WO1999023221A2 - Multivalent antigen-binding proteins - Google Patents

Abstract

A multivalent antigen binding protein comprises a single polypeptide chain comprising, in series, two or more single domain binding units which are preferably heavy chain variable domains derived from an immunoglobulin naturally devoid of light chains. Methods for their production and uses thereof, in particular for diagnosis, immunoassay and purification methods are disclosed.

Description

MU TIVAI-ENT ANTIGEN-BINDING PROTEINS

FIELD OF THE INVENTION

The present invention relates to multivalent and multispecific antigen binding proteins, methods for their production and uses thereof. In particular, the invention relates to antigen binding proteins comprising a polypeptide comprising in series two or more single domain binding units which are preferably variable domains of a heavy chain derived from an immunoglobulin naturally devoid of light chains .

BACKGROUND OF THE INVENTION

Antibodies are protein molecules belonging to a group of immunoglobulins generated by the immune system in response to an antigen. The structure of most antibody molecules is based on a unit comprising four polypeptides, two identical heavy chains and two identical light chains, which are covalently linked together by disulphide bonds. Each of these chains is folded in discrete domains. The C-terminal regions of both heavy and light chains are conserved in sequence and are called the constant regions, comprising one or more so-called C-domains . The N-terminal regions of the heavy and light chains, also known as V-domains, are variable in sequence and determine the specificity of the antibody. The regions in the variable domains of the light and heavy chains (V_L and V_H respectively) responsible for antigen binding activity are known as the hypervariable or complementarity determining regions (CDR) . Natural antibodies generally have at least two identical antigen-binding sites defined by the association of the heavy and light chain variable regions. Individual heavy or light chain domains having the capability to bind antigens have been described in the literature (Ward et al, Nature 341 (1989), 544-546) although generally most naturally occurring antibodies need both a V_H and V_L to form a complete antigen binding site and retain full immunoreactivity .

More recently, immunoglobulins capable of exhibiting the functional properties of the four-chain immunoglobulins described above but which comprise two heavy polypeptide chains and which furthermore are devoid of light polypeptide chains have been described (see European Patent Application EP-A-0584421, Casterman et al, 1994) . Methods for the preparation of such antibodies or fragments thereof on a large scale comprising transforming a mould or yeast with an expressible DNA sequence encoding the antibody or fragment are described in patent application WO 94/25591 (Unilever) .

The immunoglobulins described in EP-A-0584421, which may be isolated from the serum of Camelids, do not rely upon the association of heavy and light chain variable domains for the formation of the antigen-binding site but instead the heavy polypeptide chains alone naturally form the complete antigen binding site. These immunoglobulins, hereinafter referred to as "heavy-chain immunoglobulins" are thus quite distinct from the heavy chains obtained by the degradation of common (four-chain) immunoglobulins or by direct cloning which contribute part only of the antigen-binding site and require a light chain partner for antigen-bmdmg, thus forming a complete antigen binding site.

As described in EP-A-0584421, heavy chain immunoglobulin V_H regions isolated from Camelids (forming a complete antigen binding site and thus constituting a single domain binding site) differ from the V_H regions derived from conventional four-chain immunoglobulins in a number of respects, notably in that they have no requirement for special features for facilitating interaction with corresponding light chain domains. Thus, whereas in common (four-chain) immunoglobulins the amino acid residues at the positions involved in the V_H/V_L interaction is highly conserved and generally apolar leucine, in Camelid derived V_H domains this is replaced by a charged ammo acid, generally arginme. It is thought that the presence of charged ammo acids at this position contributes to increasing the solubility of the camelid derived V_H. A further difference which has been noted is that one of the CDRs of the heavy chain immunoglobulins of EP-A-0584421, the CDR₃, may contain an additional cyste e residue associated with a further additional cysteme residue elsewhere the variable domain. It has been suggested that the establishment of a disulphide bond between the CDR₃ and the remaining regions of the variable domain could be important in binding antigens and may compensate for the absence of light chains.

In the search for multivalent and multispecifie antigen binding proteins, attention has been directed towards the use of fragments, or portions, of a whole antibody which can nevertheless exhibit antigen binding activity. By comparison with the whole antibody, the smaller antibody fragment is advantageous for use in therapy, for example, as it is likely to be less immunogenic and more able to penetrate tissue .

Binding fragments of common (four-chain) antibodies which have been considered include Fab (light chain associated with the V_H and C_Hι domains of a heavy chain) , F_v (comprising of the V-domains of the heavy and light chains associated with each other) and ScFv (comprising a V_H domain linked to a V_L domain by a flexible peptide linker) fragments. These fragments have only one site for antigen binding compared to the two or more sites contained in the whole antibody, however, and in an attempt to overcome this problem, recombinant fragments having two or more binding sites have been proposed.

In general, those multivalent and/or multispecific constructions which have been described in the literature either comprise two or more polypeptide chains, see for example, patent application WO 94/09131 (Scotgen Limited) and WO 97/14719 (Unilever) or are based on a 'double ScFv' approach, wherein the multivalency arises when two or more monovalent ScFv molecules are linked together, providing a single chain molecule comprising at least four variable domains, as described, for example, in WO 93/11161 (Enzon Inc) and WO 94/13806 (Dow Chemical Co) . In all of these cases, the binding site is formed through the association of light and heavy chain variable domains. In WO 93/11161, reference is made to a single-chain protein comprising the binding portions of the variable regions of an antibody light (or heavy) chain but it is stated that as such proteins are comprised of two similar variable regions, they do not necessarily have any antigen-binding capability.

EP-A-0584421 (Casterman) , referred to above, discloses fragments of heavy chain immunoglobulins devoid of light chains, including fragments corresponding to isolated V_H domains or to V_H dimers linked by the hinge disulphide. Further disclosed, but not exemplified, are antibodies having different specificities on each heavy polypeptide chain which could be prepared by combining two heavy chain immunoglobulins or one heavy chain of an immunoglobulin of EP-A-0584421 with a fragment of a conventional four-chain immunoglobulin. There is no suggestion that multivalent and/or multispecific constructs may be prepared by joining together individual V_H domains. Indeed, in the absence of the inherent conformational constraints conferred on the position of the binding sites by the presence of a corresponding light chain, it might generally be expected that the binding domains in constructs of this type would sterically hinder each other, unfavourably influencing binding activity.

SUMMARY OF THE INVENTION

In a first aspect, the invention provides a multivalent antigen binding protein comprising a single polypeptide chain comprising, connected in series, two or more single domain binding units.

In another aspect, the invention provides nucleotide sequences coding for multivalent antigen binding proteins according to the invention and cloning and expression vectors comprising such nucleotide sequences. Further provided are host cells transformed with vectors comprising such nucleotide sequences and methods of producing antigen binding proteins according to the invention by expression of the nucleotide sequences in such hosts.

The invention also provides compositions comprising multivalent antigen binding proteins according to the invention.

In a further aspect, the invention provides the use of multivalent antigen binding proteins as set forth above in diagnosis or therapy or in other methods for which antibodies or fragments thereof can be used, such as in immunoassay or purification methods. Methods of treatment using the multivalent antigen binding proteins according to the invention are also provided.

In a particular embodiment of the invention, there is provided the use of said multivalent antigen binding proteins in inactivating (bacterio) phages .

By means of the invention, antigen binding proteins having the specificity and binding affinity of the whole immunoglobulin but which have the additional advantage of smaller size are obtained. Furthermore, where the multivalent antigen binding proteins of the present invention comprise variable domains having different antigen specificity, multispecific binding molecules may be obtained. Another advantage is that the constructs according to the invention may conveniently be produced at high yields economically and efficiently on a scale appropriate for industrial use. The present invention may be more fully understood with reference to the following description, when read together with the accompanying drawings. For convenience, an antigen binding protein according to the invention comprising two single binding units is herein referred to as a 'bihead'.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows a schematic representation of Camelidae IgG types.

Figure 2 shows the nucleotide sequence of the Pstl-BstEII insert of plasmid pUR4638, encoding the heavy chain variable domain of an anti-RR6 antibody (denoted R7) from a llama.

Figure 3 shows the nucleotide sequence of the Pstl-BstEII insert of plasmid pUR4640, encoding the heavy chain variable domain of another anti-RR6 antibody (denoted R9) from a llama.

Figure 4 shows the nucleotide sequence of the Pstl-BstEII insert of plasmid pUR4601, encoding the heavy chain variable domain of an anti-hCG antibody (denoted H14) from a llama.

Figure 5 shows the nucleotide sequence of the Pstl-BstEII insert of plasmid pUR4602, encoding the heavy chain variable domain or another anti-hCG antibody (denoted HI-15) form a llama. Figure 6 shows the nucleotide sequence of the Pstl-BstEII insert of plasmid pUR4603, encoding the heavy chain variable domain of an anti- Streptococcus antibody (denoted S36) from a llama.

Figure 7 shows the nucleotide sequence of the Pstl-BstEII insert of plasmid pUR4642, encoding the heavy chain variable domain of another anti- Streptococcus antibody (denoted S120) from a llama.

Figure 8 shows a map of plasmid pUR4619.

Figure 9 shows the nucleotide sequence within plasmid pUR4619, which encodes an anti-hCG-anti-RR6 bispecific biheaded antigen binding protein

(denoted H14-R9) , missing the first 4 and last 3 amino acids .

Figure 10 shows the nucleotide sequence within plasmid pUR4620, which encodes an anti-hCG-anti-RR6 bispecific biheaded antigen binding protein (denoted HI15-R7), missing the first 4 and last 3 amino acids.

