Human TNF-α transgenic (hTNFtg) mice develop erosive arthritis closely resembling rheumatoid arthritis (RA). To investigate mechanisms leading to pathological autoimmune reactions in RA, we examined hTNFtg animals for the presence of RA-associated autoantibodies including Abs to citrullinated epitopes (anti-cyclic citrullinated peptide), heterogeneous nuclear ribonucleoprotein (hnRNP)-A2 (anti-RA33), and heat shock proteins (hsp) (anti-hsp). Although IgM anti-hsp Abs were detected in 40% of hTNFtg and control mice, IgG anti-hsp Abs were rarely seen, and anti-cyclic citrullinated peptide Abs were not seen at all. In contrast, >50% of hTNFtg mice showed IgG anti-RA33 autoantibodies, which became detectable shortly after the onset of arthritis. These Abs were predominantly directed to a short epitope, which was identical with an epitope previously described in MRL/lpr mice. Incidence of anti-RA33 was significantly decreased in mice treated with the osteoclast inhibitor osteoprotegerin and also in c-fos-deficient mice lacking osteoclasts. Pronounced expression of hnRNP-A2 and a smaller splice variant was seen in joints of hTNFtg mice, whereas expression was low in control animals. Although the closely related hnRNP-A1 was also overexpressed, autoantibodies to this protein were infrequently detected. Because expression of hnRNP-A2 in thymus, spleen, brain, and lung was similar in hTNFtg and control mice, aberrant expression appeared to be restricted to the inflamed joint. Finally, immunization of hTNFtg mice with recombinant hnRNP-A2 or a peptide harboring the major B cell epitope aggravated arthritis. These findings suggest that overproduction of TNF-α leads to aberrant expression of hnRNP-A2 in the rheumatoid joint and subsequently to autoimmune reactions, which may enhance the inflammatory and destructive process.

The pathogenesis of rheumatoid arthritis (RA)4 is still unresolved. Although effects mediated by proinflammatory cytokines are pivotal in the development of this chronic destructive disorder (1), the strong association of RA with MHC class II genes, the frequent presence of autoantibodies, and the efficacy of T cell-directed therapies suggest involvement of the adaptive immune response, primarily in the initial phases of the disease but also in its subsequent course (2, 3, 4). With respect to autoimmune responses, apart from rheumatoid factor (RF), i.e., autoantibody to IgG, autoantibodies to a variety of Ags have been described in recent years (5). In particular, the discovery of autoantibodies to citrullinated proteins such as fibrin or vimentin in patients with RA was one of the most important finding in rheumatology research, pointing to the potential importance of posttranslational modifications in the generation of autoepitopes (6, 7). Another well-characterized autoantigen of interest is the heterogeneous nuclear ribonucleoprotein (hnRNP)-A2, also known as the RA33 Ag. Autoantibodies to this protein (anti-RA33) are detectable in one third of RA patients and show similar specificity for RA as RF (8, 9). Importantly, pronounced Th1-like reactivity to hnRNP-A2 has recently been described to occur in ∼50% of RA patients but not in disease controls, suggesting possible pathogenic involvement of this autoreactivity (10).

Autoantibodies may be already present very early or even years before the onset of disease, as recently demonstrated for RF and anti-cyclic citrullinated peptide (CCP) Abs (11, 12). This indicates a role of such autoimmune response in the pathogenesis of RA, but a definitive proof for this is still lacking. In contrast, there is abundant evidence for the role of cytokines in RA synovitis and the ensuing joint destruction. In particular, TNF-α, IL-1, and IL-6 appear to constitute the most important proinflammatory mediators of the RA process. The role of TNF-α is undisputed, given its abundant presence in the joint and the successful therapeutic intervention with biological agents inhibiting TNF-α (1). Another important piece of evidence for the pivotal role of TNF-α in RA pathogenesis stems from the observation that mice carrying human TNF-α as transgene develop a severe erosive inflammatory polyarthritis demonstrating typical features of human RA, such as synovial hypercellularity, inflammatory infiltrates, pannus formation, cartilage destruction, bone erosions, and, finally, crippling of paws (13). Other experimental models in which arthritis is triggered by defined Ags, such as type II collagen, and in which proinflammatory cytokines including TNF-α and IL-1 also play a major role, provide evidence for the importance of (auto)immune mechanisms. Most noteworthy, autoimmunity is clearly the trigger in the KRN × NOD model in which an autoreactive transgenic TCR drives formation of autoantibodies to the enzyme glucose-6-phosphate isomerase, leading to a severe and destructive arthritis (14, 15).

Despite these compelling data, the initial processes leading to loss of tolerance against “self” proteins in autoimmune diseases are still in the dark. Thus, it is not clear which events lead to activation of autoreactive T cells and the formation of autoantibodies and why only a limited number of Ags appear to be targeted in RA. In previous studies (16, 17), we have extensively used human TNF-α transgenic (hTNFtg) mice to investigate molecular and cellular mechanisms of tissue destruction. In the course of these studies, we became interested to know whether TNF-α-mediated inflammation could lead to loss of tolerance against self components. Therefore, we investigated sera of hTNFtg mice for the presence of autoantibodies typically found in RA including Abs to citrullinated epitopes, hnRNP-A2, and heat shock proteins (hsp). The data obtained reveal the frequent presence of IgG anti-RA33 autoantibodies in the sera of arthritic animals, whereas IgG Abs to citrullinated epitopes were not detectable, and IgG anti-hsp Abs were rarely seen. Furthermore, hnRNP-A2 was highly overexpressed in the joint of hTNFtg mice, and immunization of these mice with the recombinant protein aggravated disease, suggesting that TNF-α-driven autoimmune reactions to hnRNP-A2 may contribute to the inflammatory and destructive processes in the arthritic joint.

Tg197 hTNFtg mice carrying a 3′UTR-modified human TNF (hTNF)-α gene construct were used throughout this study (18). These mice develop a severe erosive inflammatory arthritis starting within 4–6 wk of age. In addition, sera from Tg5453 hTNFtg mice expressing only membrane-bound hTNF-α (19) and sera from the previously described Tg197 × c-fos−/− mice (16) that lack a functional c-fos gene were analyzed. Control strains included non-tg (transgenic) CBA × C57BL/6 littermates, BALB/c, C57BL/6, and DBA/1 mice obtained from Harlan-Winckelmann. All animal procedures were approved by the local ethics committee. Treatment of Tg197 mice with anti-TNF-α mAb (Infliximab; Centocor) and osteoprotegerin (OPG; Amgen) has been described in detail previously (20).

Clinical evaluation was performed in a blinded manner between 4 and 9 wk of age, as described previously (16). Briefly, joint swelling was examined using a clinical score graded from 0 to 3 (0, no swelling; 1, mild swelling; 2, moderate swelling; 3, severe swelling of toes and ankle). In addition, grip strength of each paw was analyzed on a wire 3 mm in diameter, using a score from 0 to −4 (0, normal grip; −1, mildly reduced grip strength; −2, moderately reduced grip strength; −3, severely reduced grip strength; −4, no grip strength at all).

Serial paraffin sections (1–2 μm) of hind and front paws and knees were stained with H&E for histological analyses of arthritis. Areas of inflammation and bone erosion were counted in mm2. To investigate expression of hnRNP-A2, paraffin sections were boiled in 0.01 M Na-citrate buffer (pH 6) in a microwave oven for 2 min at 780 W and for 10 min at 180 W. After cooling to room temperature, sections were stained with the 10D1 mAb (0.7 mg/ml) directed to hnRNP-A2 diluted 1/100–1/200 (10, 21). For detection of osteoclasts, sections were stained for tartrate-resistant acid phosphatase (TRAP) using the Leukocyte Acid Phosphatase staining kit (Sigma-Aldrich) as described previously (16).

