, ii) more specifically examining the amino acid residues related to those expected to be buried at the interface of the heavy and light chain domains of an immunoglobulin, iii) examining those residues identified in (ii) that would be exposed to the solvent when the single domain is used separately. An appropriate alteration to make may be identified with reference to the homologous amino acid sequences of members of a family of related substances. For example, the amino acid sequence of a molecule at the site for alteration identified as described above, may be altered to make it homologous at that site in one or more members of the family group.

The family of related substances may comprise a family of immunoglobulins, fragments and derivatives thereof. Alternatively, the family of related substances may comprise a family of proteins containing domains structurally related to the immunoglobulins ie. an immunoglobulin superfamily.

The nucleotide sequence may be altered by site directed mutagenesis using oligonucleotides designed to introduce the required alterations. Alternatively, the alteration may be achieved by use of the technique known as polymerase chain reaction.

The present invention also embraces kits having molecules comprising binding domains as herein provided. The kits may be diagnostic, purification or catalysis kits. The present invention further embraces pharmaceutical products which contain molecules comprising binding domains according to the present invention. TERMINOLOGY Binding Domain

This describes an area on the surface of a protein,

^SUBSTIT or a cavity which specifically binds to, and is therefor defined as complementary with a particular spatial an polar organisation of another molecule. A domain i folded within itself and independently of other parts o the same protein and independently of a complementar binding member. Immunoglobulin

This describes a group of structurally relate proteins consisting of two pairs of heavy polypeptid chains and two pairs of light polypeptide chain, al linked together by disulphide bonds. They have a bindin domain for another molecule such that a give immunoglobulin binds specifically to that another molecule. The protein can be natural or partly or wholly synthetically produced. The term also covers any protein having a binding domain which is homologous to an immunoglobulin binding domain.

Herein the numbering of immunoglobulin amino acid residues is according to Kabat E.A. et al. in "Sequences of Proteins of Immunological Interest" U.S. Department of Health and Human Services 1987. Antibody

This describes an immunoglobulin whether natural or partly or wholly synthetically produced. The term also covers any protein having a binding domain which is homologous to an immunoglobulin binding domain. These proteins can be derived from natural sources, or partly or wholly synthetically produced.

Example antibodies are the immunoglobulin isotypes and the Fab, F(ab^**-)2 ^{s Fv}/ ^Fv, dAb, Fd fragments. Immunoglobulin Superfamily This describes a family of polypeptides, the members of which have at least one domain with a structure related to that of the variable or constant domain of immunoglobulin molecules. The domain contains two β- sheets and usually a conserved disulphide bond (see A.F. Williams and A.N. Barclay 1988 Ann. Rev Immunol. 6_ 381- 405.

Example members of an immunoglobulin superfamily are CD4, platelet derived growth factor receptor (PDGFR), intercellular adhesion molecule. (ICAM). Except where the context otherwise dictates, reference to immunoglobulins and immunoglobulin analogs in this application includes members of the immunoglobulin superfamily and analogs thereof. Interface This describes the region on a given heavy or light chain of an immunoglobulin which associates with the complementary heavy or light chain. Framework

Each chain of an immunoglobulin has a constant (C) and a variable (V) region. Each V region is made up from three complementarity determining regions (CDR) separated by four framework regions (FR). The CDRs are variable stretches of amino acid sequences and provide the function of binding to another molecule. It is t possibility of variability which provides immunoglobuli with various binding specificities. The FRs ar substantially constant stretches of amino acid sequence which interpose the CDRs.

In order that the present invention is more full understood, it will now be described in more detail firstly in general outline and secondly with reference t specific examples provided by way of illustration onl and not by way of limitation. The following descriptio refers to the figures in which:

Figure 1 shows the nucleotide and amino acid sequence o the VH domain of the anti-lysozyme antibody D1.3 clone in pUC119; Figure 2 shows mutant oligonucleotides for substitutio of residues found in naturally occurring immunoglobuli heavy chains;

Figure 3 shows mutant oligonucleotides for substitutio with homologous residues from Thy-1; Figure 4 shows mutant sequences obtained by substitutions of Thy-1 residues into VHD1.3;

Figure 5 shows oligonucleotides for the random substitution of asparagine, serine or threonine into VHD1.3; Figure 6 shows a graph illustrating lysozyme binding activity of TGI (control), VHD1.3, VHTHY-1 and VHTHY-2; Figure 7 shows a graph illustrating lysozyme binding activity of pUC119 (control), VHD1.3, VHMutTrp, VHMutLeu 15 and VHThy-3;

Figure 8 shows a graph illustrating lysozyme binding activity of pUC119 (control), VHD1.3, VHThy-1, Thy-2, VHThy-1, Thy-3 and VHMutWD; Figure 9 shows the nucleotide sequence of the vector fdPs/Bs around the cloning site and indicates the Pstl and BstEll restriction sites; and

Figure 10 shows a graph illustrating lysozyme binding activity of vector fdPs/Bs (control), fdVHThylThy2, fdVHD1.3 and phage antibody D1.3.