Figure 11 shows the nucleotide sequence within plasmid pUR4621, which encodes an anti-hCG-anti-RR6 bispecific biheaded antigen binding protein

(denoted HI15-R9) , missing the first 4 and last 3 amino acids. Figure 12 shows the nucleotide sequence within plasmid pUR4622, which encodes a homodimeric bivalent anti-RR6 antigen binding protein (denoted R7-R7), missing the first 4 and last 3 amino acids.

Figure 13 shows the nucleotide sequence within plasmid pUR4623, which encodes a heterodimeric bivalent anti-RR6 antigen binding protein (denoted R7-R9) .

Figure 14 shows the results of an hCG/RR-6 bispecific binding assay.

Figure 15 shows the results of a RR6/RR6 bifunctional binding assay.

Figure 16 shows the binding activity and SDS-PAGE analysis of crude P. Pastoris supernatants expressing the constructs of Example 4.

Figure 17 shows a plasmid map of pHP14.3A.

Figure 18 shows the binding activity and SDS-PAGE analysis of crude H. polymorpha supernatants expressing the constructs of Example 5.

Figure 19 shows the reduction of infectivity of lactic acid bacteria phages using the constructs according to the invention.

Figure 20 shows a diagrammatic representation of the use of a biheaded antibody to form an active binding layer . Figure 21 shows the capture of I¹²⁵ labelled hCG to antibody adsorbed and double headed antibody fragment sensitised RR6-BSA wells.

Figure 22 shows a schematic representation of the use of a bispecific biheaded antibody fragment of the invention in an immunoassay to detect hCG antigen.

Figure 23 shows the assay response of latex made by adsorption of a monoclonal antibody and self- assembled bihead/RR6-BSA latex.

Figure 24 shows the assembly of a bispecific antibody fragment on a dextran surface coated with RR6-

BSA.

Figure 25 shows acidification curves of milk by the lactic acid bacterium Lactococcus lactis subsp. cremoris LM0230 with and without phage P2 and/or bihead

3850 (also referred to as bihead 3-2).

Figure 26 shows the A405 signals of an ELISA to determine bispecificity of H14-R9 biheads A405.

Figure 27 shows the nucleotide sequence within plasmid pUR4618 which encodes an anti-hcg anti-RR6 bispecific biheaded antigen binding protein

(denoted H14-R7, missing the first 4 and last 3 amino acids. DETAILED DESCRIPTION OF THE INVENTION

The invention is based on the finding that the variable domains of a heavy chain derived from an immunoglobulin naturally devoid of light chains may be joined together to form a multivalent single polypeptide which retains the antigen binding affinity of the parent whole immunoglobulin but which is much smaller in size and therefore less immunogenic, thereby providing important benefits over the use of whole antibody molecules, particularly, for example, in the area of diagnostics, therapy and targeting. Accordingly, the invention as described herein is directed to multivalent forms of antigen binding proteins, methods for preparing them and new and improved methods for their use.

As used herein, a multivalent antigen binding protein is a protein which has more than one antigen binding site.

Included within this are bivalent, trivalent, tetravalent and so on. According to one aspect, bivalent forms, that is forms having two antigen-binding sites, are preferred but it will be appreciated that higher multivalent forms may find application in certain circumstances.

A single domain binding unit means an immunoglobulin variable domain which naturally forms a complete antigen binding site.

The single domain binding units for use according to the present invention are preferably heavy chain variable domains derived from any immunoglobulin naturally devoid of light chains, such that the antigen-bmdmg site is located exclusively m the heavy chain variable domain. Preferably, the heavy chain variable domains for use m the invention are derived from immunoglobulins naturally devoid of light chains such as may be obtained from camelids as described m EP-A-0584421, discussed above.

Where the individual single domain binding units which are joined together to form the multivalent antigen binding proteins of the invention have the same antigen specificity, a binding protein which binds more than one molecule of the same type will be produced. Alternatively, multivalent and multispecific binding proteins according to the invention which are able to bind different epitopes from each other may be obtained by assembling together single domain binding units directed against different antigens. Heavy chain variable domains derived from an immunoglobulin naturally devoid of light chains having a determined antigen specificity may conveniently be obtained by screening expression libraries of cloned fragments of genes encoding camelid immunoglobulins generated using conventional techniques, as described, for example, in EP-A-0584421 and Example 1.

The multivalent antigen-bmdmg proteins of the invention may be formed by linking together the single domain binding units in series, such that each single domain binding unit is linked to at least one other variable domain.

The individual single domain binding unit may be linked sequentially by means of peptide linkers, conveniently flexible peptide linkers which allow the domains to flex in relation to each other such that simultaneous binding to multiple antigenic determinants may be achieved. It will be appreciated that the binding of the linker to the individual single domain binding unit will be such that it does not affect the binding capacity of the single domain antigen binding site. Any peptide linker which permits the single domain binding units components to be linked in such a way that each variable domain retains the binding specificity of the whole immunoglobulin from which it is derived may suitably be used. Such linkers include, e.g., peptides derived from known proteins, such as glucoamylase, cellobiohydrolase, or cell wall proteins (CWP) , or synthetic peptides which are rationally designed. The linker may suitably comprise from 1 to 400 or more amino acid residues; conveniently, the peptide linker comprises from 5 to 20 amino acid residues. This group of antigen binding proteins according to the invention with such a linker between the two single domain binding units is usually preferred because of its good production yields.

In another preferred embodiment of the invention, the individual single domain binding units may be connected directly in series without any intervening linker. In this way, the binding sites in the multivalent binding proteins according to the invention are held in much closer proximity to each other than would be the case in the whole immunoglobulin from which the immunoglobulin fragments are derived. It might generally be expected that this would give rise to unfavourable steric interactions, but surprisingly, full binding activity is found to be retained. Furthermore, these fragments with directly linked single domain binding units appear to be more stable, e.g. towards proteolytic degradation.

In an alternative embodiment, functional groups such as enzymes may be fused to the antigen binding protein.

The multivalent antigen binding proteins according to the invention may suitably find application in a wide variety of uses for which antibodies, or fragments thereof, have been proposed in the art. These uses include diagnosis, therapy, targeting, immunoassays, in agglutination, agglutination assays and purification processes, and detergents.

For use in diagnosis or therapy targeting, antigen binding proteins according to the invention having binding activity directed against both target site and the diagnostic or therapeutic agent may be constructed. Multivalent binding proteins having two or more distinct binding specificities are of particular use, for example, in the targeted delivery of therapeutic agents to their intended site of action. The binding proteins according to the invention establish the connection between therapeutic agent and target site by self-assembly, thereby avoiding the need for chemical conjugation reactions, and the therapeutic agent is guided to the target, giving increased local efficiency. Cytotoxic agents may be targeted directly to the tumour cell to be attacked, for example, by means of a bispecific, bivalent binding protein according to the invention having specificity for both the cell and the cytotoxic agent. Enzymes which are capable of generating a cytotoxic product at a target site, particularly oxido-reductases such as glucose oxidase (which catalyses the oxidation of glucose to gluconic acid, thereby producing hydrogen peroxide which exhibits cell toxicity) similarly can conveniently be delivered to the intended site of action using a binding protein according to the invention having both anti-target and anti-enzyme specificity.

Alternatively, the antigen binding proteins according to the invention may conveniently be attached to one or more appropriate diagnostically or therapeutically effective agents or carriers by methods conventional in the art.

Direct attachment of the diagnostically or therapeutically effective agent to the small antigen binding proteins of the invention runs the risk that the binding activity will be adversely affected through steric hindrance of the binding site. In a particular embodiment, therefore, the antigen binding protein has an additional polypeptide group appended to it, which additional polypeptide group does not contribute to the binding properties but which provides a "handle" for the attachment of the diagnostic or therapeutic agent.

Such antigen binding proteins with an attached polypeptide also find particular application in immunoadsorption processes, especially immunoaffinity purification processes which require that the binding protein be linked to another material, for example a label, such as an enzyme, or a solid phase, for example, a carrier material in a column.

The additional peptide group is generally attached to the antigen binding protein at or near one end of its polypeptide chain through a peptide bond, such that this polypeptide chain is prolonged by the additional peptide which now forms the terminal portion of the chain. Conveniently, the additional peptide group will be attached at its amino terminus. Suitable additional peptide linking groups and methods for their attachment are described in WO 91/08492 (Unilever). Conveniently, they comprise at least 5 but preferably not more than 20 amino acid residues and preferably include at least one lysine residue as this provides a convenient site for covalent attachment onto surfaces or proteinaceous tracers such as enzymes. A particularly suitable additional peptide linking group, particularly for coupling the antigen binding protein to a solid plastics surface, comprises the "Myc" amino acid sequence :

GLU-GLN-LYS-LEU-ILE-SER-GLU-GLU-ASP-LEU-ASN.

(see SEQ ID. NO: 1)

Coupling of the additional peptide to a solid surface such as latex particles and other structures formed from plastics material commonly used in immunoassays may conveniently be achieved by means of conventional chemical cross-linking agents. It will be appreciated that the chemical coupling site in the additional peptide should preferably be sufficiently remote from the variable domain binding sites such that the coupled molecule does not affect binding activity. Alternatively, therapeutic agents and tracers such as enzymes (for example horse radish peroxidase, alkaline phosphatase, glucose oxidase) may be covalently coupled to the additional peptide via the E-amino (epsilon) group of the lysine group. Multivalent, multispecific binding proteins according to the invention may be used to particular advantage in adsorption and purification techniques by value of their ability to exhibit specificity for two or more distinct materials. The detection and purification of ligands may conveniently be achieved using surfaces activated with constructs according to the invention. By way of illustration, a suitable support incorporating molecules for which a binding protein according to the invention has binding specificity can be activated or sensitised by coating it with a bispecific binding protein of appropriate binding specificity, the remaining binding specificity being free to bind with analyte or contaminant as appropriate depending on the intended use. Suitable molecules which can be incorporated into the support, either by adsorption or covalent bonding, include proteins, peptides, carbohydrates, DNA, RNA or conjugates thereof. Particularly preferred ligands which may be detected or purified this way include human choπonic gonadotrophm, lutemismg hormone, estrone, progesterone or metabolites thereof. The support may be particulate, planar or porous in nature. Suitable supports include those conventionally used in immuno-adsorption and purification techniques, particularly latex particles, polystyrene wells and dextran surfaces.