For cellular stimulation and immunization experiments, the B1 splice variant of hnRNP-A2 (hnRNP-A2/B1) was used, which differs from hnRNP-A2 by a 12-aa insertion close to the N terminus. For stimulation assays, His-tagged hnRNP-A2/B1 was used (10); for immunizations, highly purified untagged recombinant protein (manufactured by BioMay) was used. For characterization of Ab binding regions, three overlapping recombinant fragments were used that contained one or both RNA recognition motifs (RRM): RRM1 (aa 1–89), RRM2 (aa 80–182), and RRM1 + 2 (aa 1–182) (22). For fine mapping of epitopes, a series of 13 overlapping synthetic peptides covering the N-terminal portion of hnRNP-A2 (aa 1–206) was used as described previously (23). Two selected peptides (p50-70 and p140-160) were used for immunization of hTNFtg and control mice.

Autoreactivities against nuclear proteins were analyzed by immunoblotting, using HeLa nuclear extracts as described previously (8). Nitrocellulose membranes were cut into strips and blocked with blocking buffer (10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 0.1% Triton X-100, 3% nonfat dried milk) for 1 h at room temperature. Strips were incubated for 45 min at room temperature with serum diluted 1/50 in blocking buffer. Human autoimmune sera served as positive controls for detection of autoantibodies to hnRNPs, Sm, U1, small nuclear RNP (snRNP), Ro, La, topoisomerase I, and ribosomal P proteins. After washing, strips were incubated for 30 min with alkaline phosphatase-conjugated anti-mouse IgG or anti-mouse IgM secondary Ab (Accurate Chemical & Scientific) diluted 1/1000 in blocking buffer. To quantitate the Ab response, selected sera showing moderate to high reactivity were serially diluted (up to 1/800) and probed with hnRNP-A2 partially purified from HeLa cells, as described previously (8).

For detection of autoantibodies to stress proteins, 10 μg each of recombinant mouse hsp60, mycobacterial hsp65, rat hsp70, and hamster hsp78 (BiP) obtained from StressGen Biotechnologies were separated by SDS-PAGE on preparative minigels (Bio-Rad) and transferred to nitrocellulose membranes; transfer efficiency was controlled by staining the membranes with Ponceau S. A mAb to hsp60 (clone LK1; StressGen Biotechnologies) served as positive control.

Mouse sera were analyzed for the presence of autoantibodies to citrullinated epitopes by a commercial ELISA (Axis Shields Diagnostics) in which a CCP is used as Ag. The assay was used according to the manufacturer’s instructions, except that HRP-conjugated anti-mouse IgG or IgM (Accurate Chemical & Scientific) diluted 1/2,000 was used as secondary Ab; anti-CCP-positive human sera served as positive controls. Occurrence of Abs in mice immunized with recombinant hnRNP-A2 was monitored by ELISA, using 2 μg of Ag per well and HRP-conjugated anti-mouse IgG (Accurate Chemical & Scientific) diluted 1/2,000 as secondary Abs. Reactivities to peptides were measured by ELISA as described previously (23). Briefly, microtiter plates (Falcon) were coated overnight at 37°C with 2 μM of each peptide diluted in 0.05 M carbonate buffer (pH 9.6) and blocked by adding PBS containing 0.05% Tween 20 and 0.5% BSA. As a control, mouse sera were also tested in a noncoated well incubated with coating buffer. Serum dilution was 1/1,000, and HRP-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories) diluted 1/20,000 was used as secondary Ab.

For detection of antinuclear IgG Ab, a commercial HEp-2 test kit (Hemagen Diagnostics) was used. Sera were initially diluted 1/25 in PBS and incubated for 1 h at room temperature in a humid chamber. After washing, slides were incubated with FITC-conjugated rabbit anti-mouse IgG (DakoCytomation) diluted 1/40 for 30 min and subsequently analyzed by fluorescence microscopy. Sera were obtained from hTNFtg mice either positive (n = 6) or negative (n = 2) for anti-RA33 Ab, non-tg littermates (n = 2), BALB/c mice immunized with hnRNP-A2 (n = 2), and naive BALB/c mice (n = 2).

After removal of skin and muscles, synovial tissue from knee joints and tarsal areas of hind and front paws were mechanically homogenized in Schreiber buffer (20 mM HEPES (pH 7.9), 0.4 M NaCl, 1.5 mM MgCl2, 1 mM DTT, 1 mM EDTA, 0.1 mM EGTA, 20% glycerol) using an Ultra-Thurrax homogenizer (24). Mouse organs were homogenized in the same buffer. The tissue extracts were then centrifuged for 15 min at 14,000 rpm. Joint (300 μg) and organ extracts (200 μg) were separated by SDS-PAGE on minigels (joint extracts) or large gels (organ extracts) and transferred onto nitrocellulose membranes. Membranes were blocked for 1 h at room temperature and subsequently incubated for 1 h at room temperature under constant shaking with mAb directed to hnRNP-A2 (10D1, diluted 1/500 in blocking buffer), hnRNP-A1 (4B10 (Ref.25), diluted 1/500), the A protein of the U1 snRNP (9A9 (Ref.26), diluted 1/100), or a rabbit anti-actin Ab (diluted 1/250; Sigma-Aldrich). After washing, the membrane was incubated for 45 min at room temperature with a HRP-conjugated secondary Ab (DakoCytomation) diluted 1/2000 in blocking solution. After washing, immunostained proteins were visualized using the ECL detection kit (ECL Western Blotting Detection Reagents; Amersham Biosciences).

Tg197 mice were immunized s.c. with recombinant hnRNP-A2/B1 or synthetic peptides emulsified in CFA (Sigma-Aldrich) at 4 wk of age and boosted 3 wk later with the same amount of Ag in IFA (Sigma-Aldrich). One group of hTNFtg mice was immunized with 100 μg of recombinant hnRNP-A2/B1 protein, one with 30 μg of peptide p50-70, and one with 30 μg of peptide p140-160. Controls included mice treated with PBS or adjuvant and untreated mice. Each group consisted of five animals. Disease activity was assessed weekly by evaluating paw swelling and grip strength, as described previously (16). Nine-week-old mice were sacrificed by cervical dislocation, and blood was taken by heart puncture. In addition, immunizations were performed in mouse strains with different MHC-backgrounds BALB/c (H-2d), CBA (H-2k), C57BL/6 (H-2b), and in arthritis-prone DBA/1 mice, which were additionally immunized with chicken type II collagen.

Lymph node and spleen cells were isolated by standard procedures and resuspended at a concentration of 5 × 106 cells/ml in culture medium, and 100 μl of this suspension were added to microtiter wells (Costar) containing 100 μl of medium, with different concentrations of protein or peptides as described previously (23). Each concentration was tested in triplicate, and tests were repeated at least three times in independent experiments. After 24 h, 50 μl of supernatant were taken to analyze for the production of IFN-γ, IL-2, or IL-4 by ELISA (BD Pharmingen). After 54 h, the cultures were pulsed for 18 h with 1 μCi/well [3H]thymidine (Amersham Biosciences). The cells were subsequently harvested on filters using an automatic cell-harvesting device (Packard Instrument), and DNA-incorporated radioactivity was measured using a Matrix 9600 direct beta counter (Packard Instrument). For control stimulations, cells were incubated with 5 μg/ml Con A (Sigma-Aldrich).

Histological data are given as mean ± SEM, and group mean values were compared by unpaired two-tailed Student’s t test (see Fig. 5). Differences between groups in the prevalence of anti-RA33 Ab were calculated using Fisher’s exact test (see Table II).

FIGURE 5.

