WO2004092410A2

WO2004092410A2 - Method for diagnosing rheumatoid arthritis or osteoarthritis

Info

Publication number: WO2004092410A2
Application number: PCT/EP2003/003942
Authority: WO
Inventors: Hans-J�rgen THIESEN; Peter Lorenz; Peter Stiehl; Dirk Koczan; Peter Ruschpler
Original assignee: Thiesen Hans-Juergen; Peter Lorenz; Peter Stiehl; Dirk Koczan; Peter Ruschpler
Priority date: 2003-04-15
Filing date: 2003-04-15
Publication date: 2004-10-28
Also published as: EP1620567A1; AU2003229678A1

Description

Unser Zeichen: TH01 H04/P-WO

Method for diagnosing rheumatoid arthritis or osteoarthritis

The invention relates to a method for diagnosing rheumatoid arthritis or osteoarthritis, or of a predisposition therefor, using expression profiling data of differentially expressed genes and a biosensor chip and medical or diagnostic instrument suitable for carrying out this method, and a method for monitoring the therapeutical effect of an anti-rheumatoid arthritis or anti-osteoarthritis drug or for screening for a potential anti-rheumatoid arthritis or anti-osteoarthritis drug, and a method for the production of an anti-rheumatoid arthritis or anti-osteoarthritis drug, and the anti-rheumatoid arthritis or anti-osteoarthritis drugs obtainable by this method.

Abbreviations used in this description:

RA, rheumatoid arthritis; OA, osteoarthritis; MnSOD, manganese superoxide dismutase

A list of cited references with detailled bibliographic data can be found at the end of this description.

1 Introduction

The new enabling technologies in functional genomics and proteomics are powerful tools to study complex multifactorial diseases. A molecular inventory of diseased tissues provides a first step to gain insight into the pathophysiology of these diseases. Differentially expressed genes and their products have to be elucidated in relation to specific cell types/tissues and with respect to their expression patterns in space and time. Furthermore, the functional states of the gene products have to be described. All these approaches will provide candidate targets for in-depth functional studies that foster drug discovery processes.

Microarrays have proven very useful to acquire large data sets of differential gene expression on the transcriptional level [1-3]. However, transcript levels do not necessarily reflect the amount of protein that represents the actual functional entity of a gene [4-9]. In addition, some compartments of an organism that are of interest do not contain RNA at all, e.g. body fluids. Technologies in proteomics aim at the cataloguing of the whole protein complements of biological fluids, cells and tis- sues and at the description of specific functional states of the proteins. The complexity of protein profiles exceeds that of the RNA profiles manifold due to the posttranscriptional and posttrans- lalional generation of different protein species [10]. Two-dimensional gel eleclrophoresis coupled to MALDI-TOF mass spectrometry yields valuable insights into the molecular composition of cells and tissues. However, technical difficulties impede or even prevent the display of certain proteins and the limited dynamic range results in a bias towards abundant proteins [11-13]. These limitations have sparked efforts to find alternatives. One possibility is to make use of antibodies for protein profiling, e.g. in multi-Western blot formats or on protein microarrays [14-16].

Rheumatoid arthritis (RA) with a prevalence of about 1 % is a chronic systemic disease that involves mechanisms of autoimmunity [17, 18]. Characteristic is a severe inflammation of the joints with infiltration of activated T cells, macrophages, plasma cells and other immunocompetent cells that leads to progressive destruction of cartilage and bone. Synovial cells show signs of hyperpro- liferation and hypertrophy and the synovial tissue is transformed into a pathologic mass of cells called pannus tissue that also undergoes neovascularization. The etiology of the disease remains unresolved. Hypotheses on the ultimate cause of RA include antigen-specific autoimmune reactions and the existence of an infectious agent [19]. The significant lack of understanding of the pathogenesis of RA is exemplified by the difficulties in the molecular diagnosis and classification of the disease as well as the lack of clear prognostic markers and cure. The molecular characteriza- tion of the diseased tissues will provide means to further our understanding of the disease. Differential expression profiles of RA in comparison to other arthritic diseases like osteoarthritis (OA) will be of particular interest. OA is mainly considered a degenerative disease of cartilage, but inflammatory processes of the synovial tissue are also involved [20, 21]. OA samples provide a widely accepted comparison and have been used in many investigations (e.g. [22, 23]).

From the above the objects of the present invention result.

According to the present invention a method for diagnosing the presence or absence of rheumatoid arthritis or osteoarthritis, or of a predisposition therefor, using expression profiling data as defined in present claim 1 , a biosensor chip as defined in present claim 8, a medical or diagnostic instrument as defined in present claim 9, a method for monitoring the therapeutical effect of an anti- rheumatoid arthritis or anti-osteoarthritis drug or for screening for a potential anti-rheumatoid arthritis or anti-osteoarthritis drug as defined in present claim 10, a method for the production of an anti- rheumatoid arthritis or anti-osteoarthritis drug as defined in present claim 11 , and the anti- rheumatoid arthritis or anti-osteoarthritis drugs obtainable by this method as claimed in present claim 12 are provided. The GenBank's URL is http://www.ncbi.nlm.nih.gov/Genbank/GenbankSearch.html.

In the following the present invention is described in more detail for further illustration. This specific description is not to be construed as any limitation of the invention.

