BACKGROUND OF THE INVENTION The oral administration of a protein antigen often leads to a marked suppression of systemic humoral and cell-mediated responses to subsequent immunisation with the same antigen. This response has long been recognised as a means for inducing peripheral immunological tolerance and hence suppressing autoimmunity (for review see Weiner, H. L.

(1997) Annu. Rev. Med. 48: 341-51).

A number of studies have shown suppression of autoimmunity in animal models including experimental autoimmune encephalitis (Bitar, D. M. and Whitacre C. C. (1988) Cell.

Immunol. 112: 364-70, and Higgins, P. J. and Weiner, H. L. (1988) J. Immunol. 140: 440-5), myasthenia gravis (Wang, Z., Qiao, J. and Link, H. (1993) J. Neuroimmunol. 44: 209-14), collagen-induced arthritis (Staines, N. A., Harper, N., Ward, F. J., Thompson, H. S. and Bansal, S. (1996) Ann. NY Acad. Sci. 778: 297-305, and Nagler-Anderson, C., Bober, L. A., Robinson, M. E., Siskind, G. W. and Thorbecke, G. J. (1986) Proc. Natl. Acad. Sci. 83: 7443- 6), uveitis (Nussenblatt, R. B., Caspi, R. R., Mahdi, R., Chan, C. C., Roberge, F., Lider, O. and Weiner, H. L. (1990) J. Immunol. 144: 1689-95) and diabetes in the non-obese diabetic (NOD) mouse (Zhang, Z. J., Davidson, L., Eisenbarth, G. and Weiner, H. L. (1991) Proc. Natl.

Acad. Sci. 88: 10252-6).

Efforts to develop oral vaccines for induction of oral tolerance in humans have been stimulated by several studies showing beneficial effects of oral antigen administration to patients with autoimmune diseases including multiple sclerosis, rheumatoid arthritis, and insulin-dependent diabetes (Kagnoff, M. F. (1996) Trends Immunol. Today 17: 57-9).

However, these studies indicated the need for repeated administration of antigen in large doses, potentially limiting the therapeutic potential of systemic tolerance in autoimmunity.

Furthermore, it has been reported that conjugating antigens to the B subunit of cholera toxin allowed a significant reduction in both the antigen dose and the dosing schedule required for induction of peripheral immunological tolerance (Sun, J. B., Holmgren, J. and

Czerkinsky, C. (1994) Proc. Natl. Acad. Sci. 91: 10795-9). The B subunit, which is the non- toxic portion of cholera toxin, can potentiate the tolerogenic effect of orally administered antigens several hundred-fold. The cholera toxin B subunit consists of five identical proteins, each of about 11.6 kDa, that form a doughnut-shaped pentamer into which the toxic A subunit, which is the active agent in inducing diarrhoeal disease, is inserted (Sixma, T. K., Pronk, S. E., Kalk, K. H., Wartna, E. S., van Zanten, B. A. M., Witholt, B. and Hol, W. G. J. (1991) Nature 351: 371-7). The B subunit pentamer allows the toxin to bind with high affinity to ganglioside Gui in the gut, and the A subunit is then translocated into the cell where ADP- ribosylation of the GS subunit of adenylate cyclase induces increased cyclic AMP formation and secretion of electrolytes that leads to the life-threatening diarrhoea of cholera. The mechanism by which the cholera toxin B subunit (CTB) potentiates mucosal induction of peripheral tolerance is thought to involve a complex series of events that include enhanced presentation of the oral antigen to the mucosal surface by virtue of its binding to ganglioside GUI and hence enhanced presentation of the antigen to regulatory T cells (Holmgren, J., Lycke, N. and Czerkinsky, C. (1993) Vaccine 11: 1179-84).

One of the animal models in which potentiation of oral tolerance by conjugation to CTB has been demonstrated is the NOD mouse model of diabetes. Oral administration of insulin has been shown to delay onset of diabetes in these animals when administered at mg levels (Zhang, Z. J., Davidson, L., Eisenbarth, G. and Weiner, H. L. (1991) Proc. Natl. Acad. Sci. 88: 10252-6).

However, relative lower amounts of human insulin (microgram levels) were sufficient to effectively suppress beta cell destruction and clinical diabetes in these animals following conjugation to CTB (Bergerot et al, (1997) Proc. Natl. Acad. Sci. 94: 4610-14). In addition, the protective effect could be transferred by T cells from treated animals, and was associated with reduced lesions of insulitis, indicating protection against autoimmune diabetes could be achieved by feeding relatively small amounts of insulin conjugated to CTB. The procedure for conjugation of insulin to CTB in this study employed the heterobifunctional reagent N- succinimidyl 3- (2-pyridylthio) propionate (SPDP), and the administered conjugate was a crude, unfractionated preparation.

It is an object of the present invention to provide a novel method for preparing a product of conjugates between an antigen and a mucosal binding component which method results in a product that is more effective in inducing immunological tolerance at lower doses compared to known products.

Moreover, it is an object of the present invention to provide a product of conjugates by a method, which is more reproducible, gives higher yields, and is more economic than known methods.

SUMMARY OF THE INVENTION In accordance with the present invention there is provided a method for preparing a product of conjugates between an antigen and a mucosal binding component.

Moreover there is provided a product of conjugates between an antigen and a mucosal binding component.

The product can be administered in vivo for inducing specific immunological tolerance, inter alia peripheral immunological tolerance.

Moreover, the product can be utilise in vitro as unique research tools, inter alia, for increasing the understanding of specific immunological tolerance.

DEFINITIONS A crosslinker is a compound, which is capable of covalently binding two molecules together. After the reaction, the crosslinker, or a part of the crosslinker, form a part of the linkage between the conjugated molecules.

Molecules are referred to as being conjugated if they are covalently bond to each other through one or more crosslinker molecules.

An antigen is herein defined as a component capable of invoking immunological tolerance. The antigen may be the target antigen or a fragment or derivative thereof, or a bystander antigen.

A target antigen is an antigen for which immunological tolerance is desired. Typically, the target antigen will also be the target of an unwanted immunological response (already underway or for which the subject is at risk), and an object will be to decrease, delay, or reduce the risk of the response.

A bystander antigen is an antigen, which is antigenically distinct from the target antigen but can substitute for the target antigen in invoking specific immunological tolerance. Usually, a bystander antigen is expressed in the same tissue in the vicinity of the target antigen.

Insulin peptide as used herein refer to insulin, as well as allelic and synthetic variants, fragments, fusion peptides, conjugates and other derivatives, that contain a region that is homologous (preferably 70% identical, more preferably 80% identical and even more preferably 90% identical at the amino acid level) to at least 10 (preferably 20) consecutive amino acids of insulin, wherein the homologous region of the derivative shares with the insulin an ability to induce tolerance to the target antigen.

Insulin as used herein is bovine insulin, porcine insulin, human insulin, including any analogue thereof wherein one or more amino acids have been deleted and/or replaced by other amino acids, including non-codeable amino acids, or human insulin comprising

additional amino acids, i. e. more than 51 amino acids; and including any derivative thereof wherein at least one organic substituent is bound to one or more of the amino acids.

A mucosal binding component refers to one or more molecules, which either by themselves or in connection with each other are capable of specifically binding to the mucosal cells of the subject treated. In this specification and the appended claims the term mucosal binding component shall thus comprise both a complex of molecules capable of specifically binding to the mucosal cells, for example a multimer of identical subunits or a multimer comprising different subunits, and the individual subunits thereof even though the latter might not by themselves be able to specifically bind to the mucosal cells. If the mucosal binding component is able to bind to the mucosal cells it will also be referred to as a biologically active mucosal binding component. In some instances, the mucosal binding component has the additional characteristic of penetrating or translocating across the mucosal surface.

The cholera toxin B subunit (CTB) as used herein refer not only to the intact pentamer subunit, but also to allelic and synthetic variants, fragments, fusion peptides, conjugates and other derivatives, that contain a region that is homologous (preferably 70% identical, more preferably 80% identical and even more preferably 90% identical at the amino acid level) to at least 10 (preferably 30) consecutive amino acids of the CTB subunit, wherein the homologous region of the derivative has mucosal binding activity as well as the monomeric subunits thereof.

Crosslinker derivatives of antigen shall refer to antigen derivatised with one or more crosslinker residues. Crosslinker derivatives of antigen and crosslinker derivatives of insulin peptide may for example also be referred to using terms such as derivatised antigen or derivatised insulin peptide, respectively.

Crosslinker derivatives of mucosal binding component shall refer to mucosal binding component derivatised with one or more crosslinker residues. Crosslinker derivatives of mucosal binding component and crosslinker derivatives of cholera toxin B subunit may for example also be referred to using terms such as derivatised mucosal binding component or derivatised cholera toxin B subunit, respectively.

Activating a crosslinker derivative shall be broadly understood as any treatment of the crosslinker derivative resulting in the crosslinker part of the derivative being converted to a form which can interact with another crosslinker, for example as part of a crosslinker derivative, not treated this way. A crosslinker derivative treated in such a manner may also be referred to as an activated crosslinker derivative of for instance an antigen or an insulin peptide or simply activated antigen or activated insulin peptide.

Active immunological tolerance refers to a state in which the tolerance effect (s) are the result of an ongoing biological process; for example, down-regulation of specific effector cells by suppressor cells. Active tolerance can be demonstrated by cell mixing or cell transfer

experiments. The following two examples of experimental results (conducted with appropriate controls) are evidence of active immunological tolerance : a) when leucocytes from a tolerized animal are mixed with specific effector cells from a second animal and the activity of the effector cells is diminished; b) when leucocytes from a tolerized animal are transferred to a second animal having a autoimmune disease, and features of the disease are reduced.

Invoking immunological tolerance refers to at least one of the following effects: a) a decreased level of a specific immunological response (thought to be mediated at least in part by antigen-receptor effector T lymphocytes, B lymphocytes, antibody, or their equivalents); b) a delay in the onset or progression of a specific immunological response; or c) a reduced risk of the onset or progression of a specific immunological response.

Specific immunological tolerance occurs when immunological tolerance is preferably invoked against certain antigens in comparison with others.