Immunization of hTNFtg mice with hnRNP-A2/B1 and synthetic peptides. Animals were immunized s.c. at the age of 4 wk with either full-length recombinant hnRNP-A2/B1 or the two synthetic peptides p50-70 and p140-160 (see also Fig. 2); controls received CFA only. Animals were sacrificed at the age of 9 wk. A, Detection of Ab to HeLa nuclear proteins in sera of immunized wt and hTNFtg mice by immunoblotting. No reactivities can be seen in sera of peptide-immunized wt mice and also not in sera of hTNFtg mice immunized with p50-70, whereas pronounced reactivities are visible in sera of hTNFtg mice immunized with the complete Ag. Reactivities are visible in 2 of 5 sera from mice immunized with p140-160 and also in two sera from the CFA control group. Weak reactivity can be seen also in one serum of the untreated control group. B, Clinical evaluation of grip strength. No significant differences were seen among the four groups. However, the most pronounced decrease in grip strength was observed in mice immunized with the full-length Ag (red graph). C and D, Histological analysis of inflammation and bone erosion. The size of areas of inflammatory and eroded tissue was determined in a semiquantitative manner. As compared with CFA controls, inflammation (C) and bone erosion (D) was increased in mice immunized with hnRNP-A2/B1 or p50-70 (p < 0.05, Student’s t test).

FIGURE 5.

Immunization of hTNFtg mice with hnRNP-A2/B1 and synthetic peptides. Animals were immunized s.c. at the age of 4 wk with either full-length recombinant hnRNP-A2/B1 or the two synthetic peptides p50-70 and p140-160 (see also Fig. 2); controls received CFA only. Animals were sacrificed at the age of 9 wk. A, Detection of Ab to HeLa nuclear proteins in sera of immunized wt and hTNFtg mice by immunoblotting. No reactivities can be seen in sera of peptide-immunized wt mice and also not in sera of hTNFtg mice immunized with p50-70, whereas pronounced reactivities are visible in sera of hTNFtg mice immunized with the complete Ag. Reactivities are visible in 2 of 5 sera from mice immunized with p140-160 and also in two sera from the CFA control group. Weak reactivity can be seen also in one serum of the untreated control group. B, Clinical evaluation of grip strength. No significant differences were seen among the four groups. However, the most pronounced decrease in grip strength was observed in mice immunized with the full-length Ag (red graph). C and D, Histological analysis of inflammation and bone erosion. The size of areas of inflammatory and eroded tissue was determined in a semiquantitative manner. As compared with CFA controls, inflammation (C) and bone erosion (D) was increased in mice immunized with hnRNP-A2/B1 or p50-70 (p < 0.05, Student’s t test).

Close modal
Table II.

Anti-RA33 autoantibodies in hTNFtg mice treated with anti-TNF mAb, OPG, or lacking a functional c-fos genea

AnimalsAnti-RA33 Positive
nn%p valueb
Anti-TNF mAb treated 20 15 0.001 
OPG treated 12 0.05 
c-fos deficient 13 15 0.035 
Controls for anti-TNF trial 20 14 70  
Controls for OPG and c-fos trial 24 13 54  
AnimalsAnti-RA33 Positive
nn%p valueb
Anti-TNF mAb treated 20 15 0.001 
OPG treated 12 0.05 
c-fos deficient 13 15 0.035 
Controls for anti-TNF trial 20 14 70  
Controls for OPG and c-fos trial 24 13 54  
a

Mice were treated with either an anti-TNF-α mAb (Infliximab) or recombinant OPG, and anti-RA33 autoantibodies were measured by Western blotting at the end of the trial; in addition, autoantibodies were determined in c-fos-deficient Tg197 mice (16 ). A significant reduction of anti-RA33 autoreactivity was observed in all three groups.

b

Statistical significance vs untreated control group determined by Fisher’s exact test.

To elucidate pathways leading to loss of tolerance and formation of autoantibodies in RA, we were interested whether immune reactivities against autoantigens commonly targeted by patients with RA occur in hTNFtg mice, an animal model of inflammatory erosive arthritis that is entirely driven by TNF-α (13). Therefore, we investigated sera from two hTNFtg strains for the presence of IgG and IgM autoantibodies to citrullinated epitopes (anti-CCP), hnRNP-A2 (anti-RA33), and four different stress proteins including murine hsp60, mycobacterial hsp65, rat hsp70, and hamster hsp78 (BiP). Strains under investigation were strain Tg197 expressing hTNF-α in both soluble and membrane-bound form and strain Tg5453 expressing only the transmembrane form (18, 19). Using the commercially available anti-CCP ELISA, neither IgG nor IgM Abs to citrullinated epitopes were found (Table I). In contrast, IgM Abs to stress proteins were detected by immunoblotting in 40% of hTNFtg and control sera and were predominately directed to hsp70 and hsp78 (data not shown). IgG Abs, however, were observed in only 7% of hTNFtg sera and not at all in control sera (Table I).

Table I.

IgG autoantibodies to hnRNP-A2 (anti-RA33), citrullinated epitopes (anti-CCP), and hsp (anti-hsp) in hTNFtg micea

Mouse StrainNo.Age (weeks)Anti-RA33Anti-CCPAnti-hsp
Tg197 64 8–14 25 (39%) 0/20 ND 
Tg5453 91 8–16 55 (60%) 0/40 3/40 (7%) 
C57BL6 × CBA 20 8–14 0/20 0/15 
DBA/1-CIA 10 14 ND ND 
DBA/1-control 10 14 ND ND 
Mouse StrainNo.Age (weeks)Anti-RA33Anti-CCPAnti-hsp
Tg197 64 8–14 25 (39%) 0/20 ND 
Tg5453 91 8–16 55 (60%) 0/40 3/40 (7%) 
C57BL6 × CBA 20 8–14 0/20 0/15 
DBA/1-CIA 10 14 ND ND 
DBA/1-control 10 14 ND ND 
a

Sera from strain Tg197 (expressing both soluble and membrane-bound hTNF-α) and strain Tg5453 (expressing only membrane-bound hTNF-α), non-tg littermates (C57BL6 × CBA), and from DBA/1 mice immunized with collagen II (DBA/1-CIA) were investigated by Western blotting for the presence of anti-RA33 autoantibodies. In addition, anti-CCP Abs were measured by ELISA in 80 hTNFtg and control mice, and Abs to hsps including murine hsp60, mycobaterial hsp65, rat hsp70, and hamster hsp78 were determined by Western blotting in 55 animals. Anti-RA33 Abs were more prevalent and generally higher titered in strain Tg5453 (1/100–>1/800 in older animals), whereas titers of Tg197 mice did not exceed 1/400. IgG anti-hsp Abs were reactive with hsp70 and hsp65, IgM Abs (which were detected in 40% of hTNFtg and wt mice) reacted predominantly with hsp70 and hsp78. Data are given as numbers and percentages (in brackets) of positive sera.

In contrast to citrullinated Ags and hsps, pronounced IgG anti-RA33 reactivities were detected by immunoblotting in sera of hTNFtg mice. Interestingly, the autoantibody response was more prevalent and more pronounced in the transmembrane strain (Table I). Anti-RA33 Abs first became detectable at the age of 8 wk shortly after the onset of arthritis. Titer and incidence of Abs increased until the age of 14–16 wk when animals were sacrificed (Fig. 1, A and B). Thus, at this late stage, 50–70% of the mice were positive for IgG anti-RA33 Abs with titers ranging between 1/100 and >1/800, whereas IgM Abs were generally low-titered (1/50–1/100) and less frequently seen. Interestingly, IgG anti-RA33 Abs were only weakly reactive with the recombinant protein, whereas IgM Abs were similarly reactive with the natural and recombinant Ag (data not shown). Anti-RA33-positive sera produced a large speckled nuclear pattern on Hep-2 cells (Fig. 1,C) that is known as nuclear matrix pattern and typically produced by Abs to hnRNP proteins (27). A similar pattern was obtained with sera from BALB/c mice immunized with hnRNP-A2 (Fig. 1,C). Mouse sera recognized protein bands of 36, 38, and 39 kDa corresponding to hnRNP-A2, and its alternatively spliced variants hnRNP-B1 and -B2, and a 40-kDa protein that comigrated with the closely related hnRNP-A3 (28). Thus, hTNFtg mice showed a similar reactivity pattern as patients with RA (8). As can be seen in Fig. 1, apart from this set of proteins only a few, if any, other (nuclear) proteins were recognized. The double band migrating below the hnRNP-A2 band corresponded to the closely related hnRNP-A1 (29) and was recognized by ∼15% of the sera, most of which contained also pronounced reactivities to hnRNP-A2. Of note, nuclear Ags commonly targeted by patients with other systemic autoimmune diseases were not recognized by these sera, including Sm and U1 snRNP Ags, Ro, La or topoisomerase I. No significant difference in disease onset and severity was seen between mice with and without autoantibodies. Because in previous investigations RF had also not been detectable (18, 19), anti-RA33 appeared to be the only dominant RA-associated autoreactivity in hTNFtg animals.

