Recombinant proteins have been expressed in transgenic plants by many groups. A consistent problem is low levels of expression, which complicates purification and processing. Various approaches have been taken to overcome this, for example using different regulatory gene constructs or constructing synthetic genes, with limited success. Thus the levels of expression routinely reported are up to 0.5% total soluble protein, with many examples being less than 0.01%. One notable exception has been antibody expression in plants. Our experience here is that levels of 1% are normally expected, with 5-8% also achieved. The reason for these differences is unclear, but we believe it to be related to the inherent stability of antibody molecules in plants, as well as the manner in which these proteins are handled by the plant cell machinery.

Immune complexes comprise antibodies and antigens, and give rise to the classical immune response. Immune complexes are conventionally produced by mixing antigen with antibody and allowing the molecules to associate via the binding site of the antibody and the specific epitope of the antigen. The disadvantage of this process is that immune complexes in vitro which require careful mixing of antigen with antibody at optimal concentrations.

The present invention provides a method of increasing the production of a protein in a cell comprising expressing recombinant DNA encoding a recombinant fusion protein comprising a desired protein fused to the constant region of a heavy chain of an antibody or to part of said constant region.

For convenience, the term'heavy chain constant region'will be used throughout the following description to refer either to the full length immunoglobulin heavy chain

constant region or to a part (fragment) of this region, as the context permits. The constant region comprises a number of regions (domains) depending of the class of the antibody. For an IgG antibody the constant region comprises three regions (domains) known as CH1, CH2, and CH3; the CH3 domain being the C-terminal domain. Thus instead of using the full length constant region, truncated versions of this may be used comprising the conjoined CH1/CH2, CH2/CH3 or CH1/CH3 domains.

By using the high expression of the immunoglobulin, we have found it possible to drive up the expression levels of the recombinant fused protein.

The desired protein can be any protein including reporter molecules such as P-galactosidase, luciferase and GFP. The desired protein may also be a selectable marker such as an antibiotic.

Preferably, the desired protein is a food protein such as a storage protein, i. e. phaseolin or wheat gluten.

It is also preferably that the desired protein is a therapeutic protein. Therapeutically useful proteins include receptors, e. g. the cystic fibrosis receptor (CFTR), enzymes including prodrug activating enzymes, e. g. nitroreductases, ligands, regulatory factors, hormones, and structural proteins. Therapeutic proteins also include sequences encoding nuclear proteins, cytoplasmic proteins, mitochondrial proteins, secreted proteins, plasmalemma-associated proteins and serum proteins. The desired protein may also be selected from lipoproteins, glycoproteins and phosphoproteins, hormones, growth factors, enzymes, clotting factors such as Factor XIH and Factor IX etc., apolipoproteins, receptors, erythropoietin, drugs, oncogenes and tumor suppressors.

Specific examples of these compounds include proinsulin, growth hormone, androgen receptors, insulin-like growth factor I, insulin-like growth factor II, insulin-like growth factor binding proteins, epidermal growth factors, angiogenesis factors (acidic fibroblast growth factor, basic fibroblast growth factor, vascular endothelial growth factor and angiogenin), matrix proteins (Type IV collagen, Type VII collagen, laminin), phenylalanine hydroxylase, tyrosine hydroxylase, oncogenes (ras, fos, myc, erb, src, sis, jun), E6 or E7 transforming sequence, p53 protein, Rb gene product, cytokine receptor,

11-1, IL-6, IL-8, viral capsid protein, and other proteins of useful significance in the body. The desired protein, which can be fused, is only limited by the availability of the nucleic acid sequence encoding the protein or polypeptide to be incorporated. One skilled in the art will readily recognise that as more proteins and polypeptides become identified they can be integrated into the recombinant fusion protein of the present invention.

It is particularly preferred that the desired protein is an antigen. The term"antigen"as used herein refers to any protein that gives rise to an immune response in an animal such as a mammal, preferably a human. Preferably, the term"antigen"as used herein refers to a protein that stimulates a series of reaction in an animal that are mediated by white blood cells including lymphocytes, neutrophils and monocytes. Preferred antigens include viral antigens, bacterial antigens, protozoal antigens, parasitic antigens, tumor antigens, and proteins from viral, bacterial and parasitic organisms which can be used to induce an immune response. It is also preferred that the antigen is not a toxin.