For the present invention the inventors conducted a comparative molecular characterization of synovial tissues of patients suffering from RA versus OA on the RNA and protein levels. The initial screen profiled 12,526 transcripts by DNA oligonucleotide microarrays and 791 protein species by Western blot arrays in one RA versus one OA synovial tissue sample. The results indicated that many gene expression changes did not coincide on transcript and protein levels. Subsequently, selected genes found to be differentially expressed on the protein levels were validated by Western blotting with a higher sample number and assigned to particular cell types within the synovial tissue. In a second RNA profiling screen 44928 transcripts were compared for 6 RA and 6 OA syno- vial tissue samples by DNA oligonucleotide microarrays. The data exemplify approaches to improve our understanding of diseases that target complex tissues using global profiling technologies.

2 Materials and methods

2.1 Tissue samples

Samples of rheumatoid and osteoarthritic synovial tissue were obtained as part of surgical therapy of diseased joints during synovectomy or implantation of an endoprothesis. Collection of samples was approved by the ethical board of the University of Leipzig. Patients with rheumatoid arthritis (RA) had classical late-stage disease and met the American College of Rheumatology classification criteria. The synovial tissue was dissected, separated from associated fat, snap-frozen in liquid nitrogen and stored either in liquid nitrogen or at -80° C. All tissues were assessed by histopathology. All RA patients were under medication including steroids, disease-modifying anti-rheumatic drugs (DMARDs) and other non-steroidal anti-inflammatory drugs. Diagnosis of osteoarthritis (OA) was based on clinical examination and histopathology. OA patients typically received non-steroidal antiinflammatory drugs (NSAIDs).

2.2 Protein extraction from synovial tissue

Individual pieces of frozen tissue comparable in weight were placed into a mortar (WiTa GmbH, Potsdam, Germany) sitting in a liquid nitrogen bath. A pre-defined volume of buffer per mg tissue was added to a pre-cooled spoon-type spatula held at the surface of liquid nitrogen so that the liquid froze to a pellet. This frozen pellet was added to the tissue in the mortar and both were grinded together to a fine powder. Extracts were made with 10 mM TRIS/HCI pH 7.4, 2 mM sodium orthovanadate, 1 % SDS and protease inhibitors (Complete; Roche, Mannheim, Germany) (buffer TS). 100 mg small steel bullets (Marabu G10) were added to the powdered tissue. Then preheated TS buffer was added and the homogenate was immediately mixed and heated for 5min in a 90° C water bath with occassional further heavy vortexing. The heating procedure with TS buffer followed recommendations of the supplier of the PowerBlot™ service (see below). Finally, the tissue ho- mogenates were centrifuged for 10 min at 10° C at 20 000 x g. The supematants contained the protein fractions to be analyzed. For normalization, the protein content of the extracts were meas- ured using the BCA method (Perbio, Bonn, Germany).

2.3 Western blot array expression analysis

The Powerblot™ is a commercial Western blot format of BD Biosciences (through BD Transduction Laboratories, Lexington, KY, USA; http://bioinfo.clontech.com/powerblot/). Samples of different tissues or conditions are subjected to semi-quantitative comparative immunostaining with a panel of at that time 791 mouse monoclonal antibodies in the company's portfolio. Samples are loaded across the whole width of the gel. A manifold is then used to separate 45 lanes on the blots. Immunostaining is performed in such a way that several antibodies recognizing clearly distinguishable proteins with respect to their mobility on SDS-PAGE are added together to one lane formed by the manifold. Triplicates of each antibody reaction visualized by chemiluminiscence are evaluated by densitometry and compared between two samples. The output displays ratios of protein levels between two samples in several confidence groups depending on the quality and strength of the signal. Signals are normalized by dividing the optical density obtained for a signal through the total intensity value of all pixels in an image and multiplied with 1 ,000,000. These normalized quantities are used to compute the changes between two samples. For the Powerblot™ TS buffer extracts of patients K55/RA and K42/OA were adjusted with SDS sample buffer to give 62.5 mM Tris/HCI pH 6.8, 2% SDS, 5% glycerol, 1 % β-mercaptoethanol and 0.003 % bromophenol blue at a concentration of 1mg/ml total protein. These extracts were shipped on dry ice to the Powerblot™ supplier and directly loaded onto gels.

2.4 Western Blotting using 1D SDS-PAGE

Lammli-type 10 or 12 % SDS polyacrylamide mini-gels were run according to standard procedures. They were blotted to PVDF membrane (Polyscreen, NEN Life Sciences, Zaventem, Belgium ) by semi-dry electroblotting [24]. The blot quality was assessed by staining the membrane for 5 min- utes in 0.3% Ponceau S in 3% trichloroacetic acid and subsequent washing with H20. Blocking of the membrane was usually done overnight at 4° C with 50 mM TRIS/HCI pH 7.4, 150 mM NaCI (TBS) plus 5 % nonfat dry milk powder and 1 % bovine serum albumin. For staining a three-step protocol was used that involved primary antibodies, biotinylated secondary antibodies and Strepta- vidin-Peroxidase. Immunological reagents were usually diluted in blocking buffer and incubated at room temperature for 2 hours (primary antibodies) or 1 h (secondary antibody and tertiary reagent). The primary antibodies included rabbit anti-Statl (sc-346; Santa Cruz Biotechnology, Heidelberg, Germany; used at 1 :2000), goat anti-CD3ε (sc-1127; Santa Cruz Biotechnology, Heidelberg, Germany; used at 1 :500), mouse monoclonal anti-p47phox, anti-manganese superoxide dismutase and anti-cathepsin D (P33720, M99920, and C47620 respectively; BD Transduction Laboratories, Heidelberg Germany; used at 1 :1000, 1 :1000 and 1 :3000, respectively). All secondary antibodies and Streptavidin-Peroxidase were from Jackson/Dianova (Hamburg, Germany) and were used at 50 and 100 ng/ml, respectively. Blots intended to be measured by densitometry were developed with 4-chloro-1-naphthol as described [24]. For densitometry the blot signals were scanned using a HP desktop scanner in color mode at 150 dpi, 8bit resolution. Raw bitmap images were converted into grayscale Tiff-files and the integrated optical density, i.e. the volume of the signals, was quantitatively assessed by Phoretix 6.0.1 Advanced (Nonlinear Dynamics Ltd) without any image adjustments. Boundaries of signals were drawn by hand and background was removed by subtraction of the "average on boundary" volume. Differences between patient groups were statistically assessed by Student's t-test after calculation of the mean integrated optical density ± SD.