FIGURE 1.

Autoantibodies to hnRNP-A2 in sera of hTNFtg mice. A, Sera from wt control and hTNFtg mice were analyzed for the presence of IgG autoantibodies against nuclear proteins (from HeLa cells) by immunoblotting. Control mice showed no reactivities (lanes 1–5), whereas hTNFtg mice developed Abs against hnRNP-A2 in the course of disease: no reactivities can be seen in sera from mice 4 wk old (lanes 6–10), whereas they are visible in three sera of mice aged 8 wk (lanes 11–15) and very pronounced in older animals suffering from severe arthritis (lanes 16–20). Sera stain the characteristic hnRNP-A2/B1/B2 triplett and additionally a 40-kDa protein presumably corresponding to hnRNP-A3. This pattern is similar to that produced by the human autoimmune serum shown in the last lane. In addition, some sera (week 8, 1st serum; week 14, 2nd and 3rd serum) stain the characteristic hnRNP-A1 band migrating below the hnRNP-A2 band. B, Sequential sera of three hTNFtg mice analyzed at week 5, 7, 9, 12, and 14. All mice became reactive with hnRNP-A2, and mouse 1 also reacted with hnRNP-A1. A human serum from a patient with RA is shown at right. C, Detection of antinuclear Abs by indirect immunofluorescence. Sera from a hTNFtg mouse positive for anti-RA33, a hTNFtg mouse negative for anti-RA33, a BALB/c mouse immunized with hnRNP-A2, and a naive BALB/c mouse were analyzed using Hep-2 cells as substrate. A speckled nuclear staining pattern is produced by the serum of the positive hTNFtg mouse and by the serum of the immunized BALB/c mouse; titers of hTNFtg sera ranged from 1/25 to 1/400.

FIGURE 1.

Autoantibodies to hnRNP-A2 in sera of hTNFtg mice. A, Sera from wt control and hTNFtg mice were analyzed for the presence of IgG autoantibodies against nuclear proteins (from HeLa cells) by immunoblotting. Control mice showed no reactivities (lanes 1–5), whereas hTNFtg mice developed Abs against hnRNP-A2 in the course of disease: no reactivities can be seen in sera from mice 4 wk old (lanes 6–10), whereas they are visible in three sera of mice aged 8 wk (lanes 11–15) and very pronounced in older animals suffering from severe arthritis (lanes 16–20). Sera stain the characteristic hnRNP-A2/B1/B2 triplett and additionally a 40-kDa protein presumably corresponding to hnRNP-A3. This pattern is similar to that produced by the human autoimmune serum shown in the last lane. In addition, some sera (week 8, 1st serum; week 14, 2nd and 3rd serum) stain the characteristic hnRNP-A1 band migrating below the hnRNP-A2 band. B, Sequential sera of three hTNFtg mice analyzed at week 5, 7, 9, 12, and 14. All mice became reactive with hnRNP-A2, and mouse 1 also reacted with hnRNP-A1. A human serum from a patient with RA is shown at right. C, Detection of antinuclear Abs by indirect immunofluorescence. Sera from a hTNFtg mouse positive for anti-RA33, a hTNFtg mouse negative for anti-RA33, a BALB/c mouse immunized with hnRNP-A2, and a naive BALB/c mouse were analyzed using Hep-2 cells as substrate. A speckled nuclear staining pattern is produced by the serum of the positive hTNFtg mouse and by the serum of the immunized BALB/c mouse; titers of hTNFtg sera ranged from 1/25 to 1/400.

Close modal

To characterize the autoimmune response to hnRNP-A2 in more detail, epitope mapping studies were performed using recombinant fragments encompassing either both RRMs (aa 1–182), or RRM1 (aa 1–89), or RRM2 (aa 80–182), as well as 13 overlapping peptides covering the N-terminal portion of the protein (aa 1–206). All sera reacted with the fragment containing both RRMs in a similar manner as with the complete protein, indicating that the C-terminal portion (which could not be expressed separately due to its high glycine content) did not contain a major epitope (data not shown). Subsequent peptide mapping studies led to identification of a dominant epitope between aa 50 and 70, which was recognized by the sera of all 11 mice tested (aged between 8 and 16 wk) but not by sera from five control animals (Fig. 2). Interestingly, in previous studies, the same sequence had been found to harbor a major epitope recognized by lupus-prone MRL/lpr mice (23). Two sera were additionally reactive with peptide p90-116 located in the N-terminal part of RRM2. Of note, neither of these two peptides was recognized by non-tg control mice immunized with recombinant hnRNP-A2, confirming previous observations (23).

FIGURE 2.

Epitope mapping of hnRNP-A2. A, Schematic representation of hnRNP-A2. The protein shows a modular structure consisting of two conserved RRMs and a glycine-rich auxiliary domain. The two most conserved parts of RRM1 and RRM2 are indicated by dark gray bars. The 13 peptides overlapping by 5–10 aas used for epitope mapping are shown. The two peptides used for immunization of hTNFtg mice are in black. Numbers indicate the N- and C-terminal amino acids of each peptide. B, ELISA reactivities to p50-70 and p140-160 in Tg197 (1234567891011 ) and age-matched non-tg control mice (1213141516 ). Values are given in arbitrary units, with 1000 U corresponding to an OD value of 1.0. All Tg197 mice clearly showed reactivity to p50-70 but did not react with p140-160, which contains the same conserved motif as p50-70. None of the control mice showed reactivity to any of the 13 peptides tested.

FIGURE 2.

Epitope mapping of hnRNP-A2. A, Schematic representation of hnRNP-A2. The protein shows a modular structure consisting of two conserved RRMs and a glycine-rich auxiliary domain. The two most conserved parts of RRM1 and RRM2 are indicated by dark gray bars. The 13 peptides overlapping by 5–10 aas used for epitope mapping are shown. The two peptides used for immunization of hTNFtg mice are in black. Numbers indicate the N- and C-terminal amino acids of each peptide. B, ELISA reactivities to p50-70 and p140-160 in Tg197 (1234567891011 ) and age-matched non-tg control mice (1213141516 ). Values are given in arbitrary units, with 1000 U corresponding to an OD value of 1.0. All Tg197 mice clearly showed reactivity to p50-70 but did not react with p140-160, which contains the same conserved motif as p50-70. None of the control mice showed reactivity to any of the 13 peptides tested.

Close modal

To elucidate the effects of anti-inflammatory and antidestructive therapies on the spontaneous autoimmune response to hnRNP-A2, mouse sera collected during previous therapeutic trials were assessed for the presence of anti-RA33 autoantibodies. Sera were from mice treated with either an anti-TNF-α mAb (Infliximab) or OPG, the natural inhibitor of osteoclast differentiation and proliferation. As reported previously, pannus formation and bone destruction were largely reduced by these treatments, whereas inflammation was inhibited only by the anti-TNF-α mAb but not by OPG (20). However, both therapies affected autoantibody formation: whereas autoantibodies were observed in >50% of untreated mice, their incidence was significantly lower in both therapy groups (Table II). To get further insight into the underlying mechanism, hTNFtg mice lacking a functional c-fos gene were examined for the presence of anti-RA33 autoantibodies. Osteoclast differentiation is impaired in c-fos-deficient mice, and consequently c-fos-deficient hTNFtg mice do not develop bone erosions, whereas joint inflammation is not reduced (16). The incidence of anti-RA33 autoantibodies was indeed decreased in c-fos-deficient Tg197 mice, confirming the results obtained with OPG-treated animals. Thus, these data suggested that osteoclasts may be involved in the generation of the autoimmune response to hnRNP-A2.

