WO1993016725A1

WO1993016725A1 - Early antigen for autoimmune diabetes

Info

Publication number: WO1993016725A1
Application number: PCT/US1993/001716
Authority: WO
Inventors: C. Garrison Fathman; Lisa Paborsky; Cohava Gelber
Original assignee: The Board Of Trustees Of The Leland Stanford Junior University; Immulogic Pharmaceutical Corporation
Priority date: 1992-02-27
Filing date: 1993-02-25
Publication date: 1993-09-02
Also published as: CA2130997A1; AU3778993A; EP0627934A4; JPH07508503A; EP0627934A1

Abstract

Antigens associated with the β islet cells that activate T-cell populations in inducing type I diabetes are disclosed. These antigens are useful in diagnosis and therapy of type I diabetes.

Description

EARLY ANTIGEN FOR AUTOIMMUNE DIABETES

This work was supported in part by a grant from the National Institutes of Health. The U.S. Government has certain rights in this invention.

Technical Field

The invention relates to factors useful in diagnosis and therapy of type I diabetes. More specifically, it concerns markers in β islet cells that mediate T-cell proliferation and stimulation associated with induction of the disease.

Background Art

Type I diabetes is known to be a chronic auto- immune disease that manifests itself early in life. This condition, also called insulin-dependent diabetes mellitus (IDDM) is considered a hereditary disease although nonfamilial forms also occur. The nature of the genetic control of type I diabetes has been studied in detail. It appears clear that genes associated with the major histocompatibility complex (MHC) are involved in the hereditary transmission of the disease (Wicker, .S. et al., J Exp Med (1987) 165.:1639-1654) . Several groups have demonstrated a correlation between the absence of an aspartic residue at position 57 of the /3-chain in the human HLA-DQ class II MHC encoded glycoprotein and the occurrence of the disease (Parha , P. Nature (1990) 24 :662-664; Todd, J.A. et al., Nature (1987) 3_29_:599- 604; Morel, P.A. et al., Proc Natl Acad Sci USA (1988) £5_:8111-8115) . The most ubiquitously used model for the human form of type I diabetes is the nonobese diabetic (NOD) mouse model. The NOD strain of mice was originally selected by Makino, S. et al., Exp Anim (Tokyo) (1980) 2-2.:1, for its predisposition to develop a spontaneous form of diabetes. The disease as it manifests itself in this model is similar to that found in humans. There is a progressive infiltration of the pancreatic islets by mononuclear cells, mostly of T-cell origin, long prior to the onset of the symptoms of the disease which leads to the destruction of the insulin-producing S-cells. This occurs in the similar BB rat model system (see below) as long as 8 weeks before the onset of IDDM. It has been demonstrated that the NOD disease is controlled by at least three genes, one of which segregates with the MHC (Hattori, M. et al., Science (1986) 231:733: Prochaska, M. et al., Science (1987) 237d:286: Wicker, L. et al., ζ∑ Exp Med (1987) 165:1639 (supra)).

In the course of the disease, the insulin- producing β-cells of the pancreatic islets are selectively destroyed; other endocrine islet cells, α cells, such as those that produce glucagon or somatostatin are unaffected. The destruction of the β- cells is mediated in part by T-cell proliferation, in an induction phase, resulting in the production of effectors of the 0-cell destruction, such as cytotoxic antibodies, natural killer cells, macrophage and lymphokines (Castano, L. et al., Ann Rev Immunol (1990) £:617-679) . Thus, in general, the onset of type I diabetes is believed to involve an induction phase common to both normal and autoimmune responses, which involves an initial activation of helper CD4⁺ T-cells by engagement of the T-cell antigen receptor (TCR) with MHC class II associated peptides presented on antigen-presenting cells (APC) . This activation results in secretion of ly phokines which in turn activate the effector β-cells, cytotoxic T-cells, natural killer cells, macrophage, and the like. Indeed, it has been shown that removal of the CD4⁺ T lymphocytes from the circulation using an anti-CD4 antibody (GK1.5) blocks the development of type I diabetes in the murine model system (Shizuru, J.S. et al., Science (1988) 240:659-662) .

The involvement of T-cells in the progress of type I diabetes has also been verified by showing that autoimmune diabetes can be transferred to nonsusceptible strains using splenocytes from NOD mice (Wicker, L.S. et al., Diabetes (1986) 3_5_:855-860) ; and the showing that autoreactive T lymphocytes can cause diabetes in NOD mice (Reich, E.-P. et al., Diabetes (1989) 11:1647-1651). Pancreatic islet-specific T-cell clones have also been prepared from NOD mice by Haskins, . et al., Proc Natl Acad Sci USA (1989) ££:8000-8004 and Nakano, N. , et al., J Exp Med (1991) 173:1091-1097.

Using the NOD mouse model, Lehuen, A. et al., Immunol (1990) 144:2147-2151. demonstrated that the humoral anomalies in the NOD strain are disconnected from the occurrence of diabetes and insulitis. Nevertheless, attempts have been made to identify the antigen associated with type I diabetes by identifying proteins that bind to autoantibodies present in IDDM-affected subjects. All of these antigens are thus present at the effector level, and may or may not be associated with the progress of the condition.