The applicant has devised three related strategies for the selection of alterations to the frameworks. This invention enables the generation of antibodies and single variable domains eg. VH domains with improved properties for in vivo and in vitro use.

1. Substitution with VH Interface Residues Found in Naturally Occurring Immunoglobulin Heavy Chains

There is considerable amino acid sequence homology between different immunoglobulins. Homologies are detected by lining up different sequences one above the other, and sliding the chains along relative to one another, until the best level of identity between the different sequences is arrived at. These analyses are generally performed on a computer. As noted above, the framework residues are highly conserved, ie. particular amino acids will be present in the same positions in a series of different antibodies. Rare substitutions do occur, however, and the applicant searched for naturally- occurring substitutions of VH interface residues. Thi was performed by reference to readily availabl compilations of antibody sequences [e.g. Rabat, E.A. e al. in "Sequences of Proteins of Immunological Interest U.S. Department of Health and Human Services (1987)] This analysis enables natural variants at any position t be identified. As these variant residues occur naturall (though infrequently) in antibodies, the applican realised that they are less likely to severely disrup domain structure. These natural substitutions most ofte occur once per antibody molecule. However, the applican also combines the available substitutions from severa antibodies, together in the same molecule. It is possible that different combinations of natural substitutions can be used.

2. Substitution with Residues that have been Identified at the Homologous Position in Other Proteins of the Immunoglobulin Superfamily

Of particular interest, are proteins which have domains containing the immunoglobulin fold (a tertiary structure characteristic of antibody domains that has been found in other proteins), but which do not associate with another domain.

Examples of molecules which contain a single domain homologous to immunoglobulin variable domains include: Thyl, Po myelin, CD7, CD28 and CTLA-4 [Williams, A.F. and Barclay, A.N. Ann. Rev. Immunol. 6_, 381-405 (1988)].

Other proteins contain more than one unpaired, antibody-like domain. CD4 and MRC OX-2 each contain N- terminal domains homologous to variable domains of antibody molecules ('V'-type domains, one in MRC 0X2, two in CD4), and a C-terminal domain homologous to constant domains of antibody molecules ('C'-type domain, one each in MRC 0X2 and CD4) [Williams, A.F. and Barclay, A.N. Ann. Rev. Immunol. j>, 381-405 (1988)].

The similarity in tertiary structure between these unpaired domain proteins and the domains comprising antibodies is reflected to some extent in homologies at the amino acid sequence level. Amino acid homologies are assigned in much the same way as in 1 above, although the alignment of sequences in this case can be problematic. For Thy-1, the alignment with VH residues 37,39,91 and 93 is relatively straightforward. However, VH residues 45 and 47 have been published in two different alignments with Thy-1 [A.F. Williams and J. Gagnon, Science 216 696- 703, 1982; A.F. Williams and A.N. Barclay, Ann. Rev. Immunol. 6_ 381-405, 1988]. Exploration of alternative substitutions may be necessary to identify the most appropriate, in cases where there is more than one possible alignment.

There is a possibility that the modified VH interface exposed in the VH single domain would be antigenic in humans. This would be a disadvantage for in vivo therapeutic use. Substitution with residues which occur naturally in human Thy-1 may reduce this potential antigenic response.

SUBSTITUTE SHEET 3. Semi-Random Insertion of Polar Amino Acids a Interface Residues

This strategy enables the improvement of singl domains using a less direct method. Oligonucleotides for mutagenesis are synthesized with a mixture of bases in some positions to give ambiguities for amino acid insertion at certain triplets. For example, the present applicant has designed a strategy which will allow the random insertion of the highly polar residues asparagine, serine or threonine by the use of codons with ambiguities at the second position. These residues could be inserted for instance at the interface positions 37,45,47,93 and 103 of the VH domain. The 243 possible frameworks resulting from this mutagenesis can then be screened to identify which of the semi-random combinations have the desired properties. Strategies for screening the resulting semi-random population would include estimation of antigen binding affinity and non-specific binding by ELISA (see below). This is one example of a number of strategies which can be used to vary interface residues in a semi-random manner without recourse to the strategies outlined in 1) and 2) above.

The present invention allows the binding affinity and specificity of identified antibodies to be incorporated into single domain molecules with improved properties for in vivo and in vitro applications.