A toxin is defined herein as a protein which has a direct toxic effect on a cell and causes cell death. It is most preferred that the antigen is human immunodeficiency virus gp 120, the non-toxic C-tetanus toxin fragment or bovine respiratory syncytial virus (BRSV) F-protein.

The desired protein may be linked to the constant region or part thereof directly or indirectly through an intermediate peptide linker which may include a peptide cleavage site. This will allow separation of the desired protein from the constant region of part thereof (if required) for purification purposes. Attachment of the desired protein to the constant region or part thereof may be made via the C-terminus of the heavy chain or part thereof. Alternatively, the antigen may be fused to the antibody at a site within the constant region or part thereof.

The method of the present invention is applicable to the production of the desired protein in any eukaryotic cell. Preferably the desired protein is produced in a eukaryotic cell which is capable of producing immunoglobulins. More preferably the desired protein is produced in a mammalian cell and most preferably in a plant cell.

The method of the present invention has the benefit of allowing the production of a desired protein in a cell at a higher level than when the desired protein is not produced as part of a fusion protein comprising a constant region of a heavy chain of an antibody.

Desired proteins are generally expressed at low levels in transgenic plants, whereas antibody molecules usually accumulate to much higher levels. By genetically fusing the desired protein with a component of the antibody molecule, the antibody component acts as a carrier and stabilising molecule and allows much higher levels of production and accumulation of the desired protein.

The present invention also provides a first recombinant fusion protein comprising an antigen fused to the constant region of the heavy chain of an antibody or to a part of said constant region. The recombinant fusion protein is as defined above in connection with the method of the present invention, except that the fusion protein comprises an antigen as defined above.

The present invention also provides a second recombinant fusion protein comprising a desired protein fused to the constant region of a heavy chain of an antibody or to part of said constant region, wherein the fusion protein does not comprise a functional antigen binding domain. The recombinant fusion protein is as defined above in connection with the method of the present invention, except that the fusion protein does not comprise a functional antigen binding domain.

A functional antigen binding domain is defined herein as the antibody domain which contacts and binds an antigen. The domain comprises the complementarity determining regions (CDRs) of the antibody, which are well known to those skilled in the art and can be easily identified. The domain is preferably the variable domain of the heavy and/or light chain of an antibody. The second recombinant fusion protein of the present invention can comprise an antibody which has had one or more of its variable domains removed. Alternatively, the second recombinant fusion protein of the present invention can comprise an antibody which has one or more non-functional variable domains. The variable domains can be made non-functional by deleting the CDRs. Other methods for

producing an antibody with a non-functional antigen binding domain are well known to those skilled in the art.

The present invention also provides a recombinant antibody molecule having an antigen fused to an antibody molecule, wherein the antibody molecule has affinity for the antigen. The recombinant antibody molecule of the present invention can form an immune complex by associating with other recombinant antibody molecules of the present invention through the antigen binding sites of the antibody. Such an immune complex resembles, but is not identical to, a classical immune complex. A significant advantage of the immune complex formed from the recombinant antibody molecules of the present invention is that it is easy to prepare in contrast to conventional preparation of immune complexes in vitro which require careful mixing of antigen with antibody at optimal concentrations. In our case, the expression ratio of antibody to antigen molecules is generally fixed at 1: 1, which optimises the potential for forming immune complexes.

The recombinant antibody molecule of the present invention is highly immunogenic, especially when it is forms an immune complex, and is suitable for both systemic and mucosal immunisation without the need for an adjuvant. Adjuvants are generally incorporated into vaccine compositions to improve the immune response. As the recombinant antibody molecules of the present invention are highly immunogenic, it is possible to obtain immunisation without the need of an adjuvant.

The immune complex formed using recombinant antibody molecules of the present invention will possess substantially the same enhanced immunogenic properties as an endogenously produced immune complex or a conventionally produced immune complex.

