This invention relates to a method of combating, therapeutically or prophylactically, certain autoimmune diseases and immunological complications following transplantation.

In autoimmune diseases, the immune system of the host appears to attack certain host tissues or organs, apparently erroneously recognising as foreign specific host cell surface antigens. The origin of this anomalous behaviour has hitherto been uncertain. We have now elucidated a mechanism by which an autoimmune response may arise following exposure of the immune system to a microorganism and devised methods of combating such diseases based on this mechanism.

Chronic graft versus host (GVH) disease is a potential-ly lethal immunological complication following bone marrow transplantation. Chronic GVH disease primarily attacks epithelial cells in the skin, mucosa, intestine and liver,* several organ systems can be involved. The significance of the disease is such that morbidity in severe cases can be as high as 50% and there is no universally effective treatment.

Chronic GVH disease has properties similar to an autoimmune disease and traditionally has been regarded as an immunological reaction caused by graft-derived immuno-competent lymphocytes which mature in the transplant recipient and react against antigens of the new host.

The virus C V (cytomegalovirus) is an opportunistic virus infection that causes major morbidity and mortality in immunocompromised individuals such as patients treated with immunosuppressive drugs, for example recipients of bone marrow or organ grafts or in patients suffering from immunodeficiency such as HIV

infection. Healthy non-immunocompromised individuals usually have few clinical symptoms during the acute CMV infection. Following a primary infection, CMV remains in a latent state in some cells in the body. Most adult individuals harbour latently infected cells which can be present in blood and in bone marrow. Thus, a further complication with transplantation is that latent CMV infection is easily transferred from a healthy immune donor to a transplantation recipient. CMV is a serious, often life-threatening disease in immunocompromised patients, such as transplant recipients, treated with immuno-suppressive drugs and in HIV-infected individuals. CMV is also one of the major causes of malformations and intrauterine death of human fetuses.

In International Patent Application PCT/EP94/00188 published under No. WO 94/16724 we have- disclosed that susceptibility of cells to infection by CMV is associated with surface expression of the antigen CD13, also known as aminopeptidase N. We further disclosed that, surprisingly, the CMV virion also carries the antigen CD13 or a molecule which reacts in the same way as CD13. We believe that this molecule is not a product of the viral genome but is transferred from the CD13 positive cells when the infectious virus is released after replication therein.

Virus infection usually induces a specific humoral response directed at several of the viral proteins, and neutralising antibodies are often directed against proteins of the viral envelope. It has been found that sera from patients undergoing an acute human CMV (HCMV) infection, or in convalescence thereafter, have antibodies reactive against the CD13 antigen expressed on normal cells. It thus appears that the virus has presented a normal human cell surface protein as being foreign and thereby triggered an immune response not only against the CD13 on the CMV virons but against CD13 on normal cells. Thus, infection with a virus carrying

host proteins has led to induction of an autoimmune response against a host protein.

CD13 present on normal cells is not expected to induce an autoimmune response. However, when present on the virus, it becomes immunogenic, since viruses are taken up by or infect "antigen-presenting cells". This leads to degradation of proteins associated with the virus ^• peptides resulting from such degradation are then presented in association with host MHC- olecules to resting immunocompetent T lymphocytes which triggers the series of events that lead to a specific immune response, including the formation of antibodies against the CD13 protein. An immune response against CD13 present on the virion might therefore cause immunological damage of non-infected CD13 expressing cells, such as epithelial cells. The clinical observation that patients who have experienced a CMV infection following bone marrow transplantation have a high risk of developing chronic GVH disease with an immunological attack on cells in the intestine, liver and skin may thus be explained by a CMV-induced autoimmune reaction against CD13 expressing cells. Based on this concept, we propose that chronic GVH disease may be regarded as an autoimmune disease following a viral infection.

In 1988, S. Karuppan et al. (Karuppan, S., Lindholm, A. and Mδller, E. : Pre- and post-transplant sensitization in cyclosporin A-treated patients, Clin. Transplantation, 2: 245-251, 1988) found that some patients, who had been transplanted with allogeneic kidney grafts, developed non-HLA-specific autoreactive cytotoxic antibodies during the post-transplantation period. The formation of such antibodies was associated with viral infection in these patients. The exact specificity of these antibodies was not assessed at that time.

However, we have recently tested some other serum

samples from CMV infected transplant patients and have found that they contain antibodies against human CD13. We have recently demonstrated that sera from patients undergoing an acute CMV disease contain IgM and in some patients, subsequently IgG antibodies, specifically directed against CD13. We have also observed that patients suffering from severe GVH disease were previously observed to have produced CD13-specific autoantibodies believed to be in response to an observed CMV infection. These findings taken together demonstrate that viral infection with CD13-carrying CMV induces the formation of autoreactive antibodies of a defined specificity and suggest that such autoimmune reactivity leads to tissue damage and cause other immunological complications typical of graft versus host disease.

The above mechanism of a virus-induced autoimmune response is applicable to any case where a virus carries a cell surface antigen of the host. Other viruses known to carry host cell antigens include HIV, which carries the CD4 molecule as well as H A molecules. It has been suggested that immunity against HIV protein GP ₁₂₀ , binding to CD4, can induce an auto-anti-idiotypic immunity against CD4 which may lead to dysfunction of CD4-positive cells. However, it has not previously been suggested that immunity against viral CD4 is responsible for immunity against CD4 in HIV-infected patients and that this leads to the AIDS-related immune defect wherein CD4 positive cells and CD4 mediated function are lacking.

We now propose that other forms of autoimmune responses are triggered by infection by viruses or other microorganisms carrying host cell surface antigens leading to an immunological attack against normal host cells carrying the antigen concerned. Apart from immunological complications following transplantation, such as graft v. host, we propose also that host v.

graft reactivity is caused by reactivity against CD13 since CMV infection has been suggested to precipitate even host versus graft reactions in organ eg. kidney- allografted patients. Further we propose that a similar mechanism is responsible for the depletion of CD4- positive T cells observed in patients with HIV infection. Also other autoimmune diseases such as type I diabetes, rheumatoid arthritis, myasthenia gravis, cardiomyopathy and multiple sclerosis have been suggested to be triggered by viral or bacterial infection in genetically susceptible individuals but candidate microorganisms have not been fully identified. All these diseases are life long diseases often severely impairing the subject's quality of life and for which current therapies are not curative and have quite undesirable side effects. Insofar as these diseases may be triggered by infection with microorganisms carrying antigens present on the cells of the relevant host tissues, -they will be susceptible to treatment by the present approach to combatting autoimmune response.

Thus viewed from one aspect the present invention provides a method of combating an autoimmune disease of a type in which a characteristic protein on host cells interacts with antibodies against the characteristic protein raised in the host by an infecting organism carrying the said characteristic protein, characterised in that said antibodies or organisms or cells carrying said characteristic protein or a fragment thereof are caused to be immobilised, removed or inactivated by interaction with a binding partner therefor.

The term "combating" as used herein includes both prophylaxis and therapy.

The method of the invention is particularly applicable to damage to normal tissue that occurs during graft versus host complications and also extends to autoimmune diseases triggered by immunologically foreign microorganisms that carry host proteins, particularly

viruses. The method of the invention may also be useful in HIV infected individuals.

In another aspect, the present invention provides the use, in the preparation of a medicament for combating an autoimmune disease of a type in which a characteristic protein on host cells interacts with antibodies against the characteristic protein raised in the host by an infecting organism carrying the said characteristic protein, of a binding partner for said antibodies or for organisms or cells carrying said characteristic protein or a fragment thereof, wherein said binding partner is capable of causing said antibodies, organisms or cells to be immobilised, removed or inactivated.

In the case of graft versus host disease, whilst an association of the occurrence of this with CMV infection referred to above has been recognised (for reference see Appleton et al. , Bone Marrow Transplantation Jl 349-355 (1993), the presence of normal human host proteins on the surface of actual virions and the fact that when presented in this way, such proteins, specifically CD13, are capable of triggering an immune response had not prior to our work previously been recognised, and no parallel or correlation had been drawn to graft- associated immunological reactions.

We have however now demonstrated an exclusive association between CD13 autoantibodies in those patients who had experienced either CMV disease or CMV viraemia. Anti CD13 autoantibodies have not been detected in patients without signs of CMV infection or with other autoimmune diseases. Furthermore, we have demonstrated a clear correlation between the presence of such autoantibodies and the development of chronic graft- versus-host disease.