The frameworks generated using the model system using the VH domain of antibody D1.3 (VHD1.3) can be used as frameworks for antibodies of any specificity by replacement of part(s), or all, of its CDRs with those from an antibody molecule of desired affinity and specificity. There are numerous ways in which this can be accomplished. For instance, following the determination of the sequence of the CDRs of the antibody of desired properties (eg., binding specificity), an oligonucleotide or series of oligonucleotides is synthesized which encodes these CDRs and the framework regions containing the nucleotide substitutions necessary to make the antibody/antibody domain more polar. This oligonucleotide can then be amplified using PCR, cloned into a suitable vector such as pUC119, and the product expressed in bacteria.

Alternatively, the relevant changes at the interface can be introduced into an existing single domain antibody of desired specificity to improve the properties of that antibody. This can be achieved by a variety of methods, for example by site-directed mutagenesis, or PCR (for example using the method of Hemsley, A. et al. [Nuc. Acids Res. 1 . 6545-6551 (1989)].

In addition to this, the improved single domain antibodies show reduced non-specific. The properties of the single chain variable domains provided hereby enable selection of single domain antibodies of the desired specificity and affinity, for example, using single domains cloned into fd phage. The frameworks described

^SUBSTITUTE SHEET in this invention could be cloned into fd phage and th existing CDRs replaced by repertoires of CDRs to create new population of single domain antibody molecules tha can be screened for the desired binding specificities. Alternatively, for instance, single domain antibodies ma be isolated as described by Ward et al (1989, supra), bu may require improvements in their affinity an specificity. The CDRs from these antibodies can be cloned into the polar frameworks described in this invention and inserted into single domains cloned into fd phage. Random mutagenesis of these CDRs can then be performed and selection of antibodies of the desired affinity and specificity performed using affinity methods. Examples 15 and 16 herein show that a derivative of VHD1.3 with a more polar framework can be displayed on phage as a fusion with gene III protein (McCafferty J. et al. 1990 Nature 348 p552-554) with retention of binding activity. Display on phage would enable the generation of combinations of substitutions at framework residues by semi-random mutagenesis procedures (an example of which is given in example 11) and the subsequent selection of those with favourable binding properties.

Since the improved single domain antibodies of the present invention constitute superior versions of conventional single domain antibodies, they can be used in many of the ways as are immunoglobulins (Ig) and their superfamily of molecules or fragments. For example, Ig molecules have been used in research, therapy (e.g. cancer therapy, modulation of immune status and therapy of diseases caused by pathogens), diagnosis (for example, estimation of hormonal status), in modulation of activities of hormones or growth factors, in detection, in biosensors, in catalysis, in purification of other molecules and in screening regimes for therapeutic compounds in the pharmaceutical industries. The lower non-specific binding of the improved single domain antibodies should prove especially useful for the above applications.

The increased hydrophilicity may be of particular importance for their use as binding molecules in affinity chromatography, especially weak affinity chromatography (Zopf, D. and Ohlson, S., Nature 346 87-89, 1990).

Anti-idiotypic improved single domain antibodies can also be made. Anti-idiotypic specificities [Methods Enzymol. 178. J.J. Langone ed. Academic Press (1989)] are made in a two-stage process. Firstly, antibody A directed against a particular antigen or epitope is itself used to raise other antibodies. A proportion of the anti-A antibodies, antibodies B, will be directed against the antigen combining sites of antibody A, such that the antigen combining sites of B are complementary to that of A. In effect, the antigen combining site of antibody B, the anti-idiotype, mimics in structure the original antigen or epitope recognised by antibody A. The original antigen can be a protein or any other compound, for example a carbohydrate or a steroid, and the antibody used at any stage in the procedure could be an improved single domain antibody. The final anti- idiotypic antibody can be an improved single domain antibody produced as described herein, or a molecule of the immunoglobulin superfamily from which the anti- idiotypic determinant(s) are transferred into an improved single domain antibody framework.

Such anti-idiotypic molecules are advantageous in a variety of applications [Methods Enzymol. 178. J.J. Langone ed. Academic Press (1989)]. These include vaccines for treating cancers and diseases caused by bacteria, viruses and parasites. They may be used for blocking cellular receptors for the aforementioned pathogens as well as blocking cellular receptors for hormones. They may also be advantageous in diagnostic procedures, for example in place of antigen or peptide in ELISA. Anti-idiotypic specificities are known to be useful in the pharmaceutical industries [Methods Enzymol. 178. J.J. Langone ed. Academic Press (1989)].