The recombinant antibody molecule of the present invention can comprise any antibody molecule provided it is has affinity for the antigen to which it is fused. The antibody molecule can be a complete monoclonal antibody or an antigen binding fragment

thereof, such as a Fv, Fab or F (ab') 2 fragment. Preferably the antibody molecule comprises a constant region of a heavy chain of an antibody.

To engineer genetic constructs encoding antigen-antibody fusion proteins, there is no restriction as to the species of either of these components. The invention is of especial interest in relation to human medicine but other primate and other animal antigen/antibody combinations may be used depending on the intended biological application of the product. For example the antigens and the antibody heavy and light chains may be of human, rodent, rabbit, bovine, ovine, caprine, fowl, canine, camel, feline or primate origin.

To illustrate the range of application of the invention three model antigens have been used, namely human immunodeficiency virus gpl20, the non-toxic C-tetanus toxin fragment and bovine respiratory syncytial virus (BRSV) F-protein, with their respective specific monoclonal antibodies. For each antigen, numerous constructs are possible, in which the antigen is fused to either the C-terminus of a monoclonal antibody (Mab) full length heavy chain, or to truncated heavy chain domains consisting of 1,2 or 3 constant region domains. Alternatively, the antigen is fused to the constant region at a point within the constant region.

The present invention also provides a transgenic plant expressing a recombinant fusion protein comprising a desired protein fused to a constant region of a heavy chain of an antibody.

The present invention also provides a transgenic plant expressing the first or second recombinant protein of the present invention, or the recombinant antibody molecule of the present invention.

Methods for generating transgenic plants are well known to those skilled in the art.

The transgenic plants of the present invention may express a number of molecular forms of fusion protein for each antigen, permitting the selection of the correctly

formed molecular form of fusion protein and its expression at the highest levels and to extract and purify the fusion protein. It is also possible to characterise the plant derived fusion proteins by analysis of assembled molecular forms, recognition by and affinity for a panel of antibodies.

The advances deriving from this work are the development of a new type of engineered molecule suitable for vaccination in a range of infectious diseases. By using three different antigens, we are able to determine the efficacy of this approach for a standard bacterial immunogen, the envelope protein of an important human virus and a transmembrane fusion protein of a cattle virus that causes severe mucosal infections.

The recombinant antibody molecule resulting from the fusion of an antigen with an antibody molecule has all the components required for immune complex mediated stimulation of an immune response and our approach offers a convenient method for ensuring antigen/antibody complexing. The recombinant antibody molecules of the present invention can also form larger immune complexes in planta (see Figure 1) which can be used for immunisation. Not all parts of the antibody molecule are essential, such as the sites for FcR binding and complement activation that reside in the constant domains of IgG antibodies. Thus we also include the use of truncated antibody molecules which lack either the Cy3 domain (that is not involved in antigen recognition), or the Cyl domain.

It is also possible to use the first recombinant fusion protein of the present invention or the second recombinant fusion protein of the present invention (wherein the desired protein is an antigen) for immunisation. In particular, a fusion protein comprising an antibody molecule which does not comprise a functional antigen binding domain and an antigen can be used to deliver the antigen to phagocytes and also to initiate complement activation. By excluding the requirement for antigen recognition, it is possible to help prevent the phenomenon of modulation of antigen processing that results from antibody masking of T cell epitopes, as the antigen will be a fusion protein with the constant region of an antibody.

The use of plants is ideally suited for the production of the recombinant fusion proteins of the present invention not only because of the potential requirement for large quantities for vaccination purposes, but also because those complexes that involve full length antibody, or assembly with light chain are not readily produced in most other expression systems.

To engineer genetic constructs encoding antigen-antibody fusion proteins DNA encoding an antigen or antigenic fragment may be amplified by PCR. The genes encoding the light and heavy chains of murine, human or other mammalian monoclonal antibodies specific for these antigens are also cloned. Using these DNA sequences the gene constructs can be made encoding the desired recombinant fusion protein. For example, the following gene constructs for each of the antigens can be prepared for plant transformation : Genetic constructs, showing Ig domains that will be included in each immune complex construct.