To study spontaneous T cell responses to hnRNP-A2, Tg197, and non-tg controls were sacrificed at different stages of the disease (8 wk, 13 wk, and 15 wk). Non-tg mice immunized with hnRNP-A2/B1 served as positive controls. Spleen cells and peripheral lymph node cells were stimulated with 5 μg/ml recombinant Ag or purified protein derivative as control, and proliferation as well as production of IFN-γ was determined. As compared with cells derived from wild-type (wt) animals, the mitogen (Con A)-induced response of Tg197 cells was reduced by ∼20%. However, neither hnRNP-A2/B1 nor purified protein derivative elicited proliferative responses, and T cell cytokines such as IFN-γ or IL-4 were not detected in culture supernatants. Because chronic exposure to TNF-α renders T cells anergic (30), assays were performed also in the presence of neutralizing anti-TNF-α mAb, which, however, had no or only insignificant effects on proliferation. Thus, no Ag-specific T cell reactivity could be detected in these primary culture assays, in contrast to immunized control animals in which both T and B cell responses were readily detectable (data not shown).

Investigations previously performed in human synovial tissue had revealed hnRNP-A2 to be highly overexpressed in synovial tissue of RA patients (10). To study expression in mouse joints, tissue sections of hTNFtg and wt control mice were analyzed by immunohistochemistry. Although in the joints of control animals expression of hnRNP-A2 was hardly detectable, massive expression was seen in the inflamed joints of hTNFtg mice (Fig. 3, A and B). The protein was highly expressed in synovial macrophages and fibroblasts, particularly at sites where synovial tissue invades the bone and also in chondrocytes of articular cartilage. Pronounced expression was also seen in multinucleated TRAP-positive osteoclasts close to the cartilage-pannus junction (Fig. 3, C and D).

FIGURE 3.

Immunohistochemical analysis of hnRNP-A2 expression in the joints of wt and hTNFtg mice. A, Joint of a wt mouse showing a monolayered synovial membrane surrounding the joint and articular cartilages covering healthy bones. Expression of hnRNP-A2 is virtually undetectable. B, Joint of a hTNFtg mouse characterized by a highly inflammatory multilayered pannus invading and eroding subchondral bone at sites of bone-cartilage junctions (arrow). Massive overexpression of hnRNP-A2 (brown staining) can be seen in inflammatory pannus tissue and also in chondrocytes of articular cartilage. C and D, Serial sections stained for hnRNP-A2/B1 (brown color; C) and TRAP (purple color; D). Pronounced expression of hnRNP-A2/B1 can be seen at the site where the pannus invades into subchondral bone. The majority of stained cells appear to be synovial fibroblasts and macrophages, but staining can also be observed in some large multinucleated cells showing an osteoclast-like phenotype as revealed by positive TRAP staining (arrow). Original magnification is ×100 (A and B) and ×400 (C and D).

FIGURE 3.

Immunohistochemical analysis of hnRNP-A2 expression in the joints of wt and hTNFtg mice. A, Joint of a wt mouse showing a monolayered synovial membrane surrounding the joint and articular cartilages covering healthy bones. Expression of hnRNP-A2 is virtually undetectable. B, Joint of a hTNFtg mouse characterized by a highly inflammatory multilayered pannus invading and eroding subchondral bone at sites of bone-cartilage junctions (arrow). Massive overexpression of hnRNP-A2 (brown staining) can be seen in inflammatory pannus tissue and also in chondrocytes of articular cartilage. C and D, Serial sections stained for hnRNP-A2/B1 (brown color; C) and TRAP (purple color; D). Pronounced expression of hnRNP-A2/B1 can be seen at the site where the pannus invades into subchondral bone. The majority of stained cells appear to be synovial fibroblasts and macrophages, but staining can also be observed in some large multinucleated cells showing an osteoclast-like phenotype as revealed by positive TRAP staining (arrow). Original magnification is ×100 (A and B) and ×400 (C and D).

Close modal

To further investigate the aberrant expression of hnRNP-A2 in the joints of hTNFtg mice, protein extracts were prepared from the knees and hind paws of hTNFtg and wt mice and analyzed by immunoblotting. This analysis confirmed the immunohistochemistry data, because the 10D1 mAb stained several protein bands in extracts from hTNFtg mice, whereas no staining was seen in extracts from wt animals (Fig. 4 A). However, the expression pattern was different from that obtained with a HeLa nuclear extract: whereas in HeLa cells four bands were stained (hnRNP-A2, its splice variants hnRNP-B1/B2, and the 40-kDa protein), neither hnRNP-B2 nor the 40-kDa band were detected in the joint extract. In contrast, and in contrast to HeLa cells, a smaller variant with an estimated molecular mass of 32-kDa appeared to be almost as strongly expressed as hnRNP-A2. This protein presumably represented a recently described hnRNP-B1 splice variant lacking exon 9 (31).

FIGURE 4.

Immunoblot analysis of hnRNP-A2 expression in joint and organs of wt and hTNFtg mice. A, Expression of hnRNP-A2 in the joint. Protein extracts were prepared from knees and hind paws of wt and hTNFtg mice and separated by SDS-PAGE on minigels (300 μg/lane). For comparison, a HeLa nuclear extract is shown on the left. No expression can be seen in extracts from wt mice, whereas the mAb 10D1 stains three distinct bands in extracts from hTNFtg animals corresponding to hnRNP-A2, hnRNP-B1, and the 32-kDa splice variant of hnRNP-B1. B, Comparison of expression of hnRNP-A2 (1st panel), hnRNP-A1 (2nd panel), and the U1A protein of the U1 snRNP (3rd panel). Joint extracts from three hTNFtg and three wt mice were probed with mAb to these proteins. Expression of actin served as loading control (4th panel). Both hnRNP-A2/B1 and hnRNP-A1 (and the A1b splice variant) are overexpressed in joints of hTNFtg mice, whereas expression of the U1A protein appears to be only slightly increased in hTNFtg mice. C, Expression of hnRNP-A2 in organs. Organ extracts were prepared from spleen, thymus, lung, kidney, liver, heart, skeletal muscle, and brain, and separated by SDS-PAGE on large gels (200 μg/lane). In spleen, lung, thymus, and brain, pronounced staining of bands corresponding to hnRNP-A2/B1/B2 can be seen. In addition, a 40-kDa band is strongly stained in thymus and brain and weakly in spleen and lung. In brain a 41-kDa band is detected, which is the only band visible in liver and kidney extracts. The 31-kDa band visible in brain and the 32-kDa band visible in lung extracts represent the smaller splice variants of hnRNP-A2 and hnRNP-B1, respectively.

FIGURE 4.