The presence of host cell proteins on other species of infectious virus has been reported. Thus the presence of various host derived proteins, including

CD4, has been demonstrated on human immunodeficiency virus 1 (HIV-1) but no biological role has previously been recognised or suggested. Our proposal, the basis of the present invention, that autoimmunity is mediated by infecting microorganisms expressing on their surface host proteins, explains the observations reported by others that anti-CD4 antibodies can, inter alia, induce apoptosis in T cells with the help of HIV proteins (Westendorp et al., Nature, 375, 497-500, 1995) .

Indeed, the finding that infection with virus which carries the normal cell surface protein CD13 can trigger a specific autoimmune response is extremely surprising and is contrary to what has become known as the central dogma in immunology, namely the ability of the immune system to discriminate between "self" and "non-self". It might have therefore been expected that CD13 would not have been recognised as foreign. We have however shown the presence of anti-CDl3 antibodies in human patients-undergoing acute CMV infections and in patients undergoing severe graft versus host disease which are reactive against CD13 expressing host cells including even those which are not infected. We have also found that serum of patients with chronic graft-versus-host disease, containing anti-CD13 antibodies binds to skin biopsies and that this binding is dependent upon and mediated by the CD13 autoantibodies. Since skin is one of the organs principally affected in graft versus host disease, this represents a positive association of both cause and effect in autoimmune disease.

Whilst not wishing to be bound by theory, we propose that following uptake of virus or other microorganism carrying the host protein, (CD13 in the case of CMV) , by antigen-presenting cells, such as macrophages, monocytes, dendritic cells, Langerhans cells in the skin, B-lymphocytes, the virus particles are degraded in conventional manner and peptides, including those from CD13, associate inside the cell

with host MHC molecules which are transported to the surface of the antigen presenting cells. Again, whilst not wishing to be bound by theory, we further propose that upon infection of professional antigen-presenting with CMV, viral peptides are presented in an MHC- restricted manner to specifically reactive T lymphocytes, be they MHC class II restricted CD4 ⁺ cells or MHC class I restricted CD8 ⁺ cells. These cells then interact with B lymphocytes, reactive against CD13 by virtue of their CD13-specific Ig-receptors. Such CD13 may either be present on the antigen-presenting cells or on CMV which is internalized by binding to CD13-specific Ig-receptors. Inside the B lymphocytes, viral peptides may be presented in association with MHC class II molecules, leading to interaction between T cells reactive against viral peptides and B cells specific for CD13. T cells provide help for differentiation of CD13- reactive B cells to antibody producing cells. Thus, a continued reactivation of virus production from infected cells, such as might occur in immunocompromised individuals would help sustain autoantibody production against CD13. An analogous mechanism would operate in the case of other microorganisms carrying other host proteins e.g. HIV carrying CD4. In contrast but in accordance with the previously mentioned dogma, host proteins such as CD13 when expressed on normal host cells are non-immunogenic.

The mechanism is applicable to other microorganisms, such as viruses, carrying host proteins, such as HIV (e.g. carrying CD4) . Although viruses are likely to be the principal agents carrying host proteins which give rise to an immune response, other infectious microorganisms proliferating in a host species and transmitted by infection may acquire host cell surface proteins which, on ingestion by antigen presenting cells, would lead to an autoimmune response.

Other autoimmune diseases which may be combatted

according to the present invention include type 1 diabetes, for which one example of a characteristic protein is GAD, and which may be associated with viruses such as Echo or Coxsackie viruses. Multiple sclerosis, Sjόrgens syndrome and Wegeners granulomatoris/vasculitis have also been associated with characteristic proteins and microbial infections.

An important aspect of the invention is thus concerned with mitigating the effects of such autoimmune responses and autoantibodies raised to microorganisms, particularly virus, carrying normal host proteins.

A number of treatment strategies are possible and are encompassed by the present invention. Generally speaking, the four main approaches are to:

(i) prevent transfer of infection by the microorganism (e.g. by removing infected cells or the microorganism itself) ,*

(ii) prevent infection by the microorganism in question -e.g. by blocking its binding sites on the target cells, or by vaccinating or induction of immunological tolerance or intensified prophylaxis against de novo infection or reactivation of the microbial infection concerned;

(iii) treat patients who have the infection, for example by removal of the autoantibodies or by blocking or competition of the autoantibodies (e.g. using peptides derived from the antigen) ,*

(iv) treat patients who have progressed to autoimmune diseases, e.g. GVH disease, for example, by removal of the autoantibodies, blocking/competition etc.

In one embodiment this involves the administration to a patient possessing such autoantibodies of binding partners to these autoantibodies which will 'sequester' the antibodies and prevent tissue destruction. Such binding partners include the host protein itself or a fragment thereof. In the case of graft versus host disease, the host protein may be CD13. As far as the

fragments are concerned, soluble fragments are preferred. These may be prepared in known manner such as by proteolytic digestion or they may be prepared by recombinant DNA technology using standard techniques well known in the art and widely described in the literature. All that is necessary is that the fragments be functionally equivalent to the whole protein ie. be capable of binding to the autoantibodies.

This method may be used for the treatment of patients with infection by the microorganisms concerned e.g. CMV infection and also those which have progressed to autoimmune disease e.g. chronic graft versus host disease.

In one embodiment of this aspect of the invention, autoantibodies may be removed or depleted from the blood of a patient by essentially in vitro methods, for example by removal of blood or other body fluids, e.g. by plasmapheresis, and contacting the blood or body fluid thus removed with binding partners to the antibody concerned. Such binding partners include those referred to above, and these may conveniently be used in immobilised form. Plasmapheresis may be performed by methods known to those skilled in the art, as described for example in Lockwood et al. (Lancet 711-715, 1976) and Lockwood et al. (Adv. Nephrol. j£, 383-418, 1979) .

In another therapeutic treatment regime, antibodies or fragments thereof which themselves do not initiate an immune response may be used to compete with the endogenous autoantibodies for binding to the characteristic host protein e.g. CD13. Particularly useful in this regard are Fab monomers and dimers, which may conveniently be in water soluble or immobilised form, or antibodies of Ig classes which do not cause tissue damage including antibodies of the IgM class which lack the Fc portion, which may be prepared by the techniques of genetic engineering known to those skilled in the art. Anti-idiotypic antibodies, may also be used

and may be prepared using methods well known in the art.

In another embodiment the actual virus carrying the host protein may be prevented from triggering the autoimmune response by effectively 'neutralizing' the host antigen on the virus. Suitable agents include various forms of antibodies to the host protein which retain antigen binding properties.

A particular problem associated with graft versus host disease is that donor blood or bone marrow may be infected with CMV and thus transfer to the recipient either infectious virus carrying human CD13 to which the recipient may raise autoantibodies or infected CD13 positive cells.

Thus according to another aspect, the present invention provides a method of preventing graft versus host disease in a subject to be transfused with bone marrow, blood or a blood or plasma fraction comprising treating the bone marrow, blood, or blood or plasma fraction with an antibody to CMV or a fragment thereof to permit removal of the virus and/or infected cells prior to transplantation into the recipient.

According to another aspect, the present invention provides a method of preventing graft versus host disease in a subject to be transfused with bone marrow, blood or a blood or plasma fraction comprising treating the bone marrow, blood, or blood or plasma fraction with an antibody to CD13 or a CD13 binding fragment thereof to permit removal of the virus and/or infected cells prior to transplantation into the recipient. Preferably the antibody or fragment thereof is in water-soluble or immobilised form.

In this way, the antibodies or antibody fragments can inactivate any infectious virus or CMV infected cells and prevent the transfer of CMV infection. Marrow or blood products purged in this manner will be less capable of initiating graft versus host complications .

The method of the invention can also be used to

purge donor blood or bone marrow to remove virus particles or other microorganisms known to carry host proteins giving rise to an autoimmune response, by contact with an immobilised agent binding to the virus or other microorganism, for example anti-CD13 or indeed any antibody reacting with any surface protein of the virus or microorganism. The immobilised agent may, for example, be attached to superparamagnetic beads to aid removal.

The invention also extends to the screening of blood, plasma and products derived therefrom for the presence of microorganisms such as CMV or HIV which are associated with autoimmune conditions, using, for example, detectably labelled antibodies, in order to avoid transferring the virus itself in such body fluids to a donor.

In a further aspect, the present invention provides the use of an antibody to CMV or a fragment thereof in the manufacture of an agent for preventing graft versus host disease in a subject to be transfused with bone marrow, blood, or blood or plasma fraction by the treatment of said bone marrow, blood, or a blood or plasma fraction with said antibody or fragment thereof prior to said transfusion.