In an alternative aspect, binding proteins according to the invention may be used as cross-linking reagents. For example, a bispecific binding protein can link one phage to another, thereby leading to their mactivation. Inactivation of viruses or microorganisms may similarly be accomplished through agglutination. In a further embodiment of the invention, binding proteins may be used in detergent compositions, and the like, for the treatment of stains essentially as described in PCT/EP 98/03438. Thus, for example, a bispecific protein according to the invention can have high binding affinity for stain as one specificity and for enzyme as another one. Such a bispecific protein could fulfil the requirement of accumulating enzyme on stain either by supplying said protein together with enzyme as a pre-formed non-covalent complex or by supplying the two separately and allowing them to self-assemble either in the wash liquor or on the stain. A further important aspect is to use a binding protein that binds to several different, but structurally- related, molecules in a class of "stain substances". This would have the advantage of enabling a single enzyme species to bind (and bleach) several different stains. An example would be to use a binding protein which binds to the polyphenols in wine, tea, and blackberry.

Multivalent antigen binding proteins according to the invention may be prepared by transforming a host by incorporating a gene encoding the polypeptide as set forth above and expressing said gene in said host.

Suitably the host or hosts may be selected from prokaryotic bacteria, such as Gram-negative bacteria, for example E. coli , and Gram-positive bacteria, for example B. subtilis or lactic acid bacteria, lower eukaryotes such as yeasts, for example belonging to the genera Saccharomyces , Kl uyveromyces, Hansenula or Pi chia , or moulds such as those belonging to the genera Aspergill us or Trichoderma . Preferred hosts for use in connection with the present invention are the lower eukaryotic moulds and yeasts.

Techniques for synthesising genes, incorporating them into hosts and expressing genes in hosts are well known in the art and the skilled person would readily be able to put the invention into effect using common general knowledge.

Methods for producing antibody fragments or functionalised fragments thereof derived from the heavy chain immunoglobulin of Camelidae using a transformed lower eukaryotic host are described, for example in patent application WO 94/25591 and such techniques may suitably be applied to prepare constructs according to the present invention.

Proteins according to the invention may be recovered and purified using conventional techniques such as affinity chromatography, ion exchange chromatography or gel filtration chromatography.

The activity of the multivalent binding proteins according to the invention may conveniently be measured by standard techniques known in the art such as enzyme-linked immunoadsorbant assay (ELISA) , radioimmune assay (RIA) or by using biosensors.

The following examples are provided by way of illustration only. Techniques used for the manipulation and analysis of nucleic acid materials were performed as described in Sambrook et al, Molecular Cloning, Cold Spring Harbor Press, New York, 2nd Ed. (1989) unless otherwise indicated. HC-V denotes heavy chain variable domain,

Restriction sites are underlined.

EXAMPLES

EXAMPLE 1. Induction of humeral immune responses in llama.

5

Male llamas were immunised with a water in oil emulsion (1:9 V/V, antigen in water: Specol (Bokhout et al.) subcutaneously and intramuscularly. Per immunisation site 0.75-1.5 ml water in oil emulsion was inoculated containing

10 100:g antigen. The antigens used were: hCG (Sigma), azo-dye RR6 (ICI) which was coupled to BSA via its reactive triazine group and Streptococcus mutans HG982 cells. Immunisations were performed according to the following time table: The second immunisation was performed three weeks after the

15 first. The third was performed two weeks after the second immunisation. The immune response was followed by antigen specific ELISAs.

The anti-RR-6 response was measured by using Nunc Covalink

20 plates, which where coated with the azo-dye. After incubation with (diluted) serum samples, the bound llama antibodies were detected via a incubation with poly-clonal rabbit-anti-llama antiserum (obtained via immunising rabbits with llama immunoglobulines which were purified via ProtA

25 and ProtG columns; ID-DLO) , followed by an incubation with swine-anti-rabbit immunoglobulines (Dako) conjugated with alkaline phosphatase. Finally the alkaline phosphatase enzyme-activity was determined after incubation with p- nitro-phenyl phosphate and the optical density was measured

30 at 405nm. The anti-hCG response, was measured in essentially the same way using Nunc maxi-sorb plates coated with hCG.

The anti- Streptococcus response, was measured in essentially the same way using nunc maxi-sorb plates sensitised with Streptococcus mutans HG982.

EXAMPLE 2. Cloning, expressing and screening of llama HC-V fragments .

2.1 Isolation of gene fragments encoding llama HC-V domains .

From an immunised llama a blood sample of about 200ml was taken and an enriched lymphocyte population was obtained via Ficoll (Pharmacia) discontinuous gradient centrifugation. From these cells, total RNA was isolated by acid guanidium thiocyanate extraction (e.g. via the method described by Chomczynnski and Sacchi, 1987) . After first strand cDNA synthesis (e.g. with the Amersham first strand cDNA kit), DNA fragments encoding HC-V fragments and part of the long or short hinge region were amplified by PCR using specific primers :

PstI V_H - 2B 5 ' -AGGTSMARCTGCAGSAGTCWGG-3 ' (see SEQ. ID. NO: 2)

S = C and G, M = A and C, R = A and G, W = A and T,

HindiII Lam-07 5 ' -AACAGTTAAGCTTCCGCTTGCGGCCGCGGAGCTGGGGTCTTCGCTGTG GTGCG-3' (short hinge) (see SEQ. ID. NO: 3) HindiII Lam-08 5 ' -AACAGTTAAGCTTCCGCTTGCGGCCGCTGGTTGTGGTTTTGGTGTCTT

GGGTT-3' (long hinge) (see SEQ. ID. NO: 4)

Upon digestion of the PCR fragments with PstI (coinciding with codon 4 and 5 of the HC-V domain, encoding the amino acids L-Q) and BstEII (located at the 3 ' -end of the HC-V gene fragments, coinciding with the amino acid sequence Q-V- T) , the DNA fragments with a length between 300 and 400bp (encoding the HC-V domain, but lacking the first three and the last three codons) were purified via gel electrophoresis and isolation from the agarose gel.

2.2 Construction of Saccharomyces cerevisiae expression plasmids encoding llama HC-V domains.

Plasmids pUR4547 and pUR4548 are Saccharomyces cerevisiae episomal expression plasmids, derived from pSYl (Harmsen et al., 1993). From pSYl the PstI site, located in front of the GAL7 promoter was removed after partial digestion with PstI, incubation with Klenow fragment and subsequent blunt end ligation. After transformation the desired plasmid could be selected on the basis of restriction pattern analysis. Subsequently, the BstEII site in the Leu2 selection marker was removed by replacing the about 410bp A-flII/P-f2MI fragment with a corresponding fragment in which the BstEII site was removed via a three step PCR mutagenesis, using the primers : PCR-A :

P-lMI BOLI 1 5 ' -GGGAATTCCAATAGGTGGTTAGCAATCG

(see SEQ. ID. NO: 5)

(BstEII) BOLI 4 5 ' -GACCAACGTGGTCGCCTGGCAAAACG

(see SEQ. ID. NO: 6) PCR-B: (BstEII)

BOLI 3 5 ' -CGTTTTGCCAGGCGACCACGTTGGTC (see SEQ. ID. NO: 1 )

A fil l BOLI 2 5 ' -CCCCAAGCTTACATGGTCTTAAGTTGGCGT

;see SEQ. ID. NO:

PCR-A was performed with primers BOLI 1 and BOLI 4 and resulted in an about 130bp fragment with the PflMI restriction site at the 3 ' -end and the inactivated BstEII site at the 5 '-end. PCR-B was performed with primers BOLI 2 and BOLI 3 and resulted in an about 290bp fragment with the A fill site at the 5 '-end. The third PCR was with the fragments obtained from reaction A and B, together with the primers BOLI 1 and BOLI 2.

Finally, the about 1.8kb Sad-Hindi11 fragment was replaced with synthetic fragments, having sequences as presented below, resulting the plasmids pUR4547 and pUR4548, respectively. - Sacl/Hindlll fragment of pUR4547

Sad GAGCTCATCACACAAACAAACAAAACAAAATGATGCTTTTGCAAGCCTTCCCTT

CTCGAGTAGTGTGTTTGTTTGTTTTGTTTTACTACGAAAACGTTCGGAAGGGAA

M M L L Q A F L F |-> SUC2 ss PstI

TTCCTTTTGGCTGGTTTTGCAGCCAAAATATCTGCGCAGGTGCAGCTGCAGG

AAGGAAAACCGACCAAAACGTCGGTTTTATAGACGCGTCCACGTCGACGTCC L L A G F A A K I S A Q V Q L Q E |→

BstEII Hindlll

AGTCATAATGAGGGACCCAGGTCACCGTCTCCTCATAATGACTTAAGCTT TCAGTATTACTCCCTGGGTCCAGTGGCAGAGGAGTATTACTGAATTCGAA E S * * G T Q V T V S S * * HC-V cassette <- I

!see SEQ. ID. NO: 9 and NO: 10^' and

- Sacl/Hindlll fragment of pUR454!