Sustained immunological tolerance is immunological tolerance that measurably persists for at least 3 weeks.

DESCRIPTION OF THE INVENTION Accordingly, the present invention relates to a method for preparing a product of conjugates between an antigen and a mucosal binding component which method comprises a) reacting the antigen with a first crosslinker thereby producing a mixture of crosslinker derivatives of the antigen, b) isolating the antigen derivatised with a single crosslinker residue, c) activating the isolated crosslinker derivative of the antigen, d) reacting the mucosal binding component with a second crosslinker thereby producing a mixture of crosslinker derivatives of the mucosal binding component, e) reacting the activated crosslinker derivative of the antigen with the mixture of crosslinker derivatives of the mucosal binding component, thereby producing the conjugates between the antigen and the mucosal binding component.

In one embodiment of the present invention the mucosal binding component has GM1 binding activity when present in the product of conjugates between an antigen and said mucosal binding component. In a special embodiment the mucosal binding component is the cholera toxin B subunit (CTB). In another special embodiment the mucosal binding component is the E. coli heat-labile enterotoxin B subunit (LTB). In a further embodiment the mucosal binding component is selected among such as those disclosed in WO 98/47529,

which is incorporated herein by reference. An assay for measuring the GM1 binding activity is also disclosed in WO 98/47529.

In one embodiment of the present invention the antigen is selected from the group of autoantigens (including, but not limited to, insulin peptides, collagen peptides, PLP and myelin basic protein (MBP)), alloantigens (including, but not limited to, HLA class I and 11), xenoantigens, or allergens. The antigen may be a polypeptide, polynucleotide, carbohydrate, glycolipid, or other molecule isolated from a biological source, or it may be a chemically synthesized small molecule, polymer, or derivative of a biological material, providing it has the ability of inducing tolerance when conjugated to the mucosal binding component. In a further embodiment the antigen is selected among such as those disclosed in WO 98/47529, which is incorporated herein by reference.

In a special embodiment of the present invention the antigen is an insulin peptide. In one embodiment, the insulin peptide is insulin. In another embodiment the insulin peptide is selected among the precursor form of insulin (comprising an AAK amino acid sequence linking the B chain to the A chain), a single-chain form and a form containing a signal peptide for secretion. In a further embodiment the insulin peptide is a recombinant human insulin. In a still further embodiment the insulin peptide is the A chain of insulin alone, or the B chain of insulin alone. In a further embodiment the insulin peptide is selected among the mature B chain residues 1-12,10-22 or 11-30. In a still further embodiment the insulin peptide is selected among metabolically inactive forms of insulin, metabolically inactive insulin fragments, and metabolically inactive insulin analogues. Examples of metabolically inactive insulin analogues are such as those disclosed in WO 98/47529, which is incorporated herein by reference.

In one embodiment of the present invention the first crosslinker is a bifunctional crosslinker (i. e. with two functional groups), preferably a heterobifunctional crosslinker (i. e. with two different functional groups). In another embodiment of the present invention the second crosslinker is a bifunctional crosslinker, preferably a heterobifunctional crosslinker. In a further embodiment, the first and/or the second crosslinker is selected from the non-limiting group of N-succinimidyl (4-iodoacetyl) aminobenzoate (SIAB), N-succinimidyl-3- (2- pyridylthio) propionate (SPDP), N-succinimidyl S-acetylthioacetate (SATA), m- maleimidobenzoyl-N-hydroxysuccinimide ester (MBS), N-g-maleimidobutyryloxy-succinimide ester (GMBS). In a further embodiment the first and/or the second crosslinker is Traut's Reagent 2-iminothiolane in combination with SPDP. In a still further embodiment the first and/or the second crosslinker is succinimidyl dicarbonyl pentane or disuccinimidyl suberate.

In a further embodiment the first and/or the second crosslinker is selected among such as those disclosed in The 1999/2000 Pierce Products Catalogue (Pierce Chemical Company,

1999, USA) and the Double-Agents Cross-Linking Reagents Selection Guide (Pierce Chemical Company, 1998, USA), which are incorporated herein by reference.

In one embodiment of the present invention the isolated crosslinker derivative of the antigen is substantially free of antigen derivatised with more than one crosslinker. In a further embodiment less than 10 percent (preferably 5 percent) of the isolated crosslinker derivative of the antigen is antigen derivatised with more than one crosslinker.

In one embodiment of the present method the first crosslinker is a bifunctional crosslinker. In a special embodiment of the present method the bifunctional crosslinker is a heterobifunctional crosslinker.

In another embodiment of the present method the second crosslinker is a bifunctional crosslinker. In a special embodiment of the present method the bifunctional crosslinker is a heterobifunctional crosslinker.

In another embodiment of the present method the antigen is an insulin peptide.

In a further embodiment of the present method the mucosal binding component as present in the product of conjugates between an antigen and said mucosal binding component has GM1 binding activity.

In a special embodiment the present invention relates to a method for preparing a product of conjugates between an insulin peptide and a mucosal binding component which method comprises a) reacting the insulin peptide with a first crosslinker thereby producing a mixture of crosslinker derivatives of the insulin peptide, b) isolating the insulin peptide derivatised with a single crosslinker residue, c) activating the isolated crosslinker derivative of the insulin peptide, d) reacting the mucosal binding component with a second crosslinker thereby producing a mixture of crosslinker derivatives of the mucosal binding component, e) reacting the activated crosslinker derivative of the insulin peptide with the mixture of crosslinker derivatives of the mucosal binding component, thereby producing the conjugates between the insulin peptide and the mucosal binding component.

In one embodiment of the above method the mucosal binding component as present in the product of conjugates between an antigen and said mucosal binding component has GM1 binding activity. In a special embodiment mucosal binding component is the cholera toxin B subunit (CTB).

In one embodiment of the above method the first crosslinker is a bifunctional crosslinker, preferably a heterobifunctional crosslinker. In another embodiment of the above method the second crosslinker is a bifunctional crosslinker, preferably a heterobifunctional crosslinker. In a special embodiment, the first and the second crosslinker is a pyridyl

disulphide-containing crosslinker. In a more special embodiment the first and the second pyridyl disulphide-containing crosslinker is N-succinimidyl 3- (2-pyridyldithio) propionate (SPDP).

In a further special embodiment of the above method an insulin peptide derivatised with a pyridyl disulphide-containing crosslinker is activated by converting the pyridyl disulphide group on the derivatised insulin peptide to a thiol group using a reducing agent. In a further embodiment the reducing agent is added in an equimolar ratio to the concentration of 2- pyridyl disulphide (SSPY) groups of the derivatised insulin peptide. In another embodiment the pyridyl disulphide group on the derivatised insulin peptide is converted to a thiol group using Tris- (2-carboxyethyl)-phosphine hydrochloride (TCEP). In a special embodiment TCEP is added in an equimolar ratio to the concentration of 2-pyridyl disulphide groups of the derivatised insulin peptide. In another special embodiment TCEP is added in a molar ratio of about 1.6: 1 to the concentration of derivatised insulin peptide.

In one embodiment of the above method the derivatised insulin peptide is isolated by preparative reverse-phase HPLC (RP-HPLC).

In one embodiment of the above method the activated insulin peptide is purified before step e). In a further embodiment the activated insulin peptide is purified by preparative reverse-phase HPLC. In another further embodiment the activated insulin peptide is purified by membrane filtration, such as for instance ultrafiltration or diafiltration.

In a further embodiment of the above method the unreacted pyridyl disulphide- containing crosslinker is removed from the pyridyl disulphide-containing crosslinker derivatives of the mucosal binding component before step e). In one further embodiment the unreacted pyridyl disulphide-containing crosslinker is removed from the pyridyl disulphide- containing crosslinker derivatives of the mucosal binding component by size exclusion chromatography, such as for instance gel filtration. In another further embodiment the unreacted pyridyl disulphide-containing crosslinker is removed from the pyridyl disulphide- containing crosslinker derivatives of the mucosal binding component by use of membrane filtration, such as for instance ultrafiltration.

In one embodiment of the above method, in step e), the activated insulin peptide is added in an equimolar amount to the 2-pyridyl disulphide groups of the mixture of pyridyl disulphide protected derivatives of the mucosal binding component.

In a further embodiment of the above method the product of conjugates between the antigen and the mucosal binding component is purified after step e). In one embodiment the product of conjugates between the antigen and the mucosal binding component is purified by size exclusion chromatography, such as for instance gel filtration, membrane filtration, such as for instance ultrafiltration, or ion-exchange chromatography or a combination thereof.

In a further special embodiment the present invention relates to a method for preparing a product of conjugates between an insulin peptide and the cholera toxin B subunit (CTB) which method comprises a) reacting the insulin peptide with N-succinimidyl 3- (2-pyridyidithio) propionate (SPDP) thereby producing a mixture of pyridyl disulphide-containing crosslinker derivatives of the insulin peptide, b) isolating the insulin peptides derivatised with a single pyridyl disulphide group, c) converting the pyridyl disulphide group on the isolated insulin peptide to a thiol group thereby producing a thiolated insulin peptide, d) reacting the cholera toxin B subunit (CTB) with N-succinimidyl 3- (2-pyridyldithio)- propionate (SPDP) thereby producing a mixture of pyridyl disulphide-containing crosslinker derivatives of CTB, e) reacting the thiolated insulin peptide with the mixture of pyridyl disulphide-containing crosslinker derivatives of CTB, thereby producing the conjugates between the insulin peptide and the CTB.

In one embodiment of the above method the insulin peptide is insulin.

In a special embodiment of the above method, in step b), the isolated insulin peptide is derivatised with a single pyridyl disulphide group at the amino terminal of the B chain.

In another embodiment of the above method the molar ratio of SPDP to the insulin peptide is from about 1: 1 to about 2.5: 1. In a further embodiment the molar ratio of SPDP to the insulin peptide is from about 1: 1 to about 2: 1. In a still further embodiment the molar ratio of SPDP to the insulin peptide is from about 1.1: 1 to 1.5: 1. In a special embodiment the molar ratio of SPDP to the insulin peptide is 1.13: 1.