In a further aspect, the present invention provides the use of an antibody to CD13 or a CMV binding fragment thereof in the manufacture of an agent for preventing graft versus host disease in a subject to be transfused with bone marrow, blood or a blood or plasma fraction by treatment of said bone marrow, blood or blood or plasma fraction with said antibody or fragment thereof prior to said transfusion.

The association of autoimmunity with microorganisms which carry host proteins offers a new method of combating such diseases, by treating the microorganism with an antimicrobial agent, such as, in the case of diseases associated with viruses, an antiviral agent.

Thus viewed from a further aspect, the present invention provides the use of an antimicrobial agent in the manufacture of a medicament for use in combating an autoimmune disease of a type in which a characteristic protein on host cells interacts with antibodies against the characteristic protein raised in the host by an infecting microorganism carrying the said protein.

Many anti-microbial agents are of course well known in the art and widely described in the literature. Examples include antibiotics and the vast array of anti¬ viral agents (e.g. nucleoside analogues) which are known for the treatment of the microorganism concerned. CMV infection for example would be treated by the anti-viral agent ganciclovir.

Viewed from a further aspect, the present invention provides a method of preventing an autoimmune disease associated with the presence of a host cell protein on an infectious microorganism, comprising immunising a subject with a vaccine based on a strain of said microorganism lacking the said host protein or carrying a non-immunologically active derivative thereof.

The strain of the microorganism may be a non- infectious strain, or an avirulent strain.

In this way, the body's natural defence system is immunologically primed to destroy or inactivate infectious virus or other microorganism carrying host proteins (which would otherwise initiate autoantibody production when the body is exposed thereto) without any danger of the vaccine itself initiating the autoimmune response. Vaccines of this type can be prepared in accordance with standard techniques including use of heat-killed organisms, attenuated organisms or purified virus-specific antigens. In the case of prevention of graft versus host disease, the transplant recipient may be vaccinated prior to receiving the donor marrow or blood against native CMV, non-infectious (attenuated, avirulent) CMV, CMV proteins or peptides or any of the

above mentioned CMV types lacking CD13. The host protein on the virus may be inactivated by interaction with antibody fragments as described above or it may be biochemically inactivated. It is also possible to make virus completely lacking in host protein recombinantly with the aid of helper virus.

Viewed from a yet further aspect, the present invention provides a modified microorganism wherein said microorganism in its native infectious form carries a protein characteristic of a host cell, and wherein said characteristic host protein is absent or non-immunogenic in said modified microorganism.

In one embodiment, said modified microorganism is CMV and said characteristic host protein is CD13.

Viewed from a further aspect, the present invention provides a method of diagnosis of or of detecting a subject at risk of developing, an autoimmune disease of a type in which a characteristic protein on host cells interacts- with antibodies against the characteristic protein raised in the host by an infecting organism carrying the said characteristic protein, wherein the presence of antibodies to the characteristic host protein is detected.

In one particular embodiment this aspect of the invention provides a method of diagnosis of chronic graft-versus host disease, wherein the presence of antibodies to CD13 is detected.

Detection may be carried out using any of the aforementioned binding partners, in cell-free, soluble or immobilised form. Such immunoassays for the detection of antibodies are conventional and very well known in the art. For example an ELISA could be used for the detection of the autoantibodies, using immobilised antigen e.g. immobilised characteristic host protein, or an immobilised fragment thereof. Other tests which could be used for the detection include flow cytometry, purification of CD13 from cells for

precipitation etc.

Detection of 'at risk' subjects according to this aspect of the invention permits early intervention using, for example, any of the prophylactic measures referred to herein. This aspect of the invention could be used, for example, to identify patients who may be CD13-antibody positive and therefore at risk of developing severe chronic graft versus host disease.

Viewed from a further aspect, the present invention provides a vaccine composition for stimulating an immune response in a host against an infectious microorganism wherein the immune response is associated with the presence of a host cell protein carried on said infectious microorganism, said vaccine composition comprising a modified microorganism as hereinbefore defined together with a pharmaceutically acceptable carrier or diluent.

A vaccine composition according to the invention may be pr-epared according to methods well known in the art of vaccine manufacture. Traditional vaccine formulations may comprise the modified microorganism according to the invention together, where appropriate, with one or more suitable adjuvants eg. aluminium hydroxide, saponin, quil A, or more purified forms thereof, muramyl dipeptide, mineral or vegetable oils, vesicle-based adjuvants, non-ionic block co-polymers, or DEAE dextran, in the presence of one or more pharmaceutically acceptable carriers or diluents. Suitable carriers include liquid media such as saline solution appropriate for use as vehicles to introduce the peptides or polypeptides into the subject. Additional components such as preservatives may be included.

Likewise formulation of the binding partners used in the methods of the invention may be achieved using methods well known in the pharmaceutical and medical

arts. Thus the binding partner, which may generally be either antibodies or antigen-binding fragments thereof, capable of binding to the characteristic host protein, or host proteins or peptides or polypeptides corresponding thereto, and fragments thereof, may be formulated into compositions for- administration to the subject by any convenient administration route eg. enterally or parenterally including for example orally, by transmucosal delivery, eg. rectally, in implants or by intravenous, intramuscular, or subcutaneous injection or by infusion etc.

Actual treatment regimes or prophylactic regimes and dosages will depend to a large extent upon the individual patient and may be decided by the medical practitioner based on the individual circumstances.

The compositions prepared according to the invention may take any of the conventional pharmaceutical forms known in the art, and may be formulated in conventional manner, optionally with one or more pharmaceutically acceptable carriers or excipients. Thus for example the compositions may take the form of ointments, creams, solutions, salves, emulsions, lotions, liniments, aerosols, sprays, drops, pessaries, suppositories, tablets, capsules or lozenges.

Viewed from yet another aspect, the present invention provides a method of combating an autoimmune disease triggered by infectious microorganism associated with the presence of a host cell protein carried on an infectious microorganism comprising inducing immunological tolerance to said host protein in a subject.

Tolerance may be induced in neonates where there is evidence that the newborn may be at risk of contracting an autoimmune disease and it may also be induced in adults. Oral feeding or intravenous injection with soluble host proteins, such as CD13, or peptides thereof may be used according to conventional techniques.

The invention will now be described in more detail in the following non-limiting Examples, with reference to the Figures which show:

Figure l. Cytotoxicity of a monocytic cell line (THP-1) by a serum sample from a bone marrow transplanted patient with CD13 specific antibodies. Addition of non- complement binding monoclonal antibodies against CD13 and against CD9 caused significant inhibition of cytotoxicity, whereas antibodies against CD10, VCAM-l or CD33 did not inhibit the reaction. The total titer of the patient serum sample in a repeat experiment was 1:1024.

Figure 2a and 2b. Flow cytometric analysis of a CD13- positive serum sample on mouse 3T3 cells and of CD13 transfected 3T3 cells, showing a clearly positive reaction only of the CD13 positive cells (part A: detected-with a secondary FITC-conjugated anti-human IgM and IgG) . Part B shows that addition of F(ab) ' ₂ - fragments of a CD13 specific monoclonal antibody (MY7) causes complete blocking of binding of patients' antibodies to CD13-transfected cells, to levels similar to that of background non-transfected cells.

Fiσure 2c and 2d. Flow cytometric analysis of the same patient serum binding to mouse 3T3 cells and to CD13 transfected mouse 3T3 cells. Part C demonstrates the presence of IgM antibodies reacting only with the transfected cells, part D the presence of IgG antibodies. Immunoglobulin class determination was performed with class specific anti-Ig reagents.

Figure . Cytotoxicity of monocytic human cells by a serum sample from a young boy with a primary CMV disease, which is significantly blocked by antibodies against CD13, and partially by antibodies against CD9.

The experiment shows results when both monoclonal antibodies were added together.

Fiσure 4. Phast gel analysis of immunoprecipitation experiments of cell membrane lysates of CD13 transfected mouse cells with sera from bone marrow transplanted patients.

Lane 1 shows the positive control (anti-CD13, MY7) , lanes 2-4: three different samples from patients with CMV disease or CMV infection and lanes 5 and 6 shows results with sera from two different CMV negative patients. Molecular weight standards (M) (HMW-SDS, Pharτnacia-LKB) used for calibration of the immunoprecipitations contained myosin (212 kDa) , alpha-2 macroglobulin (170-kDa) , β-galactosidase (116-kDa) , transferrin (75-kDa) and glutamate dehydrogenase (53- kDa) .