The present invention relates to improved single domain antibodies and receptors derived from molecules of the immunoglobulin (Ig) superfamily, methods for selecting and effecting said improvements, and methods for and kits use of said antibodies or receptors in research, therapy, diagnosis, purifications, catalysis and discovery of novel therapeutics. EXAMPLE 1. PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF VALINE 37 WITH GLUTAMINE

Figure 1 shows the nucleotide and amino acid sequence of the pUC119 VHD1.3 clone used for mutagenesis studies (Ward et al, 1989, supra). The amino acid residues in the VH domain that interact with the VL have been identified as 37,39,45,47,91,93 and 103 (Amit et al, (1986) supra, Chothia, C. et al (1986) supra). Amino acids occurring naturally in VH domains were surveyed using a compilation of immunoglobulin sequences [for example, Rabat, E.A. et al. in "Sequences of Proteins of Immunological Interest" U.S. Department of Health and Human Services (1987)]. Substitutions have been found at positions 37,39,45,47,91,93 and 103 in naturally occurring heavy chains. The most polar substitutions at each of these residues in antibodies sequenced to date were chosen for replacement of interface residues by mutagenesis in the following examples.

Residue 37 is valine in 385 out of 434 sequences surveyed. In 48 other sequences, an aliphatic amino acid was substituted. In the single remaining example, a case of human heavy chain disease, a significantly more polar residue, glutamine, was substituted. An oligonucleotide was designed for incorporation of this change in VHD1.3 (VHMUTVAL, shown in figure 2). This mutation and the other mutations described in the examples below can be assembled in different combinations to make further novel derivatives. In Vitro Mutagenesis (1) The oligonucleotides detailed in fig. 2 were synthesised on an Applied Biosystems 391 DNA synthesiser and purified on Urea-acrylamide gels using standard techniques [Sambrook, J. et al. in "Molecular Cloning: a laboratory manual (second edition)". Cold Spring Harbor Laboratory Press, 11.23] prior to in vitro mutagenesis.

(2) Preparation of single stranded DNA template for mutagenesis. The VH D1.3 antibody gene (Ward et al, 1989, supra) used to exemplify this invention was carried on plasmid pUC 119 (Sambrook, J. et al. in "Molecular Cloning: a laboratory manual (second edition)". Cold Spring Harbor Laboratory Press 1.14). Single-stranded template DNA was prepared by infecting TGI cells carrying the plasmid with M13 K07 helper phage using standard techniques for growth and purification (Sambrook, J. et al. in "Molecular Cloning: a laboratory manual (second edition)". Cold Spring Harbor Laboratory Press 4.46). (3) Site-directed mutageneses were performed using the "In vitro Mutagenesis system, Oligonucleotide directed (version 2)" (Amersham International) exactly as per manufacturers instructions. Ampicillin-resistant colonies resulting from the mutageneses were grown overnight in 2YT (2YT=per litre of water, 16g Bacto- tryptone, lOg Yeast extract, 5g NaCl) containing lOOμg/ml ampicillin. These cultures were diluted in fresh 2YT and single stranded template DNA prepared by M13 K07 infection as in (2) above. Mutants were verified by DNA sequencing using the Sequenase version 2.0 kit. EXAMPLE 2 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF GLUTAMINE 39 WITH GLUTAMATE

Using similar logic and methods to those described in example 1, an oligonucleotide VHMUTGLN (fig. 2) was designed for the introduction of a glutamate residue at position 39. This substitution is found in 1 of 420 heavy chains surveyed. Glutamate is considered to be marginally more polar than glutamine [Rose et al, Science 229 834-838, 1985]. EXAMPLE 3 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORR BY SUBSTITUTION OF LEUCINE 45 WITH GLUTAMINE

Using similar logic and methods to those described in example 1, an oligonucleotide VHMUTLEU (figure 2) was designed which introduces a substitution of leucine with glutamine at residue 45 (fig. 2). This modification is found in 2 of 402 sequences surveyed (396 have leucine at this position). One of the antibodies containing glutamine at this position is a mouse antibody specific for anti-Bl, 6D-galactan. EXAMPLE 4 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF TRYPTOPHAN 47 WITH ASPARTATE Using similar logic and methods to those described in example 1, an oligonucleotide VHMUTWD (figure 2) was designed for the introduction of aspartate at position 47. This substitution is found in one of 392 heavy chains surveyed.

EXAMPLE 5 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF TYROSINE 91 WITH THREONINE Using similar logic and methods to those describe in example 1, an oligonucleotide VHMUTTYR was designe for the introduction of threonine at position 91. Thi substitution is found in one of 398 heavy chain surveyed.

EXAMPLE 6 PREPARATION OF VH D1.3 WITH A MORE POLA FRAMEWORK BY SUBSTITUTION OF ALANINE 93 WITH SERINE

Using similar logic and methods to those described in example 1, an oligonucleotide VHMUTALA (figure 2) was designed for substitution of alanine 93 with serine. This substitution is found in 4 of 410 heavy chains surveyed. One of these is in a mouse anti-B2,l fructosan.