Baekkeskov, S. et al., Science (1984) 224:1348- 1350, showed that antibodies to a 64 kd islet cell protein were present in BB rats (which spontaneously develop a type I diabetes analogous to that in humans) for as long as eight weeks before the onset of IDDM, and suggested that the antibodies could be used to predict an immune reaction against pancreatic 0-cells. The putative autoantigen was subsequently characterized by this group to be a glutamic acid decarboxylase (GAD) which is involved in the synthesis of γ-aminobutyric acid (GABA) , an inhibitory neural transmitter (Baekkeskov, S. et al., Nature (1990) 347:151-156) . In addition, a molecule cross-reactive with the 65 kd heat shock protein (hsp65) of Mycobacterium tuberculosis was shown to be a 0-cell antigen immunoreactive with diabetes-associated autoantibodies in the NOD mouse model by Elias, D. et al., Proc Natl Acad Sci USA (1990) £7:1576-1579. These authors suggested that the hsp65 antigen could be used to induce diabetes or to vaccinate against diabetes, using the principle that proteins administered in adjuvants tend to be immunogenic, while the same proteins in soluble form are tolerogenic. Indeed, administration of hsp65 in immunogenic or nonimmunogenic forms was reported to produce results consistent with this principle.

Castano, L. et al. in an unpublished manuscript, described the use of antibodies from a prediabetic patient as probes to screen a rat islet λgtll expression library and identified a colony which produced a protein immunoreactive with these antibodies. This colony was found to contain an insert which encodes a 136 amino acid fragment of carboxypeptidase-H (enkephalin convertase) . Further, the sera used to screen the library were shown to react with a 52 kd antigen corresponding to the molecular weight of the membrane form of carboxypeptidase-H.

In addition to the 64 kd GABA carboxylase, the hsp65 antigen, and the 52 kd carboxypeptidase-H, which react with autoantigens associated with type I diabetes, it is also known that autoantibodies to insulin are present in subjects affected by this condition.

Human T-cell clones which are specific for insulinoma cell antigens have been prepared by Von Vliet et al., Eur J Immunol (1989) 19_:213-216. These T-cell clones were prepared using rat insulinoma membranes as antigen, and generating the cell lines from peripheral blood mononuclear cells of patients with recent onset of the disease. In addition, it has been shown that T-cell clones from a type I diabetes patient respond to an integral membrane component of the insulin secretory granules which has been purified 5000-fold and shown to have a monomer of molecular weight 38 kd (Roep, B.O. et al., Nature (1990) 145.:632-634) .

All of the foregoing antigens are believed to be associated with the effector phase of type I diabetes. None of the foregoing has been demonstrated to be associated with the initial activation of CD4⁺ T helper cell induction phase. These antigens are found after the onset of symptoms of IDDM.

Thus, in all of these extensive studies, only antigens which are evidenced by factors present after the onset of the disease or very shortly preceding it have been identified. The present invention provides antigens which are responsible for the very early inductive stages in the development of this condition. These antigens are useful in diagnosis and design of immunotherapy.

Disclosure of the Invention

Antigens associated with the initial T-cell induction phase leading to the onset of type I diabetes have now been identified. These antigens are useful for the early diagnosis of the development of the disease and also can be used in immunotherapy of susceptible subjects.

In one aspect, the invention is directed to an antigen having a molecular weight in the range of 30-60kd normally present in mammalian, particularly human or murine, pancreatic islet 0-cell cytosol and/or membranes, and to antibodies or T-cells specifically immunoreactive with this antigen. Human proteins with this activity have molecular weights of about 37 kd, 41 kd and 51 kd, as determined by HEPC; murine proteins have molecular weights of about 36 kd, 42 kd and 55 kd as determined by gel electrophoresis. In other aspects, the invention is directed to methods to diagnose susceptibility to type I diabetes by assessing subjects for the presence or absence of antibodies or T-cells responsive to this antigen. In other aspects, the invention is directed to methods to prevent the onset of type I diabetes by rendering subjects unresponsive to the antigen of the invention and by blocking immune response to the antigen using peptide subunits thereof alone or in combination with immune modulating agents.

Brief Description of the Drawings

Figure 1 is a bar graph showing the results of a T-cell proliferation assay with respect to various insulinoma extracts.

Figure 2 is a bar graph which shows the effect of the presence and absence of antigen-presenting cells on T-cell proliferation.

Figure 3 shows a nitrocellulose blot of one-dimensional gel electrophoresis of insulinoma membrane proteins assayed for the ability of these fractions to stimulate the proliferation of T-cells from NOD mice.

Figure 4 shows an elution pattern of high performance electrophoretic chromatography (HPEC) performed on a murine insulinoma extract assayed with respect to T-cells isolated from NOD mice of various ages. Figure 5 shows the HPEC elution pattern of murine insulinoma whole cell extracts as assayed using NOD mice T-cells.

Figure 6 shows the distribution of T-cell proliferation stimulation activity between membrane and cytosol fractions of a murine insulinoma.

Figure 7 shows the results of two-dimensional gel electrophoresis of a murine insulinoma extract as assayed for the ability to stimulate proliferation of T-cells obtained from 30 day old NOD mice.

Figure 8 shows the HPEC elution pattern of the whole cell extracts of human islets as assayed by the response of NOD splenocytes.

Figures 9A and 9B show the analysis of antigen- containing extracts separated by molecular weight distribution using the T-cell clones LN-7 and LN*CM respectively.

Figures 10A and 10B show stained pancreata from a control mouse and from a mouse administered the T-cell clone LN-7.