Sad GAGCTCATCACACAAACAAACAAAACAAAATGATGCTTTTGCAAGCCTTCCTTT

CTCGAGTAGTGTGTTTGTTTGTTTTGTTTTACTACGAAAACGTTCGGAAGGAAA

M M L L Q A F L F |→ SUC2 ss

PstI TCCTTTTGGCTGGTTTTGCAGCCAAAATATCTGCGCAGGTGCAGCTGCAGG

AGGAAAACCGACCAAAACGTCGGTTTTATAGACGCGTCCACGTCGACGTCC L L A G F A A K I S A Q V Q L Q E BstEII AGTCATAATGAGGGACCCAGGTCACCGTCTCCTCAGAACAAAAACTCATC

106 + + + + + 155

TCAGTATTACTCCCTGGGTCCAGTGGCAGAGGAGTCTTGTTTTTGAGTAG S * * G T Q V T V S S E Q K L I

HC-V cassette <— | - myc tail

HindiII TCAGAAGAGGATCTGAATTAATGACTTAAGCTT 156 + + + 188

AGTCTTCTCCTAGACTTAATTACTGAATTCGAA S E E D L N * *

(see SEQ. ID. NO: 11 and NO: 12)

Both plasmids contain the GAL7 promoter and PGK terminator sequences as well as the invertase (SUC2) signal sequence. In both plasmids the DNA sequence encoding the SUC2 signal sequence is followed by the first 5 codons, (encoding Q-V-Q- L-Q = SEQ. ID. NO: 13) of the HC-V domain (including the Bstll site) , a stuffer sequence, the last six codons (encoding Q-V-T-V-S-S = SEQ. ID. NO: 14) of the HC-V domain. In pUR4547, this is followed by two stop codons, an Afill and HindiII site. In pUR4548, this sequence is followed by eleven codons encoding the myc-tag, two stop codons, an A-f-ZII and Hindi11 site.

Plasmids pUR4547 and pUR4548 were deposited under the Budapest Treaty at the Centraal Bureau voor Schimmelcultures, Baarn on 18th August 1997 with deposition numbers: CBS 100012 and CBS 100013, respectively. In accordance with Rule 28(4) EPC, or a similar arrangement from a state not being a contracting state of the EPC, it is hereby requested that a sample of such deposit, when requested, will be submitted to an expert only. Upon digesting pUR4548 with PstI and BstEII, the about 6.4kb vector fragment was isolated and ligated with the Pstl- BstEII fragments of about 350bp obtained as described above. After transformation of S . cerevisiae, via electroporation, transformants were selected from minimal medium agar plates (comprising 0.7% yeast nitrogen base, 2% glucose and 2% agar, supplemented with the essential amino acids and bases) .

2.3 Screening for antigen specific HC-V domains.

For the production of llama HC-V fragments with myc-tail, individual transformants were grown overnight in selective minimal medium (comprising 0.7% yeast nitrogen base, 2% glucose, supplemented with the essential amino acids and bases) and subsequently diluted ten times in YPGal medium

(comprising 1% yeast extract, 2% bacto pepton and 5% galactose) . After 24 and 48 hours of growth, the culture supernatant of the colonies was analysed by ELISA for the presence of HC-V fragments which specifically bind to the antigens hCG, RR6 or Streptococcus in essential the same way as described in Example 1. In this case, however, the presence of specifically bound HC-V fragments was detected by incubation with monoclonal anti-myc antibodies, followed by incubation with poly-clonal rabbit-anti-mouse conjugate with alkaline phosphatase. In this way a number of anti-hCG, anti-Streptococcus and anti-RR6 HC-V fragments have been isolated, among which are:

anti-RR6:

R7 pUR4638 (see Figure 2; SEQ. ID. NO: 15 and NO: 16) R9 pUR4640 (see Figure 3; SEQ. ID. NO: 17 and NO: 18) anti-hCG (alpha unit) :

H14 pUR4601 (see Figure 4; SEQ. ID. NO: 19 and NO: 20)

HI15 pUR4602 (see Figure 5; SEQ. ID. NO: 21 and NO: 22)

anti-Streptococcus:

S36 pUR4603 (see Figure 6; SEQ. ID. NO: 23 and NO: 24)

S120 pUR4642 (see Figure 7; SEQ. ID. NO: 25 and NO: 26)

EXAMPLE 3. Production of llama HC-V biheads by S. cerevisiae .

3.1 Construction of episomal expression plasmids encoding anti-hCG/anti-RR6 bispecific biheads.

In the anti-hCG HC-V fragments H14 and HI15 (anti-alpha- subunit) , the PstI site was removed and a Xhol site was introduced via PCR, using the primers:

MPG158WB

Xhol -GAATTAAGCGGCCGCCCAGGTGAAACTGCTCGAGTCWGGGGGA-3'

[see SEQ. ID. NO: 27] and

MPG159WB

BstEII 3 ' -CCCTGGGTCCAGTGGCAGAGGAGTGGCAGAGGAGTCTTGTTT-

!see SEQ. ID. NO: 2!

In this way the sequence: PstI

CAG GTC CAG CTG CAG GAG TCT GGG Q V Q L Q E S G

[see SEQ. ID. NO: 29 and NO: 30] became

Xhol CAG GTG AAA CTG CTC GAG TCW GGG Q V K L L E S G (see SEQ. ID. NO: 31 and NO: 32)

Upon digesting the PCR fragments with Xhol and BstEII, the about 330bp fragments were purified via agarose gel electrophoresis and isolation from the gel. The fragments were cloned into pUR4421 (see Example 1 in WO 94/25591) which was digested with the same enzymes, resulting in pJS2 (H14) and pJS3 (HI15) . Subsequently, the about 420bp EagI - HindiII fragments of pJS2 and pJS3 were isolated and ligated in the about 6.6kb BagI- HindiII vector fragment of the pSYl plasmid of which the PstI and BstEII sites were removed as described in Example 2.2. The resulting plasmids pJS7 and pJS8, respectively, were digested with BstEII and Hindlll, after which the purified vector fragment was religated in the presence of a synthetic linker having the following sequence:

BstEII PstI Hindlll

<- MPG 160 WB (49) ->

GGTCACCGTCTCCTCACAGGTGCAGCTGCAGGAGTCACTGTAATGACTTAAGCTT

CCAGTGGCAGAGGAGTGTCCACGTCGACGTCCTCAGTGACATTACTGAATTCGAA <- MPG 161 WB (48) ->

V T V S S Q V Q L Q E S L * * L K L

MPG 160 WB (49) (see SEQ. ID. NO: 33)

MPG 161 WB (48) (see SEQ. ID. NO: 34)

resulting in plasmids pJS9 and pJSlO. Finally, these plasmids were digested with PstI and Hindlll, after which the purified vector fragments of about 7.0kb were ligated with the PstI -Hindlll fragments of about 350bp of pUR4638 and pUR4640, encoding the anti-RR6 HC-V fragments R7 and R9, respectively, followed by the myc-tail. The resulting S. cerevisiae episomal expression plasmids pUR4618, pUR4619, pUR4620 and pUR4621 encode a anti-hCG-anti-RR6 bispecific bihead preceded by the SUC2 signal sequence and followed by the myc-tail.

pUR4618: SUC2 - H14 - R7 - myc (see Figure 27; SEQ. ID. NO: 35 and NO: 36) pUR4619: SUC2 - H14 - R9 - myc (see Figures 8-9;

SEQ. ID. NO: 37 and NO: 38) pUR4620: SUC2 - HI15 - R7 - myc (see Figure 10;

SEQ. ID. NO: 39 and NO: 40) pUR4621: SUC2 - HI15 - R9 - myc (see Figure 11;

SEQ. ID. NO: 41 and NO: 42)

Upon digesting these plasmids with Xhol and partially with BstEII, XhoI-BstEII fragments of about 0.7kb can be isolated and subsequently cloned into the vector fragment of pUR4547 (digested with the same enzymes) . In this way biheads can be obtained without the myc tail.

It will be appreciated that expression vectors can be constructed in which different promoter systems, e.g. the constitutive GAPDH promoter or different signal sequences, e.g. the mating factor prepro sequence are used. 3.2 Construction of episomal expression plasmids encoding antι-RR6 bivalent biheads.

Upon digesting plasmids pUR4618 and pUR4619 with BstEII (partially) and H-mdlll, DNA fragments of about ~440 and ~400 bp could be purified, respectively. These fragments were subsequently ligated with the BstEII-Hmdlll vector fragment (-6.7 kb) of pUR4638 resulting in pUR4622 and pUR4623 (see Figures 12-13; SEQ. ID. NOS : 43/44 and 45/46, respectively) , encoding a homodimeric bivalent and a hetero- dimeπc bivalent bihead, respectively.

pUR4622: SUC2 - R7 - R7 - myc pUR4623: SUC2 - R7 - R9 - myc

3.3 Production and analysis of the HC-V biheads.

After introducing the expression plasmids pUR4618 through pUR4623 into S. cerevisiae via electroporation, transformants were selected from minimal medium agar plates as described in Example 2.3. For the production of biheads, the transformants were grown overnight in selective minimal medium and subsequently diluted ten times in YPGal medium. After 24 and 48 hours of growth, samples were taken for Western blot analysis. For the immuno detection of the produced biheads via Western blot analysis, monoclonal anti- myc antibodies were used, followed by incubation with poly- clonal rabbit-anti-mouse conjugate with alkaline phosphatase.

Bi-functionality of the bispecific biheads was tested as follows : PINs coated with hCG were incubated with (diluted) medium samples. Subsequently, the PINs were incubated with a RR6- alkaline phosphatase conjugate, in which the azo-dye RR6 was coupled to the alkaline phospathase via its reactive triazine group. Finally the alkaline phosphatase enzyme- activity was determined after incubation of the PINs with p- nitro-phenyl phosphate and the optical density was measured at 405nm (see Figure 14) .