EXAMPLE 7 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF TRYPTOPHAN 103 WITH GLUTAMATE OR TYROSINE

Using similar logic and methods to those described in example 1, two oligonucleotides VHMUTTRP and VHMUTWY (figure 2) were designed for the introduction of glutamate and tyrosine respectively at position 103. These substitutions are found once each in 308 heavy chains surveyed. Glutamate is much more polar than tryptophan. Tyrosine, although more polar than tryptophan is a more conservative substitution. EXAMPLE 8 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF RESIDUES IN POSITIONS 37,39 AND 47 WITH HOMOLOGOUS RESIDUES FROM THE IMMUNOGLOBULIN FAMILY PROTEIN THY-1

SUBSTITUTE SHEET Thy-1 is a single domain protein of the immunoglobulin superfamily. Alignment of the residues of Thy-1 with those of immunoglobulin heavy chains has been performed by Williams (A.F. Williams and J. Gagnon Science 216 696- 703 1982; A.F. Williams and A.N. Barclay Ann. Rev. Immunol. 6 381-405, 1988). Although residues 37,39,91 and 93 of the VH domain interface were aligned with the same residues of Thy-1 in both publications, the residues at positions 45 and 47 were aligned with different residues reflecting the lower degree of homology of adjacent amino acids from these positions. The oligonucleotide VHTHY-1 (figure 3) was designed to incorporate the most polar residues at positions 37,39 and 47 found at homologous Thy-1 positions using the alignment published by Williams & Gagnon (1982), supra). Mutagenesis was performed as in example 1. The amino acid substitutions generated are shown in figure 4. EXAMPLE 9 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORK BY SUBSTITUTION OF RESIDUES IN POSITIONS 91,93 AND 103 WITH HOMOLOGOUS RESIDUES FROM THE IMMUNOGLOBULIN FAMILY PROTEIN THY-1

Using the strategy described in example 8, an oligonucleotide VHTHY-2 (figure 3) was designed to incorporate the most polar substitutions of VH residues 91,93 and 103 at homologous Thy-1 residues [Rabat, E.A. et al. in "Sequences of Proteins of Immunological Interest" U.S. Department of Health and Human Services (1987). Another oligonucleotide VHTHY-3 (figure 3) was

SUB designed to incorporate the residues found at these positions in rat brain Thy-1 which appears most polar overall at residues homologous to the VH interface. Mutagenesis is as described in example 1. The amino acid changes generated are shown in figure 4.

EXAMPLE 10 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORR BY SUBSTITUTION OF RESIDUES IN POSITIONS 37,39,47,91,93 AND 103 WITH HOMOLOGOUS RESIDUES FROM THE IMMUNOGLOBULIN FAMILY PROTEIN THY-1 The amino acid changes detailed in examples 8 and 9 were combined by performing a mutagenesis experiment as in example 1 using the DNA sequence of mutant protein VHTHY-1 as template and mutant oligonucleotides VHTHY-2 and VHTHY-3 (figure 3) to incorporate site directed changes.

EXAMPLE 11 PREPARATION OF VH D1.3 WITH A MORE POLAR FRAMEWORR BY SUBSTITUTION OF RESIDUES IN POSITIONS 37,45,47,93 AND 103 WITH ASPARAGINE, THREONINE OR SERINE Oligonucleotides are prepared containing the triplet GXT at each of the positions 37,39,45,47,91,93 and 103 (where X is a random mixture of the bases C,G and T; figure 5). Use of these as mutagenesis primers as in example 1 would generate the insertion of Ser, Thr and Asn respectively depending on which base was incorporated. Derivatives generated are then screened for antigen binding and improved properties. EXAMPLE 12: ASSESSMENT OF ANTIGEN BINDING STATUS OF VHTHY-1 AND VHTHY-2 MUTANT SINGLE DOMAIN ANTIBODIES VH D1.3 interface mutants VHThy-1 and VHThy-2 constructed as described in example 8 and 9 were assessed for lysozyme binding activity. Antigen binding status of mutant single domain antibodies was determined by ELISA (Enzyme Linked Immuno adsorbent Assay) according to techniques well known in the art.

This is just one of a whole range of methods that can be used to measure antigen-antibody binding. Others include Western blotting, competitive radioimmunassay and fluorescence quench.

