WO1996012742A1

WO1996012742A1 - Binding agents for treatment of inflammatory, autoimmune or allergic diseases

Info

Publication number: WO1996012742A1
Application number: PCT/EP1995/004110
Authority: WO
Inventors: Jean-Yves Marcel Paul Bonnefoy; Sybille Lecoanet-Henchoz
Original assignee: Glaxo Group Limited
Priority date: 1994-10-25
Filing date: 1995-10-20
Publication date: 1996-05-02
Also published as: DK0788513T3; JPH10507460A; IL115733A; IL115733A0; PT788513E; DE69519997D1; HUT77572A; ES2154741T3; NO971902D0; US6627195B1; FI971756A0; HK1002765A1; EP0788514A1; CA2203363A1; AU698158B2; CN1171119A; CA2203364A1; BR9509498A; JPH09511757A; NZ295158A

Abstract

Binding agents to CD11b, CD11c, CD21, CD23, a 70 to 85 KDa protein expressed on endothelial cells or a 115 KDa protein expressed on endothelial cells, can be useful in the treatment of inflammatory, autoimmune or allergic disease.

Description

BINDING AGENTS FOR TREATMENT OF INFLAMMATORY, AUTOIMMUNE OR ALLERGIC DISEASES

The present invention relates to particular binding agents which can be used in the treatment of inflammatory, autoimmune or allergic diseases.

CD23 (FCεRII) is a type II molecule of the C-lectin family which also includes the lymphocyte homing receptor (MEL-14) and the endothelial leukocyte adhesion molecule-1 (ELAM-1 ). It is a low affinity receptor for IgE. In humans a variety of haematopoietic cell types express CD23 on their surface, including follicular dendritic cells, B cells, T cells and macrophages. CD23 molecules are also found in soluble forms in biological fluids. Soluble CD23 (sCD23) molecules are formed by proteolytic cleavage of transmembrane receptors. CD23 has pleiotropic activities including mediation of cell adhesion, regulation of IgE and histamine release, rescue of B cells from apoptosis and regulation of myeloid cell growth. These functional activities are mediated through the binding to specific ligands of cell-associated CD23, or sCD23, the latter acting in a cytokine-like manner (Conrad, D.H., Annu Rev Immunol 8, 623-645 1990); Delespesse, G., et al., Adv Immunol 49, 149-191 (1991); Bonnefoy, J.Y., et al., Curr Opin Immunol 5, 944-947 (1993)).

Increased expression of CD23 has been observed in a number of inflammatory diseases. CD23 has been identified in synovial biopsies from patients with chronic synovitis, and sCD23 can be measured at concentrations exceeding the normal range in the serum and synovial fluid of patients with rheumatoid arthritis (Bansal, A.S., Oliver, W., Marsh, M.N., Pumphrey, R.S., and Wilson, P.B., Immunology 79,

285-289 (1993); Hellen, E.A., Rowlands, D.C., Hansel, T.T., Kitas, G.D., and Crocker, J.J., Clin Pathol 44, 293-296 (1991); Chomarat, P., Brioloay, J., Banchereau, J., & Miossec, P., Arthritis Rheum 86, 234-242 (1993); Bansal, A., et al., Clin Exp Immunol 89, 452-455 (1992); Rezonzew, R., & Newkirk, M.M., Clin Immunol Immunopathol 71, 156-163 (1994)). In addition, levels of serum sCD23 in rheumatoid arthritis patients are related to disease status and correlate with serum rheumatoid factor (Bansal, AS., et al., Clin Exp Rheumatol 12, 281-285 (1994)). Pro-inflammatory cytokines appear to be particularly important in rheumatoid arthritis, and a central role for TNF-α and IL-1β in the destruction of arthritic joints has been postulated (Brennan, F.M., Chantry, D., Jackson, A, Maini, R., & Feldman, M., Lancet 2, 244-247 (1989); Brennan, F.M., Maini, R.M., & Feldman M., Br J Rheumatol 31, 293-298 (1992)). It has also been postulated that CD23-CD21 interactions may play a role in the control of IgE production (Flores-Romo L. et al., Science 261 1038-1041 (1993); Aubry et al., Nature 358, 505-507 (1992)).

CD11b and CD11c are adhesion molecules that participate in many cell-cell and cell-matrix interactions. CD11 b/CD18 and CD11 c/CD18 (an association of CD11 b and CD18 and of CD11c and CD18 respectively) have been reported to bind several ligands, including CD54, fibrinogen, factor X, LPS, Con A and zymosan (Springer, T.A, Nature 346, 425-434 (1990)). The role of these binding molecules is not however completely understood. CD11b/CD18 and CD11c/CD18 are also known as MAC-1 and p150, 95 respectively. They are members of the β₂ integrin family (sometimes known as Leu-CAM, ie leucocyte cell adhesion molecules). This family also includes LFA-1 amongst its members (also known as CD11a/CD18). EP 0205405 purports to disclose Mabs to lymphocyte cellular receptors for IgE

(FCεR) cross reacting with human immunoglobulin E binding factor (IgE-BF), and derivatives thereof. WO 93/04173 purports to disclose a polypeptide which is capable of binding to one of FCEL (low affinity IgE receptor FCεRII) or FCEH (high affinity receptor FCεRI) but which is substantially incapable of binding to the other of FCEL or FCEH. Treatment of an allergic disorder is alleged with a FCEL or FCEH specific polypeptide (provided the FCEH specific polypeptide is incapable of crosslinking

FCEH and inducing histamine release).

EP 0269728 purports to disclose Mabs to the human lymphocyte IgE receptor. EP 0259585 purports to disclose Mabs recognising a surface receptor for IgE

(FCεR) on human B lymphocytes.

WO 93/02108 purports to disclose primatised antibodies for therapeutic use. The present inventors have surprisingly discovered that binding agents to CD21,

CD11b, CD11c, to a 70-85 KDa protein expressed on endothelial cells, or to a 115 KDa protein expressed on endothelial cells can be of utility in the treatment or prophylaxis of various diseases and in particular in the treatment or prophylaxis of arthritis. Prior to the present invention no data has been presented which would support such a utility, despite the publication of a large number of papers in which

CD21 , CD11 b or CD11 c have been discussed.

According to the present invention, there is provided a binding agent to CD21 , CD11b, CD11c, to an 70-85 KDa protein expressed on endothelial cells, or to a 115 KDa protein expressed on endothelial cells for use in the treatment or prophylaxis of inflammatory, autoimmune or allergic diseases. The binding agent may function by blocking the interaction between the protein and a ligand which binds to it. In vitro assays e.g. radio-immune assays may be used to study such a blocking effect. The binding agent may be in isolated form or as part of a pharmaceutical composition. Desirably it is in sterile form. Generally speaking a binding agent which is specific for CD21, CD11b, CD11c, to a 70-85 KDa protein expressed on endothelial cells, (e.g. to a 76 kDa, a 80 KDa or 85 KDa protein expressed on endothelial cells) or to a 115 KDa protein expressed on endothelial cells, is useful in the treatment/prophylaxis disclosed.

Preferred binding agents include antibodies, fragments thereof or artificial constructs comprising antibodies or fragments thereof or artificial constructs designed to mimic the binding of antibodies or fragments thereof. Such binding agents are discussed by Dougall et al in Tibtech 12, 372-379 (1994).