Fiσure 5. Binding of IgM antibodies (panel A-C) and IgG antibodies (panel D-F) in patients' sera to normal skin biopsies. Panels A and D; serum from a patient without CMV infection and without signs of cGVHD. Panels B and E; serum from a patient with CD13 specific IgG antibodies who experienced CMV disease and developed extensive cGVHD. Panels C and F; serum from a patient with CD13 specific IgM antibodies, as well as anti- nuclear antibodies, who experienced CMV disease and developed extensive cGVHD (xl65) .

Figure 6. Different staining patterns using two monoclonal antibodies against CD13, and a FITC conjugated secondary antibody, staining the epithelium (A-C) and the eccrine sweat glands (D-E) . Panels A and D; secondary antibody, panels B and E; anti-CD13 (clone MY7) , and panels C and F ^■ anti-CD13 (clone WM15) (xl65) .

Fiσure 7 Flow cytometric analysis of the binding of two different CD13 specific monoclonal antibodies to mouse cells (A and B) , mouse cells expressing human CD13 (C and D) and human lung fibroblasts (E and F) . Panels A, C and E show the results using anti-CD13 (clone MY7) , and panels B, D and F show anti-CD13 (clone WM15) .

Figure 8. Autoantibodies in a patient's serum are specifically directed against the CD13 molecule. Panel A; staining pattern of eccrine sweat glands by a serum taken from a patient who had experienced CMV infection and was suffering from extensive cGVHD. Panel B; staining pattern of the same serum after blocking with a mixture of monoclonal antibodies against CD13 (MY7, WM15 and L138) . Panel C; staining pattern of the same serum after specific preabsorption to mouse cells expressing human CD13 (xl65) .

Figure 9.. Deposition of IgM antibodies in papillary dermis in a skin biopsy from a patient known to have CD13 specific antibodies (panel A) . No deposition of IgM antibodies was found in a biopsy from a patient lacking CD13 specific antibodies (panel B) (xl35) .

Example l

Demonstration of the presence of anti CD13 antibodies in serum of patient with severe chronic GVH disease

Serum from a CMV infected patient suffering from severe GVH disease was tested for reactivity against the monocyte cell line THP-l (available from American Type Culture Collection, National Institute of Health, USA under No. TIB202) . This cell line is positive for CD13. Reactivity was measured by cytotoxicity of target cells in the presence of complement essentially according to

standard techniques (Kissmeyer, F. and Kjerbye, K.E. Lymphocytotoxicity microtechnique. Purification by lymphocyte flotation. Histocompatibility Testing 1967. Eds. Curtoni et al. p. 381-383, Munksgaard, Copenhagen) and specificity was assessed by specific blocking with non-cytotoxic antibodies against CD13 but not against other monocyte cell surface proteins.

The monocytoid THP-1 cell line was grown in tissue culture flasks and suspended in RPMI 1640 medium containing 10% inactivated human AB serum. 0.5 μl of cells at a concentration of 3 x 10 ⁶ /ml was added to flatbottomed microplates and mixed with 0.5 μl of undiluted human serum to be tested. The mixture was incubated for 30 minutes at room temperature. Blocking monoclonal antibodies at a concentration of 300 μg/ml, diluted in phosphate buffered saline in volume of 50 μl, were added to target cells, concentration 3 x l0 ⁶ /ml, and incubated at room temperature for 1 hour. The IgGl monoclonal antibodies anti-CD13 was obtained from Caltag, USA, and by Coulter, USA anti-HLA class I from Becton and Dickinson, USA, anti-CDIO from Dakopatts, Denmark, anti-CD9 from Seralab, England and anti-VCAM-l from Immunotech, France and anti-CD33 from Becton and Dickinson, USA. The cells were washed once and added to microplates and assayed as described above with human serum to be tested. Thereafter, lyophilized rabbit complement dissolved in distilled water (2 ml) , and containing 55 μl of a mixture of ethidium bromide and acridine orange, was added in a volume of 2 μl and the incubation continued for 40 minutes at room temperature. Cytotoxicity was read in a fluorescence microscope, where living cells stain green and dead cells read. Positive reactions were scored if the text mixture contained 10% more positive (dead) cells than the control.

Figure 1 shows the results, presented as inverse of log ² cytotoxic titre. It can be seen that the patients own serum has strong reactivity against the monocytoid cell line and that this reactivity is specifically blocked by antibodies to CD13 but not by antibodies against HLA class 1 molecules or against other surface structures present on the cell line, such as CD10, VCAM-l or CD33. Weak inhibition was induced by anti-CD9.

Example 2

Specificity of antibodies in CMV infected patient

In order to demonstrate the specificity of the antibodies for CD13 in the serum of the above Example, a comparative test was performed on two cell lines, identical apart from the expression by one cell line of CD13.

Mouse 3T3 cells transfected with human CD13 (PZIP cells) were utilised (Look et al, J. Clin. Invest. HI, 1299- 1307, 1989) . Sera was pre-absorbed with 3T3 cells to remove natural antibodies against mouse cells, then patient and control sera were tested for reactivity against CD13 expressing 3T3 cells and against control 3T3 cells.

Flow cytometric analysis was performed using 3T3 and PZIP cells at a concentration of 5 x 10 ⁶ /ml, mixed with 50 μl of undiluted human serum. Blocking of patient serum binding to the PZIP cell line was performed with a F(ab) ' ₂ fragment of the CD13 monoclonal antibody for 1 hour at room temperature followed by washing. Developing reagents were fluoresceinated F(ab) ' ₂ anti- human Ig heavy and light chains, - anti-human IgM and anti-human IgG, all from The Binding Site, Birmingham, England using 5 μl of a concentration of 0.2 mg/ml and

incubation for 30 minutes on ice in the dark. All analyses were performed in a Becton Dickinson FACStrak flow cytometer.

Fig. 2a shows the patient serum reacts strongly against CD13 positive cells (right hand trace) but not against control 3T3 cells (left hand trace) . Control serum was completely negative on both cells after absorption (not shown) . When the reaction was performed after blocking with anti-CD13-specific antibodies, all reactivity was removed, demonstrating that the reaction was indeed caused by CD13-specific antibodies (Fig. 2b) . Two more experiments demonstrate that the serum contains both IgM (Fig. 2c) and IgG (Fig. 2d) antibodies.

Example 3

Specificity of anti-CD13 antibodies

Western blotting experiments using membranes from mouse 3T3 cells expressing human CD13 confirmed the presence of CD13 specific antibodies against a protein of the expected size (150 kD) (data not shown) .

Example 4

Cytomegalovirus induced specific autoantibodies --σainqi ^* C-D13 in patients with chronic σraft-versus host disease

METHODS

Patient Population

From 1988 until 1994, 33 patients who underwent bone marrow transplantation (BMT) at Huddinge Hospital were selected for this retrospective study. The patients were selected according to the diagnosis of post- transplantation CMV infection and classified into three groups: (a) ten patients with CMV disease, (b) nine

patients with asymptomatic CMV infection (CMV DNAemia or CMV viremia) and (c) fourteen patients without any signs of CMV infection in the post-transplantation period (included both CMV seropositive and CMV seronegative patients) . Between one and ten sera from the different patients were available for investigation. One additional serum was from a 16 year old boy, eight weeks after a severe primary CMV infection.

Sera from 11 healthy laboratory workers served as controls. All sera were serologically screened for the presence of antibodies directed against CMV. Six were seropositive for CMV, whereas five were seronegative. Six patients on immunosuppressive therapy after organ transplantation were included as controls. Four of these patients had an active HSV-1 infection and the other two an active EBV infection. Sera from six patients with SLE disease, and from one patient with rheumatoid arthritis was also included as controls. Only one serum sample from each of the control individuals was used in this study. All sera were tested negative for HBV-, HCV- and HIV-antibodies.

Preparation for Bone Marrow Transplantation

Patients were prepared for bone marrow transplantation with cyclophosphamide and total-body irradiation or Busulphan combined with cyclophosphamide as described in Ringden et al, Bone Marrow Transplant _£, 19-25 (1992) . The day of BMT was defined as day 0. 29 patients had hematological malignancies, one severe aplastic anemia, one familial hemophagocytic histiocytosis, and two patients inherited enzyme defects. 18 patients received bone marrow from an HLA identical sibling, 4 from a mismatched family donor and 11 from HLA-A, -B and -DR genetically compatible unrelated donors. Details of transplant procedures were according to Ringden (supra) .