Modes of Carrying Out the Invention

The invention provides antigens associated with early events connected with the inductive phase of type I diabetes. As such, the antigens of the invention offer the opportunity for early screening of individuals developing the disease and offer the opportunity for intervention to prevent its development and/or perpetuation. The antigens of the invention are obtainable from mammalian, such as the exemplified murine or human, pancreatic islet β-cells or cell lines derived therefrom. Cross species reactivity of the relevant antigens has been demonstrated, and there appears to be significant homology between the murine and human antigens. Thus similar antigens are believed present and isolatable from β-islets of other mammals, such as bovine, porcine, ovine, feline and the like.

The antigens may be obtained from either the membrane fraction or the cytosol fraction of islet β- cells or their derived cell lines, or the related neuroblastoma or other neuroendocrine cells. The antigens are present in both non-susceptible individuals and in individuals susceptible to type I diabetes. Therefore, the cells used as a source of the antigens need not be derived from affected subjects.

The antigens may initially be prepared by extraction and fractionation of the native material from the 0-cells or their derived cell lines, but may more conveniently be prepared using recombinant techniques from the encoding DNA. Provision of the native protein in purified and isolated form permits the design of probes useful for retrieval of the cDNA encoding these antigens. In addition, expression libraries obtained in, for example, λgtll and transformed into iL. coli can be screened for the ability of the expression products to effect proliferation of T-cells obtained from murine models for type I diabetes, such as NOD mice. Antibodies immunoreactive with the antigens may also be used to screen expression libraries.

Preparation of Antigens

The antigens of the invention may be prepared by suitable extraction of pancreatic β islet cells from mammalian subjects, especially human or murine subjects, or from cell lines derived therefrom. The antigens are -present in the cytosol and/or membrane fractions of β- cells of both type I diabetes-susceptible subjects and subjects not susceptible to IDDM. In addition, tumor cell lines derived from jS-cells such as insulinomas may be used as starting materials. It has also been shown hereinbelow that at least one of the antigens is present in a human neuroblastoma cell line and human islets. Extraction is generally conducted by homogenizing the 5 cells in the presence of appropriate membrane buffer at about pH 7-8; removing the cellular materials by centrifugation, recovering the supernatant, and then centrifuging the supernatant at high speed to obtain the membranes, as the pellet and the cytosol as the

10. supernatant. Whole cell extracts may also be used. A general procedure for preparation of membranes is described by Fava and Cohen, J Biol Chem (1984)259:2636- 2645.

The extract is preferably then subjected to

15 separation using generally known techniques, including gel filtration, anion exchange chromatography, polyacrylamide gel electrophoresis, and other standard procedures. Active fractions are recovered.

The fractions may be assayed for activity by

20 assessing their ability to effect the proliferation of T- cells obtained from type I diabetes-susceptible subjects, including the NOD mouse or T-cell lines derived therefrom, using a standard T-cell proliferation assay. One such assay utilizes labeled thymidine incorporation

25 as a measure of proliferation and is generally conducted as follows:

The T-cell preparation is obtained from single- cell suspensions from spleen, lymph nodes or PBL taken from naive (untreated) NOD mice. Dead cells and red

30 blood cells are removed by Ficoll gradient centrifugation by spinning in the gradient for 25 minutes at 2500 rpm at room temperature. This provides an enriched lymphocyte population and contains antigen presenting cells (APC) . The T-cells and APC are plated at 0.25-0.5 x 10⁶/well in

35 96-well U-shaped plates (Costar) and incubated with the sa ple to be tested at varying concentrations for 72 hours at 37°C, 5% C0₂• The plates are then pulsed with 1 μCi of tritiated thymidine per well and incubated for an additional 16 hours. The cells are then harvested using a microcell harvester (Skatron) and counted. Enhanced uptake of the labeled thymidine indicates increased cell proliferation. In lieu of the foregoing, cloned T-cell lines can be used in the assay, along with separately added APC. The recovered fractions may then further be purified using standard protein purification techniques. A particularly useful technique which results in virtually pure protein is two-dimensional gel electrophoresis. The fractions or extracts are adjusted toll Triton X-100,.15% glycerol and 6% a pholines, pH 3-10. Isoelectric focusing is then performed according to the method of O'Farrell, P.H. J Biol Chem (1975) 250:4007-4021. The resulting one-dimensional separation is then followed by loading each region.onto a 10% SDS-PAGE gel and electrophoresis conducted according to Laemmli, U.K. Nature (1970) 227:680-685.

The fractions identified are then recovered in purified and isolated form. They can be further characterized by determination of amino acid sequence and by retrieval of the encoding gene from cDNA libraries prepared by reverse transcription of β-islet mRNA or from a genetic library. The DNA libraries are prepared using standard techniques, and can be screened using probes designed on the basis of total or partial amino acid sequence of the recovered antigen. The recovered DNA can, in turn, be used as a probe to recover DNA encoding the corresponding antigens in other species.

In addition to recovering the gene(s) encoding the antigen(s) using DNA probes, the starting cDNA library may be prepared as an expression library in, for example, λgtll, and the library then screened using techniques which detect the synthesized antigen. Two means for screening the library to detect the antigen produced are particularly preferred. In one method, antibodies prepared against the isolated antigen can be used in screening the library in conventional techniques. In a second method, the library may be screened by assessing the ability of each of the clones contained therein to produce an antigen which stimulates proliferation of T-cells obtained from type I diabetes-susceptible individuals, including NOD mice. Those colonies that produce proteins capable of stimulating this proliferation contain the gene encoding the stimulatory antigen. Upon identification of one or more clones containing a DNA encoding the antigen, as described above, DNA can be isolated from the clones and the relevant inserts sequenced and/or recovered and used in the subsequent production of the antigen. The sequenced DNA and/or its degenerate coding forms can be partially or completely independently synthesized for use in such expression systems.