A-Heavy chain I) Construct #1 Variable---Cyl---Cy2---Cy3---Linker peptide---antigen II) Construct #2 Variable---Cyl---Cy2---Linker peptide---antigen Construct #3 Cy2---Cy3---Linker peptide---antigen IV) Construct #4 Cy2---Linker peptide---antigen B-Light chain Variable---CL These constructs are shown diagrammatically in Figure 1, which also shows the final assembly of products in plants.

The linker peptide as shown in Figure 2 may conveniently be (GGGGS) 3. A conventional 6xHistidine tag and a protease cleavage site (e. g. enterokinase, Factor Xa

or thrombin) may also be introduced at the C-terminus of the antigen for purification purposes. In addition a further different protease cleavage site (e. g. enterokinase, Factor Xa or thrombin) may also be introduced between the linker peptide and the antigen.

Additional or alternative peptide tags and protease cleavage sites may be incorporated.

The gene constructs are sequenced for confirmatory purposes and inserted into plant expression vectors for plant transformation.

To generate transgenic plants expressing the 12 (four for each antigen) constructs, Nicotiana tabacum is a convenient plant host using Agrobacterium mediated transformation. For constructs I and 11, in order to generate the recombinant fusion proteins, the immunoglobulin light chain genes and the heavy chain constructs may be introduced into separate plant lines as shown in Figure 1. Following regeneration of these first generation transformants, the light chain and heavy chain transformed plants are cross-fertilised and the second generation plants screened for production of the final product. For constructs in and IV, there is no requirement for the light chain and the final product can be produced in the first generation transformed plants.

To generate stable homozygous transgenic plant lines, transgenic plants expressing correctly assembled products are used to generate homozygous plant stocks. The plants are grown to maturity, self fertilised, and the resulting seeds screened by back crossing with non-transformed plants to determine those that are homozygous. Further stocks can be generated by self-fertilisation and stored as seeds.

To characterise each molecular form of plant product and to extract and purify material for further study, the primary plant transformants are screened to determine which types of products are expressed and assembled optimally. This investigation can be performed by Western blot analysis and ELISA of crude plant extracts, using a range of antisera and monoclonal antibodies that are commercially available. Extraction and purification of the selected recombinant products may be achieved by ammonium sulphate precipitation, followed by filtration and affinity chromatography using either the immunoglobulin regions or specific peptides as affinity tags. Further

characterisation is performed to determine molecular structure, recognition of the antigen moiety by a panel of antibodies and binding affinity.

The Figures show: Figure 1 shows schematically immune complexes in plants and potential assembly arrangements.

Figure 2 shows schematically the basic molecular design.

Figure 3 shows schematically the pEXG13 principle cloning sites.

Figure 4 shows RT-PCR analysis of transgenic plants.

Figure 5 shows results of a capture ELISA for CH2-CH3-gpl20.

Figure 6 shows results of a capture ELISA for recombinant gpl20 expressed in plants Figure 7 shows a 10% SDS gel under reducing and non-reducing conditions.

EXAMPLE Production and use of recombinant fusion molecules comprising gpl20 and constant domains of a heavy chain of an antibody.

Cloning and genetic engineering of HIV gpl20 constructs: The source of DNA for the HIV gpl20 antigen was an infectious cloned isolate of HIV IIIB. Such DNA can be obtained from the Medical Research Council AIDS Reagents Project (NIBSC, Blanche Lane, South Mimms, Potters Bar, Herts EN6 3QG, UK).

The genes encoding the heavy and light chains of an HIV gpl20 specific monoclonal antibody (e. g. Gorny et al., 1991, Proceedings of the National Academy of Sciences,

USA 88: 3238-3242) were cloned as described (Ma, J. K-C., T. Lehner, P. Stabila, C. I.