2.5 Affymetrix gene chip RNA profiling

Preparation and processing of RNA as well as microarray hybridization and signal detection was performed as described earlier [25]. Briefly, total RNA was isolated from snap-frozen synovial tissue pieces by grinding the tissue together with frozen RLT buffer (RNeasy Kit, Qiagen, Hilden, Germany) supplemented with 1 % β-mercaptoethanol in the frozen state. After thawing and rigorous vortexing RNA was isolated from tissue lysates with the RNeasy kit as recommended by the manufacturer. 5μg total RNA was processed and labeled as recommended by the chip manufacturer. Hybridizations of HG-U95A or HG-U133A and HG-U133B DNA oligonucleotide microarrays (Affymetrix, Santa Clara, CA, USA) were performed according to the instructions of the manufacturer. Data were processed using Affymetrix Microarray Suite 4.0 and 5.0 and Affymetrix Data Mining Tool 3.0. The default parameter settings of the software were used. 3 Results

3.1 Comparative RNA and protein profiling of rheumatoid arthritis and osieoarthritis synovial tissues

The molecular characterization of synovial tissue on the RNA level was performed with Affymetrix DNA microarrays of the type HG-U95A containing 12526 gene sets. In a case study, synovial tissue material from one patient each with rheumatoid arthritis (RA) or osteoarthritis (OA) was compared (K55/RA vs. K42/OA). The statistics of the RNA profiling results showed 6630 (53 %) and 6963 (56 %) gene sets to be detected in the RA and OA synovial tissues, respectively (Table 1 ). These values indicate that the corresponding transcripts of 47 or 44 % of gene sets are either not expressed in synovial tissues or have not been detected due to sensitivity limits of the chip technology employed. Transcripts of 1062 genes were differentially expressed in the RA versus the OA synovial tissue, 416 overexpressed and 645 underexpressed. Not surprisingly, the overexpressed gene cluster contained genes indicative of inflammation and the activated state of immunocompe- tent cells like cytokines/chemokines and their receptors or immunoglobulins. Prominent examples of the former were interleukin 15, CXCL9/Mig, CXCL10/IP10, and CCL5/RANTES, CXCL13/B-cell chemoattractant, CCL19/Ebi1 ligand and the interleukin receptors 2γ, 7, and 15 or the chemokine receptor CCR5 (data not shown). In the RA underexpressed gene cluster the inventors found the cartilage oligomeric matrix protein that is synthesized by synovial cells and believed to be associated with OA disease [20]. Examples for other genes in this class are the adhesion molecule thrombospondin 4, growth arrest protein GAS1 and the C2H2 zinc finger protein ZIC1.

Synovial tissue from the same two patients K55/RA and K42/OA underwent protein profiling to determine proteins that are differentially expressed in both tissue samples. Total protein extracts of both samples were subjected to a comparative commercial Western blot array format (BD Power- Blot™) that used 791 monoclonal antibodies. For a few proteins more than one antibody was present in the whole panel. This set of antibodies provided an opportunity to i) detect high and low abundant proteins side by side and ii) to analyze different proteins in a number that can usually not been accomplished in a single laboratory because of time and resource constraints. From the panel of 791 antibodies 260 (33 %) detected their corresponding protein. Consequently 531 (67 %) antibodies failed to delect their corresponding antigen in RA and OA tissue extracts, either due to limited sensitivity or because these proteins were not expressed in the synovial tissue tested. Among the 260 proteins that were detectable in synovial tissues 71 were scored significantly and specifically changed (29 increases and 42 decreases). Thus, almost a third of the detected proteins showed specific changes. The results of 20 reactions were ambiguous, often due to low level signals. 3.2 Comparison of gene expression changes on the proleome and transcriptome levels

In a next step we compared the protein and RNA profiling data. Starting point was the data set with the 260 detectable protein species. Proteins and their respective transcripts had been scored in the profiling experiments as "increased", "decreased" or "not changed". The analysis revealed 16 (28 %, 16 out of 58) genes whose expression changes coincided on protein and RNA levels when looking to the proteins with unambigeous changes between the RA and OA sample. This observation stresses the relevance of expression profiling on the protein level to determine changes that have functional consequences for cells/tissues. The statistics of the detectable proteins that were unchanged between the two samples displayed also discordant expression on protein versus RNA level for 19 % (23 out of 124) of the genes. However, the majority of 81 % (101 out of 124) of the genes were scored "not changed" for both, protein and transcript expression.