Bi-functionality of the mono-specific, bivalent biheads was tested as follows:

Nunc Covalink plates, coated with RR6 were incubated with (diluted) medium samples. Subsequently, they were incubated with a RR6-alkaline phosphatase conjugate, in which the azo- dye RR6 was coupled to the alkaline phospathase via its reactive triazine group. Finally the alkaline phosphatase enzyme-activity was determined after incubation with p- nitro-phenyl phosphate and the optical density was measured at 405nm (see Figure 15) .

EXAMPLE 4. Production of HC-V biheads by P. pastoris .

4.1 Construction of integration vectors for the expression of anti-hCG/anti-RR6 bispecific biheads.

To allow the expression and secretion of the Llama bihead constructs in P. pastoris, the gene encoding the bispecific construct was fused to the alpha-mating factor leader sequence in the commercially available P. pastoris expression vector pPIC9 (Invitrogen) . The construction of the final expression vectors involved several cloning steps. Step 1 : The construction of the bispecific HCV expression vectors required the construction of two shuttle vectors, pPIC9N and pUC.HCVx2. For pUC.HCVx2 the Hindlll/EcoRI polylinker of pUC19 was replaced with a synthetic Hindlll/EcoRI fragment, destroying the original Hindlll site, introducing a Nhel site which allows the direct fusion to the alpha-Mating Factor leader sequence in pPIC9N, and introducing the Xhol and Hindlll HCVx2 insertion sites.

Synthetic insert of pUC.HCVx2:

A S Q V K L L E

AAGCTGCTAGCCAGGTGAAACTGCTCGAGCCCGGGAAGCTTGAATTC Nhel Xhol Hindlll (see SEQ. ID. NO: 47 and NO: 48)

The synthetic linker was constructed by annealing the synthetic oligonucleotides PCR.650 and PCR.651.

PCR.650: 5' -AGCTGCTAGCCAGGTGAAACTGCTCGAGCCCGGGAAGCTTG-3'

(see SEQ. ID. NO: 49)

PCR.651 : 5' -AATTCAAGCTTCCCGGGCTCGAGCAGTTTCACCTGGCTAGC-3'

(see SEQ. ID. NO: 50)

The Xhol/Hindlll gene fragments encoding the bispecific HCV fragments were excised from pUR4619 and pUR4621 (see Example 3.1) and inserted into the Xhol /Hindlll opened pUC.HCVx2 shuttle vector, thus yielding the intermediate constructs pUC.HCV.19 and pUC.HCV21. For pPIC9N the XhoI/EcoRI polylinker of pPIC9 (Invitrogen) was replaced with a synthetic XhoI/EcoRI fragment which introduces a Nhel restriction site immediately downstream of the alpha-Mating Factor leader sequence.

Synthetic insert of pPIC9N: L E K R A S

CTCGAGAAAAGAGCTAGCCCCGGGGAATTC Xhol Nhel EcoRI

(see SEQ. ID. NO: 51 and NO: 52)

The new insert was constructed by annealing the synthetic oligonucleotides PCR.648 and PCR.649.

PCR.648 : 5' -TCGAGAAAAGAGCTAGCCCCGGGG-3'

(see SEQ. ID. NO: 53)

PCR.649: 5' -AATTCCCCGGGGCTAGCTCTTTTC-3'

(see SEQ. ID. NO: 54)

Step 2 : The final expression vectors were constructed via a three point ligation. The BamHI/Nhel fragment from pPIC9N which contains the alpha-Mating Factor encoding sequence and the Nhel/EcoRI HCVx2 inserts from pUC.HCV21 and pUC.HCV19 were cloned together into a BamHI/EcoRI opened pPIC9 vector. This resulted in the isolation of the P. pastoris transformation and expression vectors pPIC.HCV19 and pPIC.HCV21 respectively.

.2 Production and analysis of HC-V biheads.

Step 1 : Transformation and selection of transformed P. pastoris cells: P. pastoris cells were transformed essentially as described by Faber et al. Briefly: P. pastoris GS115 cells were grown overnight at 30EC in 500ml YPD medium (1% Yeast Extract, 2% Peptone, 1% Glucose) to OD₆oo=l-4. The cells were spun and 5 the pellet was washed with sterile distilled water before re-suspending in 100ml KDTT buffer (50mM Potassium Phosphate pH 7.5, 25mM DTT) . After 15 minutes incubation at 37EC the cells were pelleted (3 minuntes 3000 rpm) and re-suspended in 100ml ice-cold STM buffer (92.4g Glucose/1, lOmM Tris.HCl

10 pH 7.5, ImM MgCl₂) . After 5 washes with this buffer the cell pellet was re-suspended in a final volume of 0.5ml STM buffer. Approximately 2-5 μg DNA in 2 μl H₂0 (Bglll digested pPIC constructs: DNA purified via Phenol/Chloroform extractions and precipitation) was mixed with 70 μl of fresh

15 competent P. pastoris cells (on ice) . The cells were electroporated in a 0.2cm cuvette at 1.5 kV, 400, 25 μF in a BioRad Gene-Pulser. Immediately after electroporation, 1ml of YPD medium was added to the cells. After recovery for 1 hour at 30EC, the cells were pelleted and re-suspended in

20 200 μl IM Sorbitol and plated out onto MD plates (1.34% YNB, 4xl0^~5% Biotin, 1% Glucose, 0.15% Agar). Colonies formed by transformed cells (His⁺) were visible within 48 hours incubation at 30EC. Transformed P. pastoris cells GS115 were selected essentially as recommended by the Invitrogen Pichia

25 pastoris expression manual. The plates containing the His⁺ transformants were used to screen for the Mut⁺ and Mut^s phenotype as follows: Using sterile toothpicks, colonies were patched on both an MM plate (1.34% YNB, 4xl0^"5% Biotin, 0.5% MeOH, 0.15% Agar) and an MD plate, in a regular

30 pattern, making sure to patch the MM plate first. Approximately 100 transformants were picked for each construct. After incubating the plates at 30EC for 2-3 days the plates were scored. Colonies that grow normally on the MD plates but show little or no growth on the MM plates were classified as Mut^s clones.

Step 2 : Production and evaluation of the bispecific HC-V biheads .

Transformed and selected P. pastoris clones were induced to express bispecific antibody using the protocol outlined below:

1) Using a single colony from the MD plate, inoculate 10ml of BMGY (1% Yeast Extract, 2% Peptone, lOOmM potassium phosphate pH6.0, 1.34% YNB, 4xl0^"5% Biotin, 1% Glycerol) in a 50ml Falcon tube.

2) Grow at 30EC in a shaking incubator (250 rpm) until the culture reaches an OD₆oo=2-8.

3) Spin the cultures at 2000g for 5 minutes and re- suspend the cells in 2ml of BMMY medium (1% Yeast Extract, 2% Peptone, lOOmM potassium phosphate pH6.0, 1.34% YNB, 4xl0^"~5% Biotin, 0.5% Glycerol).

4) Return the cultures to the incubator.

5) Add 20 μl of MeOH to the cultures after 24 hours to maintain induction. 6) After 48 hours harvest the supernatant by removing the cells by centrifugation.

The crude supernatants were tested for the presence of HC-V bihead fragment via analysis on 12% acryla ide gels using the Bio-Rad mini-Protean II system (Figure 16). Bispecific binding activity via shown via ELISA as follows: 1) 96 well ELISA plates (Greiner HC plates) were activated overnight at 37EC with 200 μl/well of the BSA-RR6 conjugate (see Example 1) in PBS.

2) Following one wash with PBST the wells were incubated for 1 hour at 37EC with 200 μl blocking buffer per well. Blocking buffer: 1% BSA in PBS-T.

3) Serial dilutions of test samples (100 μl) were mixed with equal volumes of blocking buffer and added to the sensitised ELISA wells. Incubated at 37EC for 1-2 hours.

4) 200 μl hCG-AP conjugate in blocking buffer was added to each well in which the hCG was coupled to the alkaline phospathase via glutaraldehyde coupling.

5) Following one wash with PBST captured hCG-AP was detected by adding 100 μl/well pNPP substrate (lmg/ml pNPP in IM diethanolamine/lmM MgCl₂) .

EXAMPLE 5. Production of HC-V biheads by H. polymorpha .

5.1 Construction of integration vectors for the expression of anti-hCG/anti-RR6 bispecific biheads.

To allow the expression and secretion of the Llama bihead constructs in H. polymorpha strain A16 (Leu-) (Hodgkins et al) , the bispecific HC-V gene construct was fused to the alpha-mating factor leader sequence and cloned downstream of the MOX promoter in the H. polymorpha transformation and expression vector pHP14.3 (See Figure 17). A culture of E. coli cells harbouring plasmid pHP14.3 was deposited under the Budapest treaty at the National collection of Type Cultures (Central Public Health Laboratory) in London (United Kingdom) under deposition number NCTC13048. The construction of the final expression vectors involved several cloning steps.

Step 1 : Construction of the pUC19 based shuttle vector pHP.l in which the Hindlll/EcoRI polylinker is replaced with a synthetic Hindlll/EcoRI fragment, destroying the original EcoRI site and introducing a BamHI, Muni and two Bglll sites :

Synthetic insert of pHP.l:

AAGCTTAGAΓCΓGGATCCCGGGCAATTGAGAΓCΓAATTC

HindiII Bglll BamHI Muni Bglll

(see SEQ. ID. NO: 55)

The new insert was constructed by annealing the synthetic oligonucleotides PCR.448 and PCR.449.