They include complete antibodies, F(ab')₂ fragments, Fab fragments, Fv fragments, ScFv fragments, other fragments, CDR peptides and mimetics. These can be obtained/ prepared by those skilled in the art. For example, enzyme digestion can be used to obtain F(ab')₂ and Fab fragments (by subjecting an IgG to molecule to pepsin or papain cleavage respectively). References to "antibodies" in the following description should be taken to include all of the possibilities mentioned above. Recombinant antibodies may be used. The antibodies may be humanized, or chimaerised.

A typical preparation of a humanised antibody in which the CDRs are derived from a different species than the framework of the antibody's variable domains is disclosed in EP-A-0239400. The CDRs may be derived from a rat or mouse monoclonal antibody. The framework of the variable domains, and the constant domains, of the altered antibody may be derived from a human antibody. Such a humanised antibody elicits a negligible immune response when administered to a human compared to the immune response mounted by a human against a rat or mouse antibody.

Alternatively, the antibody may be a chimaeric antibody, for instance of the type described in WO 86/01533.

A chimaeric antibody according to WO 86/01533 comprises an antigen binding region and a non-immunoglobulin region. The antigen binding region is an antibody light chain variable domain or heavy chain variable domain. Typically, the chimaeric antibody comprises both light and heavy chain variable domains. The non-immunoglobulin region is fused at its C-terminus to the antigen binding region. The non-immunoglobulin region is typically a non-immunoglobulin protein and may be an enzyme region, a region derived from a protein having known binding specificity, from a protein toxin or indeed from any protein expressed by a gene. The two regions of the chimaeric antibody may be connected via a cleavable linker sequence.

The antibody may be a human IgG such as IgG1 , IgG2, IgG3, IgG4; IgM; IgA; IgE or IgD carrying rat or mouse variable regions (chimaeric) or CDRs (humanised). Primatizing techniques may also be used, such as those disclosed in WO93/02108.

Other preferred binding agents (apart from antibodies or derivatives thereof) are Factor X (i.e. Factor 10), Epstein Barr Virus or a part of Epstein Barr Virus. Where intact viruses are used in medical treatment, they will generally be provided in non-virulent form. This can be achieved using techniques known to those skilled in the art, which include attenuation and mutagenesis. As will be appreciated by those skilled in the art, where specific binding agents are described herein, derivatives of such agents can also be used. The term "derivative" includes variants of the agents described, having one or more amino acid substitutions, deletions or insertions relative to said agents, whilst still having the binding activity described. Preferably these derivatives have substantial amino acid sequence identity with the binding agents specified.

The degree of amino acid sequence identity can be calculated using a program such as "bestfit" (Smith and Waterman, Advances in Applied Mathematics, 482-489 (1981)) to find the best segment of similarity between any two sequences. The alignment is based on maximising the score achieved using a matrix of amino acid similarities, such as that described by Schwarz and Dayhof (1979) Atlas of Protein Sequence and Structure, Dayhof, M.O., Ed pp 353-358.

Preferably the degree of sequence identity is at least 50% and more preferably it is at least 75%. Sequence identities of at least 90% or of at least 95% are most preferred.

It will nevertheless be appreciated by the skilled person that high degrees of sequence identity are not necessarily required since various amino acids may often be substituted for other amino acids which have similar properties without substantially altering or adversely affecting certain properties of a protein. These are sometimes referred to as "conservative" amino acid changes. Thus the amino acids glycine, valine, leucine or isoleucine can often be substituted for one another (amino acids having aliphatic hydroxyl side chains). Other amino acids which can often be substituted for one another include:

phenylalanine, tyrosine and tryptophan (amino acids having aromatic side chains); lysine, arginine and histidine (amino acids having basic side chains); aspartate and glutamate (amino acids having acidic side chains); asparagine and glutamine (amino acids having amide side chains) and cysteine and methionine (amino acids having sulphur containing side chains). Thus the term "derivative" can also include a variant of an amino acid sequence comprising one or more such "conservative" changes relative to said sequence.

The present invention also includes fragments of the binding agents or of the present invention or of derivatives thereof which still have the binding activity described. Preferred fragments are at least ten amino acids long, but they may be longer (e.g. up to 50 or up to 100 amino acids long).

The binding agents of the present invention are believed to be useful in the treatment or prophylaxis of several human diseases including arthritis, lupus erythematosus, Mashimotos thyroiditis, multiple sclerosis.diabetes, uveitis, dermatitis, psoriasis, urticaria, nephrotic syndrome, glomerulonephritis, inflammatory bowel disease, ulcerative colitis, Crohn's disease, Sjogren's syndrome, allergies, asthma, rhinitis, eczema, GVH, COPD, insulitis, bronchitis (particularly chronic bronchitis) or diabetes (particularly Type 1 diabetes). They may also be useful in studying the interactions between CD23 and various ligands e.g. between CD23 and CD21, between CD23 and CD11b, between CD23 and CD11c, between CD23 and the aforesaid 70 to 85 KDa endothelial cell protein

(which may be an 80 or 85 KDa endothelial cell protein) or between CD23 and a 115 KDa endothelial protein (which is believed to be related to the 70 to 85 KDa endothelial protein). One or more of the above interactions are believed to occur, in vivo. Antibodies or other binding agents which are capable of blocking these interactions are particularly preferred since it is believed that they may be especially suitable for reducing or alleviating cytokine mediated inflammatory effects. They may also be useful against B-cell malignancies such as chronic lymphocytic leukaemia, and hairy cell leukaemia.

Binding agents of the present invention are particularly applicable for use in the treatment or prophylaxis of rheumatoid arthritis. Without being bound by theory, the following possible explanations are put forward:- In the rheumatoid arthritis inflamed synovium, macrophages express both CD23 and the β₂ integrins CD11b and CD11c, allowing possible homotypic interactions to take place in this tissue. In addition, diffusion of soluble CD23 molecules through the synovium and their binding to the integrin ligands is also possible. Therefore, CD23-CD11b/CD11c interactions involving a positive activation loop could exist in vivo. If present in rheumatoid arthritis patients, it may explain some of the pathogenic mechanisms of disease exacerbation and chronicity, and would support the hypothesis that once localised to the joints, macrophages themselves can maintain and exacerbate inflammation via a pathway involving CD23 molecules, β₂-integrins CD11b and CD11c, as well as pro-inflammatory cytokines TNF-α, IL-1β and IL-6.

The present inventors have found that CD23 binding to CD11b and CD11c is blocked by Factor X since Factor X binds to CD11b and CD11c. Thus the present invention includes the use of Factor X or of a fragment thereof to block CD23 binding to CD11 b and/or CD11 c.

An alternative mechanism of action of anti CD23 therapy could involve the blocking of an IgE immune response. In previously published work, it has been shown that in vivo treatment of rats with anti-CD23 antibody resulted in antigen-specific inhibition of IgE production, probably by blocking the CD23-CD21 interactions necessary for complete differentiation of IgE-committed B cells (Flores-Romo et al., Science 261, 1038- 1041 (1993)).