Fux and A. Hiatt. 1994 Assembly of monoclonal antibodies with IgG1 and IgA heavy chain domains in transgenic tobacco plants. Eur. J. Immunol., 24: 131-138). RNA was extracted from relevant hybridoma cells using a standard extraction kit (e. g. Promega SV Total RNA isolation system), cDNA was prepared by using the reverse transcriptase reaction. DNA encoding the antigen, or the heavy and light chains of a monoclonal antibodies specific to the antigen, or fragments of these antibody sequences, was amplified by the polymerase chain reaction using synthetic oligonucleotides shown in Figure 2. The oligonucleotide primers 1 and 2 corresponded to the 5'and 3'ends of the antibody heavy chain and the oligonucleotide primers 3 and 4 corresponded to the 5' and 3'ends of the antigen. The 5'primer for the antigen also includes sequences encoding a linker peptide and a protease cleavage site. The primers included appropriate restriction enzyme sites (such as 5'BamHI and 3'XmaI for gpl20 ; 5'XhoI and 3'BamHI for Ig heavy chain) and extra sequences encoding linker peptides (GGGSGGGSGGGS) or protease cleavage sites (e. g. Factor Xa IEGR). If a light chain is required, the gene can be cloned according to the methodology described by Drake et al., Antibody production in plants. P. Shepherd and Dean (eds). Monoclonal Antibodies -A practical approach. Oxford University Press. 2000) Engineering of DNA constructs was carried out in standard cloning vectors- pBluescript (Stratagene) and pET32 (New England Biolabs) by standard techniques as described in Maniatis, Fritsch and Sambrook (1982)-Molecular Cloning, A Laboratory Manual. Essentially, the DNA encoding the antigen was first cloned into the vector using the multiple cloning site, followed by the DNA encoding the heavy and optionally the light chain, of an antibody or antibody fragment. In some cases, the order was reversed, with antibody genes being cloned first. Positive transformants were identified by the release of a DNA fragment of the correct size following digestion with the appropriate restriction enzymes, or PCR using the appropriate 5'and 3'oligonucleotide primers. Four chimeric immunoglobulin heavy chains constructs were engineered in which the antigen was expressed in fusion with varying portions of the immunoglobulin heavy chain as shown in Figure 1.

The specific example of cloning the Ig heavy chain CH2-CH3 domain fusion with an HIV gpl20 fragment (construct Ell in Figure 1) is shown in Figure 2. Two PCR products were amplified using oligonucleotide pairs 1-2 and 3-4. The HIV gpl20 DNA sequence encodes a truncated peptide starting at the N-terminus, up to and including the sequence KEYAL (aal47) with a stop codon immediately after. These were cloned into a vector based on pET32 which had previously been engineered with the key features of the cloning site within this vector shown in Figure 3. The gpl20 gene fragment was ligated into the BamHI-XmaI site of the vector, then the Ig heavy chain gene was cloned upstream in the NcoI-BamHI site. The entire construct was then re-amplified by PCR to include the downstream enterokinase cleavage site and His tag, and to include 5'XhoI and 3'EcoRI terminal restriction sites. This fragment was cloned into pBluescript for confirmatory sequencing.

The completed genetic constructs were then excised and cloned separately into a plant expression cassette such as that described above (Drake et al., 2000), within a plasmid (pMON530) suitable for initial screening in E. coli. Plasmids of this type are widely available, such as the pGreen system (www. green. ac. uk). The plant expression cassette in this case contains the 5'untranslated region from tobacco etch virus which stabilises mRNA in plant cells, as well as an upstream murine IgG leader sequence (Figure 3), which directs secretion of the recombinant protein out of the plant cells.

This recombinant vector was used to transform E. coli (DH5-a). Screening of transformed clones was by Southern blotting, using radiolabelled DNA probes derived from the original PCR products. Plasmid DNA was purified from positive transformants (PromegaQiaprep kit) and used for transformation of Agrobacterium tumefaciens (strain LBA4404-Gibco/BRL, UK) using a freeze/thaw procedure as follows: 1. Inoculate 50 ml Luria broth (LB) containing spectinomycin (50tg/ml) (Sigma) with Agrobacterium.

2. Shake at 28°C in the dark to an OD600 of 1.0.

3. Centrifuge the agrobacteria in sterile tubes at 3000g for 15 min.

4. Remove supernatant and, while keeping the tubes on ice, resuspend the bacterial pellet in a total of 1 ml of ice-cold l OmM CaClz.