For recombinant production of the antigens, the encoding DNA is ligated into expression systems compatible with a convenient host. A wide variety of host systems and control sequences operable in said host are now available in the art. Suitable hosts include prokaryotes such as £__•. coli and eucaryotes. such as yeast, avian cells, insect cells, mammalian cells, and plant cells; more recently, whole plant or animal organisms have also been used. Techniques for constructing expression systems and transforming appropriate hosts with the constructed systems are by now standard in the art. For production of the desired antigen, the reco binant host cells transformed with an expression system containing the gene encoding the antigen operably linked to control sequences are cultured under conditions which permit the expression of the encoding DNA, and the antigen is recovered from the culture using standard procedures. Construction systems may be employed which result in secretion of the antigen, which can then be recovered from the medium, or the antigen may be produced intracellularly, which will necessitate lysing the host cells.

The β-islet cell antigens of the invention, when isolated and characterized, provide sequence information for identification of peptides associated with interaction with the TCR on the helper T-cells. These peptide segments are contiguous sequences of at least about 7 amino acids that are^' associated with the MHC class II glycoprotein present on the surface of antigen-presenting cells, resulting in a complex that interacts with the TCR. The relevant peptides can be systematically identified by testing overlapping regions of the sequenced antigen in the presence of antigen- presenting cells in T-cell proliferation assays conducted as described above. Techniques for such a screen are included in the report by Hickling, J.K. et al., Eur J Immunol (submitted 1991, in press) . Peptides with modified structures can then be designed which retain their ability to complex with the MHC class II glycoprotein but fail to effect reaction with the TCR by assessing the ability of these modified peptides to inhibit the T-cell proliferation in the presence of known activators in this assay. Peptide modifications include extensions, deletions and substitutions, and combinations thereof. Peptides that inhibit proliferation are successful candidates. Assay

Several forms of a convenient assay are described herein, depending on whether the assay is directed to a purified antigen, a T-cell clone, or an antigen preparation. In general, all assays may be performed in microtiter wells wherein each well contains 10,000-30,000 T cells, about 10⁵-10⁶ histocompatible antigen-presenting cells, and the antigen (for crude extracts this amounts to about 0.3-20 μg/ml) . However, for more purified antigen, less antigen would be required; for splenocyte preparations used as a source of T cells, the APCs are already contained in the original preparation. Variations of the nature and amounts of T cells, histocompatible APCs, and antigen will be understood as routine matters of optimization and experimental design. If tritium uptake is to be used as a measure of proliferation, a convenient protocol is incubation for 72 hours in a suitable culture medium such as RPMI-1640 supplemented with 5% FCS, 10 U/ml 10-strep, and 200 μM L-glutamine, or other appropriate medium for T-cell proliferation as is understood in the art. The cell culture is then pulsed with, for example, 1 μCi/well of tritiated thymidine and the cells are harvested 6-16 hours later and counted. Other methods of measuring T-cell proliferation and variations of the foregoing protocol are known in the art.

Generation of T-Cell Clones Spleen or lymph node cells from naive female

30-40 day old NOD mice were stimulated in vitro with insulinoma (B23720, see below) antigen extracts or irradiated insulinoma cells. Either whole cell extracts or the extracts from membranes or cytosol were used at a protein concentration of lOμg/ml. Alternatively, irradiated insulinoma cells (5000 cells/ml) in RPMI medium containing 1% NMS, Pen-Strep, glutamine and Iμg/ml Leuko-A. can be used.

The spleen or lymph node cells used as a source of T-cells were incubated at 37°C, 5% C0₂ for 3-4 days, then washed and resuspended in RPMI medium containing 10% FCS Pen-Strep, glutamine and 15% rat Con-A supernatant or 20 U/ml rMIL-2 (Genzyme) (complete medium) , and incubated for 5 days. Stimulation of the T-cell lines was repeated 2-3 times in cycles of 8-9 days: (3-4 days with the antigen in the above "complete" medium and 5 days in complete medium + rIL-2 at 20 U/ml without antigen) . Single cell cloning of the T-cell line was performed in complete medium in the presence of 10 μg/ml antigen (or 5000/ml live irradiated insulinoma cells) with 0.5x10^ irradiated spleen cells (as APC) in 96 well plate (Costar flat bottom 1/2 area) . Seven days later the clones were restimulated again with the antigen + APC as above. 10-14 days afterwards, the wells were scored for positive growth. Growing T cell clones were transferred to 24 well Costar plates and expanded.

The cells were tested for proliferation using 20,000 T cells with 0.5 x 10⁶ irradiated splenic (2000 R) antigen-presenting cells in an assay based on labeled thymidine incorporation. Cultures of the T-cells and APC cells were set up in triplicate in the presence of antigen extracts (0.3-20 μg/ml) and incubated for 72 hours, then pulsed with 1 μCs/well of tritium-labeled thymidine (Amersham-, Inc.) and harvested 16 hours later. The incorporated radioactivity was determined using a β- plate scintillation counter and results expressed as mean cpm of incorporated thymidine. Standard deviations were less than 10%. A number of T-cell clones were obtained, including LN7 and LN*CM T-cell clones. T-cell clones were picked from the wells, restimulated as above, and expanded for specificity tests.