The 16 genes with coinciding RNA and protein level changes in the RA compared to the OA synovial tissue are grouped and displayed in table 1 with annotation of their known function. The inventors selected some of these genes for further analysis since concordance on the RNA and protein levels underscored the changes. In addition, the aspartic protease cathepsin D was included. Cathepsin D was found to be among the most prominent proteins underrepresented in RA compared to OA synovial tissues in the PowerBlot™ screen. This was observed with two different antibodies. On the mRNA level the expression of this gene was detectable in tissues derived from both diseases. However, cathepsin D mRNA levels were not significantly different as judged by the Affymetrix microarray analysis software.

One prominent gene from table 1 is the signal transducer and activator of transcription 1 (Statl ). On the RNA level, all the chip's probe sets specific for Statl are scored "increased" with a highest change of 13.3 fold. On the PowerBlot™ two different antibodies scored the Statl protein "increased". The p47phox and p67phox genes in table 1 , both overexpressed in RA, gained immedi- ate attention since their gene products cooperate in the NADPH oxidase complex. This complex is involved in superoxide production in phagocytic cells as a means of pathogen defense [26]. Interestingly, another upregulated gene presented in table 1 , the manganese superoxide dismutase (MnSOD), is involved in the metabolism of reactive oxygen species, too. 3.3 Validation of RNA and protein profiling screens

The differential expression data on RNA and protein levels described above resulted from a case study using one patient sample of each disease (RA vs. OA). In order to validate these data, Wesl- ern blots of selected gene products were performed on a panel of 16 synovial tissue samples, representing 8 cases of each disease (RA and OA) and the specific signals on the blot were quanti- tated by densitometry (Fig 1). Statl -specific antibodies detected double bands of signals in all OA and RA samples. These bands most likely represent the two Stall isoforms, Statl α and Statl β, that are known to result from alternative splicing. Densitometry was used to quantitatively compare the RA and OA conditions. The mean difference in Statl protein content was an about 3.7fold higher expression in synovial tissues from RA patients (p < 0.01 ). In contrast to Statl , cathepsin D was more abundantly expressed in OA synovial tissue. Here the mean difference was about 2.7fold (p < 0.05). Noticeably, the OA patient samples showed divergent protein levels with five extracts standing clearly out against the other three OA and the eight RA samples. The clinical data of re- spective patients did not hint at obvious reasons for these variations. The signals for p47phox consisted of an expected band at around 44/45 kD plus a faster migrating band at about 32kD that might represent a degradation product. Both bands were measured by densitometry and their optical densities added. Overall, RA synovial tissue contained about 2.8fold more p47phox than the respective OA tissue (p < 0.01 ). The higher abundance of MnSOD in the RA synovial tissue could be validated in the Western blot as well. Clear signals at about the expected size of 25 kD were obtained in all RA and most OA samples. Here, the mean increase in protein content in RA versus OA tissue was 2.7fold (p < 0.01 ). As a positive control for increased expression in RA synovial tissue, immunostaining with antibodies against the T cell marker CD3ε was performed. As expected RA tissue had significantly more CD3ε (4.3fold mean increase, p < 0.05) reflecting the higher infiltration of T cells in RA compared to the OA disease. Overall, the blot showed heterogeneity of signals and thus of respective protein contents in different patients of one disease group, most pronounced in the cathepsin D stain as described above. These deviations are likely due to differences in individual patients, different disease manifestations and/or different therapeutic treatment responses.

3.4 Extension of RNA profiling screens

The transcriptome case study was extended by subjecting synovial tissue from 6 more RA and OA patients, each, to RNA profiling on DNA oligonucleotide microarrays. Differential analysis of the 6

RA samples against the 6 OA samples resulted in 36 cross-comparisons with each OA sample as baseline for each RA sample. Concordant expression changes in 25 out of 36 cross-comparisons were taken as significant. The analysis resulted in 547 genes typically overexpressed and 773 genes typically underexpressed in the RA versus the OA synovial tissue. The genes provide additional expression signatures to distinguish RA and OA disease.

4 Discussion

Initially, two individual samples from synovial tissue representing RA or OA diseases have been taken for in depth comparative analysis on the transcriptome and proteome levels. The use of specific antibodies for protein profiling as reported here has the advantage that most if not all species of a particular gene product can be detected. In contrast, studies using two-dimensional gels for protein display might easily miss spots or fail to display certain protein species leading to an underestimation of the expression level of a particular protein.

The inventors results showed that changes in protein abundance were in many cases not reliably indicated on the transcript level. E.g., out of 58 unambiguous changes on the protein level only 16 were found on the transcript level as well (28%). Discrepancies between protein and transcript abundances reflect mechanisms of gene expression regulation that do not act at the mRNA level. Such mechanisms include translational control, post-translational processing and regulation of protein stability [27, 28]. A relatively poor overall correlation of mRNA/protein levels was pointed out before in yeast [6, 7], human liver [4] and more recently in human tumor samples [9]. In the latter case correlations of RNA/protein abundance were low not only when determined within one sample but also when the data for a particular gene were averaged across all samples. Two stud- ies noted a quite high mRNA/protein correlation for high abundant proteins [6, 29]. In contrast, a third report [9] did not. The above cited studies examined static expression levels rather than changes between two states as have done. However, recent work in yeast reported also discordant behaviour of changes of transcript/protein expression levels when carbon sources were switched [8, 30]. Altogether, available literature as well as the invention's data demonstrate that the extent of transcript expression can give a valid estimation of protein abundance for particular genes at a given situation. However, concordant mRNA/protein expression levels can not be taken as granted since numerous examples in the literature and in this invention show the opposite. This stresses the importance to generate proteome data to evaluate phenotypes of diseases.