The present invention also includes binding agents to CD21 which block such a response (e.g. the Epstein Barr virus or a part thereof). Structurally, the CD21 protein is composed of an extracellular domain of 15 (Moore et al, Molecular cloning of the cDNA encoding the Epstein Barr Virus C3d receptor (complement receptor type 2) of human B lymphocyte, Proc NatlAcad Sci USA 84: 9194 (1987)) or 16 (Weis et al, Structure of the human B lymphocyte receptor for C3d and the Epstein Barr Virus and relatedness to other members of the family of C3/C4 binding proteins, J Exp Med 167: 1047 (1988)) repetitive units of 60 to 75 amino acids, named short consensus repeats (SCRs), followed by a transmembrane domain (24 amino acids) and an intracytoplasmic region of 34 amino acids. Using CD21 mutants bearing deletions of extracytoplasmic SCRs (Carel et al, Structural requirements for C3d,g/Epstein Barr Virus receptor (CR2/CD21) ligand binding, intemalization, and viral infection J Biol Chem 265:

12293 (1990)), the present inventors have recently found that CD23 binds to SCRs 5-8 and 1-2 on CD21. The binding of CD23 to SCRs 5-8 is a lectin-like interaction, involving carbohydrates on Asn 295 and 370. In contrast, CD23 binding to SCRs 1-2 is a protein-protein interaction (Aubry et al, CD23 interacts with a new functional extracytoplasmic domain involving N-linked oligosaccharides on CD21,

J Immunol 152: 5806 (1994)). The present inventors have now tested the effect of the other ligands of CD21 (EBV, C3d,g and IFN-α) on CD23 binding to CD21 and on the regulation of Ig production in the presence of IL-4. Only EBV particles and an EBV-derived peptide were able to inhibit CD23 binding to CD21. Moreover, the EBV-peptide selectively decreased IgE and lgG4 production and increased IgM production. These data indicate that CD23 binding to the EBV-binding site on CD21 selectively regulates human Ig production in the presence of IL-4.

The present invention therefore includes within its scope the use of Epstein Ban-Virus or of a part thereof to block the binding of CD23 to CD21. A preferred part of the Epstein Barr Virus is the gp350/gp220 glycoprotein or a fragment thereof. Alternatively, an unglycosylated form of this glycoprotein or of a fragment thereof may be used.

Again without being bound by theory, it is believed that the present invention allows effective treatments to be achieved by suppressing the de novo synthesis of pro-inflammatory cytokines. This contrasts with previous uses of antibodies simply to directly neutralise the cytokine molecules already present in inflamed tissues.

It should also be noted that there are speculative publications in the art listing large numbers of antibodies as well as large numbers of possible diseases which the antibodies are said to be useful in treating, but not providing any sound evidence or data for most of the possible combinations. One such publication is WO93/02108 which is primarily directed to the production of particular chimaeric antibodies. The present invention is clearly distinguished from such publications by providing binding agents to particular molecules which are clearly indicated to be of utility in the treatment or prophylaxis of certain diseases in view of the data and explanations provided herein. Binding agents of this invention are also of particular use in the treatment or prophylaxis of allergic diseases, including non-IgE mediated diseases. They may be used in the treatment and propylaxis of ulcerative colitis. They may also be used in the treatment and prophylaxis of Crohn's disease.

The binding agents of the present invention may be used alone or in combination with immunosuppressive agents such as steroids, cyclosporin, or antibodies such as an anti-lymphocyte antibody or more preferably with a tolerance-inducing, anti-autoimmune or anti-inflammatory agent such as a CD4+T cell inhibiting agent e.g. an anti-CD4 antibody (preferably a blocking or non-depleting antibody), an anti- CD8 antibody, a TNF antagonist e.g. an anti-TNF antibody or TNF inhibitor e.g. soluble TNF receptor, or agents such as NSAIDs.

The binding agent will usually be supplied as part of a sterile, pharmaceutically acceptable composition. This pharmaceutical composition may be in any suitable form, depending upon the desired method of administering it to a patient. It may be provided in unit dosage form and may be provided as part of a kit. Such a kit would normally (although not necessarily) include instructions for use. Binding agent administrations are generally given parenterally, for example intravenously, intramuscularly or sub-cutaneously. The binding agents are generally given by injection or by infusion. For this purpose a binding agent is formulated in a pharmaceutical composition containing a pharmaceutically acceptable carrier or diluent. Any appropriate carrier or diluent may be used, for example isotonic saline solution. Stabilizers may be added such as a metal chelator to avoid copper-induced cleavage. A suitable chelator would be EDTA, DTPA or sodium citrate. They may be given orally or nasally by means of a spray, especially for treatment of respiratory disorders.

They may be formulated as creams or ointments, especially for use in treating skin disorders.

They may be formulated as drops, or the like, for administration to the eye, for use in treating disorders such as vernal conjunctivitis. For injectable solutions, excipients which may be used include, for example, water, alcohols, polyols, glycerine, and vegetable oils.

The pharmaceutical compositions may contain preserving agents, solubilising agents, stabilising agents, wetting agents, emulsifiers, sweeteners, colourants, odourants, salts (substances of the present invention may themselves be provided in the form of a pharmaceutically acceptable salt), buffers, coating agents or antioxidants. They may also contain other therapeutically active agents.

Suitable dosages of the substance of the present invention will vary, depending upon factors such as the disease or disorder to be treated, the route of administration and the age and weight of the individual to be treated. Without being bound by any particular dosages, it is believed that for instance for parenteral administration, a daily dosage of from 0.01 to 50 mg/kg of a binding agent of the present invention (usually present as part of a pharmaceutical composition as indicated above) may be suitable for treating a typical adult. More suitably the dose might be 0.05 to 10 mg/kg, such as 0.1 to 2 mg/kg. This dosage may be repeated as often as appropriate. Typically administration may be 1 to 7 times a weeks. If side effects develop the amount and/or frequency of the dosage can be reduced. A typical unit dose for incorporation into a pharmaceutical composition would thus be at least 1 mg of binding agent, suitably 1 to 1000 mg.

The present invention includes within its scope an assay for determining whether or not a particular agent which binds to CD21 , CD11 b, CD11 c or to a 70-85 or 115 KDa protein expressed on endothelial cells may be useful in the treatment of an inflammatory, autoimmune or allergic disease comprising: determining whether or not the agent is capable of blocking the interaction between CD23 and CD11b, or the interaction between CD23 and CD11c, or the interaction between CD23 and CD21, or the interaction between CD23 and a 70-85 KDa or 115 KDa protein expressed on endothelial cells.

This assay can be used for screening compounds or molecules by using cell lines expressing the appropriate molecules. Preferably CD11b is used in these assays as CD11b/CD18 and CD11c is used as CD11c/CD18. CD11b/CD18 and CD11c/CD18 can be co-expressed on cell surface.

Any appropriate assay technique can be used, e.g. protein-non protein assays (e.g. assaying the interaction of proteins with chemicals or sugars), protein-protein assays or protein-cell assays.

The present invention will now be described by way of example only with reference to the accompanying drawings; wherein: FIGURE 1a illustrates CD23-liposomes binding to CD14 positive blood mononuclear cells;

FIGURE 1b illustrates various CD23 affinity purified proteins on SDS-PAGE gels;

FIGURE 2 illustrates the percentage inhibition of CD23-liposome binding to activated blood monocytes obtained using certain monoclonal antibodies; FIGURE 3 illustrates the binding of CD23 liposomes to various transfected cells;

FIGURE 4 illustrates the effect of various substances on CD23-CD11b and CD23-CD11c interaction;

FIGURE 5 illustrates the effects on nitrite production and oxidative burst in monocytes caused by CD23 binding to CD11b and CD11c; and

FIGURE 6 illustrates that the binding of recombinant CD23 to CD11b and CD11 c specifically increases cytokine production by monocytes.