Immunoblot analysis of hnRNP-A2 expression in joint and organs of wt and hTNFtg mice. A, Expression of hnRNP-A2 in the joint. Protein extracts were prepared from knees and hind paws of wt and hTNFtg mice and separated by SDS-PAGE on minigels (300 μg/lane). For comparison, a HeLa nuclear extract is shown on the left. No expression can be seen in extracts from wt mice, whereas the mAb 10D1 stains three distinct bands in extracts from hTNFtg animals corresponding to hnRNP-A2, hnRNP-B1, and the 32-kDa splice variant of hnRNP-B1. B, Comparison of expression of hnRNP-A2 (1st panel), hnRNP-A1 (2nd panel), and the U1A protein of the U1 snRNP (3rd panel). Joint extracts from three hTNFtg and three wt mice were probed with mAb to these proteins. Expression of actin served as loading control (4th panel). Both hnRNP-A2/B1 and hnRNP-A1 (and the A1b splice variant) are overexpressed in joints of hTNFtg mice, whereas expression of the U1A protein appears to be only slightly increased in hTNFtg mice. C, Expression of hnRNP-A2 in organs. Organ extracts were prepared from spleen, thymus, lung, kidney, liver, heart, skeletal muscle, and brain, and separated by SDS-PAGE on large gels (200 μg/lane). In spleen, lung, thymus, and brain, pronounced staining of bands corresponding to hnRNP-A2/B1/B2 can be seen. In addition, a 40-kDa band is strongly stained in thymus and brain and weakly in spleen and lung. In brain a 41-kDa band is detected, which is the only band visible in liver and kidney extracts. The 31-kDa band visible in brain and the 32-kDa band visible in lung extracts represent the smaller splice variants of hnRNP-A2 and hnRNP-B1, respectively.

Close modal

To control for specificity of expression, we additionally investigated expression of two structurally and/or functionally related nuclear proteins: hnRNP-A1, which shows 70% sequence identity with hnRNP-A2 (29); and the A protein of the U1 snRNP (U1-A), which, except for the presence of two RRMs, is structurally unrelated to hnRNP-A2 but colocalizes with hnRNP proteins in spliceosomal complexes (32). This analysis revealed hnRNP-A1 to be overexpressed in the joint in a similar manner as hnRNP-A2, whereas expression of U1-A was clearly observed in joints of both wt and hTNFtg mice and appeared to be only slightly increased hTNFtg animals (Fig. 4 B).

Finally, we analyzed expression of hnRNP-A2 (and cross-reacting proteins) in mouse organs. Protein extracts from thymus, spleen, liver, kidney, brain, lung, heart, and skeletal muscle were separated by SDS-PAGE and analyzed by immunoblotting (Fig. 4,C). The most pronounced expression of hnRNP-A2 was detected in the thymus and spleen. In these extracts, the mAb recognized the characteristic triplett (hnRNP-A2/B1/B2) plus a 40-kDa protein that appeared to be identical with the 40-kDa protein observed in HeLa extracts. Remarkably, however, the 32-kDa variant was not detected. Strong expression of hnRNP-A2 was further detected in lung and brain, whereas no expression was seen in heart and skeletal muscle. In lung the 32-kDa hnRNP-B1 variant was expressed, whereas in brain a 31-kDa protein was detected, which was most likely identical with a recently described hnRNP-A2 splice variant lacking exon 9 (31). In brain a 41-kDa band was seen additionally, which appeared to be selectively expressed in kidney and liver, whereas expression of hnRNP-A2/B1/B2 was undetectable in these two organs. Interestingly, in these analyses no differences were observed between organs from wt and hTNFtg mice (data not shown). Thus, overexpression of hnRNP-A2 appeared to exclusively occur in the joint of hTNFtg mice and, furthermore, resulted in a unique expression pattern (hnRNP-A2/B1 and the 32-kDa variant) that was not seen in any of the organs analyzed (Table III).

Table III.

Expression profiles of hnRNP-A2 and related proteins reactive with mAb 10D1 in mouse organsa

Molecular Mass kDaProtein hnRNPSpleenThymusBrainLungLiverKidneyJoint (normal)Joint hTNFtg
41 A3b − − ++ − ++ − − 
40 A3b +/− ++ +/− − − − − 
39 B2 +++ +++ ++ − − − − 
38 B1 +++ +++ ++ − − − 
36 A2 +++ +++ ++ ++ − − − ++ 
32 B1bc − − +/− ++ − − − ++ 
31 A2bc − − ++ − − − − − 
Molecular Mass kDaProtein hnRNPSpleenThymusBrainLungLiverKidneyJoint (normal)Joint hTNFtg
41 A3b − − ++ − ++ − − 
40 A3b +/− ++ +/− − − − − 
39 B2 +++ +++ ++ − − − − 
38 B1 +++ +++ ++ − − − 
36 A2 +++ +++ ++ ++ − − − ++ 
32 B1bc − − +/− ++ − − − ++ 
31 A2bc − − ++ − − − − − 
a

Note the unique expression pattern (A2/B1/B1b) that was exclusively observed in the joints of hTNFtg mice and the absence of these proteins in the joints of normal control animals.

b

Assumed to be identical with hnRNP-A3 isoforms based on migration in SDS-PAGE (28 ).

c

Smaller splice variants of hnRNP-A2/B1 lacking exon 9 (31 ).

To investigate whether immunization could induce disease in non-tg mice, arthritis-prone DBA/1 mice were immunized with hnRNP-A2/B1 or peptide p50-70 or collagen II as positive control. Although a strong Ab response was evoked, no signs of arthritis emerged within the observation period (up to 6 mo), and spreading of the Ab response to other nuclear proteins was also not observed (data not shown).

Because we assumed that the failure to induce arthritis was due to the low expression of hnRNP-A2 in the joints of healthy mice, we next investigated whether immunization with hnRNP-A2/B1 would influence the progression of arthritis in hTNFtg mice. Animals (five mice per group) were immunized s.c. at the age of 4 wk, i.e., before onset of clinical disease, with either the recombinant protein, with peptide p50-70 or with peptide p140-160. As described above, p50-70 contains a major B cell epitope, whereas p140-160, despite its structural similarity with p50-70, is not targeted by autoantibodies. Control animals received adjuvant only. Blood was taken after 9 wk when the animals were sacrificed. As can be seen in Fig. 5 A, a pronounced Ab response against hnRNP-A2 was observed in 4 of 5 hTNFtg mice immunized with the full-length Ag: sera recognized the characteristic set of protein bands between 36 and 40 kDa, and similar responses were seen in wt mice (data not shown). In contrast, none of the hTNFtg mice immunized with p50-70 reacted with any of these bands while showing pronounced reactivity in the peptide-specific ELISA (data not shown). Remarkably, 2 of 5 hTNFtg mice immunized with p140-160 and one mouse immunized with CFA developed a relatively strong reactivity to hnRNP-A2, which was much more pronounced than the reactivities seen in untreated mice.

To study the effects of immunization on the development of arthritis, clinical assessment of grip strength and paw swelling was performed weekly (Fig. 5,B). To evaluate histological effects, the area of inflammatory synovial tissue and the area of bone erosions was assessed in the joints and tarsal area of the hind paws. Although clinical evaluation did not show significant differences among the four groups, histological analyses, performed blinded for treatment groups, revealed a significantly higher (p < 0.05) degree of inflammation in mice immunized with either full-length Ag or p50-70, whereas animals immunized with p140-160 did not differ from controls that had received CFA only (Fig. 5,C). In line with the increased inflammatory response, areas of erosions were twice as large as those of controls (Fig. 5 D). Thus, immunization with the full-length protein or the major B cell epitope aggravated disease, suggesting potential involvement of the anti-RA33 autoimmune response in TNF-driven erosive arthritis.

A remarkable feature of systemic autoimmune diseases is the production of autoantibodies directed to ubiquitously expressed proteins. Most of the major autoantigens are part of large nuclear complexes composed of DNA or RNA and multiple proteins such as nucleosomes, spliceosomal complexes, or ribosomes. Although autoantibodies against these structures may be generated in a rather disease-specific manner, the pathogenetic role of most of these Ags has remained elusive (33, 34, 35). RA differs from other systemic autoimmune diseases both clinically and serologically. Thus, RA is characterized by the occurrence of high-titered RF and autoantibodies against citrulline-containing proteins such as fibrin or vimentin, whereas nuclear Ags are rarely targeted (5, 7, 36, 37). An exception is hnRNP-A2, which seems to form a relatively disease-specific target for autoantibodies and autoreactive Th1-like cells (8, 9, 10). Despite this finding, and although RF and anti-CCP Abs are significantly associated with a more severe disease course, it is still unclear which role these autoimmune responses play in the pathogenesis of erosive arthritis, which is a unique feature of RA that is not seen in other rheumatic disorders.