PCR.448 : 5' -AGCTTAGATCTGGATCCCGGGCAATTGAGATCT-3' (see SEQ. ID. NO: 56)

PCR.449: 5' -AATTAGATCTCAATTGCCCGGGATCCAGATCTA-3'

(see SEQ. ID. NO: 57)

Step 2: The alpha-mating factor leader-bispecific HC-V gene from the pPIC9-HCV19 and pPIC9-HCV21 vectors were excised as BamHI-EcoRI fragments and inserted into the BamHI/MunI opened shuttle vector pHP.l giving pHPl.HCV19 and pHPl.HCV21 respectively.

Step 3: In the final cloning step, the Nhel/Bglll inserts from the intermediate constructs pHPl.HCV19 and pHPl.HCV21 were inserted into the Nhel/Bglll opened H. polymorpha transformation vector pHP14.3 yielding pHP14.HCV19 and pHP14.HCV21 respectively.

5.2 Production and analysis of the HC-V biheads.

Step 1 : Transformation and selection of transformed H. polymorpha cells:

H. polymorpha cells (strain A16) were transformed essentially as described under 4.2 except that all culturing was done at 37EC and the pHP constructs were digested with

Sfil/NotI before transformation. The plates containing the

Leu⁺ transformants were used to screen for the Mut⁺ and Mut- phenotype as described under 4.2.

Step 2 : Production and evaluation of the bispecific HC-V biheads :

Transformed and selected H. polymorpha clones were induced to express bispecific antibody using the same protocol used to express HC-V bihead in P. pastoris as described under 4.2. The crude supernatants were tested for the presence of HC-V bihead fragment via analysis on 12% acrylamide gels using the Bio-Rad mini-Protean II system (Figure 18). Bispecific binding activity via shown via ELISA (see 4.2).

EXAMPLE 6. Production of llama HC-V Triple-heads by S. cerevisiae .

Upon digesting pUR4603 and pUR4642 with BstEII and Hindlll, the about 6.8kb vector fragment can be isolated and religated in the presence of the oligonucleotides MPG160WB and MPG161WB (see Example 3.1). From the resulting plasmids, an about 0.4kb PstI fragment can be isolated, encoding the anti- Streptococcus HC-V fragments, from which the first five amino acids are lacking at the N-terminus and are fused to the C-terminus . Upon digesting either one of the plasmids pUR4618 - pUR4621 with PstI, and subsequently religating the vectors with the about 0.4 PstI fragments obtained as described above, a new set of yeast expression plasmids can be obtained. The orientation of the about 0.4 PstI fragment in the newly obtained plasmid can be determined via a digestion with BstEII. The proper orientation will result in two fragments of about 0.4kb. The wrong orientation will result in a fragment of about 0.75 and a small fragment of about 0.05kb. Plasmids with the PstI fragment in the proper orientation will encode tripleheads consisting of an HC-V fragment binding to HCG, directly followed by an HC-V fragment binding to Streptococcus, directly followed by an HC-V fragment binding to RR6 and finally the myc-tail.

It will be appreciated that it is possible (for those skilled in the art) to design alternative construction routes .

EXAMPLE 7. Reduction of the infectivity of Lactic Acid Bacteria phages by the use of llama HC-V biheads .

7.1 Coupling of RR6-dye molecules to the phage

To coat the phages with RR6 molecules 10 μl P2 phage stock (~10¹²phages/ml) + 890 μl coupling buffer ( 0. IM sodium tetraborate, 0.15M NaCl, pH 8.5) + 100 μl RR6 solution in coupling buffer, were mixed and incubated for 1 hour at 37EC (ration of 5*10⁴ RR6 molecules to 1 phage) . To inactivate non-reacted chloro-atoms of RR6, 5 μl blocking buffer ( 1.0M Tris blocking buffer pH 9, with an excess of primary amino groups) was added and incubated 30 minutes at 37EC.

7.2 Effect of biheads on the infectivity of RR6-coated phages .

To test the neutralising effect of the anti-RR6 bihead produced by S. cerevisiae containing plasmid pUR4623, this bihead was mixed (in a range of 0-300ng) with 5.5*10⁵ phages in 200 μl total volume and incubated for 0.5 hours at 37EC.

From this mixture, 100 μl was added to 100 μl of an overnight culture of Lactococcus lactis subsp. cremoris LM0230, grown in M17 (1*10⁹ cfu/ml) and spread on a plate of

M17 containing 0.5% glucose and lOmM CaCl₂. Plates were incubated overnight at 30EC. In Figure 19 it is shown that

14 ng bivalent antibody fragments give a reduction of infection of over 90%. EXAMPLE 8. Reduction of the infectivity of Lactic Acid Bacteria (LAB) phages by the use of llama HC-V biheads.

8.1 Raising heavy chain antibodies against the LAB phage P2 and obtaining antigen specific HC-V fragments

An immune response directed against the LAB P2 phages was induced and followed in essentially the same way as described in Example 1. The llama was injected several times with about 0.2 mg phage protein. From the immunised llama an enriched lymphocyte population was obtained and subsequently HC-V gene fragments were obtained as described in Example 2.1. The construction and screening of a yeast HC-V library was performed essentially as described in Examples 2.2 and 2.3.

In this way a number of anti-LAB-phage fragments were obtained. The sequence of three of these are presented below:

pUR3823:

QVQLQESGGG LVQTGGSLRL SCAASGRTSS DYSVGWFRQA PGKEREFLAV

MMLSGTGTYY ADSVKGRAAI SRDLAKNTVY LEMNSLKPED TAVYYCALDR AGWLRTEENV YDYWGQGTQV TVSS

(see SEQ. ID. NO: 58) pUR3824:

QVQLQESGGG LVQPGGSLRL SCAVSGAPFR ESTMAWYRQT PGKERETVAF

ITSGGSKTYG VSVQGRFTIS RDSDRRTVLL QMNNLQPEDT AVYYCHRALS NTWGQGIQVT VSS

(see SEQ. ID. NO: 59) pUR3825 :

QVQLQESGGG LVQPGGSLRL SCVVSGEGFS NYPMGWYRQA PGKQRELVAA

MSEGGDRTNY ADAVKGRFTI SRDNAKKTVY LQMSSLKPED TAVYYCNAAR

WDLGPAPFGS WGQGTQVTVS S

(see SEQ. ID. NO: 601

8.2 Construction of episomal expression plasmids encoding anti-LAB-phage bivalent biheads

Episomal expression plasmids encoding bivalent anti-LAB phage biheads were constructed essentially as described in Examples 3.1 and 3.2, using the above mentioned fragments as starting material. In this way amongst others the plasmids pUR3843 and pUR3850 were constructed encoding the bihead preceded by the SUC2 secretion signal.

pUR3843: SUC2 - 3823 - 3825 PUR3850: SUC2 - 3825 - 3824

The HC-V biheads 3843 and 3850 were produced by yeast transformants obtained essentially as described in Example 3.3.

8.3 Effect of biheads on the infectivity of LAB-phages

To test the neutralising effect of the anti-LAB-phage biheads produced by S. cerevisiae containing plasmids pUR3843 or pUR3850, the biheads were mixed (45 μg of 3843 and 24 μg of 3850) with 10³, 10⁶ or 10⁸ phages in 200 μl total volume. After an incubation for 0.5 hours at 37EC, 100 μl of an overnight culture of Lactococcus lactis subsp. cre-7.oris LM0230, grown in M17 (1*10⁶ cfu/ml) was added to these mixtures. Subsequently the mixture was spread on a plate of M17 containing 0.5% glucose and lOmM CaCl₂. Plates were incubated overnight at 30EC after which the number of pfu' s was estimated.

Table 1

Number of plaque forming units

Bihead (μg) 10^J phages 10^b phages 10^d phages

3843 (45) 0 < 10^J Confluent 3850 (24) 0 0 < 10³

Effect of biheads on the acidification of milk

10

In a subsequent experiment the acidification of milk upon inoculation with lactic acid bacteria at 30°C was followed by the simultaneous registration of the pH with a HP-3852A Data Acquisition logger. To this end 100 ml XVM-glucose

15 medium (= skim milk containing 0.35% yeast extract, 0.35% peptone and 1% glucose) was inoculated with 100 μl of an overnight culture of Lactococcus lactis subsp. cremoris LM0230 in XVM-glucose (10⁹ cfu/ml) and incubated for 17 h at 30°C after addition of 240 μg bihead 3850. The XVM is

20 acidified by the culture in a period of 4 h (Figure 25, graph 1) and is not influenced by the presence of the bihead. When 10³ pfu/ml P2 phage was added with the LM0230 culture in a parallel experiment, no acidification occurred during the whole period of 17 h (Figure 25, graph 2) . When

25 10³, 10⁶, or 10⁸ pfu/ml P2 phage was added, together with 240 μg bihead 3850, the acidification by the culture can be completely (in the case of 10^J cfu/ml phage) or partially restored (Figure 25, graphs 3, 4, 5). EXAMPLE 9. Activating sur aces using HC-V biheads via self assembly, for detection and purification of analytes .

9.1 Use of bispecific HC-V bihead to form an active binding layer on polystyrene wells by self assembling onto pre-adsorbed molecules.

In summary, this example shows how double headed antibody fragments can be used to form an active binding layer on polystyrene wells by self assembling onto pre-adsorbed molecules. Figure 20 shows a diagrammatic representation of this. The adsorbed antibody surface is shown (A) as is a surface sensitised with double headed antibody fragment (B) made by the self assembly of double headed antibody fragment onto a pre-adsorbed RR6-BSA surface (C) . These two surfaces are then able to bind hCG (D) .