FIGURE 7a illustrates the inhibition of CD23-liposome binding to RPMI 8226 cells by various CD21 ligands. FIGURE 7b illustrates the inhibition of IL-4 induced IgE and IgG4 production by an EBV peptide binding to CD21.

FIGURE 7c illustrates the modulation of immunoglobulin production by an EBV peptide binding to CD21. FIGURE 7d illustrates the absence of inhibition of IgE production with a C3 peptide binding to CD21.

FIGURE 8 illustrates the inhibition of the binding of CD23 liposomes to certain endothelial cells due to the presence of an anti-CD23 MAb.

EXAMPLES

(In some of the following examples the terms "ip", "id" and "n" are used. These mean "intraperitoneal", "intradermal" and "number of animals" respectively.)

EXAMPLES 1-6 Interaction between CD23 and CD11b and between CD23 and CD11c

Examples 1 to 6 and the accompanying Figures (see later) illustrate the interaction of CD23 with CD11b and/or CD11c.

In these Examples, full-length recombinant CD23 incorporated into fluorescent liposomes was shown to bind to COS cells fransfected with cDNA encoding either

CD11 b/CD18 or CD11c/CD18 but not with transfectants expressing CD11 a/CD18. The interaction between CD23-liposonr.es and CD11b/CD18 or CD11c/CD18-transfected COS cells was specifically inhibited by anti-CD11b or anti-CD11c, respectively, and by anti-CD23 monoclonal antibodies. The functional significance of this ligand pairing was demonstrated by triggering CD11b and CD11c on monocytes with either recombinant CD23 or anti-CD11b and anti-CD11c monoclonal antibodies to cause a marked increase in nitrite (NO₂-), oxidative products (H₂O₂) and proinflammatory cytokines (IL-1β, IL-6 and TNFα). These CD23-mediated activities were decreased by Fab fragments of monoclonal antibodies to CD11b, CD11c and CD23. These results demonstrate that the surface adhesion molecules CD11b and CD11c are receptors for CD23 and that this novel ligand pairing regulates important activities of monocytes. The following discussion explains briefly the experimental design and the rationale behind Examples 1 to 6 (which follow):-

Total blood mononuclear cells were incubated with recombinant full-length CD23 incorporated into fluorescent liposomes and analysed by flow cytometry (Pochon, S. et al. J. Exp. Med. 176, 389-398 (1992)). A fraction bound CD23-liposomes

(Example 1, Fig. 1a) which was then shown by double staining to consist of CD14-positive cells (i.e. monocytes). To confirm that monocytes were able to bind CD23-liposomes, blood mononuclear cells were FACS-sorted into CD14-positive and CD14-negative populations (Example 1, Fig. 1a). CD23-liposomes were shown to bind only to the CD14-positive population (Example 1, Fig. 1a). Since monocytes were found to express neither membrane IgE nor CD21 (not shown), the known ligands for CD23, it was investigated whether monocytes express a different receptor for CD23. Monocytes were lysed and cell extracts purified over an affinity column coupled with recombinant soluble CD23. SDS-PAGE and silver staining analysis of the eluted material revealed bands of around 80 and 160 kDa MW

(Example 1 , Fig. 1 b). Antibodies identifying antigens within this range of MW and reported to be expressed on monocytes were tested by FACS for their capacity to inhibit CD23-liposome binding to monocytes (Example 2, Fig. 2). Anti-CD11b and anti-CD11c monoclonal antibodies both inhibited CD23-liposome binding to monocytes, with varying degrees of potency (Example 2, Fig. 2). Anti-CD13, anti- CD49d, anti-CD21 (not expressed on monocytes) and anti-CD11a (the third member of the β2 integrin family of adhesion molecules) had no significant effect (Example 2, Fig. 2). Antibodies against MHC Class I, Class II, CD14 and CD45, all of which highly expressed on monocytes, were also tested for their effect on CD23- liposome binding. None however had any effect (not shown). Anti-CD18 monoclonal antibody gave a partial inhibition of CD23 binding. This could be due either to steric hindrance or to the induction of a conformational change in the CD11b and CD11c molecules upon anti-CD18 Mab binding. The monocytederived proteins eluted from the CD23-affinity column were immunoreactive with anti-CD11c (Example 1, Fig. 1b) and anti-CD11c/CD18 antibodies (not shown).

To confirm that the α chain of CD11b/CD18 and CD11c/CD11b were receptors for CD23, full-length cDNAs encoding CD11b and CD11c were transiently ccfransfected with CD18 cDNA into COS cells. Transfectants expressing

CD11b/CD18 and CD11c/CD18 were both shown to bind CD23-liposomes, in contrast to transfectants expressing CD11a/CD18 (Example 3, Fig. 3). This might be explained by the higher degree of homology between CD11 b and CD11 c when compared to their homology to CD11a. The specificity of the interaction was demonstrated by inhibiting CD23-liposome binding using anti-CD11b, anti-CD11c and anti-CD23 monoclonal antibodies. The same results were obtained using BHK cells expressing CD11b/CD18 and CD11c/CD18 (not shown). As further proof of the specificity of the CD23 interaction, activated blood monocytes from a Leukocyte Adhesion Deficiency patient, lacking β2 integrin expression due a mutation in the gene encoding the β subunit were unable to bind CD23-liposomes

(not shown). Together, these data demonstrate that CD23 interacts with CD11b and CD11c on normal human monocytes and on transfectants.

CD11b and CD11c are adhesion molecules that participate in many cell-cell and cell-matrix interactions. The examples show that CD11 b/CD18 and CD11 c/CD18 may exhibit an additional adhesive function by virtue of their ability to bind CD23. CD23 seems to identify an epitope close or identical to factor X as observed by the capacity of factor X to inhibit in a dose dependent manner CD23-liposome binding (Example 4, Fig. 4) without affecting surface expression of CD11b or CD11c on monocytes (not shown). None of the other ligands tested had any effect. CD23 may be acting as a C-type lectin in its interaction with CD11b and CD11c. EDTA decreases CD23 binding to monocytes (Example 4, Fig. 4) by chelation of Ca²⁺ which is necessary to CD23 binding and/or by chelation of the divalent cations which are necessary for the binding of ligands to CD11b and CD11c (Altieri, D.C.

J. Immunol. 147, 1891-1898 (1991)). CD23-CD11b/CD11c interaction seems to involve sugars, but not sialic acid, as observed by the capacity of tunicamycin, but not neuraminidase, to decrease CD23 binding to monocytes. CD23 bears extracellularly a triplet of amino acids (Asp, Gly, Arg) (Kikutani, H. et al. Cell 47, 867-885 (1986)), which in the reverse orientation is a common recognition site for the integrin receptors. Therefore, the effect of a polyclonal antibody directed against this tripeptide was tested for its capacity to inhibit CD23 binding to monocytes. No inhibition was observed, confirming the absence of inhibition obtained with fibrinogen (Example 4. Fig. 4). IgE which is binding in the lectin domain of CD23, partially inhibits CD23 binding to monocytes (Example 4, Fig. 4).

Those results indicate that CD23 would seem to be acting as a C-type lectin recognising partly sugar and protein structures, reminiscent of what has been observed for CD23 interaction with CD21 (Aubry, J-P. et al. J. Immunol. 152, 5806-5813 (1994)).