In contrast to systemic lupus erythematosus (SLE), no spontaneous animal model of RA exists, and, therefore, several inducible or tg models are being used that cover only certain aspects of the disease. Thus, hTNFtg mice are an excellent model to study cellular and molecular mechanisms triggered by TNF-α, i.e., by a state of chronic inflammation, whereas they are less suitable to study the initial events that lead to TNF-α overproduction (13). In contrast, because autoimmune reactions may arise as a consequence of chronic inflammatory processes, we considered this model quite useful for investigating this aspect. Thus, it is now widely assumed that chronic inflammation may lead to aberrant expression of self proteins thereby creating neoepitopes, which may become targets of autoreactive T and B cells. Potential mechanisms include overexpression over longer periods leading to antigenic overload, aberrant localization, overexpression of minor splice variants, or posttranslational modifications (7, 38, 39, 40, 41, 42). In this respect TNF may play a dual role, being able to both promote and abrogate pathological autoimmune reactions (43).

The data presented in this study further confirm that such mechanisms exist, and that they can lead to the induction of an autoimmune response that is characteristically observed in patients with RA. Thus, abundant overexpression of hnRNP-A2 was detected in the joints of hTNFtg mice, and the animals developed autoantibodies to this protein, which increased in titer as disease progressed. In contrast, therapies reducing joint destruction, and particularly bone erosion, also significantly reduced the autoantibody response, even if inflammation (and synovial overexpression of hnRNP-A2) persisted. This was most impressively seen in c-fos-deficient hTNFtg mice, which developed no erosions at all, although the inflammatory state of their joints was as severe as in hTNFtg animals (16).

Apart from hnRNP-A2 and its major splice variant hnRNP-B1, a 32-kDa protein was expressed that presumably corresponded to a recently described smaller splice variant reported to be expressed in testis, brain, and skin (31). Remarkably, in our analyses this variant was undetectable in lymphoid organs. This suggests that aberrant expression of the smaller variant in the inflamed joint might form the molecular basis for pathological autoimmune reactions to an Ag that, due to its high expression in lymphoid organs, is normally tolerized by the immune system. An analogous observation was recently made in rats susceptible to experimental autoimmune encephalitis, where T cell autoimmunity was directed to a brain-specific splice variant of proteolipid protein that was not expressed in the thymus (44). In this context, the identification of a major epitope in RRM1 (p50-70) was of particular interest because the same epitope had previously been found to be targeted by MRL/lpr mice, one of the most widely used models of SLE (23). In contrast to other lupus models, MRL/lpr mice may develop erosive arthritis and RF and suffer from a SLE/RA overlap disease with SLE symptoms predominating (45). Interestingly, the epitope was not recognized upon immunization with the full-length Ag and thus may represent a cryptic epitope that under normal conditions is not exposed to the immune system (46, 47, 48, 49, 50).

Remarkably, Ab to hnRNP-A1, which shows ∼70% identity with hnRNP-A2 (29, 32), were rarely detectable, although this protein was as highly expressed in the joint as hnRNP-A2. Of note, hnRNP-A1 is also infrequently targeted in human RA and if so mostly by cross-reacting anti-hnRNP-A2 Ab (8). Thus, even if aberrant expression of a self protein may be necessary for induction of autoimmune reactions, this is certainly not sufficient, and other factors including posttranslational modifications, expression of unusual splice variants, and, last but not least, immunogenetic factors may substantially contribute. The fact that posttranslational modifications of hnRNP-A2 (e.g., phosphorylation or methylation) may play an important role in the generation of the autoantibody response is strongly suggested by the weaker reactivity of IgG anti-RA33 Ab with the recombinant protein. Thus, autoantibodies may be primarily directed to modified epitopes, or (and more likely) the modifications may induce structural changes that might reveal cryptic epitopes such as p50-70. HnRNP-A2 is known to become reversibly phosphorylated and methylated in vivo, and methylation has been recently shown to be essential for nuclear localization of this protein (51).

Interestingly, hTNFtg animals developed neither RF nor anti-CCP Abs and, apart from a few exceptions, no IgG Ab to stress proteins, although IgM anti-hsp reactivities were clearly seen in ∼40% of hTNFtg and control mice. Stress proteins are of particular interest because autoimmunity to them has been suggested to play a (possibly beneficial) role in the pathogenesis of RA (52, 53, 54), and overexpression of hsp60 and other stress proteins has been observed in the joints of mice with adjuvant arthritis as well as in patients with RA (52, 55, 56, 57). In contrast, anti-hsp autoimmunity does not seem to be specific for any (autoimmune) disease and may commonly occur during infections and even in healthy individuals (58, 59, 60). Because it is plausible to assume that hsps were also overexpressed in the inflamed joint of hTNFtg mice, it was somewhat unexpected that the immune system of the animals appeared to be tolerant to these proteins, despite the presence of preformed IgM autoreactivities. This finding indicates that hTNFtg mice were unable to mount a potent T cell response to stress proteins, which might be predominantly of an anti-inflammatory nature as described for patients with juvenile idiopathic arthritis (53, 61).

So far, anti-CCP Abs have been exclusively found in humans and not in animal models of arthritis and other inflammatory diseases, although synovial expression of citrullinated proteins was observed in various arthritis models (62, 63). In patients with RA, anti-CCP Abs seem to be closely linked to the presence of the shared epitope, a pentameric sequence found in RA-associated HLA-DR alleles (64, 65). Thus, it may be possible that citrullinated epitopes cannot be efficiently presented by murine MHC class II molecules. However, it should be taken into consideration that the anti-CCP assay was developed for detection of human autoantibodies, and, therefore, it is possible that mice may generate an autoimmune response to a citrullinated epitope that does not cross-react with the cyclic peptide used in the anti-CCP assay. However, sera of hTNFtg mice were also not reactive with citrullinated (human) filaggrin and fibrinogen (our unpublished observation), supporting the assumption that these mice do not mount an autoimmune response to citrullinated epitopes. Although these findings cannot be completely extrapolated to the human disease, they may allow the hypothesis that anti-CCP Ab as well as RF are not induced by an inflammatory process, which is bolstered by the very early appearance of these Abs often years before clinical symptoms manifest (11, 12). Nevertheless, further studies using murine proteins are required to definitely answer this question.

Taken together, these observations suggest a cascade of events leading to the generation of pathological autoimmune reactions against hnRNP-A2 in the inflamed joint: 1) TNF-triggered inflammation induces aberrant expression of a protein that under normal conditions is not or only weakly expressed in the joint; 2) aberrant expression leads to aberrant presentation of the Ag by Ag-presenting cells and subsequently to the activation of autoreactive T cells and the generation of autoantibodies; 3) this process is enhanced by tissue destruction leading to the release of large amounts of autoantigens; 4) autoantibodies and autoreactive T cells may further enhance the inflammatory and destructive processes by forming immune complexes, activation of the complement cascade, and recruitment of macrophages and other inflammatory cells, establishing a vicious circle in which inflammation and autoimmunity mutually enhance each other until the target tissue is destroyed.

Thus, even if the autoimmune response to hnRNP-A2 arises secondarily to tissue destruction, it may nevertheless contribute to the pathophysiology of erosive arthritis. The arthritis-enhancing effect of immunization with either the complete protein or the peptide containing the major B cell epitope supports such assumption. Nevertheless, our data do not exclude a primary role for this autoimmune reaction in human disease, at least in those patients who show autoantibodies to hnRNP-A2 very early in the course of their disease (66, 67).