Antibody Production

The antigens including protein antigens of the invention and the relevant peptide fragments can be administered to mammalian subjects in standard immunization protocols to prepare antibodies specifically immunoreactive with the antigen or peptide subunit thereof. Techniques for conferring immunogenicity on peptide subunits by conjugation to carriers is well known in the art. The protein or carrier-coupled peptide is injected into a suitable subject, preferably in the presence of adjuvants, and the progress of immunization can be monitored by detection of antibody titers in the plasma or serum. Standard ELISA or other immunoassays may be used with the immunogen as antigen to assess the levels of antibodies. The antisera are separated from the red blood cells and can then be used as polyclonal preparations or antibody-secreting cells from the immunized host may be immortalized using standard techniques to obtain cell lines that secrete monoclonal antibodies immunospecific for the antigens or subunit thereof. The antibodies per se may also be separated from other plasma proteins. In addition to the polyclonal and monoclonal preparations, immunologically reactive and specific fragments such as Fab, Fab', and the like, are also useful in immunoassay techniques and are included in the antibodies of the invention. These antibodies or fragments may be purified using standard techniques. Procmosis for Onset of Type I Diabetes

The antigens of the invention can be used in early detection of subjects who are developing type I diabetes. The antigens themselves are present in both normal and susceptible individuals; however, only those individuals developing IDDM produce helper T-cells whose proliferation is stimulated by the presence of these antigens. Thus, in one approach to early prognosis, T-cells obtained from the subject to be tested are used in the standard T-cell proliferation assays described hereinabove to test for their response to the antigens of the invention. Subjects whose T-cells proliferate in these assays are developing the disease.

It is not at present known whether these individuals produce antibodies to the early antigens of the invention; availability of these antigens will permit the presence or absence of these antibodies to be ascertained. In the event that subject individuals are shown to produce antibodies prior to the onset of the condition, the use of the antigen to detect the presence or absence of these antibodies can also be used as a prognostic tool. Standard immunoassays for the detection of the presence or absence of these antibodies are well known, and a variety of protocols involving a variety of labels can be used.

Therapy

The availability of the antigens of the invention offers the possibility of novel therapeutic methods for blocking the development and preventing the onset of type I diabetes. Current strategies are extremely aggressive and employ immune system depressants in general. A more benign therapy is offered by the invention antigens which can be administered under non- immunogenic conditions that render the subject unresponsive to the antigens rather than eliciting an immune response thereto. General techniques for administration of tolerizing doses of antigens or relevant peptides thereof are known in the art, including introduction in the absence of adjuvant and administration in soluble form. Use of the identified peptides that form complexes with MHC class II glycoprotein on the surface of antigen presenting cells and activate T-cells may also be employed for inducing unresponsiveness, and administration of small peptides may be more amenable to this technique.

In addition to the known routes of injection subcutaneously, intravenously, or intraperitoneally, the antigen or peptides of the invention may be administered using ot'her modes of formulation suitable for peptide- containing compositions including transmucosal and transdermal forms of administration, and when properly formulated, by oral dosage. Suitable formulations that include pharmaceutically acceptable excipients for introducing intact peptides or proteins to the bloodstream by other than injection routes (as well as by injection) can be found in Remington's Pharmaceutical Sciences, (latest edition) Easton, PA. In particular, pumps which provide the active ingredient in situ may be used.

The antigens of the invention may be administered alone or in concert with anti-CD4 antibodies or other CD4 blockers and/or other immune modulating substances. This approach to conferring tolerance is disclosed in U.S. Patents 4,681,760 and 4,904,481. In this approach, the antigen and the anti-CD4 antibodies or immunoreactive fragments are administered concomitantly. By "concomitant" administration is meant within a time frame which permits the anti-CD4 component to block the helper T-cell response to the antigen. The nature of "concomitant" in this sense is described in the above- referenced U.S. patents, incorporated herein by reference.

Finally, therapeutic methods that utilize modified peptides that behave as antagonists capable of binding the MHC class II glycoprotein but resulting in a complex which is not interactive with the T helper cells can also be used. Modes of administration and formulation for these peptides are similar to those described above.

As the antigens of the invention provide a relatively benign mode of intervention to prevent the onset of IDDM, the methods of prognosis also provided may be employed as a screening tool applied universally to infants and/or children. The assay methods described herein require only a small blood sample; this provides a relatively noninvasive screen. Individuals who test positive by virtue of the ability of their T-cells or antibodies to respond to the antigens of the invention can then be treated as described above to prevent the progression of the disease.

The following examples are intended to illustrate but not to limit the invention.