After the initial screen of one RA and OA sample on the transcriptome and proteome levels, the inventors strategy was to validate candidates for differential expression between RA/OA by Western blotting using 8 RA against 8 OA samples. Furthermore, additional 6 RA and OA samples were analyzed on the transcript levels leading to characteristic RNA expression signatures for the two diseases.

The differences in gene expression between RA and OA synovial tissues likely reflect i) the differ- ences in the pathological phenotypes of resident synovial cells ii) the different extent of infiltration of in particular lymphoid cells into the synovium and iii) the different disease-specific interplay between all cell types in the arthritic microenvironment. A good example for case i) are the synovial fibroblasts. Such fibroblasts from RA patients display an activated/transformed phenotype, secrete a distinct pattern of cytokines and are mainly responsible for the massive synovial hyperplasia [31]. The activated RA synovial fibroblasts are e.g. a source of the pro-inflammatory interleukin 15 [32] that was among the inventors list of RA overexpressed transcripts (see Results). With respect to case ii) RA synovium is infiltrated to a higher extent by mononuclear cells like T and B lymphocytes or macrophages. Thus, differences in cellular composition will certainly contribute to expression differences when complex tissues were compared. However, infiltrating cells do also change their phenotype in the arthritic joint. Immigrating monocytes, e.g, can differentiate into mature macrophages [33]. Such changes will also contribute to differences in gene expression. Lastly, case iii), all the pathological cell-cell and cell-matrix interactions between synovial resident cells and infiltrated cells will be mirrored in the gene expression profiles of RA against OA synovial tissues. Whatever the reasons for differential expression patterns are, global expression studies on the RNA as well as protein level performed on complex tissues provide subsets of transcripts and proteins for further investigation. A first means to look into the involvement of particular cell types is the use of immunohistochemistry. This was exemplified with two proteins that were found to be differentially expressed between RA and OA in this report. The usefulness of gene expression profiling for identification of novel target genes in complex diseased tissues was recently demonstrated for multiple sclerosis [34].

In the present invention Statl was identified to be significantly more abundant in RA compared to OA synovial tissue on RNA as well as protein levels. Stat proteins are important mediators of cyto- kine signalling and Statl is at the center of interferon signal transduction [35]. The inventors results demonstrated that increased Statl protein could be mainly attributed to macrophages and T lymphocytes although the mass of fibroblast-like pannus cells also displayed weak Statl signals. The pro-inflammatory chemokine RANTES, that is overexpressed in RA synovial tissue ( [23], see also the inventors data in the Results section and below), is known to activate Statl in T cells [36]. Statl is also considered to be an important transcription factor in macrophages, especially in response to bacterial lipopolysaccharide [37, 38]. So far studies of Statl in the subject of arthritis are rudimentary. Two reports analysed Statl activation in synovial fluid cells, but came to different conclusions [39, 40]. In preliminary experiments analyzed the activation state of Statl protein immunoprecipi- tated from RA and OA synovial tissues through determination of its phosphorylation status. In both diseases found tyrosine 701 and serine 727 phosphorylated Statl species (data not shown). The investigation of quantitative differences in phosphorylation states between both diseases awaits further experimentation.

Transcriptional upregulation of interferon-inducible genes was to be expected as consequence of higher Statl protein levels and potential activation. Indeed, the inventors RNA profiling results identified a number of interferon-inducible genes with more abundant transcript levels in RA than OA synovial tissue. Prominent examples are cytokines/chemokines interleukin 15, CXCL9/Mig, CXCL10/IP10, and CCL5/RANTES that are also well-established genes abundantly expressed in inflammatory disorders including rheumatoid arthritis (fore reviews see [41 , 42]). The Statl gene itself can also be induced by interferons [43]. The suppressor of cytokine signaling (SOCS)1 gene was also among the interferon-inducible genes whose transcripts were overexpressed in RA versus OA in the inventors screen. SOCS proteins are important negative regulators of cytokine sig- naling [44]. In contrast the mRNA expression of the related SOCS3 was more abundant in OA (data not shown). This might be of importance since a recent report highlighted a strategy to treat RA disease by upregulation of SOCS3 protein [45].

Three gene products that found to be more abundant on RNA and protein levels in RA compared to OA synovial tissue are components of the metabolism of reactive oxygen species. The p47phox and p67phox proteins are part of the NADPH oxidase system of phagocytic cells that is essential for the innate immune response to pathogens [26], The NADPH oxidase system is further considered an important mediator of inflammatory disorders like ischemia-reperfusion injury in the brain, heart and liver [46]. Thus, the upregulation of p47phox and p67phox in synovial tissue of RA pa- tients described in this invention pointed to a possible role of NADPH oxidase produced superoxide in RA in humans. The inventors RNA profiling data also contained expression information on other genes whose products can be found in the NADPH oxidase complex: Both, p40phox and the small GTPase Rac2 were found to be upregulated (1.9fold and 11fold, respectively) in RA compared to OA tissue. In contrast, gp91 phox was neither detected on the RNA level nor on the protein level in the Powerblot™ of both tissues (data not shown). Strikingly, the p47phox gene has recently been shown to be an arthritis susceptibility gene in the rat, regulating the severity of the disease [47]. A natural occuring polymorphism was found that resulted in decreased NADPH oxidase activity and promoted activation of arthritogenic T cells. Activation of NADPH activity by pharmocological substances consequently led to amelioration of arthritis. This was surprising as increases in reactive oxygen radicals like superoxide are normally considered to play a promoting role in arthritis [48, 49]. The inventors are not aware of any study examining the available p47phox^{" "} knockout mice with respect to development of arthritis. However, a knockout-mouse of gp91 phox developed arthri- tis with similar kinetics and severity as control mice [50]. Clearly, more work is necessary to solve the exact role of NADPH oxidase in arthritis.