To evaluate the functional significance of the interaction of CD23 with CD11b or CD11c, we investigated whether CD23-CD11b/CD11c interaction could trigger monocytes to release proinflammatory mediators such as nitric oxide, H₂O₂ and cytokines. Triggering of adherence-activated normal monocytes using recombinant soluble CD23, anti-CD11b or anti-CD11c antibodies increased the generation of NO₂- indicating the activation of the NO pathway (Moncada, S., Palmer, R.M.J. & Higgs, E.A. Pharmacol. Rev. 43, 109-144 (1991)). The effect of CD23 on nitrite production was inhibited by Fab fragments of anti-CD23 monoclonal antibodies and by nitroarginine, a specific inhibitor of the NO synthase pathway (Example 5, Fig. 5a). The oxidative burst was also shown to be regulated through CD11b and CD11c since recombinant soluble CD23, anti-CD11b and anti-CD11c monoclonal antibodies all caused oxidation of hydroethidine to ethidium bromide in monocytes (Example 5, Fig. 5b). This confirms and extends the finding that anti-CD11b monoclonal antibodies induce an oxidative burst in monocytes

(Trezzini, C, Schüepp, B., Maly, F.E. & Jungi, T.W. Brit. J. Haematol. 77, 16-24 (1991)). CD23 binding to CD11 b and CD11 c was associated with an early specific Ca²⁺ flux in blood monocytes (not shown). Since activated macrophages are an important source of proinflammatory cytokines, we evaluated the effect of recombinant soluble CD23 and of anti-CD11b and anti-CD11c monoclonal antibodies on the production of such cytokines by monocytes. Recombinant soluble CD23, anti-CD11b and anti-CD11c monoclonal antibodies were potent stimulators of IL-1β, IL6 and TNFα. Again, the specificity of this induction was demonstrated by using Fab fragments of anti-CD11b, anti- CD11c and anti-CD23 monoclonal antibodies (Example 6, Fig. 6). Interestingly, IL-1 and TNFα were potent inducers of CD23-liposome binding to monocytes (not shown), suggesting a potential cytokine autocrine loop through CD11b and CD11c stimulation and regulation.

EXAMPLE 1

a) CD23-liposomes bind to CD14-positive blood mononuclear cells (See Fig.

1a).

Blood mononuclear cells were stained with anti-CD14 monoclonal antibody (Becton Dickinson, Erembodegem, Belgium) followed by sheep FITC-conjugated

F(ab')₂ antibodies to mouse IgG and IgM (Bioart, Meudon, France), both diluted in PBS, 0.5% BSA and 0.05% sodium azide prior to FACS-sorting (FACStar Plus, Becton Dickinson) into CD14-positive and CD14-negative cell populations. Separated cells were then stained with CD23-liposonr.es or control (glycophorin A)- liposomes diluted in 0.5% BSA, 0.1% sodium azide, 2 mM CaCl₂, 140 mM NaCl, 20 mM Hepes, pH 7 and incubated for 2 h at 4°C (Pochon, S. et al., J. Exp. Med. 176 389-398 (1992)). After washes, cells (5,000 events/condition) were analysed by FACS. b) Apparent molecular weight of CD23-affinity purified blood monocyte proteins and immunoreactivity with an anti-CD11c monoclonal antibody (See Fig.

1b).

Lysates of blood monocytes were affinity purified on a CD23-column, eluted proteins separated on SDS-PAGE gels and transferred onto nitrocellulose. Mr markers are shown on the left. The gel was silver stained (left lane). Filters were incubated with either an isotype matched antibody (middle lane) or with an anti-CD11c monoclonal antibody (BU-15, right lane), then with horseradish peroxidaseconjugated goat anti-mouse antibody (Kpl; Gaithersburg, Massachusetts).

EXAMPLE 2

Anti-CD11b and anti-CD11c monoclonal antibodies decrease CD23-liposome binding to activated blood monocytes (See Fig. 2).

Monocytes were enriched from mononuclear cells by Ficoll and overnight adherence to plastic in RPM1 1640 (Seromed, Berlin, Germany) supplemented with

2 mM glutamine and 10% heat-inactivated FCS (Flow Laboratories, Irvine, Scotland). Activated monocytes were then incubated with CD23-liposomes in the presence of different monoclonal antibodies (αCD) or isotype-matched controls (CTRL) (Becton Dickinson), all tested at 10 μg/ml. Anti-CD11a monoclonal antibodies 25.3 and B-B15 were obtained from Immunotech (Luminy, France) and

Serotec (Oxford, UK), respectively. Anti-CD11b monoclonal antibody 44 was from Serotec, mon.gran 1 was from Janssen (Beerse, Belgium), Leu-15 was from Becton Dickinson (Erembodegem, Belgium) and (Bear-1) was from Sera-Lab Ltd (Sussex, GB). Anti-CD11c monoclonal antibody 3.9 was from Serotec, SL9 was from Sera-Lab and BU-15 was from The Binding Site (Birmingham, UK). Anti- CD13 (SJ1D1), anti-CD18 (BL5), anti-CD23 (mAb25) and anti-CD49d (HP2.1) monoclonal antibodies were from Immunotech. Anti-CD21 monoclonal antibody BL13 was from Immunotech, OKB7 from Ortho and BU-33 was obtained from Dr. MacLennan (Birmingham University, UK), HB-5 from ATCC, OKB7 from Ortho

Diagnostics System Inc (Raritan, NJ). Anti-CD14, anti-CD3, anti-CD16 and anti-CD20 monoclonal antibodies were from Becton-Dickinson. Cells were analysed by FACS and mean fluorescence intensity (MFI) measured. Data of a representative experiment are presented. MFI of cells stained with control-liposomes was 6.5 and with CD23-liposomes was 84.5. Percentage inhibition using arithmetic linear MFI values is calculated according to the following formula:

EXAMPLE 3

CD23-liposomes bind to α chains of CD11b/CD18 and CD11c/CD18 on recombinant transfectants (See Fig. 3). cDNAs coding for CD11a (Corbi, AL., Miller, L.J., O'Connor, K., Larson, R.S. & Springer, T.A EMBO J. 6, 4023-4028

(1987)) was recloned in pCDNAI (Invitrogen, San Diego, CA). cDNA for CD11b (Corbi, A.L., Kishimoto, T.K, Miller, L.J. & Springer, T.A. J. Biol. Chem. 263, 12403-12411 (1988)) and CD18 (Kishimoto, T.K, O'Connor, K, Lee, A, Roberts, T.M. & Springer, T.A. Cell 48, 681-690 (1987)) were recloned in pCDM8 (Seed, B., Nature 329 840-842 (1987)). 20 μg aliquots of DNA were fransfected in COS-7 cells (ATCC) by electroporation (260 V, 960 μFD) using a Gene Pulser device (Bio-Rad, Richmond, CA) and 0.4 cm cuvettes in 20 mM Hepes pH 7.4, 150 mM NaCl. Co-transfections of CD11a, b or c with CD18 were performed in order to get expression of the β2 integrins at the cell surface. Controls were done with single chain transfections. 48 h after transfection, COS cells were stained with anti-CD11a, anti-CD11b and anti-CD11c monoclonal antibodies or isotype-matched monoclonal antibodies (control) followed by FITC-labelled sheep anti-mouse antibody. Between 10 to 15% of the cells were shown to express CD11a, b, c or CD 18 by staining with the respective monoclonal antibodies. Prior to staining with

CD23-liposomes, CD18-positive fransfected COS cells were then FACS-sorted in order to increase the percentage of cells expressing β2 integrins. CD11a/CD18, CD11b/CD18 and CD11c/CD18 transfectants were then incubated with CD23-liposomes (trace 2) or control (glycophorin A)-liposomes (trace 1). The specificity of CD23 interaction with CD11b and CD11c was demonstrated by inhibition of

CD23-liposome binding to CD11b/CD18 and CD11c/CD18 transfectants using anti-CD11b (trace 4), anti-CD23 (trace 5) and anti-CD11c (trace 6) monoclonal antibodies, respectively. No binding of CD23-liposomes was observed on CD11a/CD18 transfectants and no effect of anti-CD11a monoclonal antibody was found (trace 3).