In a further embodiment of the above method, in step e), the thiolated insulin peptide is added to the pyridyl disulphide-containing crosslinker derivatives of CTB in a molar ratio of thiolated insulin peptide to pentameric derivatised CTB of about 4: 1.

In a still further special embodiment the present invention relates to a method for preparing a product of conjugates between an insulin peptide and the cholera toxin B subunit (CTB) which method comprises a) reacting the insulin peptide with N-succinimidyl 3- (2-pyridyldithio) propionate (SPDP) thereby producing a mixture of pyridyl disulphide-containing crosslinker derivatives of the insulin peptide, b) isolating the insulin peptides derivatised with a single pyridyl disulphide group by preparative reverse-phase HPLC, c) converting the pyridyl disulphide group on the isolated insulin peptide to a thiol group thereby producing a thiolated insulin peptide using Tris- (2-carboxyethyl)-phosphine

hydrochloride (TCEP), d) reacting the cholera toxin B subunit (CTB) with N-succinimidyl 3-(2-pyridyidithio)- propionate (SPDP) thereby producing a mixture of pyridyl disulphide-containing crosslinker derivatives of CTB, e) reacting the thiolated insulin peptide with the mixture of pyridyl disulphide-containing crosslinker derivatives of CTB, thereby producing the conjugates between the insulin peptide and the CTB.

In a further aspect, the present invention relates to a product obtainable by any one of the methods defined above.

In a still further aspect, the present invention relates to a product of conjugates between an antigen and a mucosal binding component wherein the individual conjugate is characterised by consisting of one mucosal binding component conjugated to one or more antigens.

In one embodiment of the above product, the antigen is an insulin peptide.

In another embodiment of the above product, the mucosal binding component as present in the product of conjugates between an antigen and said mucosal binding component has GM1 binding activity. In one embodiment the mucosal binding component is the cholera toxin B subunit (CTB).

In a special embodiment the present invention relates to a product of conjugates between an insulin peptide and the cholera toxin B subunit (CTB) wherein the individual conjugate is characterised by consisting of one cholera toxin B subunit (CTB) conjugated to one or more insulin peptide units. In one embodiment the product of conjugates has an insulin to CTB ratio of from about 1: 1 to about 1: 20. In a more special embodiment the product of conjugates has an insulin to CTB ratio of from about 1: 2 to about 1: 10, preferably from about 1: 3 to about 1: 5, more preferably of about 1: 4 (the insulin to CTB ratios are based on the pentamer of CTB).

In a still further aspect the present invention relates to the use of an insulin specific T- cell hybridoma assay for the characterization of the antigen presenting potentiation of a conjugate between an antigen and a mucosal binding component.

In a further aspect the present invention relates to a pharmaceutical composition comprising, as an active ingredient, any one of the above-defined products together with a pharmaceutically acceptable carrier or diluent. In one embodiment, the composition in unit dosage form, comprises from about 0.01 to 100 mg of the product.

In a still further aspect the present invention relates to a pharmaceutical composition for inducing specific immunological tolerance, the composition comprising, as an active

ingredient, any one of the above defined products together with a pharmaceutically acceptable carrier or diluent.

In a still further aspect the present invention relates to a pharmaceutical composition for treating an autoimmune disease or allergic condition, the composition comprising, as an active ingredient, any one of the above-defined products together with a pharmaceutically acceptable carrier or diluent.

In a special embodiment the pharmaceutical composition is for oral, nasal, or pulmonary administration.

In a further aspect the present invention relates to a method of inducing specific immunological tolerance of a mammal, the method comprising administering to said mammal an effective amount of any one of the above-defined products or any one of the above- defined compositions.

In one embodiment of the method the effective amount of the product is in the range of from about 0.00001 to about 10 mg/kg body weight per day.

In another embodiment of the method the administration is carried out by the oral, nasal, or pulmonary route.

In a still further aspect the present invention relates to the use of any one of the above- defined products for the preparation of a medicament.

In a further embodiment the present invention relates to the use of any one of the above-defined products for the preparation of a medicament for inducing specific immunological tolerance in a mammal.

In a further embodiment the present invention relates to the use of any one of the above-defined products for the preparation of a medicament for treating an autoimmune disease or allergic reaction in a mammal. In a special embodiment the autoimmune disease is selected among such as those disclosed in WO 98/47529, which is incorporated herein by reference.

In a still further aspect, the present invention relates to the use of any of the above defined products or pharmaceutical compositions for inducing tolerance in the context of tissue transplantation. In one embodiment, the product can be used for decreasing the risk of rejection in a recipient of a tissue graft transplanted from a donor, by administering the product or composition to a mucosal surface of the recipient. Both xenografts and allografts are included. In another embodiment, the product can be used for decreasing the risk of graft-versus-host disease in a recipient of a tissue graft transplanted from a donor, by administering the product or composition to a mucosal surface of the recipient. In a special embodiment the antigen is a single HLA antigen, an HLA antigen cocktail, or a cell or cell

derivative from one or more individuals. Thus, in a further embodiment the product can be used for improving transplantation of a tissue graft from a donor into a recipient.

In a further aspect, the present invention relates to the use of any of the above defined products or pharmaceutical compositions for inducing immunological tolerance to target antigens on the mucosal surface. Of particular interest are target antigens preferentially associated with mucosal disorders, exemplified by inflammatory bowel disease, irritable bowel syndrome, ulcerative colitis, Celiac disease, and Chron's disease.

Any novel feature or combination of features described herein is considered essential to this invention.

PHARMACEUTICAL COMPOSITIONS The products of the invention (including the products prepared according to the methods of the invention) may be administered alone or in combination with pharmaceutically acceptable carriers or excipients, in either single or multiple doses. The pharmaceutical compositions according to the invention may be formulated with pharmaceutically acceptable carriers or diluents as well as any other known adjuvants and excipients in accordance with conventional techniques such as those disclosed in Remington: The Science and Practice of Pharmacy, 19t"Edition, Gennaro, Ed., Mack Publishing Co., Easton, PA, 1995.

The pharmaceutical compositions may be specifically formulated for administration by any suitable route such as the oral, rectal, nasal, pulmonary, topical (including buccal and sublingual), transdermal, intracisternal, intraperitoneal, vaginal, and parenteral (including subcutaneous, intramuscular, intrathecal, intravenous and intra-dermal) route.

Preferably the pharmaceutical composition is intended for mucosal administration.

Selection of a route of administration of a pharmaceutical composition to a mucosal surface will in turn depend, inter alia, on the general condition and age of the subject to be treated, on nature of the clinical condition being treated, on the conjugate chosen, and the ease of administration to a particular surface. The most typical mucosal surfaces used are those of the gastrointestinal tract, the nasal mucosa, the vaginal mucosa, and the airway mucosa.

Administration to the gastrointestinal tract may be performed by oral administration, suppositories, intubation, endoscopy, or any other suitable technique.

Pharmaceutical compositions for oral administration include solid dosage forms such as capsules, tablets, dragees, pills, lozenges, powders, and granules. Where appropriate, they can be prepared with coatings such as enteric coatings, or they can be formulated so as to provide controlled release of the active ingredient such as sustained or prolonged release according to methods well known in the art.

Liquid dosage forms for oral administration include solutions, mulsions, suspensions, syrups and elixirs, and solid forms suitable for dissolution or suspension in liquid prior to use.

Nasal administration typically involves the use of a free flowing liquid, cream, or gel containing an effective concentration in a comfortable volume. Because the nasal mucosa is relative more quiescent and there is a relative paucity of proteolytic enzymes, an effect may in some instances be obtained using a lower amount of conjugate.

Administration to the mucosa of the airway typically involves the formation and inhalation of an aerosol. The aerosol may either be a finely dispersed liquid, or a powder.

Apparatus and methods for forming aerosols are described in Kirk-Othmer,"Encyclopedia of Chemical Technology", 4t"Ed Vol. 1, Wiley NY USA, pp 670-685,1991; and Newman, "Aerosols and the Lung", Clarke & Davia, eds, Buttersworths, London, England, pp 197-224, 1984. The reader may also consult US patents No 4,624,251; 3,703,173; 3,561,444; and 4,627. Portable inhalers permit dosages to be conveniently administered a number of times a day, where necessary.

A typical oral dosage is in the range of from 0.00001 to about 10 mg/kg body weight per day, preferably from about 0.0001 to about 1 mg/kg body weight per day administered in one or more dosages such as 1 to 3 dosages. The exact dosage will depend upon the expected volume of distribution of the composition before reaching the intended site of action, the degree of degradation and penetration expected for the mode of administration, the frequency of administration, the sex, age, weight and general condition of the subject treated, the nature and severity of the condition treated and any concomitant diseases to be treated and other factors evident to those skilled in the art.

The formulations may conveniently be presented in unit dosage form by methods known to those skilled in the art. A typical unit dosage form for oral administration one or more times per day such as 1 to 3 times per day may contain from about 0.001 to about 1000 mg, preferably from about 0.01 to about 100 mg of the product.

Suitable pharmaceutical carriers include inert solid diluents or filles, sterile aqueous solution and various organic solvents. Examples of solid carriers are lactose, terra alba, sucrose, cyclodextrin, talc, gelatine, agar, pectin, acacia, magnesium stearate, stearic acid, or lower alkyl ethers of cellulose. Examples of liquid carriers are syrup, peanut oil, olive oil, phospholipids, fatty acids, fatty acid amines, polyoxyethylene, or water. Similarly, the carrier or diluent may include any sustained release material known in the art, such as glyceryl monostearate or glyceryl distearate, alone or mixed with a wax. The pharmaceutical compositions formed by combining the present compounds and the pharmaceutically acceptable carriers are then readily administered in a variety of dosage forms suitable for the

disclosed routes of administration. The formulations may conveniently be presented in unit dosage form by methods known in the art of pharmacy.

Formulations of the present invention suitable for oral administration may be presented as discrete units such as capsules or tablets, each containing a predetermined amount of the active compound, and which may include a suitable excipient. These formulations may be in the form of powder or granules, as a solution or suspension in an aqueous or non-aqueous liquid, or as an oil-in-water or water-in-oil liquid emulsion.