As prophylaxis against graft-versus-host disease (GVHD) , cyclosporine (CsA) was combined with a short course of methotrexate. One patient received high dose acyclovir as CMV prophylaxis, no one received ganciclovir. Most patients were given pre-emptive therapy with ganciclovir or foscarnet immediately upon diagnosis of a CMV infection either by a rapid isolation technique or PCR using CMV specific primer pairs (Ehrnst, A., Barkholt, L. Lewensohn-Fuchs, I., Ljungman P., Teodosiu, 0., Staland, A., Ringden, 0.. Johansson, B., "CMV PCR monitoring in leukocytes of transplant patients", Clinical and Diagn. Virology, 1, 139 (1995)) .

The clinical diagnosis of GvHD disease was confirmed in all patients by appropriate biopsies from affected tissues and graded as 0 through IV with the criteria previously described for acute GvHD (Thomas et al, N. Engl. J. Med 292. 895-902 (1975), and Deeg et al, Ann. Rev. Med __$., 11-24 (1984)) and as "none", "limited" or "extensive" for chronic GvHD (Shulman et al, Am. J. Med, 69 204-17 (1980) ) .

Definitions of CMV disease

CMV pneumonia was defined as a clinical syndrome characterized by radiographic evidence of pulmonary interstitial infiltrates, hypoxia and CMV verified from broncho-alveolar lavage fluid or autopsy material. Verification of CMV was performed either by virus isolation, typical histology, immunohistochemical staining with CMV specific antibodies on histological material or by DNA PCR using CMV specific primer pairs. CMV gastroenteritis was defined as gastrointestinal symptoms combined with the identification of CMV in biopsy material obtained by upper or lower endoscopy. The diagnosis of CMV encephalitis was determined from a combination of CMV viremia and characteristic

neurological symptoms in the absence of any other cause of encephalitis. CMV syndrome was defined as CMV viremia combined with at least 50% reduction in absolute neutrophil or platelet counts. In addition, the neutrophils should be <1 x 10 ⁹ /1 or the platelets <50 x 10 ⁹ /1. No patient was diagnosed with CMV hepatitis. The day of CMV disease was defined as the day when the diagnosis was verified by an invasive procedure.

Cells

THP-1 cells, (acute monocytic leukemi cell line, ATCC, Rockville, Maryland, U.S.A.), were maintained in bicarbonate-free RPMI 1640 medium supplemented with 25 mM HEPES [4- (2 hydroxyethyl) -1-piperazine ethanesulfonic acid] , 10% heat inactivated fetal calf serum, L- glutamine (2 mM) , penicillin (100 U/ml) and streptomycin (100 ug/ml) (all from GIBCO BRL, Grand Island, N.Y) . Human embryonic lung fibroblasts (HL734, provided by V.- A. Sundquist) and mouse NIH-3T3 cells (ATCC, Rockville, Maryland, USA) were maintained in bicarbonate-free minimal essential medium with Hank's salts (GIBCO BRL) with the same supplements as described. The NIH 3T3 transfectants expressing human aminopeptidase N (hAPN- 3T3, provided by A.T. Look, St Jude Children's Research Hospital, Memphis, Tenn.) (Look et al, J. Clin. Invest. Hi,1299-1307, (1989)) were maintained in the same medium with the addition of 800 μg of G418 (Geneticin,* GIBCO BRL) per ml. In some experiments, we used fresh monocytes from buffy coats. Peripheral blood mononuclear cells were isolated by density gradient centrifugation on Lymphoprep (Nycomed, Oslo, Norway) , washed and resuspended in culture medium (RPMI) . Monocytes were collected after adherence to plastic at 37°C and used after washing in PBS.

AntibnrHPB

Pooled sera containing alloantibodies from patients immunized by multiple blood transfusions or organ transplantation were used as a positive control (Karuppan et al, Transplantation £-3., 666-673 (1992)) and sera from healthy nontransfused AB blood group males served as negative controls.

For the blocking assays, the following mouse monoclonal antibodies were used: anti-HLA class I (ATAB, Scarborough. U.S.A.), anti-human HLA class II (BU25, The binding site, Birmingham, U.S.A.), anti-CD9 (Sera-Lab, Crawley Down, United Kingdom) ,* anti-CDIO (Dakopatts, Glostrup, Denmark) ,* anti-CD14 (Becton Dickinson, Mountain View Calif) ,* anti-CD33 (Becton Dickinson) anti- VCAM-1 (Immunotech S.A, Marseille, France) or the following anti-CDl3 monoclonals MY7 (Coulter, U.S.A.), WM47 (Sera-Lab) or Leu-M7 (Becton Dickinson) . These antibodies were also used for cytofluorimetric testing of the different cell lines used.

Microcytotoxic assay

All sera were screened for the presence of cytotoxic antibodies directed against fresh monocytes and THP-1 cells. All positive monocyte-reactive sera were then serially diluted and further tested against the THP-1 cells. For confirmation of the results, human lung fibroblasts, mouse 3T3 cells and mouse cells expressing human CD13 (hAPN-3T3) were also used in some experiments as target cells.

Inhibition assay

In triplicates, 0.5 μl of a THP-1 cell suspension of 4- 5xl0 ⁶ cells/ml were incubated with the following noncytotoxic mouse monoclonal antibodies: anti-HLA class

I, anti-HLA class II, anti-CD33, anti-CD9, anti-CDIO, anti-CD14, anti-VCAM-l and positive and negative controls (Karuppan et al, supra) for 1 hour at 22°C. All tested antibodies were confirmed by flow cytometry to be directed against cellular antigens present on the THP-1 cells. The selected monocyte reactive sera from the cytotoxicity assay and doubling dilutions were then added to the cells and the standard cytotoxicity assay was performed as described above. The results of the inhibition of monoclonal antibodies were compared to the results with undiluted serum. Based on these experiments, the criterion chosen for positive inhibition was that cytotoxicity should be completely removed or reduced by at least two log ² dilution steps.

Preabsorhtion of human sera with mouse cells

Since human sera contain natural antibodies, all sera were preabsorbed with mouse N1H-3T3 cells prior to testing. The absorbed sera were nonreactive in flow cytometry assay against 3T3 cells.

Binding of human sera to mouse cells expressing human □211

For flow cytometric investigations, 5xl0 ⁵ mouse 3T3 cells or 5xl0 ⁵ mouse cells expressing human CD13 (hAPN-3T3) , were incubated with 50 μl patient's serum (preabsorbed to N1H-3T3 cells) for 30 min at 22°C. The cells were washed three times with PBS containing 0.1% sodium azide. 5 μl of fluoresceinated (FITC) F(ab) ' ₂ fragments of goat anti-human IgG+IgM antibodies (Jackson Immuno Research Laboratories Inc, U.S.A.) or goat anti-human IgG and IgM antibodies separately (both either from Southern Biotechnology Associates, Inc. Birmingham, U.S.A. or TAGO Inc., Burlingame, C.A.) were added and incubated at 4°C on ice in the dark for 25 min. The cells were washed, resuspended in PBS and then analysed

on a Becton Dickinson flow cytometer (FACStrak) . For staining of THP-1 cells, HL734, fresh monocytes, NIH-3T3 cells or hAPN-3T3 cells, a total of 5-10xl0 ⁵ cells in PBS were incubated with the primary antibody (anti-CD13, - CD10, -CD9, -CD14, -CD33, -VCAM-l, -HLA class I or HLA class II) for 30 min at 4°C, washed twice in ice cold PBS and incubated with a FITC conjugated rabbit anti-mouse F(ab) ' ₂ fragment for 25 min at 4°C in the dark. The cells were washed twice, resuspended in PBS and analysed in the cytofluorimeter (FACSort) .

Cytofluorimetric analysis

A fluorescence activated cell analyser (FACStrak or a FACSort, Becton Dickinson, Mountain View, California) with an argon laser producing 400 mW of light at 488 nm was used for all analyses. A positive signal was defined by Kolmogorow-Smirnov statistics as a greater than a 20-channel increase in the mean fluorescence for the serum or the monoclonal antibody as compared to the negative control.

Preparation of F(ah '- fragments for the determination of CD13 specific antibodies in human sera

To obtain F(ab) ' ₂ fragments of the mouse monoclonal antibodies directed against CD13, MY7 (IgGl) was digested with pepsin at a concentration of 4 mg/lOO mg of the IgG concentration for approximately 18 h at 37°C. The digested protein was then dialysed against a phosphate buffered saline containing 0.02% Na-azide over night at room temperature. The F(ab)' ₂ fragments were thereafter eluted after passing through a protein A column to remove any undigested protein or the Fc fragments. The purified F(ab) ' ₂ fragments were concentrated and used in a saturated concentration in the inhibition assay.