Example 1

Ability of fl-cell Extracts to Effect

Proliferation of NOD T-Cells Test extracts were prepared using the method of Fava and Cohen, J Biol Chem (1984) 259:2636-2645. Briefly, confluent cell layers were washed three times with phosphate buffered saline in the absence of calcium and magnesium ion. The remaining liquid was aspirated after the' last wash and membrane buffer which consists of 20 mM HEPES, pH 7.5, 1.5 mM MgCl₂; 1 mM EGTA; 1 mM PMSF and 1 μg/ml Leupeptin was added. The cells were scraped into the buffer and homogenized (Dounce) and the hqmogenate was centrifuged at 2500 rpm in a Sorval S31 Centrifuge for 10 minutes at 4°C. The pellet was discarded. The supernatant was used as a "whole cell extract" or centrifuged for 30 minutes at 48,000 xg to separate the membrane from the cytosol. The membrane- containing pellet was suspended in 20 mM HEPΞS and frozen at -70°C. The cytosol supernatant was also retained for testing. Similar extractions to obtain membrane fractions and cytosols were conducted with respect to the murine insulinoma cell line B23720 the glucagonoma pancreatic α-cell line and the human neuroblastoma line SY5Y (Goya, L. , et al., Neurochem Res (1991) 1£:113-116) . The cytosol fraction of the insulinoma cells was further subjected to high performance electrophoretic chromatography (HPEC) using the Applied Biosystems, Inc. HPEC 230A system. (The extracts are believed to contain proteins derived from membrane as well as cytosol, and contain 400 μg total protein as subjected to HPEC.)

For HPEC the extracts are adjusted to a final concentration of 7.5 mM tris-phosphate pH 7.5, 0.25% SDS, 15% glycerol, and then loaded onto a 10% SDS-tris phosphate tube gel (3.5 x 10 cm) and electrophoresed in a tris-phosphate buffer system. The proteins were eluted from the bottom of the gel into 7.5 mM tris-HCI, pH 7.5; assayed for protein concentration and analyzed on a 12.5% SDS polyacrylamide gel. 80% of the protein was recovered in these fractions. These fractions that induced the proliferative response in the assay described below were pooled. These fractions corresponded to a molecular weight range of 30- 60 kd and are labeled C-pool I in Figure 1. The C-pool I and the corresponding pools obtained from the "cytosol" extracts (also containing some membrane portion) of the ^■ or-cells and human neuroblastoma cells (labeled α-pool and NB-pool in Figure 1, respectively) were compared using the T-cell proliferative assay wherein T-cells were prepared from the spleens of unprimed NOD female mice of 3-30 days of age. The T-cell preparation was obtained as described hereinabove wherein single cell suspensions were prepared, in this case from spleen and lymph nodes, and clarified by removing dead cells and red blood cells using Ficoll gradient centrifugation. The resulting preparation contains a full range of T-cell types, as well as antigen presenting cells.

As described above, the lymphocyte preparation at each age was plated at 0.25-.5 x 10⁶ cells per well in 96 well U-shaped plates. The lymphocytes were obtained from unfractionated (untreated) spleen and contained B- cells and macrophages as a source for APC. The splenocytes for each age group represented a pool of 3-8 mice. The wells were incubated with the protein fractions (10 μg/ml) . The reaction mixtures were incubated for 72 hours at 37°C at 5% C0₂ and the plates were then pulsed with 1 μCi of tritiated thymidine per well and incubated for an additional 16 hours. The cells were harvested and counted.

The results are shown in Figure 1. As shown in the figure, RPMI medium alone did not stimulate thymidine uptake in this assay under any conditions, nor did any of the extract pools stimulate the proliferation of T-cells derived in the manner described above from either BALB/c or C57B1 mice. This shows that response is specific to T-cells from NOD mice. Further, the pool from the alpha cells shows no stimulation. These mice do not develop type I diabetes-like symptoms. The susceptibility to simulation of T-cells from NOD mice was tested at various ages from 8-28 days. As shown in Figure 1, the C-pool I derived from B23720 murine insulinoma (? islets) was stimulatory to T-cells from very young mice as well as from older mice. The human neuroblastoma-corresponding pool stimulated proliferation at all ages. However, the α-pool was not capable of stimulating NOD T-cell proliferation, showing a β-cell-specific antigen. Controls using PHA as an inducer showed the expected levels of stimulation of T- cells from all of NOD, BALB/c and C57B1/6 strains. The T-cell assay described above was modified by removing antigen-presenting cells from the test NOD T-cell preparation and assessing the effect of the various extracts in the presence and absence of APC as well as in the presence of fixed APC. In this T-cell preparation, a single-cell suspension of spleen cells from NOD mice was passed over a nylon wool (Robin Lab) column (Julius et al., Eur J Immunol (1973) 2:645) to enrich for T-cells and deplete B-cells, plasma cells and accessory cells. The cells were incubated in the column for 45 min at 37°C in complete medium (RPMI, 10% FCS) and then washed slowly with a large volume of medium. An average of 15-25% yield was obtained, and the resulting cells contained no effective APC. For samples run in the presence of APC, the APC were prepared from irradiated (4000 R) NOD spleen cells; fixed APC were prepared by treating APC with 0.1% glutaraldehyde for 60 sec.

As shown in Figure 2, the presence of antigen- presenting cells is required to elicit a response either in the presence of β islet cytosol extracts or human neuroblastoma cell cytosols. Extracts from α-cells failed to elicit significant response both with and without the presence of APC. Using whole-cell extracts as the source of antigen, either APC or APC that had been fixed only 4 hours after incubation with antigen were required for the stimulation of T-cell proliferation by either the insulinoma extract or the human neuroblastoma SY5Y extract.

As shown from the foregoing results, human antigen (derived from the neuroblastoma line) is cross- species-reactive with the NOD T-cell preparation, and the antigen derived from both β islets and the human cell line requires the presence of antigen-presenting cells in order to be effective in stimulating thymidine uptake.