5. Transfer 100 ml aliquots of the bacterial resuspension into sterile microcentrifuge tubes and then place into liquid nitrogen.

6. Use the frozen agrobacteria directly for DNA transformation or store at-80°C.

7. Pipette 10 ml of binary plasmid from a DNA miniprep onto the surface of 100 ml competent frozen agrobacteria.

8. Incubate the DNA-bacteria mixture for 5 min at 37 ° C in a water bath.

9. Add lml of LB to the bacteria and shake at 28°C for 4 h.

10. Centrifuge the bacteria for 2 min at 12,000 x g and resuspend the pellet in 100 ml of LB.

11. Spread 50 ml of the bacterial suspension onto LB medium (made semi-solid with 15 gl-'agar) containing spectinomycin required for selection of agrobacteria carrying binary and disarmed Ti helper plasmid. Seal the plates and incubate for 2-3 days at 28°C.

12. Re-streak resulting colonies on selective LB medium in separate Petri dishes and incubate at 28°C for 2 days.

Plant transformation and regeneration: All gene constructs were introduced into Nicotiana tabacum, var. xanthii. Tobacco transformation with A. tumefaciens was by standard procedures. Leaf discs are cut from surface sterilised tobacco leaves and incubated with a culture of the recombinant A. tumefaciens, containing cDNA inserts.

The infected discs are transferred to culture plates containing a medium that induces regeneration of shoots, supplemented with kanamycin and carbenicillin (Sigma, UK).

Shoots developing after this stage are excised and transplanted onto a root inducing medium, supplemented with kanamycin. Rooted plantlets are transplanted into soil after the appearance of roots. The detailed methodology was as follows: 1. Remove Agrobacterium (containing binary vector) from-80°C and streak onto semi-solid LB medium (with appropriate antibiotics) in a 9 cm Petri dish. Incubate at 28°C for 2 d.

2. Inoculate Agrobacterium from Petri dish into 10 ml LB.

3. Shake at 28°C to an OD600 of 1.0.

4. Remove 4 ml of Agrobacterium suspension and add to 16 ml of sterile distilled water in a sterile 9 cm Petri dish.

5. Cut 0.5-1.0 cm leaf discs from surface sterilised leaves and immerse for 5 min in the diluted Agrobacterium suspension.

6. Briefly dry leaf discs on sterile filter paper.

7. Place leaf discs on shoot regeneration medium, 10-15 discs per 20 ml of medium in each 9 cm Petri dish. Incubate for 2 d at 25°C with a 16 h photoperiod.

8. Transfer leaf discs to shoot regeneration medium containing 500 mg/1 carbenicillin and appropriate concentration of selective agent for transformed plant cells. Incubate for 21 d at 25°C with a 16 h photoperiod.

9. Transfer leaf discs to shoot regeneration medium in sterile 175 ml glass jars containing 500 mg/1 cefotaxime and 200 mg/1 kanamycin. Incubate at 25°C with a 16 h photoperiod.

10. Developing shoots are removed when they reach a convenient size (approx. 0.5 cm in length) and transferred to rooting medium (3-4 shoots/40 ml medium/175 ml glass jar). Incubate for 14 d at 25°C with a 16 h photoperiod.

11. Shoots lacking roots are trimmed at the base and replaced in fresh rooting medium for an additional 21 d or until roots appear.

12. Rooted shoots are transferred to compost in plant pots, watered and supplied with nutrients. Plants are kept in seed trays and covered with a lid for 24 h after transfer to compost, to minimise initial water loss.

Regenerated plants were screened for expression of immunoglobulin chains and each of the antigens by Western blot and ELISA of crude leaf extracts using available antisera and monoclonal antibodies (e. g. from the MRC AIDS Directed reagents programme, address given above). Transgenic plants were self-fertilised to establish homozygous plant lines and cross fertilised to generate antibody producing plants.