If a solid carrier is used for oral administration, the preparation may be tablette, placed in a hard gelatine capsule in powder or pellet form or it can be in the form of a troche or lozenge. The amount of solid carrier will vary widely but will usually be from about 25 mg to about 1 g. If a liquid carrier is used, the preparation may be in the form of a syrup, mulsion, soft gelatine capsule, or sterile injectable liquid such as an aqueous or non-aqueous liquid suspension or solution.

A typical tablet, which may be prepared by conventional tabletting techniques, may contain: Core: Product of conjugates 0.5 mg Lactosum Ph. Eur. 67.8 mg Cellulose, microcryst. (Avicel) 31.4 mg Amberlite 1.0 mg Magnesii stearas Ph. Eur q. s.

Coating: HPMC approx 9 mg Mywacett 9-40 T* approx. 0.9 mg *Acylated monoglyceride used as plasticizer for film coating.

If desired, the products of the invention may be administered in combination with further pharmacologically active substances, including agents that enhance the tolerogenic effect of the conjugate at the mucosal surface. An example of an additional active component is a cytokinine, e. g. IL-4, IL-10, or TGFß.

Since the compositions are intended for mucosal administration, it is useful to prepare compositions that are not only stable for the expected shelf life, but also resistant to the pH extremes, enzymes, and other assauts of the mucosal environment.

The present invention is further illustrated by the following methods and examples which are, however, not intended to limit the scope of the invention in any way.

MATERIALS AND METHODS: Materials Human insulin produced recombinantly in yeast was obtained from Novo Nordisk (Bagsvaerd, Denmark). Recombinant CTB produced in a mutant strain of Vibrio cholerae deleted of the cholera toxin A subunit gene was obtained from SBL Vaccin (Stockholm, Sweden). SPDP, Tris (2-carboxyethyl) phosphine hydrochloride (TCEP), o-phenylene diamine (OPD) and SuperBlockTM were obtained from Pierce (Illinois, USA). Bio-Gel P-2 was obtained from Bio Rad (California, USA). SDS-gels, Mark 12 protein molecular weight standards, and Multi-Mark molecular weight standards were obtained from Novex (San Diego, CA). Coomassie Brillant Blue R staining solution and ganglioside GM1 were from Sigma (Missouri, USA). 1, (DMPC) was purchased from Avanti Polar Lipids (Alabaster, AL). The insulin-specific T cell hybridoma H-11, which was isolated from a NOD mouse, was kindly provided by Dr Dale Wegmann (Barbara Davis Center for Childhood Diabetes, University of Colorado, Denver, Colorado). The interleukin-2 (IL-2) ELISA kit was obtained from Pharmingen (San Diego, CA) and used according to the manufacturer's instructions. Sequencing grade trypsin was from Boehringer Mannheim, (Indianapolis, IN). RPMI was obtained from Gibco, Life Technologies (Gaithersburg, MD) and PSN (Penicillin-Streptomycin) were from Gibco.

Antibodies For western blotting experiments the antibodies used were as follows : for insulin detection a monoclonal antibody to human proinsulin was produced at Zymogenetics and a goat anti-mouse antibody coupled to horseradish peroxidase from Rockland Immunochemicals (Gilbertsville, PA) was employed for visualisation; for CTB detection a goat anti-choleragenoid antibody was purchased from List Biological Laboratories (Campbell, CA) and was used with a donkey anti-goat antibody coupled to horseradish peroxidase obtained from Jackson Immunochemicals (West Grove, PA).

Antibodies employed in ELISAs were as follows : for the Gml/insulin ELISA a monoclonal antibody against insulin was obtained from Sigma (Missouri, USA) and a goat anti-mouse antibody coupled to horseradish peroxidase from Biosource International (Camarillo, CA) was used as the secondary antibody; the GM,/CTB ELISA employed a goat anti-choleragenoid antibody purchased from List Biological Laboratories (Campbell, CA) and a donkey anti-goat antibody coupled to horseradish peroxidase from Jackson Immunochemicals (West Grove, PA) was used as the secondary antibody; the sandwich

ELISA employed the Sigma monoclonal antibody against insulin, a chicken antibody against CTB which was prepared by ZymoGenetics, and a goat anti-chicken antibody coupled to horseradish peroxidase from Promega (Madison, WI) for visualisation of immunoreactivity.

Biacore analysis employed monoclonal antibodies prepared at ZymoGenetics against human proinsulin and human thrombopoietin.

Methods For step a) of a method according to the present invention, the antigen and the first crosslinker should be reacted under conditions, which for a person skilled in the art are appropriate for the specific antigen and crosslinker. The first crosslinker may be added to the antigen in a ratio of crosslinker to antigen of from about 1: 1 to about 2.5: 1. A ratio of crosslinker to antigen of about 1.13: 1 is preferred.

In step b) of a method according to the present invention, antigen derivatised with a single crosslinker residue may be isolated by use of methods known in the art such as for example size exclusion chromatography, RP-HPLC, membrane filtration, ion-exchange chromatography and other methods known by a person skilled in the art to be suitable for isolation of such a compound. RP-HPLC is preferred for the purpose of the present invention.

The isolation of antigen derivatised with a single crosslinker residue reduces potential problems with aggregation in connection with the later conjugation to the derivatised mucosal binding component (derivatised with varying number of crosslinker residues).

Following the isolation the derivatised antigen may be worked up using for instance lyofilisation, precipitation, resolubilisation and the like and then, in step c) of a method according to the present invention, activated in a manner suitable for the activation of the chosen first crosslinker. The conditions for the activation should be chosen so that the activation of as many of the crosslinker derivatives as possible, preferably all, is achieved.

The resulting activated crosslinker derivative may be isolated by use of any method known by a person skilled in the art to be suitable for isolation of such a compound such as for example size exclusion chromotography, such as for instance gel filtration, membrane filtration, such as for instance gel filtration, and ion-exchange chromatography. Gel filtration is preferred for the purpose of the present invention.

For step d) of a method according to the present invention, the mucosal binding component and the second crosslinker should be reacted under conditions, which for a person skilled in the art are appropriate for the specific mucosal binding component and crosslinker. If the mucosal binding component is of a multimeric structure this should be taken into consideration when choosing the molar ratio of crosslinker to the mucosal binding component. The molar ratio of crosslinker to the mucosal binding component, as the mucosal

binding component appears in the product of conjugates, should be chosen to ensure that each biologically active mucosal binding component will be able to be conjugated to at least one antigen residue. The upper limit for the amount of crosslinker conjugated to the mucosal binding component (which eventually will decide the amount of antigen conjugated to the mucosal binding component) will be a result of among others steric considerations, the number of sites on the mucosal binding component available for derivatisation, the nature and potency of the antigen and economic considerations.

To concentrate the resulting derivatised mucosal binding component and/or remove any unreacted second crosslinker, the resulting reaction mixture may be subjected to such treatments as size exclusion chromatography, such as for instance gel filtration, membrane filtration, such as for instance ultrafiltration, ion-exchange chromatography or the like.

Ultrafiltration is preferred for the purpose of the present invention.

For step e) of a method according to the present invention, the derivatised mucosal binding component and the activated antigen should be reacted under conditions, which for a person skilled in the art are appropriate for the specific derivatised mucosal binding component and the specific activated antigen. For some conjugates of antigen and mucosal binding component, where the mucosal binding component has a multimeric structure, it might be preferable that the mucosal binding component is dissociated into monomeric subunits prior to the conjugation and reassociated after the conjugation.

The amount of activated antigen to use should be calculated according to the amount of crosslinker residues on the derivatised mucosal binding component. The molar ratio of activated antigen to multimeric derivatised mucosal binding component should be based on the degree of derivatisation of the derivatised mucosal binding component (i. e. the amount of crosslinker residues on the multimeric derivatised mucosal binding component as described in the previous paragraph) as exemplified in Method B.

The pyridine thione released by the conjugation reaction may be removed by use of such treatments as size exclusion chromatography, such as for instance gel filtration, membrane filtration, such as for instance ultrafiltration, ion-exchange chromatography or the like. For the purpose of the present invention ultrafiltration is preferred.

In the following, the method of preparing a product of conjugates between an antigen and a mucosal binding component according to the present invention is exemplified by the following two methods of preparing a product of conjugates between insulin and CTB.

Method A Conjugation of insulin to CTB a) Crosslinker derivatisation of insulin

Insulin, dissolve in 0.1 M acetic acid at 10 mg/ml (s = 0.855 x 103 M-'cm-'at 280 nm used to determine the concentration), is diluted with 0.5 M Hepes, pH 8.5 and the pH is adjusted such that the final solution is 5 mg/ml insulin in 0.1 M Hepes, pH 7.0. SPDP is dissolve in ethanol at about 10 mg/ml, and its concentration confirmed by measuring the increase in absorbance at 260 nm upon addition of sodium bicarbonate and release of N- hydroxysuccinimide (s = 8.2 x 103 M-'cm-'at 260 nm). SPDP is added in a 2: 1 molar ratio to insulin, and the solution incubated 1 hour at room temperature with gentle rocking. b) Isolation of insulin derivatised with a single crosslinker The insulin derivatised with 2-pyridyl disulphide (SSPY) at the amino terminus of the B chain, referred to as insulin-SSPY, is isolated by preparative reverse-phase HPLC (RP- HPLC) using a Waters Delta Prep HPLC at 60 ml/min with a 52 x 250 mm Vydac C4 Column (10-15 um, 300 A) and a gradient from 33% acetonitrile/0. 1 % TFA to 36% acetonitrile/0. 1 % TFA over 30 minutes. The pool of insulin-SSPY containing the 2-pyridyl disulphide at the amino terminus of the B chain is identified by LC/MS following tryptic digestion.

