Mitrea Cristina Mirela
Clinical Universitary Hospital "N.Malaxa"
Host Laboratory: Laboratory of Molecular Genetics � Hospital San Raffaele, Milano, ITALY
Supervisors: Prof. Pierangelo BONINI, Prof. Maurizio FERRARI, Dr. Vito LAMPASONA

Download as a PDF here

Background

Type 1 diabetes is a chronic disorder due to the destruction of insulin producing pancreatic islet beta cells by the patient immune system. Diabetes autoimmunity is characterised by the presence of circulating autoantibodies to islet beta cell antigens. Prediction of type 1 diabetes is based upon the measurement of these autoantibodies and the identification and cloning of autoantigens is therefore important for prediction and eventual prevention of the disease. Identified autoantigens in type 1 diabetes include GAD, the protein-tyrosine phosphatase IA-2 and insulin. However, there exists evidence that several patients, especially those with slowly developing autoimmunity, have antibodies to still unidentified autoantigens.

Several proteins have been suggested to be target of these autoantibodies, however the majority of these putative targets have been difficult to confirm as true autoantigens in experienced laboratories. The major drawback in numerous studies apparently relates to the insufficient sensitivity and specificity of autoantibody tests based either on ELISA or western blot assays and the use of recombinant autoantigens expressed in bacteria. Autoantibodies are frequently low in titre compared to antibody responses to pathogens and recognise mostly conformational epitopes. Immobilisation of antigens on plates or nitro-cellulose membranes easily results in the loss of conformational epitopes and in the exposure of cryptic epitopes that can be bound aspecifically by low affinity circulating antibodies. Moreover, bacterially expressed recombinant autoantigens are often poor autoantibody targets, as a consequence of the frequent inability of bacteria to properly fold eukaryotic proteins, and also entails the risk of detecting aspecific antibody responses to contaminant bacterial proteins.

Rationale and specific aims

Before specific autoantibody routine testing is introduced in the clinical laboratory, putative autoantigens are in need of validation by sensitive and specific techniques, other than those usually applied to screening for novel protein target of autoantibodies. A consensus has been achieved in the type 1 diabetes research community that the most reliable assays are those based on the immunoprecipitation of radio-labelled recombinant antigens expressed either in anin vitrosystem, like rabbit reticulocytes, orin vivo, as a more cumbersome alternative, in a eukaryotic cell system, like recombinant baculovirus infected insect cell lines.

We decided to focus our attention on two different proposed type 1 diabetes autoantigens still lacking definitive confirmation. These are GLUT-2, a low affinity glucose transporter expressed at the cell surface of pancreatic islets beta cells, hepatocytes, small intestine and kidney epithelial cells, and ICA12/SOX13 an ubiquitous transcription factor highly expressed in the endocrine pancreas. Autoantibodies directed to GLUT-2 were initially reported in 1990 by JH Johnson et al.(1). In this study an inhibitory activity on glucose uptake by cultured rat pancreatic islets, but not rat hepatocytes or human erythrocytes, was observed after incubation with purified IgG immuno-globulins from type 1 diabetes patients. The same group later identified the target of these antibodies as the GLUT-2, as indicated by selective inhibition of glucose uptake in GLUT-2 transfected cell line after incubation with patients' sera (2). The studies by another group is also supportive of this observation and reports the presence of autoantibodies to GLUT-2 detectable in western blot (3). The identification of a type 1 diabetes autoantigen named ICA12 was initially reported also in 1992 by D.U. Rabin et al. (4) upon screening of a pancreatic islets cDNA expression library with patients' sera. Autoantibodies to the ICA12 antigen were detected by western blot although at a relatively low frequency (5 out of 12 patients were found positive). No other study on ICA12 has reached publication until the recent identification of SOX13 a novel transcription factor of the HMG (high mobility group) family which share a very high degree of homology with ICA12.

The goal of our research project was therefore to develop assays for the measurement of autoantibodies to GLUT-2 and ICA12/SOX13 based on radio-labelled recombinant antigens expressedin vitroorin vivoin a eukaryotic system and to validate the presence of these two putative autoantigens in a cohort of type 1 diabetes patients.

Material and Methods

Sera: Sera from 100 newly diagnosed type 1 diabetes patients and 57 non diabetic age matched controls were used in the pilot test experiments. As a positive control serum for the GLUT2 antibody test a polyclonal rabbit anti-human GLUT2 antibody was used (Chemicon).

Cloning:

The ligation reaction was transformed into competent E. Colii cells of the X.L. Blue MRF' strain (Stratagene) and plated on agar plate containing ampicillin as selective agent. Several bacterial clones were subsequently grown in LB medium and plasmid DNA extracted with Quantum-prep spin columns (Biorad). Clones containing cDNA were identified by DNA restriction analysis.

The GLUT2 cDNA was re-amplified with appropriate primers for subcloning into the pSPUTK plasmid vector. This vector contains an optimised leader sequence downstream of the SP6 phage promoter and allows the efficient transcription and translation of cloned cDNAsin vitro.

The GLUT-2 cDNA was also cloned into the pIZT/V5-His (Invitrogen) and pFastBac (GIBCO) vectors for subsequent expressionin vivo in insect cell lines and positive recombinant clones identified by restriction analysis. For the pFastBac experimental procedure recombinant baculovirus genomes, or bacmids, were generated upon transformation of identified pFastBac-GLUT2 clones into E. Colii of the DH10Bac strain. Selected GLUT2-bacmid clones were then grown in LB medium and bacmid DNA extracted with a modified alkaline lysis method. A full length cDNA encoding ICA12/SOX13 was obtained by cutting with the Sph 1 restriction enzyme and ligation of the two partially overlapping original cDNA clones.