The third overexpressed gene product in RA versus OA that has functions in the metabolism of superoxide is the manganese superoxide dismutase (MnSOD). The superoxide dismutase enzyme family is essential for the elimination of tissue damaging superoxide. Thus, this enzyme family can be considered a counter-player to the superoxide producing NADPH oxidase system. MnSOD, a mitochondrial matrix protein, is the principal scavenger of superoxide that is leaking from mito- chondrial respiratory chains [51]. The fact that the inventors screen turned up this gene as well stresses the importance to study the functions of reactive oxygen species in arthritis. The anti- inflammatory effects of increased superoxide dismutase have been observed after gene transfer of the extracellular enzyme in the collagen-induced arthritis mouse model where it improved the clinical and histological score [52].

Proteases play important roles in tissue damage and destruction in the arthritic joint [53, 54]. The lysosomal and secreted aspartate protease cathepsin D has not been studied in much detail with respect to arthritis in contrast to e.g. cathepsins B and L. The inventors found higher protein but not RNA levels in the OA compared to RA synovial tissue. Cathepsin D was mainly observed in tissue macrophages by immunohistochemistry (data not shown). In the literature, elevated cathepsin D transcripts were described in interstitial and in part also in perivascular synovial regions in RA [55]. The protein abundance was not examined. High cathepsin D activity accompanied cartilage erosion in an in-vitro culture system of RA synovial cells in contrast to OA cultures [56]. There appear to be some contradictions to the inventors results, but conditions and what exactly was tested (transcript/protein/activity) were also quite different in all the studies.

In conclusion, in the present invention the inventors have described a number of differentially expressed genes in RA versus OA synovial tissues by expression profiling on the RNA and protein levels. Prominent examples included Statl , p47phox, MnSOD and cathepsin D. Discordant gene expression of respective transcripts and proteins emphasize the need of proteome analysis to un- derstand the pathological phenotype. The search for differentially expressed genes is an important means to find markers of disease for diagnosis and starting points for evaluation of pathophysi- ological pathways. Such differentially expressed genes are also candidates for studies on the post- translational level, functional investigations in animal models, e.g. employing knock-out or trans- genic strategies, and for searches for genetic polymorphisms. Table 1. Compilation of concordant expression changes between K55/RA and K42/OA synovial tissues on protein and transcript levels

Data based on Affymetrix decision matrix or BD Powerblot software evaluation. "Change", relative change of gene expression of K55/RA against K42/OA as baseline (I increase, D = decrease) Ace. no., Genebank accession number

Fiq. 1 :

Protein levels of Statl , Cathepsin D, p47phox, manganese superoxide dismutase (MnSOD) and CD3ε in synovial tissue extracts of eight patients each with osteoarthritis (OA, white bars) or rheumatoid arthritis (RA, black bars). Western blots with specific antibodies against respective proteins were analyzed by densitometry. Plotted is the mean volume of the signals ± S.D. for each patient group. The significance of the differences between the two patient groups is indicated by the as- teriks ^* (p < 0.05) and ** (p< 0.01 ).

Summing up, the global scale molecular profiling of diseased tissues is an important first step to unravel candidate target molecules that are involved in the pathogenesis of a disease. The inventors have performed a comparative molecular characterization on the transcriptome (microarray with 12526 gene specificities) and proteome level (multi-Western blot Powerblot™ with 791 antibodies) of synovial tissue from rheumatoid arthritis (RA) compared to osteoarthritis (OA) patients. From the panel of 791 antibodies 260 (33 %) detected their corresponding protein. Out of 58 unambigu- ous changes on the protein level only 16 coincided on the transcript level (28%>). Statl , p47phox and manganese superoxide dismutase were shown to be reproducibly overexpressed in RA versus OA synovial tissue by Western blots with a panel of 8 RA versus 8 OA samples. Cathepsin D was among the most prominent proteins scored to be underexpressed in RA by the Powerblot™ whereas no differences of the respective transcript was observed. The lower abundance of cathep- sin D protein in RA compared to OA tissue was reproduced in other patient samples as well. Im- munohistochemistry assigned the Statl protein in RA synovial tissue mainly to macrophages and T lymphocytes and the p47phox protein in particular to macrophages. In conclusion, the inventors approach provides new candidate molecules for further analysis of rheumatic diseases and stressed the importance of studies on the protein level.

In present claim 1 the 773 gene transcripts numbered 1 to 773 represented by GeneBank accession numbers AA017245 to Z24725 are expressed at a reduced level for RA in comparison to OA and are marked "D" ("decreased"), whereas the 546 gene transcripts numbered 774 to 1320 with GeneBank accession numbers AA005023 to Z83819 are expressed at increased levels for RA in comparison to OA and are marked "I" ("increased"). In present claim 1 the 88 gene products numbered 1321 to 1408 represented by GeneBank accession numbers AA888001 to Z18859 are expressed either at a reduced protein level for RA and then marked "D" ("decreased") or at an elevated level compared to OA and then marked "I" ("increased"). 5 References

[I] Brown, P. O., Botstein, D., Nat. Genet., 1999, 21 , 33-7. [2] Lipshutz, R. J., Fodor, S. P., Gingeras, T. R., Lockhart, D. J., Nat. Genet.. 1999, 21, 20-4.