RNA extraction: Total RNA was extracted from plants less than a month old, using the Promega plant RNA extraction kit (Rneasy Plant mini kit). Alternatively, leaf tissue was frozen in

liquid nitrogen and rapidly ground to a fine powder. This was mixed with 2 volumes of ice cold guanidine hydrochloride buffer (8M guanidine hydrochloride, 20mM MES, 20mM EDTA, 50mM p-mercaptoethanol, pH7). After agitation it was added to 1 volume of phenol: chloroform: isoamyl alcohol (25: 24: 1), mixed thoroughly, and centrifuged at 10,000 rpm for 45 minutes. The upper (aqueous) phase was collected, mixed with pre-cooled ethanol (0.7 volumes) and 1M acetic acid (0.2 volumes), and incubated at-200°C for 16 hours. After centrifugation, the precipitate was washed three times with 3M sodium acetate, pH5.2, and once with 70% ethanol. The pellet was dissolved in sterile RNAse free water (Sigma, UK) containing RNAse free DNAse (Promega, UK), and incubated at 37°C for 1 hour, then at 70°C for 5 minutes.

RNA concentration and purity were assessed using the GeneQuant II RNA/DNA Calculator (Pharmacia BioTech, UK). RNA was stored in sterile RNAse free water at -20°C.

RT-PCR was performed using appropriate oligonucleotide primers to determine that the correct RNA transcript was being made by putative transformed plants.

Characterisation and selection of recombinant immune complexes from plants: To confirm the expression of each engineered constructs in plants, leaf extracts were examined by ELISA and Western blot analysis. Samples were extracted in 150mM NaCl and 20mM tris, pH8 (TBS) with leupeptin (lOmg/ml) (Calbiochem). Capture ELISA analyses were performed by incubating the plant extracts on microtiter plates that had been pre-coated with a monoclonal antibody specific for HIV gpl20 (ADP401 from the MRC Aids Directed Program) and blocked with 5% non-fat dry milk in TBS.

After overnight incubation at 4°C, the plates were washed in TBS with 0.05% Tween 20, then incubated with either a horseradish peroxidase labelled Sheep anti-Mouse IgG antiserum (The Binding Site, UK), or one of a panel of gpl20 specific monoclonal antibodies (MRC Aids Directed Program), followed by the relevant horseradish peroxidase antiserum (The Binding Site, UK). The assay was developed using the TMB substrate (Sigma, UK) and absorbance was read at 450nm.

For Western blot analyses, the plant extracts were boiled in 75mM tris-HCl (pH6. 8) and 2% SDS under either non-reducing or reducing conditions (by addition of 0.1% P mercaptoethanol). SDS-polyacrylamide gel electrophoresis (PAGE) in 10% acrylamide or 4-20% gradient gels was performed. The gels were blotted onto nitro-cellulose paper and blocked in TBS containing 0.05% Tween 20, and 1% non-fat dry milk. The blots were then incubated in appropriate antiserum for 2 hours at 37°C. After washing, an appropriate second-layer alkaline phosphatase-conjugated antiserum was applied for 2 hours at 37°C. Antibody binding was detected by incubation with nitroblue tetrazolium (300mg/ml) and 5-bromo-4-chloro-3-indolyl phosphate (150mg/ml) (Promega, UK).

Glycosylation is determined by Western blot, examining binding to lectins such as concanavalin A, or by using glycans specific antisera (kindly provided by Dr. Loic Faye, University of Rouen). Functional studies of antigen binding affinity to available monoclonal antibodies may be performed using surface plasmon resonance techniques.

Recombinant protein purification: Purification was performed using a procedure which we have previously determined for IgG extraction from transgenic plants. Following initial precipitation from crude plant extract with ammonium sulphate, the recombinant antibody was concentrated by stirred cell filtration using a YM30 molecular weight cut-off filter (Amicon, UK). Purification was by affinity chromatography using agarose coupled to anti-mouse or human IgG antibodies (Sigma, UK) as the ligand. Elution was in 0. 1M glycine-HCl pH 2.5.

The recombinant heavy chains may also be designed to allow a further affinity purification step using the 6xHis fusion peptide tag with immobilised metal chelate affinity chromatography, which can subsequently be cleaved by thrombin.

The expression of these fusion proteins is not limited to plants, and could be carried out in any other eukaryotic cells, especially cells that can produce immunoglobulins such as mammalian cells.