Blocking of antibody h n-iing to CD13 expressing mouse cells by F.ab) '. fragments of CD13 specific antibodies

5x10 ^s cells were incubated with the CD13 specific F(ab) ' ₂ fragments for one hour at 22°C. The cells were washed in PBS and then incubated with the sera and a FITC conjugated antibody was added as described above. The cells were resuspended in PBS, analysed in the FACStrak and histograms were generated similarly as for the untreated cells. Based on these experiments, the criterion chosen for positive inhibition was complete blocking or reduction by at least 20 channels in mean fluorescence compared to controls.

Preparation of cell membranes

Membranes of mouse 3T3 cells or mouse cells expressing human CD13 were collected by scraping monolayers of cells into PBS, pelleted followed by homogenization in PBS containing 8 μg/μl respectively of leupeptine, aprotinin, soybean trypsin inhibitor and 1 mM PMSF (all from SIGMA) . The homogenate was centrifuged, the pellet rehomogenized and centrifuged repeated times. The supernatants from low speed centrifugations were combined and ultracentrifuged at 30,000 rpm for 45 min at 4°C. The pellet was resuspended in PBS with protease inhibitors and stored at -20°C until use. The total protein concentration of the membrane preparations was estimated using the bicichoninic acid protein assay method with BSA as standard and using Pierce (Rockford, IL) reagents according to the manufacturer's instructions.

Immunoprecipitation

The samples used for immunoprecipitation were NP40- solubilized membrane preparations of mouse 3T3 cells or

mouse cells transfected with cDNA for human CD13 (hAPN- 3T3) . Approximately 50 μg total protein of the membrane preparations were solubilized in 300 μl immunoprecipitation buffer containing 1% NP40, 50 mM Tris (pH 7.4), 150 mM NaCl and 2mM EDTA for 30 minutes at 4°C. The sample was thereafter centrifuged at maximum speed and the cleared supernatant was transferred to a clean tube, mixed with 1 μl of the patient's serum, which had been preabsorbed by mouse 3T3 cells, and 50 μl of a 1:1 slurry (vol./vol.) of protein A-Sepharose CL4B (Pharmacia) was added to the tube. Incubation was performed first for 10 min on ice followed by an additional 16 h at 4°C with continuous rotation. The pellets were washed once in 0.2% NP40, 10 mM Tris-HCl (pH 7.5), 150 mM NaCl and 2 mM EDTA, once in 0.2% NP40, 10 mM Tris-HCl, 500 mM NaCl and 2 mM EOTA and finally twice in 10 mM Tris-HCl. The precipitates were solubilized by 5 min boiling in SDS-PAGE sample buffer and centrifuged before application to a Phast gel (prefabricated polyacrylamide gradient gels. 10-15%, Pharmacia-LKB, Uppsala: Sweden) . The gels were run, fixed and stained in a silver nitrate solution according to the manufacturer's instructions.

Immunoblot analysis

Equivalent amounts (100 μg) of 3T3 or hAPN-3T3 membrane proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) . For immunoblot analysis, the proteins were transferred electrophoretically from the gel to a nitrocellulose membrane (Hybond ECL- Western, Amersham International, United Kingdom) . The membranes were washed in PBS containing 0.3% Tween 20 and blocked using a 2% fish gelatine (Amersham Int. United Kingdom) solution in PBS/0.3% Tween 20 for 1 h at room temperature, and thereafter washed for 5 min three times in PBS/0.3% Tween 20 with continous agitation.

The membranes were cut into strips and incubated with the patient's serum (preabsorbed to NIH-3T3 cells) in dilution 1:20 in PBS containing 0.3 % Tween 20 and 0.25% fish gelatine for 20 h at room temperature. The membranes were washed as described before and incubated with a peroxidase (HRP) conjugated rabbit-anti-human immunoglobulin G or M (Dakopatts, Denmark) . Molecular weight standards used for calibration of the immunoblots contained a mixture of biotinylated proteins obtained from Bio-Rad (Richmond, CA.) which contained myosin (200-kDa) , β-galactosidase (116-kDa) , phosphorylase B (97kDa) and serum albumin (66-kDa) . The molecular markers were detected using an avidine-horseradish peroxidase conjugate, whereas the immunoreactive bands were identified with the ECL-substrate (Amersham Int.) according to the manufacturer's instructions. The membranes were exposed to a Hyperfilm-ECL (Amersham Int.) with intensifying screens for 10 sec at room temperature.

RESULTS

33 allogenic BMT patients were studied. During the course of the experiments, the investigators doing the laboratory assays had no access to specific clinical information about the patients included in the study. Altogether 19 out of 33 patients had developed CMV viremia or CMV DNAemia after BMT. Among these, 10/19 patients were diagnosed with CMV disease. Two patients had both CMV pneumonia and gastroenteritis, two patients only CMV pneumonia, one patient CMV gastroenteritis, one patient CMV encephalitis, and four patients had CMV syndrome alone.

Characteristics of the study population regarding presence of acute and chronic GvHD is shown in table 1. There was no significant difference between the

different patient groups regarding acute GvHD. However, there was a significant difference in the rate of chronic GvHD. Six of 10 patients with CMV disease had extensive chronic GvHD and one limited chronic GvHD. The remaining patients could not be evaluated for chronic GvHD, since they died early after BMT. Only one of 19 patients developed chronic GvHD in the other two groups (p=0.0002, Fisher's exact test) . Five patients could not be evaluated.

Detection of CD13 specific antibodies

All sera were screened for the presence of CD13 specific antibodies by using the different methods described above. Among the 33 BMT patients tested, sera from five patients showed cytotoxicity against the monocytic cell line THP-1. The results obtained for these sera were identical if fresh monocytes, human lung fibroblasts or hAPN-3T3 -cells were used as target cells. The specificity was determined by blocking of different cell surface molecules by monoclonal antibodies. The tested blocking antibodies were confirmed to react with the target cells by immunostaining and flow cytometric analysis, as described in the Materials and Methods (data not shown) . Figure l shows a serum from one patient with CMV disease, where both CD13 specific and CD9 specific monoclonal antibodies blocked cytotoxicity. None of the other monoclonal antibodies tested including anti-CDIO, -CD14, -CD33, -VCAM-l, -HLA class I- or -HLA class II antibodies, inhibited this reaction. The same inhibitory pattern was observed in all these patients, but the highest dilution of the different sera that resulted in cytotoxicity varied between patients (1:16- 1:1024) . If CD9 and CD13 specific antibodies were used together, an additative inhibitory effect was observed in sera from two patients (data not shown) . These data suggests the presence of both CD13 and CD9 specific

antibodies in sera from some patients with CMV disease. Since only IgM, IgGl and IgG3 antibodies are known to bind complement and therefore would be detected by this method, further analyses were needed to investigate whether other classes of CD13 specific antibodies were present. Each serum was then tested for binding to CD13-transfected hAPN-3T3 cells or the control 3T3 cells and analysed by flow cytometry. Blocking of CD13 with F(ab)' ₂ fragments of a CD13 specific antibody was performed with all positive sera to confirm the presence of CD13 antibodies (figure 3) . This method confirmed the presence of CD13 specific antibodies in four out of the five THP-1 reactive sera. The fifth patient had only CD9 specific antibodies. Sera from eleven additional patients were found to have CD13 specific antibodies reactive with CD13-transfected mouse fibroblasts. This sensitive method also gave us the possibility to distinguish between CD13 specific antibodies of IgG and IgM classes (figure 2a and 2b) . Only sera from three of the patients had IgG CD13 specific antibodies. CD13 specific antibodies of both IgM and IgG classes were also found in the convalescence serum from a patient with a primary CMV infection (figure 2c and 2d) . However, in none of the 24 control individuals could CD13 specific antibodies be detected (table 2) . Several of these individuals had THP-1 reactive cytotoxic antibodies, but no reactions were blocked by CD13 specific antibodies and no sera reacted with the hAPN-3T3 cells in flow cytometry following absorbtion with 3T3 cells.

To further substantiate our findings, we performed immunoprecipitation experiments on sera from all patients with CD13 specific antibodies and on sera from six negative patients. As a positive control we used monoclonal antibody directed against human CD13 (MY7) . This antibody precipitated a protein of 150 kD, which

agrees with the expected size of human CD13 (Look et al, supra) . In sera from 13/15 patients containing CD13 specific antibodies, we were able to detect a band on the Phast gel of approximately the same size. Yet another protein with a band size of approximately 100 kDa was precipitated by all of the sera. In four of the sera also another band of approximately 75 kDa was detected. These latter two bands were not found in the positive control sample precipitated by mouse-anti-human CD13 (MY7) (figure 4) . None of the six previous negative sera precipitated any proteins visible on the Phast gel. These data clearly indicate that the antibodies found in the CMV infected patient's sera, indeed were CD13-specific autoantibodies.