Following lyophilisation the insulin-SSPY is resolubilised in 0.1 M citrate, pH 4.0 containing 30% acetonitrile, and the concentration determined by measuring the absorbance at 280 nm and correcting for the contribution of SSPY groups to the absorbance (Grassetti et al, (1967) Biochem. Biophys. 119: 41-49). Briefly, the molar concentration of SSPY groups in the preparation is determined by measuring the change in absorbance at 343 nm upon addition of 100 mM dithiothreitol (c = 8.08 x 103 M-'cm-'at 343 nm, Stutchbury et al., (1975) Biochem J. 151: 417-432). This concentration is multiplie by 5.1 x 103 and subtracted from the absorbance measured at 280 nm to give the absorbance due to protein at 280 nm. The insulin concentration is then calculated using e = 0.855 x 103 M-'cm-'. c) Activation of derivatised insulin The SSPY groups on insulin are converted to thiol groups by addition of TCEP in an equimolar ratio to the 2-pyridyl disulphide concentration. The mixture is rocked for 30 minutes at room temperature and the resulting thiolated insulin (insulin-prop-SH) purified by preparative RP-HPLC as outlined above. The purified insulin-prop-SH is then lyophilised. d) Crosslinker derivatisation of CTB Three separate reactions are carried out in order to prepare CTB derivatised with SSPY (CTB-SSPY) with differing amounts of incorporated cross-linker. Recombinant CTB, prepared at 5 mg/ml in 0.1 M borate, pH 8.0 (using E = 1.04 x 103 M-'cm-'at 280 nm), is incubated with freshly prepared SPDP at various concentrations (table 1) for 1 hour at room temperature. Unreacted SPDP is removed by size exclusion chromatography on Bio-Gel P-2, using 0. M borate, pH 8.0 as buffer. The concentration of 2-pyridyl disulphide groups in the

preparations is determined as outlined above and the protein concentration of the CTB- SSPY, is calculated using e = 1. 04 x 103 M-1 cm-1 at 280 nm after correcting for the contribution of pyridyl disulphide to the absorbance.

Analysis of CTB-SSPY preparations by RP-HPLC is carried out using a Hewlett Packard 1100 HPLC at 0.5 ml/min with a 3.2 x 150 mm Vydac C4 column (5, u, 300 A) at 40°C and a gradient from 30% acetonitrile/0. 1 % TFA to 50% acetonitrile/0. 1 % TFA over 20 minutes. e) Conjugation of insulin and CTB For each CTB-SSPY preparation solid insulin-prop-SH is added in an equimolar amount to the 2-pyridyl disulphide groups in the CTB-SSPY, as indicated in table 1.

Following a 2-hour incubation at room temperature the released pyridine thione is removed by size exclusion chromatography on Bio-Gel P-2 using 0.35 M sodium bicarbonate as buffer.

Method B Conjugation of insulin to CTB a) Crosslinker derivatisation of insulin Insulin is suspended in 4 mM EDTA to a concentration of 10 mg/ml. 1 M triethanolamine is added to a concentration of triethanolamine of 10 mM and the pH adjusted to 7.5. SPDP is dissolve in ethanol at about 7.5 mg/ml. SPDP is added in a 1.13: 1 molar ratio to insulin over a period of 15 to 30 minutes and if necessary the pH is adjusted to 7.5.

The reaction mixture is left at room temperature for at least 1 hour. b) Isolation of insulin derivatised with a single crosslinker The insulin derivatised with 2-pyridyl disulphide (SSPY) at the amino terminus of the B chain (in the B1 position), referred to as insulin-SSPY, is isolated by preparative reverse- phase HPLC (RP-HPLC) using a C18 substituted Fuji Davison silica (15 um, 200 A) Using a a 250 mm column, elution is carried out in a buffer containing 28. mM Na2SO4/10 mM citric acid, pH3. The pool of insulin-SSPY containing the 2-pyridyl disulphide at the amino terminus of the B chain is identified by LC/MS following tryptic digestion. The eluate from the RP-HPLC is cooled to a temperature of about 4°C under stirring and the resulting precipitate is then resuspended to a final concentration of insulin- SSPY in the aqueous suspension of about 10 mg/ml.

The pH of the aqueous suspension is adjusted to 7.4 and the precipitate is dissolved under gentle stirring. The solution is filtered on 0.45 pm filter and 1 M NaCI is added to a concentration of NaCI of about 30 mM.

c) Activation of derivatised insulin The SSPY groups on insulin are converted to thiol groups by addition of TCEP in an molar ratio of TCEP to insulin-SSPY of about 1.6: 1 at a pH of 3. The resulting insulin-prop- SH is then subjected to gel filtration on a Pharmacia Sephadex G-25 Medium column using 50 mM NaCI/2 mM citric acid, pH 3.1/NaOH as the eluent. The eluate is immediately cooled to a temperature of about 4°C and stored at that temperature. d) Crosslinker derivatisation of CTB This procedure is carried out in order to prepare CTB-SSPY with approximately 4 SSPY groups per CTB pentamer. 1 mol/kg triethanolamine, pH 8.0 is added to recombinant CTB (approximately 10 mg/ml in 150 mM NaCI, 22 mM NaH2PO4, pH 7.4) to a concentration of triethanolamine of 25 mmol/kg and pH is adjusted to 8.0. Freshly made SPDP (7.5 mg/ml in ethanol) is added in a molar ratio of 8.75 SPDP to 1 CTB pentamer over a period of about 15 minutes. The reaction mixture is left at room temperature for at least 1 hour.

To concentrate the CTB-SSPY and to remove low molecular reaction products, the resulting reaction mixture is subjected to diafiltration on a Biomax 5 membrane (5 kD) using 50 mmol/kg NaCI/10 mmol/kg citric acid/pH 7.0. The degree of derivatisation of CTB-SSPY is verified by use of RP-HPLC. e) Conjugation of insulin and CTB The pH of the CTB-SSPY from step d) is adjusted to about 3.4 with 1 M HCI which causes the CTB-SSPY to dissociate into monomers. Since the CTB-SSPY is prepared with approximately 4 SSPY groups per CTB pentamer, the insulin-prop-SH from the final gel filtration in step a) is added thereto in a molar ratio of insulin-prop-SH to pentamer CTB- SSPY of 4: 1 under gentle stirring at a temperature of between 18 and 25 °C over a period of 0 to 60 minutes.

The resulting mixture is kept at room temperature for no longer than about 48 hours and is then subjected to size exclusion chromatography on Pharmacia Superdex 75 under eluation with 50 mM NaCI/10mM citric acid, pH 3. The pH of the eluate is adjusted to 7.8 with 0.5 M NaOH over a period of about 1 hour, which causes the reassociation of the conjugated product into pentamers. Alternatively, the pH of the resulting mixture is adjusted to 7.8 over a period of 0 to 60 minutes and then the mixture is subjected to size exclusion chromatography an a Pharmacia Superdex 200 under eluation with 50 mM NaCI/10 mM phosphate/pH 7.4.

Characterisation and analysais of CTB-insulin Size Exclusion Chromatooraphy Analytical size exclusion chromatography is carried out using a Hewlett Packard HP1050 HPLC equipped with a diode array detector, binary pump system and Chemstation software. A 10 x 300 mm Superdex-200 column (Pharmacia, USA) is used with 50 mM sodium phosphate, pH 7.0 containing 100 mM NaCI as buffer, and a flow rate of 0.6 ml/min.

ELISA Three different ELISAs are used to characterise CTB-insulin (CTB conjugated to insulin). Two of these involve binding of conjugate to immobilised ganglioside GM1, as outlined by Svennerholm and Holmgren (1978), with antibodies against either insulin or CTB used for detection. 96-well ELISA plates are coated with 0.1 mi 5.0 pg/ml GM1 in 0.1 M sodium carbonate, pH 9.0, at 4°C for 16 hours. Following five washes with 0.2 mi PBS/0.05% Tween-20 (ELISA buffer) unbound sites are blocked by incubating for 10 minutes at room temperature with SuperBlockTM. Following a further five washes 0.1 mi of each sample diluted in 1% (w/v) bovine serum albumin in ELISA buffer is added to the plates in duplicate.

Plates are incubated 1 hour at 37°C with shaking and washed as outlined above. For detection of GM,-bound insulin sequential incubations at 37°C for 1 hour with 0.1 ml of 6.9 mg/ml anti-insulin monoclonal antibody and 0.1 mi of a 1: 1,000 dilution of goat anti-mouse antibody coupled to horseradish-peroxidase are carried out with washes in between as outlined above. For detection of GM,-bound CTB a 1: 2, 500 dilution of goat anti-CTB antibody and a 1: 10,000 dilution of donkey anti-goat antibody coupled to horseradish-peroxidase are used. Antibodies are diluted in 1 % (w/v) bovine serum albumin in ELISA buffer. Following washing immobilised antibodies are visualised by addition of 0.1 ml development reagent (0.4 mg/ml OPD, 0.024% hydrogen peroxide in 0.1 M citrate, pH 5.0). After 10 minutes at room temperature the reaction is terminated by addition of 0.1 ml 1 M sulphuric acid and absorbance measured at 490 nm.

The third ELISA relies on direct capture of CTB-insulin conjugates using antibodies against insulin and detection with antibodies directed against CTB. 96-well plates are coated at 4°C for 16 hours with 0.1 ml of 5. ug/ml anti-insulin monoclonal antibody in 0.1 M sodium carbonate, pH 9.0. Washing, blocking of unbound sites, and application of samples are carried out as outlined above. Plates are then incubated for 1 hour at room temperature with 0.1 ml per well of chicken anti-CTB antibody diluted to 0.5 mg/ml in 1 % (w/v) bovine serum albumin in ELISA buffer. Following washing and incubation for 1 hour at 37°C with 0.1 ml 0.5 pg/ml goat anti-chicken antibody coupled to horseradish peroxidase immobilised conjugate is visualised as outlined above.