The immunoglobulin chain sequences can be exchanged for sequences from other classes of immunoglobulins.

Other sequences can be incorporated into the construct in addition to the epitope tags and protease cleavage sites. For example, these could code for other functional regions, such as markers, enzyme activity, or sites for chemical coupling to other active molecules.

In addition to HIV gpl20, two further examples have been developed. Similar constructs have been prepared using tetanus toxin non-toxic C-fragment (kindly donated by Dr. Neil Fairweather at Imperial College, London) and a tetanus toxin specific murine monoclonal antibody (kindly donated by Dr. Claus Koch at the State Serum Institute, Copenhagen). In addition, we have prepared corresponding constructs for the F-protein of bovine respiratory syncytial virus and a specific murine monoclonal antibody the genes for which were kindly provided by Dr. K. G. Madsen at the Danish Veterinary Institute for Virus Research.

Among the benefits of the invention, the following are noteworthy:- (i) Co-expression of antibody with antigen increases the levels of expression of antigen.

(ii) Antigen and whole antibody are genetically linked to form a type of immune complex. In other constructs, antigen and antibody fragments (that may or may not include the antigen binding site) are linked to form fusion proteins.

(iii) The recombinant antibody molecule and the antigen are automatically expressed at 1: 1 ratio. In the case of constructs involving whole IgG, larger complexes may form spontaneously.

(iv) Antigen-antibody fusion molecules are easy to produce.

(v) The use of plants to make these molecules significantly reduces the cost, particularly if production is increased to agricultural scale.

(vi) For oral vaccines, consumption of edible transgenic plants offers an alternative, safe mode of delivery for immune complex vaccines.

(vii) There is potentially no need for addition of an adjuvant as the engineered molecule is already highly immunogenic. The CH2 and CH3 immunoglobulin domains include the site for complement activation and opsonisation of phagocytes Results: Results have been obtained for two of the described constructs-CH2-gpl20 and CH2-CH3-gpl20. RT-PCR confirme that specific mRNA of the desired size was being made by transgenic plants (Figure 4).

Analysis by capture ELISA demonstrated that both constructs are being expressed by plants. We have gone on to characterise CH2-CH3-gpl20 more fully and demonstrated that by ELISA that it contains IgG2a epitopes as well as gpl20 epitopes as shown in Figures 5 and 6. Moreover, expression of the recombinant gpl20 is greater in plants expressing the CH2-CH3-gpl20 construct, than those expressing either gpl20 alone, gpl20 with a C-terminal HDEL tag, or CH2-gpl20. Preliminary results indicate an expression level of gpl20 of approximately 0.8%.

Purification of plant derived material has been carried out and Western blot analysis demonstrates the presence of protein bands that are recognised by specific antibodies to gpl20 and murine IgG2a (Figure 7). Under reducing conditions, the most prominent band has a molecular size consistent with that expected of approximately 50Kd. Under non-reducing conditions, higher molecular weight forms approaching 100Kd are detected. This is consistent with the formation of homodimers, through association of the CH2 and CH3 domains of the heavy chain.

A preliminary investigation of immunogenicity has been carried out in a single rhesus monkey. Following 2 administrations of purified plant derived CH2-CH3-gpl20 (150g per dose) covalently linked with the adjuvant mycobacterial heat shock protein 70 using the SPDP bifunctional reagent, specific antibody and cellular immune responses were detected. After 3 immunisations, very significant T-cell proliferative responses were detected, not only to the plant CH2-CH3-gpl20 construct, but also to

purified recombinant gpl20 expressed in Chinese hamster ovary cells. This indicates that the plant recombinant gpl20 is correctly expressed and folded in an immunogenic conformation. Furthermore, four macaques have been immunised by the mucosal intra-vaginal route, and four by the targeted iliac lymph node route (Lehner et al., Nature Medicine, 2: 767-775,1996). After two immunisations, induction of interferon-y and interleukin-4 has been detected, indicating that both TH1 and TH2 responses are elicited. Further immunisation studies are being carried out in rhesus monkeys as well as mice.