Preparation of a reactive red 6 bovine serum albumin conjugate

A conjugate of reactive red 6 (RR6) and bovine serum albumin (BSA) was made by incubating 200 μl of RR6 (lOmg/ml in distilled H₂0) with 1ml of BSA (10mg/ml in phosphate buffered saline) with constant mixing for 3 hours at room temperature. A 200 μl solution of ethanolamine (IM in distilled H₂0) was added and the resulting solution mixed constantly for 15 minutes at room temperature. The BSA-RR6 conjugate was separated from free RR6 by application of 0.75ml to a PD10 column (Pharmacia) and the column eluted with phosphate buffered saline containing 0.1% sodium azide. The eluent was collected as 1ml fractions. Fractions 4 and 5 were red in colour and were pooled (the RR6-BSA conjugate elutes before unreacted or free RR6) . Fractions 10 onwards were also red in colour but these fractions contain the unbound RR6 and so were discarded.

Preparation of polystyrene wells adsorbed with RR6-BSA

Individual wells from Greiner high binding plates were isolated from a 96 well plate by sawing. Into these wells, 100 μl aliquots of 10 μg/ml of RR6-BSA or monoclonal antibody recognising hCG was added and incubated for 3 hours at room temperature. Both the antibody adsorbed and RR6-BSA adsorbed wells were then washed 3 times with PBSTA.

Self assembly of a double headed antibody fragment onto RR6- BSA wells Following the PBSTA wash of RR6-BSA wells, 100 μl of affinity purified double headed antibody fragment (HI15-R9: pPIC.HCV21) at 20 μg/ml in PBSTA was added per well. After 1 hour the RR6-BSA wells were washed 3 times with PBSTA.

Capture of I¹²⁵ labelled hCG to antibody adsorbed and double headed antibody fragment sensitised RR6-BSA wells hCG (2500 IU/ml in PBSTA) was spiked with 10 μCi I¹²⁵ hCG (Amersham) and then diluted to 500, 100, 20 and 4 IU/ml of hCG with PBSTA. Aliquots (100 μl) of these dilutions were incubated in the wells for 1 hour at room temperature after which time the wells were extensively washed with PBSTA. Wells were then counted on a gamma counter. Figure 21 shows the amount of dpm captured by the monoclonal antibody adsorbed wells (A) and by the double headed antibody fragment sensitised wells (B) over a range of hCG concentrations . The double headed antibody fragment sensitised wells (HI15- R9 : pPIC.HCV21) bound approximately twice as much hCG than the adsorbed antibody wells at saturating hCG concentrations. This would indicate that the double headed 5 antibody fragment sensitised wells possessed a higher density of active hCG binding sites than the adsorbed antibody wells.

9.2 Use of bispecific HC-V bihead to self assemble on 10 latex particles for the detection of human chorionic gonadotrophin (hCG)

Preparation of a RR6-BSA latex

To 950 μl of lOmM borate buffer, 0.01% erthiolate, pH 8.5 a

15 50 μl aliquot of Duke blue latex (10% solids) was added and mixed by inverting. The diluted latex was then centrifuged at 8,000 2 f°^r 10 minutes at room temperature, the supernatant removed and the pellet vortexed briefly. The pellet was re-suspended in a solution made up of 900:1 of

20 borate buffer (as above) and 100 μl of the previously prepared RR6-BSA conjugate. Latex particles were sonicated for 10 s using a sonic probe. The solution containing the latex was mixed for 30 minutes at room temperature. Following this the latex was pelleted as before and re-

25 suspended in 1 ml of the borate buffer.

Self assembly of a double headed antibody fragment onto RR6- BSA latex

By virtue of its specificity the double headed antibody

30 fragment (HI15-R9: pPIC.HCV21) self assembles on the surface of a RR6-BSA latex. This was achieved by incubating 5:1 of the RR6-BSA latex with 3:1 of supernatant from Pichia pastoris expressing the bihead HI15-R9 (pPIC.HCV21) made up to 40 μl with phosphate buffered saline containing 0.1% sodium azide and 0.1% Tween 20 (PBSTA) for 15 minutes at room temperature.

Assessing the self assembled double head for use in hCG assay hCG (10 μl at various concentrations) was added to the self assembled double head RR6-BSA latex and incubated for 15 minutes at room temperature. The mixture was then allowed to flow up a nitrocellulose strip. The nitrocellulose strip had a line of antibody recognising a different epitope to hCG than the double headed antibody fragment (Figure 22 shows a schematic representation of the principle) . The amount of latex binding at the antibody line was determined by scanning the intensity of the line using an autoreader. The results are shown in Figure 23. By way of comparison a latex was made by adsorption of a monoclonal antibody, specific for hCG, using a similar methodology as that for the RR6-BSA latex. This latex was incubated with hCG (10 μl at the various concentrations) and also subjected to the same evaluation on a nitrocellulose strip (results shown in Figure 23) .

Figure 23 shows that the self assembling latex compares well with the adsorbed antibody latex. In fact, the hook effect seen with the adsorbed latex (labelled A in Figure 23) at the higher hCG concentrations is less pronounced with the self assembling latex (labelled B in Figure 23) giving the assay a higher range of hCG detection. The most likely explanation for this is an increased number of hCG binding sites on the self assembling latex compared with the adsorbed latex.

9.3 Assembly of bispecific HC-V antibody fragments to a dextran surface coupled with RR6-BSA conjugate

Preparation of a Bialite chip coupled with RR6-BSA

A new CM5 biosensor chip (Biacore AB) was docked into a

Bialite biosensor (Biacore AB) . The flow rate, of HEPES buffered saline (HBS) was set to 10 μl/min. The RR6-BSA conjugate was amine coupled to the CM5 chip using an amine coupling kit (Biacore AB) according to the manufacturer's instructions. Briefly, two 40 μl injections of NHS/EDC were performed to activate the biosensor chip surface. Following activation, two 40 μl injections of RR6-BSA (diluted 1:10 in lOmM sodium acetate, pH 4.0) were performed. The biosensor chip surface was then blocked by two injections of 40:1 ethanolamine (IM).

Preparation of a binding surface by self assembly of double headed antibody fragment and subsequent detection of hCG Double headed antibody fragment (HI15-R9: pPIC.HCV21) was assembled onto the RR6-BSA coupled dextran surface by injection of 20 μl purified double headed antibody fragment (HI15-R9: pPIC.HCV21, 100 μg/ml in HBS). This can be seen as an increase in response units (RU) on the Bialite sensorgram shown in Figure 24 labelled A. hCG was then injected (20 μl of a 10 IU/ml solution made up in HBS) . Detection of hCG can be seen by an increase in RU (Figure 24, labelled B) . EXAMPLE 10 Anti-hCG/anti-RR6 bispecific biheads containing a linker peptide .

10.1 Construction of S. cerevisiae episomal expression plasmids encoding anti-hCG/anti-RR6 bispecific biheads containing a linker peptide.

Between the H14 and the R9 encoding DNA fragments synthetic linkers were introduced encoding different linker peptides. To this end the about 50 bp long BstEII-Hindlll fragment of pJS7 (see Example 3.1) was replaced by an about 50 bp long BstEII-Hindlll fragment having the following sequence:

MVaJA Bs tEI I Xbal Drai n PstI Hindl l l

5' GTCACCGTCTCTAGATGGCCACCAGGTGCAGCTGCAGGAGTCAACTTA 3'

(see SEQ. ID. NO: 61)

MVbJA 3' GCAGAGATCTACCGGTGGTCCACGTCGAGCTCCTCAGTTGAATTCGA 5 '

(see SEQ. ID. NO: 62)

This resulted in pSJ7a. In this plasmid the about 20 bp Pstl-Hindlll fragment was replaced with the about 370 bp Pstl-Hindlll fragment encoding the anti-RR6 HC-V fragment R9 and/with the myc-tail of pUR4640 (see Example 3.1) and resulting in pSJ7b.

Upon digesting plasmid pSJ7b with Xjal and Drain the about 7 kb vector fragment was ligated with five synthetic oligo nucleotide linker fragments presented below: MV01JA 5' CTAGTGGTACTTCCGGTTCCCAG 3'

(see SEQ. ID. NO: 63)

MV02JA 3' ACCATGAAGGCCAAGG 5'

(see SEQ. ID. NO: 64) S G T S G S Q

(see SEQ. ID. NO: 65)

MV03JA 5' CTAGTTCTTCATCTGCTTCTGCCTCTTCAGCCCAG 3'

(see SEQ. ID. NO: 66)

MV04JA 3' AAGAAGTAGACGAAGACGGAGAAGTCGG 5'

(see SEQ. ID. NO: 67)

S S S S A S A S S A Q (see SEQ. ID. NO: 68)

MV05JA 5'CTAGTGGTTCTCCAGGTTCACCAGGTCAG 3'

(see SEQ. ID. NO: 69) MV06JA 3' ACCAAGAGGTCCAAGTGGTCCA 5'

(see SEQ. ID. NO: 70)

S G S P G S P G Q

(see SEQ. ID. NO: 71)

MV07JA 5' CTAGTGCTACTACAACTGGTTCTTCACCAGGTCCAACTCAG 3'

(see SEQ. ID. NO: 72)

MV08JA 3' ACGATGATGTTGACCAAGAAGTGGTCCAGGTTGA 5' (see SEQ. ID. NO: 73)

S A T T T G S S P G P T Q

(see SEQ. ID. NO: 74)

MV09JA 5' CTAGTGCTAATCATTCTGGTAATGCTTCTCAG 3'

(see SEQ. ID. NO: 75)

MV10JA 3' ACGATTAGTAAGACCATTACGAAGA 5'

(see SEQ. ID. NO: 76)

S A N H S G N A S Q (see SEQ. ID. NO: 77) The oligonucleotide linker fragments encode the last amino acid of the N-terminal HC-V fragment (S) and the first amino acid of the C-terminal HC-V fragment, intersected by the connecting linker peptide. This resulted in plasmids pUR5330 to 5334, respectively.