LC-MS LC-MS analysis of the CTB-insulin conjugates employed a Michrom BioResources Magic 2002 HPLC system (Michrom BioResources, Inc., Auburn, CA) equipped with a 1.0 x 150 mm PLRP-S 1000A 8u column (Polymer Laboratories, Amherst, MA), which is used at a flow rate of 50 ml/min and a column temperature of 30°C. 5. ug of protein in 50 ul of 0.35 M NaHCO3, pH 8.1, is injected onto the column using a 100 pI sample loop with an injection time of 200 msec. The column is equilibrated at 30% B and a linear gradient from 30 to 50% B over 90 minutes is immediately initiated (A: 2% acetonitrile + 0. 1 % TFA, B: 90% acetonitrile + 0.09% TFA). The outlet from the HPLC UV detector is plumbed directly into a Finnigan LCQ Ion Trap Mass Spectrometer (Thermoquest Corp., San Jose, CA) with no flow splitting, a heated capillary temperature of 200°C, and a sheath gas flow of 80 (arbitrary units). The source voltage is 5.60 kV and the capillary voltage is 43.67 V. Mass spectra from 400-2000 m/z are recorded continuously during the gradient with 3 microscans per full scan.

In some cases tryptic digestion is carried out prior to LC-MS to allow identification of sites of derivatisation, as well as estimation of the degree of derivatisation. Insulin-SSPY and CTB-SSPY are digested with trypsin at 1: 50 (w: w) trypsin: protein in 100 mM Tris, pH 8.5 containing 100 uM TPCK for 18-24 hrs at 37°C. Prior to trypsinisation CTB-SSPY is reduced in 6.0 M guanidine/HCI, 150 mM Tris, 0.25 mM EDTA, pH 7.6 with a 25-fold molar excess of dithiothreitol over the number of cysteine residues for 1 hr at 37°C. The reduced protein is then alkylated by addition of a 5-fold molar excess of 4-vinylpyridine followed by incubation in the dark at room temperature for 2 hours. The pyridylethylated CTB-SSPY is immediately desalted using RP-HPLC and lyophilised prior to trypsinisation.

Amino Acid Analysis Amino acid analysis uses the Waters AccQ-Tag@ chemistry according to the manufacturer's instructions. To calculate the concentration, the corrected average pmol for each amino acid is multiplie by the molecular weight of that amino acid. These values are then summed and corrected for the amount of sample hydrolyse and the percentage of the hydrolyse sample injected to the analyser as well as for destruction of Cys, Met and Trp during hydrolysis. To calculate the molar ratio of insulin to CTB monomer (CTB,) the experimentally determined pmol levels of three amino acids whose compositions are the same in CTB, and insulin are used to determine total moles present in the sample as follows : [total moles = 1/3 (moles Leu/6 + moles Phe/3 + moles Val/4)].

The moles of insulin and CTB, are determined using the pmol determined for the amino acids Asx, Glx, Ala and lie, whose compositions differ in CTB and insulin, according to the followingequations: moles insulin = [a. a.-moles CTB, (a. a. c)]/a. a. i.

moles CTB, = [total moles (a. a. i.)-a. a.]/ [a. a. i.-a. a. c.] where a. a. is the moles of amino acid (determined experimentally for the sample), a. a. = moles of that amino acid in insulin (from the amino acid composition of insulin) and a. a. = moles of that amino acid per CTB, (from the amino acid composition of CTB,). The molar ratio of insulin to CTB, is calculated by dividing moles insulin by moles CTB, for each of the four determinations, and averaging the resulting values. For the purpose of defining the molar insulin : CTB, ratio, we assume all the insulin present in the final material is conjugated to CTB.

Surface Plasmon Resonance Biomolecular Interaction Analysis Analysis of the biomolecular interaction between CTB or CTB-insulin conjugates and GM1 is carried out using a BlAcore Model 1000 with upgrade and sensor chips coated with GM,-containing liposomes. The liposomes are prepared by mixing 0.5 mM DMPC and 2% ganglioside GM1 in PBS. Following 10 cycles of freeze-thawing in an ethanol/dry ice bath the mixture is passed through a Mini Extruder (Avanti, Alabaster, AL) nine times according to the manufacturer's instructions. Following degassing an HPA Sensor Chip (BlAcore, Inc., Uppsala, Sweden) is coated with liposomes according to the manufacturer's instructions. A flow cell that has been coated with DMPC alone is used as a negative control surface.

Following cleaning the efficiency of flow cell coating is tested to confirm that flow cells accumulated 1000-2000 RU of liposomes and less than 100 RU from a 100 ug/ml bovine serum albumin solution.

Binding of recombinant CTB (positive control) and CTB-insulin conjugates to flow cells is carried out at a final concentration of 20 ug/ml in PBS. Following binding of the samples, their ability to bind antibodies against insulin is tested by flowing a monoclonal antibody against proinsulin (20 ug/ml in PBS) across the flow cell. A monoclonal antibody against thrombopoietin (20 pg/ml in PBS) is used as a negative control to confirm that binding of the anti-insulin antibody is specific.

SDS-PAGE and Western Blot Analvsis SDS-PAGE is carried out using 10-20% Tris-Tricine gels and a Novex Xcell II mini-gel box according to the manufacturers instructions with two exceptions. First, a voltage of 95 V is used during electrophoresis, which is lower than recommended and is necessary to prevent heating of the gel, which causes dissociation of the CTB pentamer (CTB5). Second, the samples are not heated following addition of sample buffer, again to avoid dissociation of the CTB5. 50-200 ng of each sample are loaded on the gel for western blotting and 1-5 pg for coomassie blue staining.

Following electrophoresis the gels are either stained with coomassie blue using the staining solution according to the manufacturers directions, or the samples are electroblotted to nitrocellulose for 2 hours at 400 mA in a Hoefer Mighty Small tank using 25 mM Tris, pH 7.4,192 mM glycine, 20% methanol as transfer buffer. Unbound sites on the nitrocellulose are blocked with western buffer A (50 mM Tris, 5 mM EDTA, 150 mM NaCI, 0.05% Igepal CA-630,0.25% gelatine) for 1 hour at room temperature. The membrane is then incubated overnight at 4°C with primary antibody diluted in western buffer A. For insulin detection the mouse monoclonal antibody is used at a 1: 1000 dilution and for CTB detection the goat anti- choleragenoid antibody is used at a 1: 10,000 dilution. Following three 20 minute washes in western buffer A the membranes are incubated for 1 hour at room temperature with secondary antibodies in western buffer A at a 1: 4000 dilution for the anti-mouse and a 1: 20,000 dilution for the anti-goat antibodies for insulin and CTB detection respectively. The membranes are then washed three times with western buffer B (50 mM Tris, pH 7.4,5 mM EDTA, 5% NP-40, 150 mM NaCI, 0.05% Igepal CA-630,0.4% SDS) for 20 minutes each time, and then incubated with the Amersham ECL reagent according to the manufacturers instructions. Proteins are visualized by exposing the membranes to X-ray film.

T Cell Activation Assay The ability of the various conjugates to activate T cells is determined by monitoring IL-2 production by the insulin specific T cell hybridoma H-11 in the presence of spleen cells derived from NOD mice. NOD mice are killed by cervical dislocation and their spleens removed and cleaned of any remaining pancreas and fat tissue. Each spleen is homogenised into a single cell preparation between two microscope slides and diluted into 50 mi of RPMI 1640 containing 10% FBS and PSN. The cells are washed in 50 mi media, and the red blood cells lysed by addition of 9 volumes of water followed immediately by 1 volume of 10 X PBS. The cells are then washed again in 50 ml media and suspended at 10 x 106 cells/ml. The hybridoma T cells are also washed in media and added to the spleen cells at 10 x 105 cells/ml. Insulin or CTB-insulin conjugates are diluted into 100 pl of media and mixed with 100 ul of cells containing 1 x 106 freshly isolated spleen cells and 1 x 105 hybridoma T cells. After 24 hours incubation 100 NI supernatant is removed and assayed for IL-2 by ELISA.

LIST OF FIGURES Figure 1: Preparation of insulin-prop-SH.

(A) Purification of insulin-SSPY. Insulin was reacted with a 2-fold molar ratio of SPDP at pH 7.0 for 1 hour at room temperature. Insulin derivatised at the N terminus of the B chain was

purified by preparative RP-HPLC on a 52 mm column. LC-MS following reduction and tryptic digestion was used to identify peaks. Some underivatised insulin remained (peak 1), as well as insulin derivatised at the N-terminus of the A chain or at lysine 29 of the B chain (peak 11), at the N terminus of the B chain (peak lit), and at two or three primary amino groups (peak IV).

(B) Purification of insulin-prop-SH. The material in panel A eluting between 33.0 and 34.6 minutes, which represents insulin derivatised with SSPY at the N-terminus of the B chain, was pooled, lyophilised, reduced with TCEP, and purified by RP-HPLC on a preparative 52 mm column. Mass spectrometric analysis indicated the major peak eluting at 32.3 minutes is insulin-prop-SH (peak 1), while the peak at 33.9 minutes is insulin-SSPY that failed to be reduced in this reaction (peak 11). Material eluting between 31.4 and 33.0 minutes was pooled and lyophilised for conjugation to CTB-SSPY.

Figure 2: Preparation of CTB-SSPY.

CTB was derivatised with SPDP at various molar ratios and unreacted SPDP was then removed by gel filtration. The number of 2-pyridyl disulphide groups on the CTB, was determined, and the purified CTB-SSPY was analysed by RP-HPLC as outlined in methods.

Chromatograms A-D show CTB containing 0,0.41,1.43, or 2.82 2-pyridyl disulphide groups per monomer respectively. LC/MS analysis indicated the peak eluting at 28 minutes corresponds in mass to underivatised CTB monomer, while later peaks correspond in mass to CTB, derivatised with 2-pyridyl disulphide SSPY groups at various primary amines, with retention time increasing as the number of SSPY groups increased.

Figure 3: Size exclusion chromatography of CTB-insulin conjugates.

CTB-insulin conjugates prepared with CTB-SSPY preparations containing 0.43,1.43 or 2.82 pyridyl sulphide groups per CTB monomer (panels A-C respectively) were analysed by size exclusion chromatography on a Superdex-200 column as outlined in methods. Fifty ut of each conjugate was applied to the column, which is equivalent to 31.5,29.5 and 30. ug CTB for the three conjugates in A-C, respectively. One-minute fractions were collected from 14 to 32 minutes for each run and analysed by three different ELISAs as shown in figure 6.

Recombinant CTB elutes from this column at 28 minutes and insulin elutes at 32 minutes.

Figure 4: SDS-PAGE analysis of CTB-insulin.