Example 2 Purification and Characterization of the 8 Islet Extract

A. Murine Insulinoma. The membrane proteins extracted from B23720 insulinoma were subjected to size separation using SDS- PAGE, using 2 mg total load per gel. The elution pattern is shown as a nitrocellulose blot in Figure 3. As shown in Figure 3, discrete peaks at molecular weights 37.8, 41.9 and 55 were observed; it is believed that the small peak at 108.7 is a multiple of the 55 kd peak. The elution pattern shown in Figure 3 is presented in terms of the counts per minute observed of thymidine uptake by the spleen-derived T-cells from 30-40 day old NOD mice. The cytosol fraction from mouse insulinoma

B23720 prepared as described in Example 1 was subjected to high performance electrophoretic chromatography (HPEC) , conducted as described in Example 1. These results are shown in Figure 4. Elution patterns assayed with respect to T-cells derived from NOD mice of various ages is shown. As seen in Figure 4, the proliferation response for the various fractions increases steadily over a period of 17-54 days.

The whole-cell extract of the rat insulinoma was also subjected to HPEC under the conditions described above. Figure 5 shows the elution pattern as determined by the proliferative response of the insulinoma-specific T-cell line NOD-F40 prepared from the spleen of a 40 day old female NOD mouse using the method described hereinabove. Three peaks of activity are shown.

As is apparent from the foregoing results, the antigen appears to be present in what are purportedly cytosol extracts, whole-cell extracts and membrane extracts of the murine insulinoma. Figure 6 shows a comparison of the relative activities of the antigen in each extract as assessed using NOD-F40 T-cell line proliferation. As shown in Figure 6, the majority of the antigen appears to be present in the membrane.

The membrane fraction of the insulinoma prepared as described in Example 1 was subjected to two- dimensional electrophoresis. The samples were adjusted to 1% Triton X-100, 15% glycerol and 6% ampholines, pH 3- 10. Isoelectric focusing in one dimension was performed according to O'Farrell, P.H., J Biol Chem (1975) 250:4007-4021. and after this procedure, the gels were removed from their glass tubes and loaded onto a 10% SDS phage electrophoresed according to Laemmli, U.K., Nature (1970) 227:680-685. The results are shown in Figure 7.

B. Human Islets

Using the procedures described in Example 1, whole-cell extracts of human islets were obtained. HPEC performed on these extracts using the methods described above resulted in the elution profile shown in Figure 8, confirming that the corresponding human antigen(s) have molecular weights in the range of 30-60 kd. The elution profile of Figure 8 was determined using fresh splenocytes from 40 day old female NOD mice. As shown in Figure 8, the human islet extract contains three distinct peaks; one at about 37 kd, one at about 41 kd, and one at about 51 kd.

The distinct nature of the three proteins was studied using a panel of T-cell clones prepared as described hereinabove. First, T-cell lines established from lymph node cells of 30-day-old NOD female mice, selected on whole cell extracts of the insulinoma, exhibited a T-cell reactivity similar to that of unprimed NOD lymphocytes -- that is, there were three peaks of response corresponding to the three antigen peaks in the 30-60 kd region. However, following two cycles of antigenic restimulation, T-cell lines cloned by limiting dilution showed differences in the antigen to which they responded. The T-cell clone LN*CM responded to the antigen of approximately 51 kd (Figure 9A) , and the clone LN7 responded to the antigen of approximately 37 kd (Figure 9B) . Thus, the individual peaks of T-cell activity presumably correspond to distinct proteins as opposed to multimers or degradation products.

Example 3 To distinguish the antigens of the invention from other proteins obtainable from islets, the standard thymidine incorporation assay described above was used, with the test antigens at 10 μg/ml and cultures of 1-3 x 10⁵ cells per well of single cell female murine NOD splenocyte suspensions prepared in the RPMI-based medium. The results are shown in Table 1. TABLE 1.

³H-Thymidine incorporation

Antigens CPM±S.D. whole cell extracts beta insulinoma 31553±1862 alpha glucagonoma 1012±38 islets (NOD mice) 13416±2178 islets (human) 29032±2090 pancreatic hormones rat insulin 899±105 bovine insulin 958±87 C-l peptide (mouse) 854±119 C-l peptide (rat) 997±68 glucagon 1002±95 somatostatin 765±74 recombinant proteins hsp65 1078±329 hsp70 932±119 carboxypeptidase-H 559±121

PM-1 669±98

GAD-65 99±25

GAD-67 155±17 peripherin 67±78

The data in Table 1 show that β-insulinoma, islets from NOD mice, and islets from humans are successful in stimulating thymidine incorporation into splenocytes. Extracts of a-glucagonoma are not. Other pancreatic hormones, including insulin, C-peptides, glucagon and somatostatin, are also not active. Also inactive are heat shock protein (hsp) 65, hsp70, carboxypeptidase-H, PM-l, GAD-65, GAD-67 and peripherin. None of these proteins are capable of showing the stimulatory effect of the islet extracts.