The ELISA for lysozyme binding by mutant single domains was undertaken as follows:

Overnight cultures of ampicillin-resistant clones were diluted 1 in 10 into fresh 2YT (with lOOμg/ml ampicillin) and grown for lhr at 37^βC. Isopropyl β-D- thiogalactopyranoside (IPTG) was added to ImM final concentration and the cells cultured for a further 24-30 hrs at 37°C. Supernatants were prepared by centrifugation for use directly in the ELISA. 1) The plates (Falcon microtest III flexible plate) were coated with 200μl per well of lmg/ml hen egg lysozyme in 50mM NaHC03, pH 9.6 overnight at room temperature.

2) The wells were rinsed with three washes of phosphate buffered saline (PBS), and blocked with 300μl per well 2% skimmed milk powder in PBS for two hours at 37°C.

3) The wells were rinsed with three washes of PBS and 200μl of culture supernatant were added and incubated for two hours at room temperature.

4) The wells were washed three times with 0.05% Twee 20 in PBS and three times in PBS.

5) 200μl of a suitable dilution (1 in 1000) of rabbit polyclonal antiserum against the Fv fragment in 2% skimmed milk powder in PBS was added to each well and incubated for two hours.

6) Washes were repeated as in (4).

7) 200μl of a suitable dilution (1 in 5000) of goat anti-rabbit antibody (Sigma) coupled to horseradish peroxidase, in 2% skimmed milk powder in PBS, was added to each well and incubated at room temperature for one hour.

8) Washes were repeated as in (4). 9) 200μl 2,2' azino-bis (3-ethylthiazolinesulphonic acid) [Sigma] (0.55mg/ml in citrate buffer (citrate buffer comprises: 54mls 50mM citric acid plus 46mls 50mM trisodium citrate per lOOmls), ^'with lμl 30% hydrogen peroxide:water per 10 mis) was added to each well and the colour allowed to develop for up to ten minutes at room temperature.

The reaction was stopped by adding 0.05% sodium azide in 50mM citric acid pH4.3. ELISA plates were read at 405nm in a Titertek Multiskan M.C. to give an optical density reading for each well. The optical density reading is in proportion to the amount and affinity of the primary antibody used in the ELISA, in this case a single domain antibody. The result shown in figure 6 demonstrates that these mutants had retained their ability to bind lysozyme. The VHThy-1 mutant appears to have higher affinity/quantity of lysozyme binding activity than the parent VH D1.3, whereas mutant VHThy-2 has slightly less.

Given the complexity of the problem (see above), this result was surprising and reinforces the point that the properties of VH single domain antibodies can surprisingly be improved without unduly compromising antigen binding, as taught in this document.

EXAMPLE 13 ASSESSMENT OF ANTIGEN BINDING STATUS OF VHMUTTRP, VHMUTLEU AND VHTHY-3 MUTANT SINGLE DOMAIN ANTIBODIES

VHD1.3 interface mutants VHMutLeu, VHMutTrp and VHThy-3 constructed as described in examples 3,7 and 9 respectively were assessed for lysozyme binding activity., Antigen binding status of mutant single binding domains was determined by ELISA as described in example 12. The result shown in figure 7 demonstrates that these mutants had retained their ability to bind lysozyme.

EXAMPLE 14 ASSESSMENT OF ANTIGEN BINDING STATUS OF VHTHY- 1, THY-2; VHTHY-1, THY-3 AND VHMUTWD MUTANT SINGLE DOMAIN ANTIBODIES VHD1.3 interface mutants VHThy-1, Thy-2; VHThy-1, Thy-3 and VHMutWD constructed as described in examples 10 and 4 respectively were assessed for lysozyme binding activity. Antigen binding status of mutant single binding domains was determined by ELISA as described in example 12. The result shown in figure 8 demonstrates that these mutants had retained their ability to bind lysozyme. Thus, extensive changes (6 amino acid substitutions) can be made at the VH interface increasing the polarity of the domain without affecting the ability to bind lysozyme. The approach used to select substitutions taught in this document may be expected to be applicable to any VH domain.

EXAMPLE 15 CLONING OF GENE ENCODING VHTHY-1, THY-2 INTO fdPs/Bs FOR DISPLAY ON PHAGE

The gene encoding the derivative VHThy-1, Thy-2, generated in example 10 by using the DNA sequence of the mutant protein VHThy-1 as template and the mutant oligonucleotide VHThy-2 to incorporate site directed changes, was subcloned into the vector fdPs/Bs for display of this VH domain on phage as a fusion with gene III protein. The vector fdPs/Bs is similar to fdCATl (McCafferty, J. et al, 1990 Nature 348 p552-554) except that it contains Pstl and BstBll restriction sites for cloning (Figure 9).