[3] Lockhart, D. J., Winzeler, E. A., Nature. 2000, 405, 827-36.

[4] Anderson, L., Seilhamer, J., Electrophoresis, 1997, 1_8, 533-7.

[5] Anderson, N. L., Anderson, N. G., Electrophoresis, 1998, 19, 1853-61.

[6] Gygi, S. P., Rochon, Y., Franza, B. R., Aebersold, R., Mol. Cell. Biol.. 1999, 19, 1720-30. [7] Futcher, B., Latter, G. I., Monardo, P., McLaughlin, C. S., et al., Mol. Cell. Biol., 1999, J9, 7357-

68.

[8] Griffin, T. J., Gygi, S. P., Ideker, T., Rist, B., et al., Mol. Cell. Proteomics. 2002, 1, 323-33.

[9] Chen, G., Gharib, T. G., Huang, C. C, Taylor, J. M., et al., Mol. Cell. Proteomics, 2002, 1, 304-

13. [10] Klose, J., Electrophoresis. 1999, 20, 643-52. [pii].

[I I] Gygi, S. P., Corthals, G. L., Zhang, Y., Rochon, Y., et al., Proc. Natl. Acad. Sci. U. S. A.. 2000, 97, 9390-5.

[12] Rabilloud, T., Proteomics, 2002, 2, 3-10.

[13] Corthals, G. L., Wasinger, V. C, Hochstrasser, D. F., Sanchez, J. C, Electrophoresis, 2000, 21, 1104-15. [pii].

[14] Castedo, M., Ferri, K. F., Blanco, J., Roumier, T., et al., J. Exp. Med., 2001 , 194, 1097-110.

[15] Melnick, M., Chen, H., Min Zhou, Y., Jaskoll, T., BMC Dev. Biol.. 2001 , 1, 15.

[16] Templin, M. F., Stoll, D., Schrenk, M., Traub, P. C, et al., Trends Biotechnol., 2002, 20, 160-6.

[17] Feldmann, M., Brennan, F. M., Maini, R. N., CeH, 1996, 85, 307-10. [18] Feldmann, M., Nat. Immunol.. 2001 , 2, 771-3.

[19] Weyand, C. M., Goronzy, J. J., J. Mol. Med., 1997, 75, 772-85.

[20] Pelletier, J. P., Martel-Pelletier, J., Abramson, S. B., Arthritis Rheum., 2001 , 44- 1237-47.

[21] Poole, A. R., Front. Biosc 1999, 4, D662-70.

[22] Sprott, H., Pap, T., Rethage, J., Wintersberger, W„ et al., J Rheumatol. 2000, 27, 2347-50. [23] Konig, A., Krenn, V., Toksoy, A., Gerhard, N., et al., Virchows Arch., 2000, 436, 449-58.

[24] Lorenz, P., Pepperkok, R., Ansorge, W., Pyerin, W., J. Biol. Chem., 1993, 268, 2733-9.

[25] Ibrahim, S. M., Mix, E., Bottcher, T., Koczan, D., et al., Brain, 2001 , 124 , 1927-38.

[26] DeLeo, F. R., Quinn, M. T., J. Leukoc. Biol., 1996, 60, 677-91.

[27] Dever, T. E., CeH, 2002, 108, 545-56. [28] Hershko, A., Ciechanover, A., Annu Rev Biochem, 1998, 67, 425-79.

[29] Orntoft, T. F., Thykjaer, T., Waldman, F. M., Wolf, H., et al., Mol Cell Proteomics. 2002, 1, 37-

45. [30] Ideker, T., Thorsson, V., Ranish, J. A., Christmas, R., et al., Science, 2001 , 292, 929-34.

[31] Pap, T., Muller-Ladner, U., Gay, R. E., Gay, S., Arthritis Res., 2000, 2, 361-7.

[32] Harada, S., Yamamura, M., Okamoto, H., Morita, Y., et al., Arthritis Rheum., 1999, 42, 1508-

16. [33] Kinne, R. W., Brauer, R., Stuhlmuller, B., Palombo-Kinne, E., et al., Arthritis Res., 2000, 2,

189-202.

[34] Lock, C, Hermans, G., Pedotti, R., Brendolan, A., et al., Nat Med, 2002, 8, 500-8.

[35] Levy, D. E., Darnell, J. E., Jr., Nat. Rev. Mol. Cell Biol., 2002, 3, 651-62.

[36] Wong, M., Fish, E. N., J. Biol. Chem.. 1998, 273, 309-14. [37] Kovarik, P., Stoiber, D., Eyers, P. A., Menghini, R., et al., Proc. Natl. Acad. Sci. U. S. A., 1999,

96, 13956-61.

[38] Ohmori, Y., Hamilton, T. A., J. Leukoc. Biol., 2001 , 69, 598-604.

[39] Wang, F., Sengupta, T. K., Zhong, Z., Ivashkiv, L. B., J. Exp. Med., 1995, 182, 1825-31.

[40] Yokota, A., Narazaki, M., Shima, Y., Murata, N., et al., J. Rheumatol., 2001 , 28, 1952-9. r411 Gerard. C. Rollins, B. J., Nat Immunol., 2001 , 2, 108-15.