Expressionin vitro:

The plasmid DNA of isolated clones was used for in vitro coupled transcription and translation in the presence ³⁵S-methionine (Amersham) with the TnT rabbit reticulocyte system (Promega). Recombinant radio-labelled proteins were then purified of unincorporated ³⁵S-methionine by size-exclusion chromatography on a NAP-5 column (Pharmacia), their incorporated radioactivity measured in a liquid scintillation beta counter (Kontron), and analysed by polyacrilamide gel electrophoresis under denaturing conditions followed by autoradiography to confirm expression of recombinant GLUT2 and ICA12/SOX13 proteins of the appropriate molecular weight.

Immunoprecipitation assays:

Expressionin vivo:

SF9 insect cells were cultured in SF900 serum free medium (GIBCO) until early log phase. 0.5 x 10⁶ cells were then plated in 6-well plates and transfected with 5 micrograms of recombinant pIZT/V5-His-GLUT2 plasmid using Insectin-Plus (Invitrogen) liposome mixture. After 4 hours incubation cells the transfection mixture was removed and replaced with fresh serum free medium. Cells from replicate wells were harvested at day 2, 3, 4 post transfection and lysed with HEPES pH7.4, Triton X-100 1% buffer containing a protease inhibitor cocktail (SIGMA). Efficiency of transfection was also monitored by fluorescence-microscopy, based on the co-expression of green fluorescent protein from the same vector. After day 4 for the pIZT/V5-His-GLUT2 experiment the antibiotic Zeocin was added to the culture medium to select for stable expression of GLUT2 from resistant cell lines. For the pFastBac experiment 0.5 x 10⁶ cells were plated in 6-well plates and transfected with 1 micrograms of GLUT2-bacmid DNA using CellFectin liposomes (GIBCO). After 4 hours incubation cells the transfection mixture was removed and replaced with fresh serum free medium. Cells from replicate wells were harvested at day 2, 3, 4 post transfection and lysed with HEPES pH7.4, Triton X-100 1% buffer containing a protease inhibitor cocktail. At day 4 the supernant from transfected cells was collected, diluted 1 to 4 with fresh medium and added to SF9 cells in early LOG phase for amplification of eventual Baculovirus stock..

Western blot:

Results GLUT2 expressionin vitroand immunoassay:

GLUT2 expression was efficiently achievedin vitro from the pSPUTK vector, as demonstrated from incorporated radioactivity after coupled transcription and translation. However, upon autoradiography a protein product of apparent molecular size much larger than expected was observed. This recombinant protein was nevertheless immunoprecipitated by the polyclonal GLUT2 antibody. An immunoprecipitation assay based on this recombinant protein failed to show differences in binding of antibodies between patient and control sera.

GLUT2 expressionin vivoand western blots:

ICA12/SOX13 expressionin vitroand immunoassay:

ICA12/SOX13 expressionin vitro was achieved from both amplified and cloned open reading frames and the full length construct. The recombinant proteins observed upon autoradiography showed the expected molecular size. Binding to both ICA12/SOX13 constructs after immunoprecipitation was observed in a proportion of type 1 diabetes sera. A minority of control sera reacted with only one of the construct. Immunoprecipitation of the full length ICA12/SOX13 protein indicated presence of autoantibodies in a minority of type 1 diabetes patients.

Discussion

Cloning and expression of GLUT2 using two different approaches could not be satisfactorily reached. In one case expressionin vitroof cloned GLUT2 cDNA resulted in a product of unexpected molecular size, although still recognised by a polyclonal anti-GLUT2 antibody. It is unclear at the moment whether this large protein is a multimeric form of GLUT2 which includes several GLUT2 proteins, possibly covalently bound together upon translation, or whether protein other than GLUT2, present in the translation reaction, are then cross-linked to the recombinant GLUT2. Lack of binding from type 1 diabetes sera in immunoprecipitation could therefore either be ascribed to poor antigenicity of the recombinant GLUT2 or simply absence of autoantibodies. In thein vivoexpression systems GLUT2 expression could not be convincingly demonstrated nor stable expression GLUT2 insect cell lines established. Lack of expression in this system is usually associated with inappropriate leader sequences in the recombinant cDNA and therefore inefficient translation in cells or to a secondary negative effect on cell growth or viability due to the recombinant protein in expressing cell. Both possibilities remains to be investigated.

ICA12/SOX13 expressionin vitroproved more straightforward and preliminary experiments yielded two partial recombinant protein. Both these protein could be bound by antibodies present in sera from type 1 diabetes patients although with different backgrounds in the control sera group. Antibody testing with the full length ICA12/SOX13 clone now available in our laboratory indicated the presence of autoantibodies to ICA12/SOX13 in a minority of patients with type 1 diabetes. Whether this autoantibodies are associated with phenotypic differences in patients, like HLA antigens or other autoimmune pathologies, remains to be elucidated.

Acquired Technical Experience:

• Polymerase chain reaction

• Cloning in plasmid vectors

• Plasmid DNA extraction and restriction analysis

• In vitro

  transcription and translation

   

• Immunoprecipitation assay

• SDS-PAGE

• Western blot assay

• Insect cells culture and transfection