A miniprep of pUC119 VHThy-1, Thy-2 DNA was prepared using standard procedures (Sambrook et al, 1989 supra). The VHThy-1, Thy-2 encoding sequences were amplified by PCR using the primers RVHTHYFOR and KSJ6. RVHTHYFOR 5' TGA GGA GAC GGT GAC CGT GGT GCC TTG GCC AGT G 3' This incorporates a BstEll site at the 3' end of the VHThy-1, Thy-2 gene. KSJ6

5' AGG TGC AGC TGC AGG AGT CAG G 3' This incorporates a Pstl site at the 5' end of the VHThy- 1, Thy-2 gene.

PCR was performed using 20mM Tris (pH7.3 at 70^βC), 50mM KC1, 4mM MgCl₂, 0.01% gelatin with lOμM each oligonucleotide, ImM each dNTP, 5 units Taq polymerase and approximately 50ng pUC119VHThy-l, Thy-2 DNA in a total volume of lOOμl. The product of the PCR reaction was ethanol precipitated and resuspended in 20μl lOmM Tris, pH8.0, 0.1 mM EDTA, A lOμl portion was digested using Pstl (20 units) and BstEll (20 units) in NEB buffer 2 in a total volume of 50μl at 37°C for 2h (restriction enzymes obtained from New England Biolabs, CP Labs, Bishops Stortford). Following digestion, the reaction mixture was phenol extracted and ethanol precipitated. The product was electrophoresed on a 1% agarose Tris- acetate-EDTA gel and the band of approximately 350bp excised and the DNA purified using Geneclean (Bio 101, La Jolla, California). Vector DNA (fdPs/Bs RF form) was prepared using standard procedures (Sambrook et al, 1989 supra). This DNA (1.2μg) was digested with Pstl and BstEll (50 units) in lOOμl NEB buffer 3 at 37°C for 90 min. The products were phenol extracted and ethanol precipitated and the resuspended DNA was phosphatased as described by Sambrook et al, (1989 supra). A preparative 0.7% Tris-borate-EDTA agarose gel was performed and th band of approximately 9kb excised and the DNA purifie using Geneclean and resuspended in 10 μl lOmM Tris, pH8.0, O.lmM EDTA. Ligation was performed using 5μl eac of the digested vector and insert DNA using 200 units o T4 DNA ligase in lOμl NEB ligase buffer. The ligatio mixture (8μl) was transformed into competent E.col MC1061 cells prepared according to Sambrook et al, 1989 supra) and the mixture plated on 2YT agar containing 20μg/ml tetracycline. Colonies were picked, single stranded DNA was prepared (Sambrook et al, 1989 supra) and the DNA was sequenced using a Sequenase 2.0 kit (United Sates Biochemical, Cleveland, U.S.A.). The sequence of the insert corresponded to VHThy-1, Thy-2. The derivative has been named fdVHThy-1, Thy-2.

A clone of VHD1.3 in fdPs/Bs was prepared starting from pUCll9VHD1.3. The insert encoding VHD1.3 was prepared by digestion of pSWl-VHD1.3-TAGl (Ward E.S. et al., 1989 supra) with Pstl and BstEll. Other procedures were as above. This derivative has been named fdVHD1.3. EXAMPLE 16 ELISA ASSAY OF VHTHY-1, THY-2 DOMAIN DISPLAYED ON PHAGE

The fdVHThy-1, Thy-2 phage constructed in example 15 was shown to be functional in the binding of the antigen, lysozyme, using an ELISA assay.

Viral particles were prepared by growing E.coli MC1061 cells containing fdVHThy-1, Thy-2; fdVHD1.3; phage antibody D1.3 (displaying scFvD1.3; McCafferty, J. et al, 1990 Nature 348 p552-554) or fdPs/Bs in 50ml 2YT medium containing 15μg/ml tetracycline for 16 to 24h. The culture supernatant was collected by centrifugation for 10 min at lOOOOrpm in an 8 x 50ml rotor. Phage particles were precipitated by adding 1/5 volume 20% polyethylene glycol (PEG)/2.5M NaCl and leaving at 4^βC for lh. Phage particles were pelleted by centrifugation for 15 min as described above and the pellets resuspended in sterile lOmM Tris, pH8.0 ImM EDTA containing 1% gelatin to 1/40 the of the original volume.

1. ELISA plates were coated with lysozyme and blocked with PBS containing skimmed milk powder as described in example 12.

2. Wells wer*-*¹ rinsed with PBS. 3. Concentrated phage (200μl) was added to each well as appropriate and incubated at room temperature for 2h.

4. The wells were washed three times with 0.5% Tween 20 in PBS and three times with PBS.

5. Sheep anti-M13 serum (200μl; 1 in 1000) in PBS containing 2% skimmed milk powder was added to each well and incubated for lh.