The proteins listed in Table 1 as recombinant proteins were prepared as follows. The cDNA encoding human hsp65 was cloned by PCR using polyA⁺ RNA isolated from human EBV-transformed B cell line, post-incubation at 42°C for 2 hr. Following reverse transcription, the hsp65-encoding fragment was amplified by PCR using the primers 5' -CGGGGATCCGCC-AAAGATGTAAAATTTGGTGCAGATGCC and 5'-GTCCTCGAGTTAGAACATGCCACCTCCCATACCACCTCC (30 cycles of 30 sec at 94°C, 30 sec at 55°C and 1 in at 72°C) . The cDNA encoding human carboxypep-tidase-H, PM-1 (gifts from G. Eisenbarth) and human hsp70 (ATCC/clone pH 2.3), were cloned into expression vector pTrc99A (Aman, E., et al., Gene (1988) £2:301-315) (His6) which was constructed by insertion of a synthetic DNA fragment encoding six histidine residues into the polylinker of the pTrc99A expression vector, to encode protein tagged with six histidine residues at the N-terminus. Plasmid constructs were transformed into E. coIi-Tgl (supE hsdΔslac- proAB)F' [traD36proAB+lacIqlacZΔ M15] and protein expression was induced by addition of IPTG to the culture medium. Bacteria were lysed in 100 mM Tris pH 8.0, 6M GuHCl, and insoluble material was removed by centrifugation at 40 Kg for 30 min. Recombinant proteins were purified using Ni-NTA-agarose (Qiagen, Chatsworth, CA) in the presence of 6M GuHCl and dialyzed against PBS. Protein concentration was determined using BCA assay (Pierce). Mouse GAD-65, GAD-67 and peripherin expressed in baculovirus system were gifts from Dr. Roland Tisch (H. McDevitt lab, Stanford) .

Example 4 Ability of T-Cell Clones to Stimulate Insulitis

The T-cell line LN-7 was stimulated with insulinoma antigen, and three days later 2-5 x 10° stimulated cells were injected intraperitoneally into 3- week-old NOD female mice. Four weeks later, the pancreata were removed, fixed in formalin buffer, and stained with hematoxylin and eosin. Pancreata from age- matched NOD female mice were used as controls.

The results of this experiment are shown in Figures 10A and 10B. Figure 10A represents the control; Figure 10B represents the mouse injected with LN-7 T cells. The injected mice showed acceleration of destructive insulitis, as compared to controls. Similar results were obtained with LN*CM T cell clones.

Claims

1. An antigen in purified and isolated form, which antigen is isolatable from pancreatic islet β-cells and which antigen has a monomer molecular weight in the range of about 30-60 kd and is capable of stimulating the proliferation of NOD murine T-cells in the presence of antigen-presenting cells.

2. The antigen of claim 1 which is isolatable from human pancreatic islet /3-cell membranes, and has a molecular weight of about 37 kd, 41 kd or 51 kd as determined by HPEC; or which is isolatable from murine pancreatic islet jS-cell membranes and has a molecular weight of about 36 kd, 42 kd or 55 kd, as determined by gel electrophoresis.

3. A peptide fragment of a protein antigen of claim 1 which fragment binds to the MHC class II-encoded glycoprotein on antigen presenting cells (APC) to form a complex wherein said complex is recognized by NOD murine T-cells or a modified form thereof which binds to the MHC class II-encoded glycoprotein on APC to form a complex, wherein said complex is not recognized by NOD murine T- cells.

4. A recombinant DNA in purified and isolated form which encodes a protein antigen of claim 1; or a recombinant DNA which encodes a protein antigen of claim 1 included in an expression system capable of expressing said encoding DNA when contained in a host cell, which expression system comprises said encoding DNA operably linked to control sequences compatible with said host.

5. A recombinant host cell which contains the expression system of claim 4.

6. A method to produce a protein useful in therapy and diagnosis of type I diabetes which protein is isolatable from pancreatic islet β-cells and which protein has a monomer molecular weight in the range of

30-60 kd and is capable of stimulating the proliferation of NOD murine T-cells in the presence of antigen- presenting cells, which method comprises culturing the cells of claim 5 under conditions which permit the expression of said encoding DNA; and recovering the protein from the cell culture.

7. Antibodies or immunologically reactive fragments thereof specifically immunoreactive with the antigen of claim 1, which antibodies or fragments are in purified and isolated form; or are contained in a composition substantially free of red blood cells; or are substantially free of plasma proteins; or are monoclonal antibodies.

8. A method to identify an individual developing type I diabetes which method comprises contacting T-cells obtained from said individual with the antigen of claim 1 under conditions of a T-cell proliferation assay; and determining the ability of said T-cells to proliferate in the presence of said antigen; so as to identify an individual whose T-cells proliferate under these conditions as developing type I diabetes.

9. A method to identify an individual developing type I diabetes which method comprises contacting serum or plasma obtained from said individual with the antigen of claim 1 under conditions whereby antibodies immunoreactive with the antigen of claim 1 will form a complex; and detecting the presence or absence of said complex; so as to identify an individual providing serum or plasma capable of forming said complex as developing type I diabetes.

10. The antigen of claim 1 in nonimmunogenic form or the peptide of claim 3 for use in preventing development or progression of Type I diabetes.

11.. The antigen of claim 1 in nonimmunogenic form or the peptide of claim 3 in combination with an immune modulator for use in preventing development or progression of Type I diabetes.

12. A pharmaceutical composition useful in preventing the development or progression of type I diabetes which comprises as active ingredient the antigen of claim 1 in nonimmunogenic form or the peptide of claim 3 in admixture with a pharmaceutically acceptable excipient, and optionally further containing an immune modulator.

13. The composition of claim 12 wherein said immune modulator comprises anti-CD4 antibodies or immunoreactive fragments thereof.

14. A composition free of red blood cells which is enriched in T-cells responsive to the antigen of claim 1.