CTB-insulin conjugates were electrophoresed on 10-20% tricine gels and either stained with Coomassie Blue (A) or transferred to nitrocellulose and visualised by western blot analysis with antibodies against CTB (B) or insulin (C). The relative migration position of molecular weight markers is indicated.

(A) Lanes 1 and 7 contained molecular weight markers while lanes 2,3, and 4 contained conjugates prepared with CTB-SSPY preparations containing 0.43,1.43, or 2.82 pyridyl

sulphide groups per CTB monomer respectively. Lane 5 contained recombinant CTB while lane 6 contained recombinant insulin.

(B) Lanes 1,2, and 3 contained conjugates prepared with CTB-SSPY preparations containing 0.43,1.43, or 2.82 pyridyl sulphide groups per CTB monomer respectively, and lane 5 contained recombinant CTB.

(C) Lanes 1,2, and 3 contained conjugates prepared with CTB-SSPY preparations containing 0.43,1.43, or 2.82 pyridyl sulphide groups per CTB monomer respectively.

Figure 5: LC-MS analysis of CTB-insulin.

Five ug of conjugates with ratios of insulin to CTB, of 0.33,1.0, and 1.79 (panels A-C respectively) were analysed by RP-HPLC with on-line MS as outlined in"Methods". Labelle peaks were identified as (I) CTB,, (II) insulin-prop-SH dimer, (III) CTB,-insulin" (IV) CTB,- insulin2, (V) CTB,-insulin3, and (VI) CTB,-insulin5.

Figure 6: ELISA of unfractionated CTB-insulin.

Conjugates with ratios of insulin to CTB, of 0.33 (), 1.0 and 1.79 (-) were assayed in the sandwich ELISA (A) and in GM1 binding ELISAs with antibodies to CTB (B) or insulin (C) for detection at different dilutions. The undiluted conjugates contained similar concentrations of CTB (see table 1).

Figure 7: ELISA of fractionated CTB-insulin.

Fifty ul of each conjugate, containing approximately 0.6 mg CTB/ml (see table 1), with ratios of insulin to CTB, of 0.33 (), 1.0 (), and 1.79 (-) were fractionated by size exclusion chromatography as shown in figure 3. One-minute fractions were collected from 14 to 32 minutes. Each fraction was then assayed in the sandwich ELISA (A) and in GM1 binding ELISAs with antibodies to CTB (B) or insulin (C) for detection.

Figure 8: Biacore analysis of CTB-insulin.

Binding of CTB and CTB-insulin to a Biacore sensor chip coated with GM1 was carried out as outlined in methods. The arrows indicate the points of addition to the sensor chip of (1) the sample, (2) a buffer wash, (3) antibodies against a non-specific antigen, (4) a buffer wash, (5) antibodies against insulin, and (6) buffer wash.

In panel A the amount of sample applied to the chip contained 20 ug of CTB, in panel B 20 pg of total protein was applied to the chip, and in panel C the amount of sample applied contained 20 ug of insulin. Samples applied to flow cells in which the lipid bilayer contained GM1 were recombinant CTB (), conjugate with a molar insulin to CTB1 ratio of 0.33 (---- ---), conjugate with a molar insulin to CTB, ratio of 1.0 (------), and conjugate with a molar insulin to CTB, ratio of 1.79 (----). The solid line in each panel indicates binding of conjugate to a flow cell in which GM, was not added to the bilayer.

Figure 9: Activation of T cells by CTB-insulin.

Insulin (+) or CTB-insulin conjugates with ratios of insulin to CTB1 of 0.33 (#), 1.0 () or 1.79 (-) at various concentrations were incubated with freshly isolated spleen cells and hybridoma T cells for 24 hours and supernatants were then assayed for IL-2 by ELISA as outlined in"Methods".

EXAMPLES The process for preparing the conjugates and the conjugates are further iliustrated in the following examples, which however, are not to be construed as limiting.

Example 1 Production of the thiolated insulin (insulin-prop-SH) Insulin contains three primary amino groups that may be derivatised with SPDP: two are at the NH2 termini of the A and B chains while the third is the £-amino group of lysine 29 of the B chain.

To prepare the conjugate in which insulin is conjugated to CTB at the N-terminus of the B chain, insulin was reacted with SPDP at pH 7.0 for 1 hour at room temperature. The reaction products were resolved by RP-HPLC and identified by LC-MS following tryptic digestion and reduction. Although the major product was insulin derivatised at the N terminus of the B chain, some underivatised insulin remained. The reaction also generated insulin derivatised at two or three primary amino groups, as well as at the N-terminus of the A chain or lysine 29 of the B chain (figure 1A). Following preparative reverse phase HPLC purification insulin derivatised with SSPY at the N-terminus of the B chain was recovered in a 35-40% yield.

Reaction with higher concentrations of SPDP, reaction at increased temperatures, or reaction for longer periods of time did result in a decrease in the amount of underivatised insulin. However, this was accompanied by an increase in the amount of insulin containing more than one SSPY group, and little change in the amount of insulin derivatised with SSPY at a single site.

The omission of the purification step resulted in high molecular weight aggregates and solubility problems in the final preparation, as well as limitations as far as characterization of the product, which was increased in complexity.

The lyophilised insulin-SSPY was resuspended at pH 4.0 prior to reduction to the reactive insulin-prop-SH molecule with an equimolar ratio of TCEP to 2-pyridyl disulphide groups. This reducing reagent was chosen since, unlike classic reducing reagents such as dithiothreitol, it is rendered unreactive after reduction of substrate has occurred and hence only stoichiometric reduction may take place.

LC-MS analysis confirmed that the reaction resulted in specific reduction of the SSPY groups to the reactive prop-SH while maintaining the internal disulphide bonds in insulin. The specificity of the reduction derives from the fact that the SSPY group is more reactive than

the internal insulin disulphide bonds. Subsequent purification of the insulin-prop-SH by reverse phase HPLC resulted in approximately a 70% stepwise yield (figure 1 B).

Example 2 Production of CTB-SSPY CTB was reacted with varying amounts of SPDP in order to generate conjugates with varying amounts of insulin coupled to CTB. As the molar ratio of SPDP to CTB, in the reaction mixture was increased, the number of 2-pyridyl disulphide groups on the CTB, increased. Three reactions were carried out, and generated CTB containing 0.41,1. or 2.82 2-pyridyl disulphide groups per monomer (table 1, see next page).

Unreacted SPDP was removed by size exclusion chromatography and the products examined by RP-HPLC with mass spectrometric analysis for identification of resolved species (figure 2). Since the CTB pentamer dissociates at acidic pH into its monomeric form, and the RP-HPLC analysis is carried out in 0. TFA, CTB monomers are visualized in this analysis. The chromatograms show that underivatised monomeric CTB decreases as the number of pyridyl disulphide groups incorporated into each monomer increases. This is accompanied by an increase in peaks corresponding to CTB monomers containing 1 or more 2-pyridyl disulphide groups.

Multiple peaks corresponding in mass to derivatisation of the CTB with a single pyridyl disulphide group were observed in each preparation. Presumably these peaks represent derivatisation at different primary amines within the CTB monomer. Similar patterns were seen with higher stoichiometric levels of pyridyl disulphide incorporated into the CTB, again indicating derivatisation at multiple sites. This was confirmed by mass spectrometric analysis following proteolytic digestion, which indicated 6 of the 9 lysine residues in CTB, may be derivatised with SSPY, as well as the NH2 terminus. Hence, the product of the reaction is heterogeneous, and may be manipulated with respect to the number of SSPY groups added to each pentamer. The purified CTB-SSPY is typically generated with an 80% yield.

Table 1 : Preparation of conjugates with differing insulin:CTB1 ratios mg insulin- SPDP:CTBa [ CTB-SSPY ]b SSPY/CTB1b prop-SH [Protein]d Insulin:CTB1d [CTB]d (mg/ml) conjugatedc (mg/ml) (mg/ml) 0.5 2.09 0.43 1.1 0.64 0.33 0.63 1.5 2.53 1.43 3.6 0.88 1.0 0.59 3.0 2.51 2.82 7.1 1.16 1.79 0.61 a In order to prepare CTB with differing levels of pyridyl disulphide derivatisation SPDP was added to 5 mg/ml CT<BR> molar ratios as outlined in methods.<BR> b The concentration of CTB-SSPY and the number of SSPY groups per CTB monomer was determined following<BR> unreacted SPDP by size exclusion chromatography as outlined in methods.<BR> c Solid insulin-prop-SH was added to 5 mg CTB-SSPY in equimolar ratios to the number of pyridyl disulphide grou<BR> SSPY.<BR> <P>D Amino acid analysis was used to determine the total protein concentration, the ratio of insulin to CTB, and the co<BR> insulin and CTB in the three conjugates following removal of released pyridine thione by size exclusion chromatogr

Example 3 Conjugation of insulin-prop-SH to CTB-SSPY Conjugates were prepared using each of the CTB-SSPY preparations by mixing insulin-prop-SH and CTB-SSPY in ratios designed to minimize the amounts of unreacted components in the final conjugates. Hence, the number of SSPY groups in the purified CTB- SSPY solutions were quantitated and, since each SSPY group may react with one insulin- prop-SH molecule, insulin-prop-SH was added to the CTB-SSPY solution in a 1: 1 molar ratio to SSPY groups (table1).

Solid insulin-prop-SH was added to the CTB-SSPY solutions and the pyridine thione released in each reaction was removed by size exclusion chromatography.

Analytical size exclusion chromatography of each of the purified conjugates provided evidence that conjugation of CTB-SSPY to insulin-prop-SH had occurred (figure 3). These analyses indicated the majority of the UV-absorbing material eluted from the column between 15 and 27 minutes after the beginning of the run, which is earlier than the elution positions of either CTB5 (28 minutes) or insulin (32 minutes). Thus the bulk of the material is larger in mass than either CTB5 or insulin, suggesting the presence of conjugate. In addition, the peaks corresponding in elution position to CTB-SSPY and insulin-prop-SH were either not visible, or greatly diminished in height compared to those expected for analysis of equivalent amounts of these reactants as was used in the conjugation reactions. This indicates the CTB-SSPY and insulin-prop-SH reactants have been used up in the conjugation reactions.