Sera from nine of the BMT patients were analysed further with the Western blot technique. By this method, we showed that IgM antibodies recognised a band of the same size as in the positive control, i.e. 150 kD (data not shown) . Again, the 100 kD band was detected in all these experiments and two of the four samples also reacted with a band of approximately 75 kDa (data not shown) . Sera from the three patients with both IgG and IgM specific antibodies (determined by flow cytometry) , were confirmed with the Western blot technique to contain both IgM and IgG antibodies that recognized the 150 kD band (data not shown) .

CD13 specific antibodies were exclusively found in patients that experienced CMV infection

The different groups of patients were classified according to their clinical CMV record. Patients transplanted before 1992 were screened for CMV with a rapid culture technique and after January 1, 1992, all patients were routinely screened for the presence of CMV DNA by PCR. CD13 specific antibodies were detected in

sera from 15 out of the 33 BMT patients, (always detected by at least two different methods) .

None of these patients belonged to the patient group without CMV disease and/or CMV viremia/DNAemia. Six of the nine patients with CMV DNAemia and nine of the ten patients with CMV disease developed CD13 specific antibodies (table 1) .

Only in one serum from a patient with CMV disease were we unable to detect CD13 specific antibodies. However, this patient died of liver failure early after BMT and only one serum sample was available for testing. There was a statistical difference in the frequency of CD13 antibodies between the CMV negative group and the viremia group (p=0.008) or the CMV disease group (p<0.000 1) . There was no difference in the frequency of CD13 antibodies between the viremia/DNAemia group and the CMV disease group. Of the three patients who switched to IgG production, one had CMV DNAemia and the other two had CMV disease.

CMV was always detected prior to the occurrence of CD13 specific antibodies

We were able to analyse between six and ten sera from some patients collected before and at different time points after BMT. CD13 antibodies could never be detected in sera collected before three weeks prior to the diagnosis of CMV infection. In all the three patients with IgG antibodies, IgM antibodies could be detected in earlier samples. The IgM antibodies disappeared and could not be detected in later sera from these patients. Data in table 3 show the kinetics of CD13 antibodies of IgM and IgG classes in one patient with CMV disease who developed chronic GvHD directly after the course of acute GvHD. The data show that IgM

CD13-antibodies were detected at the time of diagnosis of CMV DNAemia. Later IgG antibodies were formed as the patient developed extensive chronic GvHD. Of all cytotoxic antibodies in the patient serum samples, anti- CD13 blocked the major part of reactivity, indicating that no polyclonal non-specific activation was induced by CMV, but rather a response mainly to CD13. In the other 12 patients, only CD13 specific antibodies of the IgM class have been detected. In one of them CD13 specific IgM antibodies disappeared and were not followed by IgG antibodies.

CMV disease and CD13 specific antibodies predispose for the development of extensive chronic GVHD

Of 25 bone marrow transplant patients who could be evaluated for the development of cGvHD, seven had extensive cGvHD; of these six were CD13 autoimmune (p=0.0002, and shown in table 1) of whom all had experienced CMV disease. Four of the seven patients with extensive cGvHD received bone marrow from an HLA identical sibling. Conversely, 18 patients had no or limited chronic GvHD, seven of these patients had CD13 antibodies and only one had CMV disease. Among the ten patients with CMV disease, seven patients could be evaluated for cGvHD, all had CD13 antibodies, six extensive cGvHD. Only one patient lost CD13 reactivity; this patient had limited cGvHD.

In four of the six patients with extensive chronic GvHD, we were able to detect CD13 specific antibodies prior to the diagnosis. The other two patients had the progressive form of cGvHD and thus no clear interval between the acute and the chronic GvHD.

The immune response to CD13 (ie. CD-13 specific antibodies) was demonstrated in certain patients,

specifically those bone-marrow transplanted patients who developed chronic graft-versus-host disease. No CD13 specific antibodies were found in healthy individuals whether CMV seropositive or not or in patients with other autoimmune diseases or other viral infections.

Table 1. The relation between CMV infection, CD13 -autoantibodies and chronic GVHD

Number of Anti-CD13 Chronic GVHD Chronic GVHD Patients patients (absent/ (extensive) dying limited) <100 days*

No CMV infections 14

CMV viremia 9 or DNAemia

CMV disease 10

* patients who died before day 100 after BMT were not evaluated for chronic GVHD

Table 2. Studies of CD13-autoantibodies in controls

Patient category Number of Presence of Contribution by Reactivity against CD13 patients cytotoxic anti -CD13 reactivity CD-13 transfected cells antibodies (blocking of cytotoxicity) ( flow cytometry)

Healthy control 11 0 no no individual

SLE no no

HSV-l infection no no

EBV infection no no

Rheumatoid no no arthritis

Table 3. Kinetics of CD13 antibodies of IgM and IgG classes in a patient with CMV disease who developed extensive chronic GVHD directly after the course of the acute GVHD

Weeks prior to Cytotoxic Recidual Presence Presence Comments or post BMT antibodies titer after of IgM of IgG

(titer) CD13 blockade anti-CD13 anti-CD13

-19 +2 +4 Acute GVHD +5 o +6 CMV DNAemia +7 + +11 >1/512 ND + +14 >1/512 1/4 + Chronic GVHD +15 CMV pneumonia +25 1/1024 1/4 ND ND +27 1/1024 1 /4 + +30 1/1024 (ND) + +33 1/256 ( 1 /64 ) +

Example 5

The contribution of a TMV induced autoimmune response tn tissue damaσe in chronic σraft-versus-host-disease

The skin and oral mucosa are organs commonly attacked in chronic GVH disease. Cellular targets in these organs include keratinocytes, fibroblasts and endothelial cells, all of which are CD13 positive. In order to illustrate our proposal that autoimmunity is at least in part responsible for tissue damage associated with chronic GVH disease, we conducted further studies on certain of the patients in the study presented in Example 4, looking at the antibody profile in biopsies from skin and oral mucosa, and the binding properties thereof.

Methods:

Patients

28 samples from (11/33) patients either with cGVHD (6 oral mucosa and 8 skin biopsies) or without signs of cGVHD (6 oral mucosa and 6 skin biopsies) , and oral ucosal biopsies from two control individuals (obtained from the Department of Oral Pathology, School of Dentistry, Huddinge, Sweden) were studied. The patient biopsies were taken as part of the follow-up procedure after bone marrow transplantation (BMT) and were processed routinely for histopathological examination at the Pathology department.

Detection of IσG and iσM antibodies in biopsies from skin and oral mucosa

To investigate the presence of IgM and IgG antibodies, immunohistochemical staining was performed. Formalin fixed tissue, embedded in paraffin, was sectioned, deparaffinized in xylene, treated with H ₂ 0 ₂ in methanol,

rehydrated through a series of graded alcohols, and washed in PBS. Pronase treatment and non-specific blocking in PBS containing 5% BSA preceded immunostaining. The sections were incubated with rabbit anti-human IgG or IgM antibodies (both from Dakopatts, Denmark) . After washing, the specimens were incubated with a biotin-conjugated swine anti-rabbit antibody, followed by horse radish peroxidase-conjugated AB complex (Dakopatts, Denmark) . The presence of IgM or IgG immunoglobulins was detected using the DAB substrate (Sigma) . Sections were counter-stained in hematoxyline, washed, dehydrated and mounted before examination in a light microscope. The results were classified as 0, l+, 2+ or 3+ according to the intensity of the staining.

Detection of specific antibodies in patients' sera binding to normal skin biopsies

To test if antibodies in sera from patients with or without extensive cGVHD would bind to structures in normal skin sections, we obtained punch biopsies, 4 mm in diameter, from skin of the right buttock of three healthy individuals (one male and two females) . The biopsies were frozen in isopentane chilled by dry ice, 4 μm sections were cut, and placed on glass slides. Indirect immunofluorescence staining with sera from six patients diagnosed with extensive cGVHD, eight patients with limited cGVHD, four patients with no signs of cGVHD, two patients with SLE, and two healthy individuals, were performed on the skin biopsies. The sections were incubated with 50 μl of a ten-fold dilution of each serum for 30 minutes at room temperature. After washing, bound antibody was identified by FITC conjugated goat anti-human IgG or IgM antibodies (both from TAGO, Inc., Burlingame, CA, USA) . The biopsies were examined in a fluorescence microscope.