After transformation of S. cerevisiae with these plasmids, the production levels of the biheads were determined via Western blot analysis and a anti-hCG ELISA using anti-myc Ab for detection of the bound bihead (see Example 2.3). Production levels are presented in Table 2 below:

Table 2

The production levels of the biheads in which the two HC-V domains are separated by a linker peptide (consisting of between 5 and 11 amino acids) were found to be 3 to 5 times higher as found for the bihead in which the two HC-V fragments are connected without a peptide linker.

It is therefore expected that other linker peptides, e.g. the short hinge regions found in the heavy chain antibodies are equally suitable and give even better production yields .

Finally, the bispecificity of the biheads was demonstrated using the ELISA as described in Example 3.3, the results of which are presented in Figure 26.

Literature referred to in the Examples

Chomczynnski, P. and Sacchi, N. (1987) Single step method of RNA isolation by acid guanidium thiocyanate-phenol- chloroform extraction. Analytical Biochem. 162: 156-159.

Bokhout, B. A. , Van Gaalen, C. , and Van Der Heijden, Ph. J. , (1981), A selected water-in-oil emulsion: composition and usefulness as an immunological adjuvant. Vet. Immunol. Immunopath., 2: 491-500

Bokhout, B. A., Bianchi , A. T. J. , Van Der Heijden, Ph. J. , Scholten, J. . and Stok, . , (1986), The influence of a water-in-oil emulsion on humoral immunity. Comp. Immun. Microbiol. Infect. Dis., 9: 161-168.

Giuseppin , M.L .F. , Lopes , M.T . S . , Planta, R. J. , Verbakel , J.M.A. , Verrips, C.T. (1991) Process for preparing a protein by a yeast transformed by multicopy integration of an expression vector. PCT application WO 91/00920 (UNILEVER)

Harmsen, M.M. , Langedijk, A.C., van Tuinen, E., Geerse, R.H., RauP, H.A., Maat, J. , (1993) Effect of pmrl disruption and different signal sequences on the intracellular processing and secretion of Cyamopsis tetragonoloba galactosidase by S. cerevisiae. Gene 125 115-123

Faber, K.N., Haima, P., Harder, W. , Veenhuis, M and Geert, A.B. , (1994) Highly efficient electrotransformation of the yeast Hansenula polymorpha . Current Genetics, 25: 305-310

Hodgkins , M. , Mead, D., Ballance, D.J., Goodey, A. and Sudbery, P., (1993) Expression of the Glucose Oxidase Gene from Aspergillus niger in Hansenula Polymorpha and its use as a reporter gene to isolate regulatory mutations. Yeast, 9:625-635

BUDAPEST TREATY ON THE INTERNATIONAL

RECOGNITION OF THE DEPOSIT OF MICROORGANISMS

FOR THE PURPOSES OF PATENT PROCEDURE

INTERNATIONAL FORM

_T0Dr Paul van der Logt 1 Head of BioRecognition Unit RECEIPT IN THE CASE OF AN ORIGINAL DEPOSIT Unilever Research issued pursuant to Rule 7.1 by the Colworth House INTERNATIONAL DEPOSITARY AUTHORITY Sharnbrook identified at the bottom of this page

IDENTIFICATION OF THE MICROORGANISM

Identification reference given by the Accession number given by the DEPOSITOR: INTERNATIONAL DEPOSITARY AUTHORITY: p HP 14.3a in E. coli XL-1 Blue NCTC 13048

II. SCIENTIFIC DESCRIPTION AND/OR PROPOSED TAXONOMIC DESIGNATION

The microorganism identified under I above was accompanied by: I I a scientific description

r^*γ i a proposed taxonomic designation

(Mark with a cross where applicable)

III. RECEIPT AND ACCEPTANCE

This International Depositary Authority accepts the microorganism identified under I above, which was received by it on 26th August 1997 (date of original deposit) 1

IV. INTE-WATIO-.AI. DEPOSITARY AUTHORITY same: National Collection of Type Culture. Signature (s) of person (s) having the power to represent the International Depositary Authority or of authorized official (β) :

Adress: Central Public Health Laboratory S 61 Colindale Avenue Date : ^18/11/97 B a-ta-.

LONDON, NW9 5HT . Clinical Scientist

Where Rule 6.4(d) applies, such date is the date on which the status of International depositary authority was acquired; where a deposit made outside the Budapest Treaty after the acquisition of the status of international depositary authority is converted into a deposit under the Budapest Treaty, such date is the date on which the microorganism was received by the international depositary authority.

Form BP/4 (sole page) BUDAPEST TR ATY ON THE INTERNATIONAL

RECOGNITION OF THE DEPOSIT OF MICROORGANISMS

FOR THE PURPOSES OF PATENT PROCEDURE

INTERNATIONAL FORM

Unilever Research Laboratoπum Vlaardingen RECEIPT IN THE CASE OF AN ORIGINAL DEPOSIT

Postbus 114 issued pursuant to Rule 1 1 by the

3130 AC VLAARDINGEN INTERNATIONAL DEPOSITARY AUTHOR ITY

Nederland identified at the bottcm j. -hi . p& .

name and address of depositor

I . IDENTIFICATION OF THE MICROORGANISM

Iden I fi ;at ion reference 3i.%n by the Accession nunc-: giver by the DEPOSITOR INTERNATIONAL DEPOSIT.'PY ^UTHOP.IT.

E. coli JM109-pUR4547 CBS 100012

II . SCIENTIFIC DESCRIPTION AND/OR PROPOSED TAXONOKIC DESIGNATION

The mi -toorganism identified under I above was accompanied by: X a scientific description j a proposed ta.ononu designation imaric with a cross where applicable)

III. RECEIPT AND ACCEPTANCE

This International Depositary accepts the microorganism identified under I above, wh I ;h was receiv e d by it on Monday, 18 August 1997 -be o iginal JepoiiC

IV. RECEIPT OF REQUEST FOR CONVERSION

The microorganism identified under I above was received by this Interna ional De os nary Authority on not applicable .dace of the original deposit* and a request to convert the original deposit to a deposit under the Budapest Treaty was :c-.-.ιv.- by it on not applicable (date of receipt of re iest for conversion!

V . INTERNATIONAL DEPOSITARY AUTHORITY

Name : Centraalbureau voor Schimmelcultures Signatureis) ot perεon(s) having the power to represent the International Depositary Authority or ;>£ authori-zed official Is) -

Ad ress : Oosterstraat 1 P.O. Box 273 3740 AG BAARN Mrs F.B. Snippe-Claus dr R.A.& Samson The Netherlands Date: Friday, 29 August 1997 rfhere Rule 6.4(d) applies. such o -, is the date on which the sc_.u<; of interna ion l depositary authority was acquired.

Focm BP-4 (Sule page) BUDAPEST TREATY ON THE INTERNATIONAL

RECOGNITION OF THE DEPOSIT OF MICROORGANISMS

FOR THE PURPOSES OF t-ATENT PROCEDURE

INTERNATIONAL FORM

Unilever Research Laboratorium Vlaardingen RECEIPT IN THE CASE OF AN ORIGINAL DEPOSIT

Postbus ll4 Issued pursuant to Rule 7 . 1 by the

3130 AC VLAARDINGEN If-rTERNATIONAL DEPOSITARY AUTHORITY

Nederland identified at the- bottύtri of this pa i-

name and address of depositor

'where Rule ό.-l(d) applies, such Jat- it- rhe date on which the status j internet icna ! depositar. authority was acquit"*..

:ora BP/ . i ee page i • ^■p.- ?1

Claims

Claims

1. A multivalent antigen binding protein comprising a single polypeptide chain comprising, in series, two or more single domain binding units.

2. A protein according to claim 1, wherein the single domain binding units are heavy chain variable domains derived from an immunoglobulin naturally devoid of light chains.

3. A protein according to claim 1 or claim 2, wherein the single domain binding units are heavy chain variable domains derived from a Camelid immunoglobulin.

4. A protein according to ^"any ones of claims 1 to 3, comprising a bivalent antigen binding protein.

5. A protein according to any one of claims 1 to 4, wherein the single domain binding units are connected directly in series.

6. A protein according to any one of claims 1 to 4, wherein the single domain binding units are connected in series by a peptide linker.

7. A protein according to any one of claims 1 to 6, wherein the protein has an additional peptide appended to it.

8. A protein according to any one of claims 1 to 7, wherein the protein has an enzyme fused to it. y

9. A protein according to any one of claims 1 to 8, wherein the single domain binding units have different antigen specificity from each other.

10. A protein according to any one of claims 1 to 8, wherein the single domain binding units have the same antigen specificity as each other.

11. Nucleotide sequences coding for the polypeptide of the multivalent antigen binding protein of any one of the preceding claims.

12. An expression vector comprising a nucleotide sequence according to claim 11.

13. A host cell transformed with a vector according to claim 12 and capable of expression of the nucleotide sequences to produce the polypeptide of the multivalent antigen binding protein.

14. A process for preparing a multivalent antigen binding protein according to any one of claims 1 to 10 comprising transforming a host by incorporating a gene encoding said polypeptide and expressing said gene in a host.

15. Use of a multivalent antigen protein according to any of claims 1 to 10 in diagnosis, therapy, targeting, immunoassays, cross-linking methods, including agglutination, or for purification processes.

6. Use of a multivalent antigen protein according to any one of claims 1 to 10, for inactivation of (bacterio) phages or viruses.