6. Washing was repeated as in 4.

7. Peroxidase conjugated rabbit anti-goat immunoglobulin (200μl; 1 in 5000; Sigma) was added and incubated for lh.

8. Washes were repeated as in 4.

9. Peroxidase substrate was added as in example 12 and colour allowed to develop for lh. Both fdVHThy-1, Thy-2 and fdVHD1.3 gave ELIS signals 4 to 5 times above the value obtained wit fdPs/Bs, whereas with phage antibody D1.3, the signal wa approximately 8 times that with fdPs/Bs (Figure 10). Thus modification of the VH interface residues does not affect the ability of the domain to bind lysozyme when displayed on phage.

It will be appreciated by those skilled in the art that the present invention has been described above by way of example only, and that considerable modifications to the procedure may be made to effect a similar outcome without departing from the scope of the invention.

Claims (18)

1. A single chain variable domain, which is a synthetic analogue of another single chain variable domain of a member of an immunoglobulin family or superfamily, and in which analogue, one or more interface amino acid residues of the domain is altered as compared to the said another domain, in which a said altered amino acid is substituted with a residue which occurs in an analogous position in a member of an immunoglobulin family or superfamily, so that the analogue is more hydrophilic than the said another domain.

2. A single chain variable domain according to claim 1 in which a said altered amino acid residue is in a framework region.

3. A single chain variable domain according to claim 1 or claim 2 in which a said altered amino acid residue is in a complementarity determining region.

4. A single chain variable domain according to any one of claims 1 to 3 wherein the synthetic analogue has essentially the same binding activity as the said another domain.

5. A single chain variable domain according to claim 2 or claim 3 in which the amino acid sequence of a complementarity determining region is additionally altered by way of amino acid substitution, deletion, addition, or inversion, to alter the specificity and/or binding characteristics of the analogue as compared to the natural domain.

6. A single chain variable domain according to any one of claims 1 to 5, which is a synthetic analogue of a single variable immunoglobulin heavy chain domain.

7. A single chain variable domain according to claim 6, in which one or more of the amino acid residues 37, 39, 45, 47, 91, 93 and 103 is altered.

8. A single chain variable domain according to claim 6 or claim 7 in which the amino acid alterations comprise one or more of the following: i) substitution of valine 37 with glutamine or threonine; ii) substitution of glutamine 39 with glutamate; iii) substitution of leucine 45 with glutamine; iv) substitution of tryptophan 47 with aspartate or glycine; v) substitution of tyrosine 91 with threonine, serine or methionine; vi) substitution of alanine 93 with serine or glutamate; vii) substitution of tryptophan 103 with glutamate tyrosine or threonine; viii) substitution of valine 37, leucine 45, tryptophan 47, alanine 93 and/or tryptophan 103 with any of asparagine, threonine or serine; ix) substitution of valine 37 with threonine and glutamine 39 with glutamate and tryptophan 47 with glycine; x) substitution of tyrosine 91 with serine or methionine and alanine 93 with glutamate and tryptophan 103 with threonine. .

9. A single chain variable domain according to any one of claims 1 to 8 which is coupled to a further molecular moiety.

10. An immunoglobulin single chain variable domain according to claim 9 wherein the further molecular moiety is an enzymic-, fluorescent-, or radio-label, or a portion of an immunoglobulin.

11. A diagnostic kit which comprises a single chain variable domain according to any one of claims 1 to 10, together with one or more ancillary reagents for carrying out the diagnostic test.

12. A therapeutic composition which comprises at least a single chain variable domain according to any one of claims 1 to 10.

13. A method for making a single chain variable domain which is a hydrophilic synthetic analogue of another single chain variable domain of a member of an immunoglobulin family or superfamily, which comprises:

(i) inspecting the interface regions of a said single

SUBSTITUTE SHEET chain variable domain to identify hydrophobic amino acid residues; and

(ii) producing a said analogue of said single chain variable domain in (i) in which one or more of said hydrophobic residues is substituted with a less hydrophobic residue which occurs in an analogous position in a member of an immunoglobulin family or superfamily.

14. A method according to claim 13 which comprises:

(a) obtaining the nucleotide sequence encoding one or more of the identified hydrophobic amino acid residues;

(b) using site directed mutagenesis to alter the nucleotide sequence to introduce a triplet coding for the substitute amino acid, (c) using the altered nucleotide sequence in a recombinant expression system to express the synthetic analogue.

15. A method according to claim 13 or claim 14 wherein more than one amino acid residue is substituted.

16. A method according to any one of claims 13 to 15 wherein the substitute amino acids are derived from naturally monomeric members of the immunoglobulin superfamily.

17. A method according to claim 16 wherein the naturally monomeric member is Thy-1.

18. A method according to any one of claims 13 to 17 wherein the synthetic analogue has essentially the same binding activity as the said another domain.

SUBSTITUTE SHEET