EXAMPLE 4

Structural characterisation of CD23-CD11b. CD11c interaction (See Fig. 4).

(a) Involvement of sugars and divalent cations.

Purified activated blood monocytes were treated or not with tunicamycin (10 μg/ml) for 48 h or with neuraminidase (0.1 U/ml; both from Boehringer Mannheim, Mannhein, Germany) for 45 min. Cells were then incubated with CD23-liposorr.es or control-liposomes in the absence or presence of EDTA (5 mM; top left panel), Ca²⁺ or Mn²⁺ (1 to 10 mM; top right panel).

(b) Factor X does inhibit CD23 binding to monocytes. Purified activated blood monocytes were incubated with CD23-liposomes in absence or presence of factor X (0.1 to 10 U/ml; Sigma) (bottom left panel), fibrinogen (50 μg/ml; Sigma), purified recombinant ICAM-1 (produced in our laboratory), LPS (1 μg/ml; Sigma), human serum opsonised-zymosan (1 mg/ml; Sigma), IgE (50 μg/ml; The Binding Site, Birmingham) or polyclonal antibody to RGD peptide (1/500, ATCC) (bottom right panel). Cells were analysed by FACS and MFI measured. Percentage inhibition was calculated as for Example 2.

EXAMPLE 5

Recombinant CD23 by binding to CD11b and CD11c specifically increases a. the nitrite product and b. the oxidative burst by monocytes.

Monocytes were incubated a, for 4 days at 37°C or b, overnight in the absence or presence of recombinant soluble CD23 (Graber P. et al., J. Immunol. Methods 149

215-226 (1992)) (50 ng/ml), anti-CD11a (clone 25.3), anti-CD11b (clone 44), anti-CD11c (clone BU-15) monoclonal antibodies (all at 10 μg/ml).

To assess the amount of NO produced (which is shown in Fig. 5a), the culture supernatants were assayed for the stable end products of NO, NO₂- and NO₃- according to Green et al., Annu. Rev. Immunol. 2 199-218 (1984). The specificity of CD23-mediated increase of NO₂- production was demonstrated by inhibition of NO₂- production by Fab fragments of anti-CD23 monoclonal antibodies (mab25) (tested at 10 μg/ml) and by inhibition with nitroarginine (N-Arg at 1 mM) (Sigma).

Activated monocytes were incubated with hydroethidine (Molecular probes, Eugene, OR) (0.3 μg/ml) for 30 min at 37°C (Rothe G. et al., J. Leukoc. Biol. 47 440-448 (1990)) and analysed by FACS. Percentage increase in red fluorescence of stimulated monocytes is shown in comparison to untreated monocytes (See Fig. 5b). Monocytes which had undergone an oxidative burst shown an increase of red fluorescence signals compared to untreated monocytes reflecting oxidation of hydroethidine to ethidium bromide (Lacal P.M. et al., Biochem. J. 268 707-712 (1990)). MFI values of monocytes alone were 159+/-10. Mean+/-SD values of 6 experiments are presented. Con A, which is known to induce a respiratory burst in monocytes, was used as a positive control. The specificity of the CD23 interaction with CD11b and CD11c was demonstrated by inhibition of CD23-mediated increase of H₂O₂ production by Fab fragments of anti-CD11b (clone 44), anti-CD11c (clone BU-15) and anti-CD23 (mAb25) monoclonal antibodies (tested at 10 μg/ml).

EXAMPLE 6

Binding of recombinant CD23 to CD11b and CD11c specifically increases cytokine production by monocytes (See Fig. 6).

Monocytes were incubated overnight at 37°C in the absence or presence of recombinant soluble CD23 (Graber P. et al., J. Immunol. Methods 149 215-226 (1992)) (50 ng/ml), anti-CD11a (clone 25.3), anti-CD11b (clone 44), anti-CD11c (clone BU-15), anti-CD23 (mAb 25 - this antibody is available from Immunotech. It is discussed in published European Patent Application EP-A-0269728) monoclonal antibodies, Con A (Sigma) (all at 10 μg/ml), LPS (1 ng/ml) (Sigma) or

PMA (5 ng/ml) (Calbiochem, La Jolla, CA). Cytokines were measured in the culture supernatant by specific ELISA The ELISA's limit of sensitivity is 0.05 ng/ml for IL-1β (Ferrua et al., J. Immunol. Methods 114 41-48 (1988)) 0.01 ng/ml for TNFα (Medgenix, Biotechnie, Rungis, F) and <0.01 ng/ml for IL-6 (Manie et al., Eυr. Cytokine Netw. 4 51-56 (1993)). The specificity of CD23 interaction with

CD11b and CD11c was demonstrated by inhibition of CD23-mediated increase of cytokine production by Fab fragments of anti-CD11b (clone 4), anti-CD11c (clone BU-15) and anti-CD23 (mAb25) monoclonal antibodies (tested at 10 μg/ml). Mean+/-SD values of 4 experiments are presented.

EXAMPLE 7

The following materials and methods were used in this example:

Cell Lines Tonsil or blood mononuclear cells were separated into T and B cell subpopulations by rosetting with sheep red blood cells.

The B cell line RPMI 8226 was obtained from the American Type Culture Collection (ATCC, Rockville, MD) and cultured in RPM1 1640 complete medium.

Peptides and CD21 ligands

Two peptides, from gp350 of EBV and C3, known to bind to CD21 (Servis et al, C3 synthetic peptides support growth of human CR2-positive lymphoblastoid B cells, J Immunol 142: 2207 (1989)) were synthesized. PepEBV (TGEDPGFFNVEIC-NH2) was produced on an ABI 431 A synthesizer using FastMoc chemistry and PepC3 (GKQLYNVEATSYAC-NH2) was obtained from Neosystem (Strasbourg, France). Aggregated C3d,g was prepared as described previously (Carel et al (1990) supra). Sucrose gradient purified EBV was obtained from Advanced Biotechnologies (Columbia, ML) and IFN-α was obtained from Sigma (St Louis,

MO).

Liposome preparation

CD23-liposorres were made as previously described (Pochon et al, Demonstration of a second ligand for the low affinity receptor for immunoglobulin E (CD23) using recombinant CD23 reconstituted into fluorescent liposomes, J Exp Med 176: 389 (1992)) using 10 μmoles of the synthetic phospholipids POPC (Avanti Polarlipids Inc. Alabaster, AL) mixed with 50 nmoles of fluorescent dye DiO18 (Molecular Probes, Eugene, OR) and then dialyzed against HEPES buffer together with purified recombinant CD23 or with glycophorin A (0.2 μmoles each) referred to as control protein.