As the number of pyridyl disulphide groups on the CTB-SSPY used in the reaction increased, the apparent molecular weight of the conjugate increased as evidenced by earlier elution from the size exclusion column.

Example 4 Biochemical characterisation of CTB-insulin coniugates-SDS-PAGE analysis The CTB-insulin conjugates were analysed by SDS-PAGE with Coomassie Blue staining and by western blotting with antibodies against both insulin and CTB (figure 4). In all cases bands larger than CTB5 were visualized, with apparent molecular weights between approximately 60 and 180 kDa. In each case two major bands were observed, which increased in size as the number of 2-pyridyl disulphide groups on the CTB-SSPY used in the reaction increased. Presumably these two major bands correspond to the two major peaks observed by size exclusion chromatography (figure 3). The bands observed by Coomassie Blue staining were also visualised by western blotting using antibodies against both insulin and CTB, indicating that these bands contain both insulin and CTB, and hence represent conjugates between the two molecules.

Example 5 Biochemical characterisation of CTB-insulin coniuqates-amino acid analysis The total protein concentration of the final conjugates was determined by amino acid analysis. This data was also used to calculate the molar ratio of insulin to CTB monomers in the preparations, and was found to be 0.33,1.0 and 1.79 for the conjugates prepared using derivatised CTB containing 0.43,1.43 or 2.82 pyridyl disulphide groups per monomer (table 1). Both results were then used to calculate the concentrations of insulin and CTB in each conjugate (table 1).

Example 6 Biochemical characterisation of CTB-insulin coniuqates-mass spectrometric analysis Mass spectrometric analysis of the preparations was carried out following reverse phase HPLC at acidic pH, which causes dissociation of the CTB pentamer into its monomeric state. The RP-HPLC separated underivatised CTB, from unconjugated insulin-prop-SH, as well as CTB, containing varying amounts of insulin conjugated to it. On-line mass spectrometric analysis of the resolved components allowed confirmation of their identity based on mass. Thus, the reverse phase HPLC of the preparation with a molar insulin : CTB ratio of 0.33 revealed primarily a single peak corresponding in mass to underivatised CTB, and multiple peaks of a mass equivalent to CTB, containing one insulin molecule (figure 5A).

The multiple peaks observed corresponding in mass to CTB monomer derivatised with a single insulin presumably represent conjugation through multiple primary amines on the CTB, as outlined above. In addition there was a small peak corresponding in mass to a dimer of insulin-prop-SH, which presumably forms in solution before the reactive sulphydryl group can cross-link to the pyridyl disulphide on the CTB-SSPY.

LC-MS analysis of the preparation with a molar insulin : CTB ratio of 1.0 indicated relatively less underivatised CTB, and more CTB, containing one insulin molecule than was observed for the preparation with a lower molar insulin : CTB ratio (figure 5B). A few small peaks corresponding in mass to CTB, containing 2-pyridyl disulphide groups that had not reacted with insulin-prop-SH were also visible, as was CTB, that had conjugated two insulin molecules. The peak corresponding in mass to dimers of insulin-prop-SH had also increased relative to the conjugate containing less insulin. These observations were even more striking upon analysis of the conjugate with a molar insulin : CTB, ratio of 1.79 (figure 5C). Relatively little underivatised CTB, or CTB, conjugated to one insulin molecule was observed in this preparation. However, the peaks corresponding in mass to CTB, containing two, three and four insulin molecules were increased relative to the other conjugates, as was expected since the molar ratio of insulin-prop-SH to CTB-SSPY was greater. In addition, the peak of

insulin-prop-SH dimer was also large compared to the conjugates with lower insulin : CTB, ratios. Clearly the concentration of insulin-prop-SH added to the reaction is proportional to the dimerisation of the insulin-prop-SH.

Example 7 Biochemical characterisation of CTB-insulin coniuqates-ELISA In order to examine the effect of altering the molar insulin to CTB, ratio by dilution ELISA, the samples were normalise to a constant concentration of CTB. As shown in table 1, the conjugates with insulin to CTB, ratios of 0.33,1.0 and 1.79 all contained approximately 0.6 mg CTB/ml, allowing direct comparison of each following dilution to the same extent.

Three different ELISAs were used for characterization of the CTB-insulin preparations in this manner.

The sandwich ELISA, in which antibodies against insulin are bound to the ELISA plate and immobilized conjugate is detected with antibodies against CTB, detects the physical presence of conjugate. As expected the response in this ELISA increased as the molar ratio of insulin : CTB, in the conjugate increased from 0.33 to 1.0, with a small additional increase observed at a molar insulin : CTB, ratio of 1.79 (figure 6A). A different pattern was observed with the ELISAs that detect binding to ganglioside Gml. Using antibodies against CTB, which detect both underivatised CTB as well as CTB-insulin conjugates bound to GM1, there was a dose response for all three conjugates (figure 6B). However, the response in this ELISA decreased as the amount of insulin conjugated to CTB increased. ELISA of the unfractionated conjugates using antibodies against insulin, which detects only CTB-insulin conjugates since free insulin is unable to bind GM1, showed a similar pattern (figure 6C).

Thus, the dilution of conjugate at which half maximal response occurred in both assays increased as the amount of insulin in the conjugate decreased, consistent with conjugation of insulin to CTB decreasing the affinity of the CTB for GM1 Each conjugate was also analysed by ELISA following fractionation of similar amounts (based on the concentration of CTB present) by size exclusion chromatography. Analysis of fractions collected from the Superdex-200 column by the sandwich ELISA indicated the presence of conjugate eluting between 17 and 29 minutes (figure 7A). As was observed for the unfractionated conjugate, the response in this ELISA increased as the molar ratio of insulin : CTB, in the conjugate increased from 0.33 to 1.0, with no additional increase observed at a molar insulin : CTB, ratio of 1.79. This trend is consistent with greater insulin antibody immunoreactivity as the amount of insulin in the conjugate increased, up to a point where the assay became saturated. In addition the apparent size of the conjugate, observed

as an increase in immunoreactivity at earlier elution times upon size fractionation on Superdex-200, increased as the amount of insulin increased.

GM, ELISA analysis of the conjugates with insulin : CTB, ratios of 0.33 and 1.0 following size fractionation indicated two peaks of immunoreactivity (figure 6B and 6C). The peaks observed for the conjugate with a molar insulin : CTB, ratio of 1.0 eluted slightly earlier than those observed for the molar ratio of 0.33, again indicative of an increase in mass at the higher insulin conjugation level. For both GM1 ELISAs the immunoreactivity of the fractionated conjugate with a molar insulin : CTB, ratio of 0.33 was higher than that observed for the conjugate with a molar insulin : CTB, ratio of 1. even though the reverse was true for the sandwich ELISA (figure 7A). In addition, the conjugate with a molar insulin : CTB, ratio of 1.79 showed virtually no immunoreactivity in these ELISAs following fractionation, even though a reasonable response had been observed in the sandwich ELISA. This substantiates the observation that conjugation of insulin to CTB lowers the affinity of the CTB for GM1 Example 8 Biochemical characterisation of CTB-insulin conjugates-surface plasmon resonance biomolecular interaction analysis Surface plasmon resonance analysis involving real-time measurement of the specific biomolecular interaction between CTB in the conjugates and ganglioside GM1 was carried out using sensor chips that had been coated with liposomes containing GM1 The sensorgrams obtained upon BlAcore analysis of 20 ug of CTB and 20 ug of CTB-insulin conjugates is shown in figure 8A. The response observed for CTB alone was faster than that seen with conjugates, suggestive of CTB having a lower affinity for GM1 once insulin has been immobilized on its surface. In addition, the response observed upon binding the conjugate to GM1 decreased as the amount of insulin in the conjugate increased, even though the amount of CTB presented to the GM1 was constant. These observations are in agreement with the response seen in the GM,/CTB ELISA outlined above.

Subsequent washing of the chip surface (beginning at 375 seconds) did not remove the responses, indicating the interaction is specific and strong. Addition of an antibody against a non-specific antigen resulted in a small increase in response, which was completely abolished by washing. This indicates that antibodies do not interact non-specifically with the immobilized CTB and CTB-insulin conjugates. However, subsequent washing of the chip surface with antibodies against insulin resulted in a significant response for the flow cells containing conjugate, indicating insulin was immobilized on the surface of the chip and hence conjugated to CTB. In contrast, the antibodies to insulin do not give a specific response when flowed over the cell containing CTB alone.

When the BlAcore analysis was carried out presenting a constant amount of total protein to the immobilised GM1. the response was inversely proportional to the amount of insulin in the conjugate (figure 8B), and the same effect was observed when the amount of conjugate presented was a constant level of insulin (figure 8C). Hence, even though the conjugates with higher insulin : CTB, ratios are being presented to the GM, at increased levels, an increased response is not observed. These observations also indicate that affinity of the conjugate for GM, is suppressed as insulin is conjugated to the CTB.

Example 9 Biochemical characterisation of CTB-insulin coniuqates-induction of immunological response An in vitro assay was developed to assess the ability of the various conjugates to activate T cells. The assay employed a T cell hybridoma specific for insulin that had been isolated from a NOD mouse, and freshly isolated spleen cells served as a source of antigen presenting cells. Upon addition of insulin or conjugate the T cells are activated and release IL-2. In this model the level of IL-2 is a direct indicator of the ability of the conjugate to activate T cells. Figure 9 shows that insulin alone increases IL-2 production by the T cells in a dose-dependent manner. However, the CTB-insulin conjugates cause a similar degree of activation at doses that are several orders of magnitude lower. This indicates that conjugation of insulin to CTB dramatically increases the ability of insulin to activate the T cells, presumably by enhanced presentation of the insulin to the T cells. In addition, the amount of insulin conjugated to CTB affects the activation of the T cells. Several-fold lower amounts of the conjugate with a molar insulin : CTB, ratio of 0.33 were required to achieve a similar IL-2 release to that seen with conjugate containing a molar insulin to CTB, ratio of 1.79 (figure 9).