To investigate the specificity of the antibodies in the patients' sera, we performed blocking experiments. The specimens were incubated with a mixture of three CD13 specific monoclonal antibodies (MY7, WM15 and L138) for 1 hour at room temperature. Thereafter, the sections were assayed for the binding of the patients' sera as described above. In parallel experiments, CD13 or CD9 specific monoclonal antibodies bound to cells were detected with FITC conjugated rabbit anti-mouse F(ab) ' ₂ fragme ts.

To further substantiate our findings, non-specific antibodies were removed by absorption to mouse NIH-3T3 cells, whereas absorption of CD13 specific antibodies also were performed to mouse cells transfected with human CD13 (Look et al, J. Clin. Invest. jJJL, 1299- 1307 (1989)) . Briefly, 100 μl of three of the tested positive sera were incubated with approximately lxlO ⁶ cells for three hours two-four times. The sera were cleared by centrifugation and assayed for the binding to normal skin biopsies as described above.

Routine screening for autoantibodies

Analyses for the presence of autoantibodies in patients' sera were performed (1:10 dilution) with sera from five patients with extensive cGVHD, three patients with no signs of cGVHD, one SLE patient and one healthy individual. Anti-nuclear antibodies, anti-basal membrane antibodies, and anti-interstitial substance antibodies were tested against kidney, liver, ventricle (all rat) and esophagus (monkey) . Sera were also screened for the presence of rheumatoid factor. Bound antibody was detected using FITC conjugated sheep anti- human IgG+IgM+IgA antibodies (Swedish Institute for Infectious Disease Control, Stockholm, Sweden) .

_l______T_i

Antibodies in sera recognise certain cells in skin biopsies from normal individuals

To investigate if sera from these patients contained autoreactive antibodies, sera were tested for binding to cryosections of skin biopsies from normal individuals. A nuclear staining (Figure 5C,* this patient also had anti-nuclear antibodies) , a cell membrane related staining concentrated to the basal keratinocytes (Figure 5E) or a diffuse granular cytoplasmic staining (data not shown) in keratinocytes were seen. These epidermal staining patterns were always combined with a reaction in the papillary dermis (Figure 5C) as well as a strong reaction at basal membranes around eccrine sweat glands (Figure 8a) , whereas the endothelium only showed a weak reaction (data not shown) . In addition, cells in eccrine sweat glands were stained. All six patients, that had CD13 specific antibodies and later developed extensive cGVHD, showed a specific staining of the basal membranes around eccrine sweat glands (Figure 8a) . Weaker but similar staining pattern was observed by sera from CMV viraemic patients who developed limited cGVHD (data not shown) . None of the sera from three patients with limited cGVHD who lacked CD13 specific antibodies, sera from four patients without cGVHD, sera from two patients with SLE or sera from two healthy individuals, stained any of the described structures. A serum obtained from one of the SLE patients showed a distinct nuclear staining of keratinocytes, and one control serum demonstrated a diffuse granular cytoplasmic staining pattern of keratinocytes.

Antibodies in sera from patients with extensive r- VHD bind to the CD13 molecule in normal skin biopsies

The staining patterns of the papillary dermis, the basal membrane around eccrine sweat glands as well as the endothelial cells were found to be specific for CD13. Monoclonal antibodies against CD13 gave rise to two different staining patterns (Figure 6) . Clone WM15 (Figure 6C) and L138 (data not shown) labelled the papillary dermis, the basal membrane around eccrine sweat glands (Figure 6F) , and the endothelium (data not shown) , whereas clone MY7 showed an intense cell membrane related staining of the basal layer of epidermal keratinocytes (Figure 6B) . In addition, MY7 demonstrated a similar staining of basal membranes around eccrine sweat glands (Figure 6E) as was found with the other monoclonal antibodies against CD13. These three antibodies and two additional CD13 specific monoclonal antibodies were therefore characterised further by flow cytometric analysis for binding to mouse NIH-3T3 cells, to mouse NIH-3T3 cells expressing human CD13, and to human lung fibroblasts, which are known to express high levels of CD13. The data obtained showed no binding of MY7 antibodies to mouse cells, but high levels of binding to the two types of CD13 expressing cells (Figure 7) . The same results were obtained using three other CD13 specific monoclonal antibodies (WM47, 3D8 and L138, data not shown) . However, clone WM15 showed similar levels of binding to HL and mouse cells expressing human CD13 as the other CD13 monoclonals tested, but also recognised a protein on mouse NIH-3T3 cells, suggesting that the CD13 molecule may exist in different isoforms present on different cells. Therefore, we suggest that WM15 recognises a, yet unidentified, mouse molecule which might be a CD13 equivalent. Staining the skin biopsies with our control antibody against CD9 resulted in an interstitial staining of the epidermis which demonstrated a completely different pattern compared to what was seen with the CD13 monoclonal antibodies (data not shown) .

To block the CD13 epitopes in the skin biopsies, we used a mixture of unconjugated mouse monoclonal antibodies to human CD13 (MY7, WM15 and L138) . In all three of the tested positive sera, preincubation of the skin biopsies with monoclonal antibodies against CD13 significantly decreased the binding of the serum antibodies to the structures (Figure 8B) . In addition, preabsorption of patients' sera containing CD13 specific antibodies to mouse cells expressing human CD13 abolished specific binding of serum antibodies to skin biopsies (Figure 8C) .

Due to suspected similarities between the CD13 staining pattern seen in the papillary dermis in skin biopsies and a pattern expected when staining for elastin, formalin fixed skin biopsies were stained for elastin with polychrome metyleneblue according to Weigert (Zbl. allg. Path. Path. Anat. (1898), 2., 289) and analysed in a light microscope. Completely different staining patterns were observed for elastin and CD13 monoclonals or patients' sera containing CD13 specific antibodies (data not shown) .

Differences in I9M deposits in skin and oral mucosa biopsies from patients with cGVHD compared to patients without signs of cGVHD

All skin and oral mucosa biopsies examined for the presence of IgG showed high non-specific staining of different cell types in epidermis, dermis and the adnexa. Therefore, no significant differences in IgG deposits was found in patients with extensive cGVHD or in patients without any signs of cGVHD. It is a well- known problem that most skin and oral mucosa biopsies also from healthy individuals demonstrate this non¬ specific staining pattern.

However, skin and oral mucosa sections, when evaluated for in vivo IgM deposits, showed a difference in intensity of the staining when comparing patients diagnosed with extensive cGVHD to patients without cGVHD. In biopsies from patients with extensive cGVHD, the subepithelial stroma and the endothelial cells in skin (Figure 9) , and the basal membranes around blood vessels and salivary glands in oral mucosa (data not shown) were strongly (2+ or 3+) stained. Lower (1+) intensity or no (0) staining was observed in biopsies taken from patients without cGVHD or from controls. The difference of staining intensity in the subepithelial stroma between the two patient groups, was not as significant in the skin as in the oral mucosa biopsies, which was in concordance with the clinical localisation of cGVHD. However, when analysing biopsies containing endothelial cells in this small patient material, all six patients with cGVHD had 2+ or 3+ staining intensity in skin biopsies and all four patients analysed had 2+ or 3+ staining in oral mucosa biopsies. This was co-mpared to patients or controls without cGVHD where 2/5 had 2+ or 3+ staining intensity in skin and 1/4 had 2+ or 3+ staining in oral mucosa biopsies.

Few patients had autoantibodies detected by routine immunological analysis

Analyses for the presence of autoantibodies were performed. One patient had anti-nuclear antibodies, two patients and two controls had low titres of antibodies to intercellular substances, whereas no sera contained anti-basal membrane antibodies. One patient had rheumatoid factor.

No staining pattern similar to the one obtained by sera containing CD13 specific antibodies, was observed in the animal biopsies, nor did monoclonal antibodies against

human CD13 recognise any structures in the animal sections (data not shown) .

We have demonstrated deposits in vivo of IgM antibodies in skin and oral mucosa biopsies. Unfortunately, as stated above, specific staining for IgG was not possible due to non-specific binding of the anti-IgG reagents. The surface epithelium, subepithelial stroma, endothelium, and the basal membrane surrounding the vessels and the eccrine sweat glands showed particularly high deposition of IgM antibodies. These locations together with epithelial cells in the intestine, and in the bile cannaliculi of the liver are often affected in cGVHD. We have also demonstrated that all affected tissus structures in the skin express high amounts of CD13. Binding of CD13 specific antibodies to cells and structures expressing high levels of CD13 correlates with and contributes to the tissue damage and the chronic inflammation typical of cGVHD.