Flow cytometry

Liposome binding assay Cells (10⁵) were resuspended in 50 μl of the liposome suspension, diluted 10 times in 0.5% BSA, 0.1% NaN₃, 2 mM CaCl₂, 140 mM NaCl, 20 mM Hepes, pH 7.0 and incubated for 2 h at 4°C. Cells were washed twice before analysis on a FACStar plus (Becton Dickinson,

Erembodeggen, Belgium).

Competition of CD23-liposomes with EBV, EBV peptide, IFN-α, C3 peptide and C3d,g

RPMI 8226 cells were co-incubated with glycophorin-liposomes or CD23-liposomes and EBV (1×10⁵ to 1×10⁹ particles/ml), EBV peptide and C3 peptide (50 nM to 50 μM), aggregated C3d,g (4 ng/ml to 1 μg/ml) and IFN-α (1000 U/ml) for 2 h at 4°C. Cells were analyzed as described above. IL-4-induced Ig production assays

Cells were incubated at 10⁶/ml for 14 days in Iscove's medium enriched with transferrin, bovine insulin, oleic acid, linoleic acid, palmitic acid, BSA (all from Sigma) and 10% FCS (Flow Laboratories, Irvine, Scotland) as described by Claassen et al (A cell culture system that enhances mononuclear cells IgE synthesis induced by recombinant interleukin-4, J Immunol Methods 126: 213

(1990)). Assays were performed using total PBMNC with IL-4 alone (200 U/ml) or IL-4 plus anti-CD40 (1 μg/ml) (Serotec Ltd. Oxford, UK), or using purified tonsillar B cells with IL-4 and anti-CD40. IgE, G, A and M were quantified by specific ELISA as previously described (Bonnefoy et al, Inhibition of human interleukin-4-induced IgE synthesis by a subset of anti-CD23/Fc epsilon RII monoclonal antibodies, Eur J Immunol 20: 139 (1990)). lgG4 was measured by ELISA as follows. A mouse anti human lgG4 antibody (Southern Biotechnology, Birmingham) diluted at 10 μg/ml in bicarbonate buffer, pH 9.6 was coated overnight in 96 well plates (100 μl/well). Saturation was then performed with PBS plus 1% BSA (200 μl/well) for 2 h at RT. Samples to be tested were diluted in PBS plus 0.5% BSA and 0.1% Tween (100 μl/well) and incubated overnight at 4°C. After washes with PBS plus Tween, a peroxidate-labelled sheep anti human lgG4 antibody (Vital products, St Louis, MO) diluted 1/5000 in PBS/BSA plus Tween was added for 1 h at RT. After washes with PBS plus Tween, o-phenylene diamine (Sigma) is added and the colorimetric reaction was stopped with 2M H₂SO₄. Plates were finally read at 492 nm.

The results obtained are discussed below:

Human CD21 has been previously described to be a receptor for the C3d,g and iC3b proteins of the complement system (Weis et al, Identification of a 145,000 Mr membrane protein as the C3d receptor (CR2) of human B lymphocytes Proc NatI Acad Sci USA 81: 881 (1984)), for the gp350/220 envelope glycoprotein of EBV (Nemerow et al, Identification of gp 350 as the viral glycoprotein mediating attachment of Epstein Barr Virus (EBV) to the EBV/C3d receptor of B cells : sequence homology of gp350 and C3 complement fragment C3d, J Virol 61: 1416 (1987); Tanner et al, Epstein Barr Virus gp 350/220 binding to the B lymphocyte C3d receptor mediates adsorption, capping and endocytosis, Cell 50: 203 (1987)) and for IFN-α (Delcayre et al, Epstein Barr Virus/complement C3d receptor is an interferon α receptor, EMBO J 10: 919 (1991 )). We have therefore tested all these CD21 ligands for their ability to inhibit CD23-liposome binding to the CD21-expressing cells, RPMI 8226 cells (Pochon et al (1992) supra). Intact particles of EBV were able to inhibit CD23 binding to CD21 in a dose dependent manner. This is shown in Figure 7a, which illustrates the inhibition of CD23-liposome binding to RPMI 8226 cells by some CD21 ligands. [RPMI 8226 cells were coincubated for 2 h with CD23-liposomes or glycophorin-liposomes and various concentrations of EBV (particles/ml), PepEBV and PepC3 (μM). Percentage of inhibition is calculated as follows:

The MFI of glycophorin-liposomes was subtracted from the MFI of CD23-liposomes. Inset= FACS profiles of RPMI 8226 cells stained with glycophorinliposomes (traces 1) or CD23-liposon.es (traces 3) alone or in presence of EBV particles (top), PepEBV (medium) and PepC3 (bottom) (traces 2). Results are taken from a representative experiment.]

Of the other CD21 ligands tested for inhibition of CD23 liposome binding, only EBV decreased the binding of CD23. A complete inhibition of CD23 binding was observed with EBV intact particles, although EBV is reported to bind to SCR 2 of CD21 and not to SCRs 5-8 where CD23 binds to sugars in this latter region (Aubry et al (1994) supra). This complete inhibition of CD23 binding could be due to the size of the virus particles or to the fact that EBV may modify the conformation of the CD21 molecule. In order to exclude that inhibition was due to steric hindrance by the virus particles, we then tested the effect of a peptide of gp350/220, known to bind to CD21 (Servis et al (1989) supra). This EBV peptide was able to inhibit CD23 binding to CD21 in a dose dependent manner, with a maximum of 55% inhibition (Fig 7a). These experiments suggest that the EBV peptide binding is close to the CD23 binding site and partially blocks CD23 binding. These data confirm our previous finding that CD23 does bind to CD21 (Aubry et al, CD21 is a ligand for CD23 and regulates IgE production, Nature 358: 505 (1992)) and extend them by showing that SCR2 is probably a region interacting with CD23. In contrast, a C3 peptide corresponding to the CD21 binding site on C3d (Servis et al (1989) supra) (Fig 7a), aggregated C3d,g and IFN-α (data not shown) were unable to inhibit the binding of CD23 to RPMI 8226 cells. This suggests that SCRs 1 and 3-4, where C3 and IFN-α bind respectively, may not be involved in CD23 binding. The EBV binding to CD21 does not require glycosylation of SCfR2 of CD21 (Moore et al, Inhibition of Epstein Barr Virus infection in vitro and in vivo by soluble CR2 (CD21) containing two short consensus repeats, J Virol 67: 3559 (1991 )). Likewise, CD23 binding to SCR2 region is independent of sugars (Aubry et al (1994) supra). This is in line with our observation that a non-glycosylated synthetic peptide is able to decrease CD23 binding to CD21. Therefore, CD23 binds to a binding site in SCR2 on CD21 that is close or identical to the EBV binding site which differs from the binding sites previously described for C3d,g and IFN-α.

CD23 was previously shown to positively regulate IgE production by binding to CD21 on B cells (Aubry et al (1992) supra). Based on the observation that an EBV peptide blocked CD23 binding to CD21, we investigated the effect of this EBV peptide on IgE production. The EBV peptide was able to inhibit IL-4-induced IgE production in a dose-dependent manner. This is shown in Figure 7b, which illustrates the inhibition of IL-4-induced IgE and lgG4 production by an EBV peptide binding to CD21. [PBL or purified tonsillar B cells (10⁶/ml) were incubated for 14 days with 200 U/ml of IL-4 alone or in the presence of anti-CD40 antibody (1 μg/ml) and increasing concentrations of EBV peptide. IgE and lgG4 were measured by specific ELISA and mean values +/- SD of one representative experiment are presented (n=4)].