[42] Choy, E. H., Panayi, G. S., N Enql J Med, 2001 , 344, 907-16.

[43] Lehtonen, A., Matikainen, S., Julkunen, I., J. Immunol., 1997, 159, 794-803.

[44] Alexander, W. S., Nat. Rev. Immunol., 2002, 2, 410-6.

[45] Shouda, T., Yoshida, T., Hanada, T., Wakioka, T., et al., J. Clin. Invest, 2001 , 108, 1781-8. [46] Walder, C. E., Green, S. P., Darbonne, W. C, Mathias, J., et al., Stroke, 1997, 28, 2252-8.

[47] Olofsson, P., Holmberg, J., Tordsson, J., Lu, S., et al., Nat. Genetics, 2003, 33, 25-32.

[48] Miesel, R., Kurpisz, M., Kroger, H., Free Radic. Biol. Med., 1996, 20, 75-81.

[49] Bauerova, K., Bezek, A., Gen. Phvsiol. Biophys.. 1999, 18, 15-20.

[50] Wipke, B. T., Allen, P. M., J. Immunol., 2001. 167. 1601-8. [51] Fridovich, I., Annu. Rev. Biochem., 1995, 64. 97-112.

[52] lyama, S., Okamoto, T., Sato, T., Yamauchi, N., et al., Arthritis Rheum.. 2001 , 44, 2160-7.

[53] Keyszer, G., Redlich, A., Haupl, T., Zacher, J., et al., Arthritis Rheum., 1998, 41, 1378-87.

[54] Cunnane, G., FitzGerald, O., Hummel, K. M., Youssef, P. P., et al., Arthritis Rheum., 2001 , 44,

1744-53. [55] Keyszer, G. M., Heer, A. H., Kriegsmann, J., Geiler, T., et al., Arthritis Rheum., 1995, 38, 976-

84.

[56] Ermis, A., Muller, B., Hopf, T., Hopf, C, et al., J. Rheumatol., 1998, 25, 208-13.

Claims

Method for diagnosing the presence or absence of rheumatoid arthritis or osteoarthritis, or of a predisposition therefor, using expression profiling data comprising the steps of:

(a) determining in cells obtained from a sample of an individual to be diagnosed for rheumatoid arthritis or osteoarthritis, or for a predisposition therefor, qualitatively and quantitatively the expression of a suitable number of genes selected from the group consisting of genes coding for:

and allelic forms or mutants or splice variants thereof,

to obtain expression profiling data of these genes which are representative for the current status of the individual to be diagnosed,

(b) comparing the expression profiling data obtained in step (a) with the expression profiling data of normal individuals and/or with the expression profiling data of individuals suffering from rheumatoid arthritis or osteoarthritis, or having a predisposition therefor, and

(c) excluding or diagnosing rheumatoid arthritis or osteoarthritis, or a predisposition therefor, on the basis of the results of the comparison made in step (b).

2. Method according to claim 1 , wherein in step (a) the expression profiling data of said genes are determined using nucleic acid probes specifically hybridizing with said genes and/or using antibodies specifically binding to proteins encoded by said genes.

3. Method according to claim 2 wherein said nucleic acid is a DNA or RNA.

4. Method according to claim 3 wherein said DNA or RNA is an oligonucleotide or an ap- tamer.

5. Method according to claim 2, wherein said antibodies are polyclonal, monoclonal, chimeric or single-chain antibodies or functional fragments or derivatives of such antibodies.

6. Method according to anyone of claims 2 to 5, wherein for detecting successful nucleic acid hybridization or antibody binding a colored, fluorescent, bioluminescent, chemolumines- cent, phosphorescent or radioactive label, an enzyme, an antibody or a functional fragment or derivative thereof, a protein A/gold based system, a biotin/avidin/streptavidin based system or an enzyme electrode based system is used.

7. Method according to anyone of the preceeding claims, wherein in step (a) as defined in claim 1 said sample is selected from blood, synovial tissues or synovial fluids.

8. Biosensor chip comprising immobilized on its surface an addressable pattern of nucleic acid probes specifically hybridizing with the genes defined in claim 1 or an addressable pattern of antibodies specifically binding to proteins encoded by said genes.

9. Medical or diagnostic instrument comprising a biosensor chip according to claim 8.

10. Method for monitoring the therapeutical effect of an anti-rheumatoid arthritis or anti- osteoarthritis drug or for screening for a potential anti-rheumatoid arthritis or anti-osteoarthritis drug using the method of anyone of claims 1 to 7, or a biosensor chip according to claim 8, or a medical or diagnostic instrument according to claim 9, comprising the steps of:

(a) preparing expression profiling data characteristic for a normal healthy status,

(b) preparing expression profiling data characteristic for a rheumatoid arthritis or osteoarthritis status in the absence and in the presence of an anti-rheumatoid arthritis or anti-osteoarthritis drug or of a potential anti-rheumatoid arthritis or anti-osteo- arthritis drug, and (c) evaluating the therapeutic effect by comparing the expression profiling data obtained in steps (a) and (b).

11. Method for the production of an anti-rheumatoid arthritis or anti-osteoarthritis drug compris- ing the steps of the method of claim 10 and then mixing the compound(s) identified, or a pharmaceutically acceptable salt, metal complex or ester thereof, in a therapeutically effective amount with a pharmaceutically acceptable carrier and/or diluent and/or adjuvant.

12. Anti-rheumatoid arthritis or anti-osteoarthritis drug obtained by a method according to claim 11.