This effect was observed in T-cell dependent and also in T-cell independent IgE production systems (Fig 7b), in which T-cell help is replaced by anti-CD40 Ab. This confirms our previous observation (Henchoz et al, Stimulation of human IgE production by a subset of anti-CD21 monoclonal antibodies : requirement of a co-signal to modulate ε transcripts, 81: 285 (1994)) that CD23-CD21 can regulate IgE production even in absence of T-cells by a homotypic B-B cell interaction, since B cells can express both CD23 and CD21 molecules. Intact EBV has been reported to provide the permissive signal for IgE switching (Thyphronitis et al, IgE secretion by Epstein-Barr virus infected purified human B lymphocytes is stimulated by interleukin-4 and suppressed by interferon-γ, Proc Natl Acad Sci USA 86: 5580 (1989)), like T-cells or CD40L. In contrast to the EBV particles, the EBV peptide is probably unable to crosslink membrane CD21 and is therefore unable to increase IgE production. The EBV peptide is rather inhibitory, decreasing IgE production by preventing the CD23-CD21 interaction.

Since IL-4 is known to induce lgG4 as well as IgE (Lundgren et al, lnterleukin-4 induces synthesis of IgE and lgG4 in human B cells, Eur J Immunol 19: 1311

(1989)), we investigated the effect of the EBV peptide on IL-4-induced lgG4 production. As shown in Fig 7b the EBV peptide was also able to inhibit lgG4 in a dose dependent manner. This observation suggests that the CD23-CD21 interaction also controls lgG4 production.

The effect of the EBV peptide was then investigated on the production of other Ig classes. No significant effect was found on polyclonal IgG and IgA production. This is shown in Figure 7c, which illustrates the modulation of immunoglobulin production by an EBV peptide binding to CD21. [PBL or purified tonsillar B cells (10⁶/ml) were incubated for 14 days with 200 U/ml of IL-4 alone or in the presence of anti-CD40 antibody (1 μg/ml) and increasing concentrations of EBV peptide. Ig were measured by specific ELISA and mean values +/- SD of one representative experiment are presented (n=4)]. In contrast, IgM production was significantly increased in the presence of the EBV peptide, especially when whole PBL were used (Fig 7c).

Not all CD21 ligands can regulate lgE/IgG4 production. A C3 peptide binding CD21 did not inhibit IgE and lgG4 production (not shown). This is shown in Figure 7d, which illustrates the absence of inhibition of IgE production with a C3 peptide binding to CD21. [Purified tonsillar B cells (10⁶/ml) were incubated for 14 days with 200 U/ml of IL-4 and anti-CD40 antibody (1 μg/ml) and increasing concentrations of C3 peptide or EBV peptide. IgE was measured by specific ELISA and mean values +/- SD of one representative experiment are presented (n=4)].

The C3 peptide did not inhibit CD23 binding (Fig 7a). These results highlight again the correlation between CD23-CD21 pairing and lgE/lgG4 production. IFN-α was not tested since it is already known that IFN-α inhibits IgE production (Pene et al, IgE production by normal human lymphocytes is induced by IL-4 and suppressed by α-interferon, γ-interferon, and prostaglandin E2, Proc Natl Acad Sci USA 85: 8166 (1988)), although IFN-α had no effect on CD23 binding to CD21 (not shown). In conclusion, this study therefore shows that an EBV peptide decreases CD23 binding to CD21 and selectively decreases IgE and lgG4 production by human B cells.

EXAMPLE 8 - CD23 binds to endothelial cells

An endothelial cell line (LT2, Endolethium vol 2, p 191-201 , 1994) or purified human umbilical venule endothelial cells were incubated with CD23-liposonr.es (CD23L), or glycophorine-liposomes (L-Gly) as a control. Specificity of the binding was demonstrated by the inhibition of CD23-liposome binding by an anti-CD23 mAb (mAb25 = a-CD23). Cells were analysed by FACS and MFI measured.

The results are shown in Figure 8. LT2 - derived proteins were purified on a CD23 - affinity column. two bands of 115 and 76 K Da were identified

Claims

1. A binding agent to CD21 , CD11 b, CD11 c, to a 70-85 KDa protein expressed on endothelial cells, or to a 115 KDa protein expressed on endothelial cells, for use in the treatment or prophylaxis of inflammatory, autoimmune or allergic diseases.

2. A binding agent according to claim 1, wherein the binding agent is an antibody, a fragment thereof, an artificial construct comprising an antibody or comprising a fragment thereof, a mimetic, or a derivative of any of these binding agents.

3. A binding agent according to claim 1 or 2 which is a humanised or chimaerised antibody.

4. A binding agent according to claim 1 which is Epstein Barr Virus, Factor X, a part of the Epstein Barr Virus (which may be in glycosylated or unglycosylated form), a fragment of Factor X, or a derivative of any of these binding agents.

5. A binding agent according to claim 4, which is the gp350/220 glycoprotein of

Epstein Barr Virus, the corresponding protein in unglycosylated form, a fragment of the aforesaid glycoprotein or protein, or a derivative of any of these binding agents.

6. A binding agent according to claim 5 which is a peptide with the amino acid sequence TGEDPGFFNVEIC, a fragment thereof, or a derivative of said peptide or fragment.

7. A binding agent according to any preceding claim which blocks interaction between CD23 and ligands which bind to it in vivo.

8. A binding agent according to any preceding claim for use in the treatment of arthritis, lupus erythematosus, systemic lupus erythematosus, Mashimotos thyroiditis, multiple sclerosis, diabetes, uveitis, dermatitis, psoriasis, urticaria, nephrotic syndrome, glomerulonephritis, inflammatory bowel disease, ulcerative colitis, Crohn's disease, Sjogren's syndrome, allergies, asthma, eczema, GVH, COPD, bronchitis, insulitis, rhinitis or diabetes.

9. A binding agent according to any of claims 1 to 7 for use in the treatment of arthritis, allergis, ulcerative colitis or Crohn's disease.

10. A binding agent according to claims 9 for use in the treatment of rheumatoid arthritis.

11. The use of a binding agent to CD21, CD11b, CD11c, to a 70 to 85 KDa protein expressed on endothelial cells, or to a 115 KDa protein expressed on endothelial cells for the manufacture of a medicament for the treatment of arthritis, lupus erythematosus, systemic lupus erythematosus, Mashimotos thyroiditis, multiple sclerosis, diabetes, uveitis, dermatitis, psoriasis, urticaria, nephrotic syndrome, glomerulonephritis, inflammatory bowel disease, ulcerative colitis, Crohn's disease, Sjogren's syndrome, allergies, asthma, eczema, GVH, COPD, bronchitis, insulitis, rhinitis or diabetes.

12. The use of a binding agent according to claim 11 for the manufacture of a medicament for the treatment of arthritis, allergies, ulcerative colitis, or Crohn's disease.

13. The use of a binding agent according to claim 12 for the manufacture of a medicament for the treatment of rheumatoid arthritis.

14. A pharmaceutical composition comprising a binding agent according to any of claims 1 to 7 and a pharmaceutically acceptable carrier.

15. A binding agent according to claim 1 substantially as hereinbefore described.

16. A method of treating an inflammatory, autoimmune or allergic disease comprising administering a pharmaceutically effective amount of a binding agent to CD21 , CD11b, CD11c, to a 70 to 85 KDa protein expressed on endothelial cells, or to a 115 KDa protein expressed on endothelial cells, to a patient.