So far, there is only limited knowledge available on the functions of hnRNP-A2 and its variants. Functions ascribed to hnRNP-A2/B1 include roles in regulation of alternative splicing, mRNA transport, and translation (68, 69, 70), whereas nothing is known about the smaller variants that show a more restricted expression pattern. They appear to be developmentally regulated because they are mainly expressed in young animals (71). Therefore, their presence in inflamed tissue of older mice and patients with RA (our unpublished observation) is intriguing, pointing to an unusual role in cells exposed to proinflammatory and stressful conditions. Of note, several studies from other investigators have reported overexpression and cytoplasmic accumulation of hnRNP-A2 in various kinds of cancers, suggesting a functional role in the altered cellular metabolism of transformed cells (72, 73, 74).

In summary, the data obtained in the course of our investigations are in line with the hypothesis of the “altered” self, suggesting that altered expression of an autoantigen in the course of an inflammatory process may lead to altered processing, altered Ag presentation, and, finally, to activation of autoreactive T and B cells. Thus, our observations demonstrate a molecular link between inflammation, ensuing tissue destruction and generation of a potentially pathogenic autoimmune response in an animal model of RA. However, altered self per se is obviously not sufficient to induce an autoimmune response, and it will be a challenging task for the future to elucidate the mechanisms that render only a few of numerous potential candidate autoantigens immunogenic in patients with autoimmune diseases and the respective animal models.

We thank Elisabeth Höfler for expert technical assistance with autoantibody detection, Jean-Paul Briand (Institut de Biologie Moleculaire et Cellulaire, Centre National de la Recherche Scientifique, Strasbourg, France) for the synthesis of hnRNP-A2 peptides, and Walther van Venrooij (Radboud University of Nijmegen, Nijmegen, The Netherlands) for providing the mAb 9A9.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by the Center of Molecular Medicine of the Austrian Academy of Sciences, Vienna.

4

Abbreviations used in this paper: RA, rheumatoid arthritis; RF, rheumatoid factor; hnRNP, heterogeneous nuclear ribonucleoprotein; CCP, cyclic citrullinated peptide; hsp, heat shock protein; hTNF, human TNF; tg, transgenic; OPG, osteoprotegerin; TRAP, tartrate resistant acid phosphatase; RRM, RNA recognition motif; snRNP, small nuclear RNP; wt, wild type; SLE, systemic lupus erythematosus.

1
Feldmann, M., R. N. Maini.
2003
. Lasker Clinical Medical Research Award: TNF defined as a therapeutic target for rheumatoid arthritis and other autoimmune diseases.
Nat. Med.
9
:
1245
-1250.
2
Weyand, C. M..
2000
. New insights into the pathogenesis of rheumatoid arthritis.
Rheumatology
39
: (Suppl. 1):
3
-8.
3
Kremer, J. M., R. Westhovens, M. Leon, E. Di Giorgio, R. Alten, S. Steinfeld, A. Russell, M. Dougados, P. Emery, I. F. Nuamah, et al
2003
. Treatment of rheumatoid arthritis by selective inhibition of T-cell activation with fusion protein CTLA4Ig.
N. Engl. J. Med.
349
:
1907
-1915.
4
Smolen, J. S., S. Hayer, G. Schett, K. Redlich, M. Aringer, G. Kollias, E. Wagner, G. Steiner.
2004
. Autoimmunity and rheumatoid arthritis.
Autoimmun. Rev.
3
: (Suppl. 1):
S23
5
Steiner, G., J. Smolen.
2002
. Autoantibodies in rheumatoid arthritis and their clinical significance.
Arthritis Res.
4
: (Suppl 2):
S1
-S5.
6
Doyle, H. A., M. J. Mamula.
2001
. Post-translational protein modifications in antigen recognition and autoimmunity.
Trends Immunol.
22
:
443
-449.
7
Vossenaar, E. R., W. J. van Venrooij.
2004
. Citrullinated proteins: sparks that may ignite the fire in rheumatoid arthritis.
Arthritis Res. Ther.
6
:
107
-111.
8
Hassfeld, W., G. Steiner, A. Studnicka Benke, K. Skriner, W. Graninger, I. Fischer, J. S. Smolen.
1995
. Autoimmune response to the spliceosome: an immunologic link between rheumatoid arthritis, mixed connective tissue disease, and systemic lupus erythematosus.
Arthritis Rheum.
38
:
777
-785.
9
Steiner, G., K. Skriner, J. S. Smolen.
1996
. Autoantibodies to the A/B proteins of the heterogeneous nuclear ribonucleoprotein complex: novel tools for the diagnosis of rheumatic diseases.
Int. Arch. Allergy Immunol.
111
:
314
-319.
10
Fritsch, R., D. Eselbock, K. Skriner, B. Jahn-Schmid, C. Scheinecker, B. Bohle, M. Tohidast-Akrad, S. Hayer, J. Neumuller, S. Pinol-Roma, et al
2002
. Characterization of autoreactive T cells to the autoantigens heterogeneous nuclear ribonucleoprotein A2 (RA33) and filaggrin in patients with rheumatoid arthritis.
J. Immunol.
169
:
1068
-1076.
11
Rantapaa-Dahlqvist, S., B. A. de Jong, E. Berglin, G. Hallmans, G. Wadell, H. Stenlund, U. Sundin, W. J. van Venrooij.
2003
. Antibodies against cyclic citrullinated peptide and IgA rheumatoid factor predict the development of rheumatoid arthritis.
Arthritis Rheum.
48
:
2741
-2749.
12
Nielen, M. M. J., D. van Schaardenburg, H. W. R. Reesink, R. J. van de Stadt, I. E. van der Horst-Bruinsma, M. H. de Koning, M. R. Habibuw, J. P. Vandenbroucke, B. A. Dijkmans.
2004
. Specific autoantibodies precede the symptoms of rheumatoid arthritis: a study of serial measurements in blood donors.
Arthritis Rheum.
50
:
380
-386.
13
Kollias, G., E. Douni, G. Kassiotis, D. Kontoyiannis.
1999
. On the role of tumor necrosis factor and receptors in models of multiorgan failure, rheumatoid arthritis, multiple sclerosis and inflammatory bowel disease.
Immunol. Rev.
169
:
175
-194.
14
Kyburz, D., M. Corr.
2003
. The KRN mouse model of inflammatory arthritis.
Springer Semin. Immunopathol.
25
:
79
-90.
15
Monach, P. A., C. Benoist, D. Mathis.
2004
. The role of antibodies in mouse models of rheumatoid arthritis, and relevance to human disease.
Adv. Immunol.
82
:
217
-248.
16
Redlich, K., S. Hayer, R. Ricci, J. P. David, M. Tohidast Akrad, G. Kollias, G. Steiner, J. S. Smolen, E. F. Wagner, G. Schett.
2002
. Osteoclasts are essential for TNF-α-mediated joint destruction.
J. Clin. Invest.
110
:
1419
-1427.
17
Zwerina, J., S. Hayer, M. Tohidast Akrad, H. Bergmeister, K. Redlich, U. Feige, C. Dunstan, G. Kollias, G. Steiner, J. Smolen, G. Schett.
2004
. Single and combined inhibition of tumor necrosis factor, interleukin-1, and RANKL pathways in tumor necrosis factor-induced arthritis: effects on synovial inflammation, bone erosion, and cartilage destruction.
Arthritis Rheum.
50
:
277
-290.
18
Keffer, J., L. Probert, H. Cazlaris, S. Georgopoulos, E. Kaslaris, D. Kioussis, G. Kollias.
1991
. Transgenic mice expressing human tumour necrosis factor: a predictive genetic model of arthritis.
EMBO J.
10
:
4025
-4031.
19
Georgopoulos, S., D. Plows, G. Kollias.
1996
. Transmembrane TNF is sufficient to induce localized tissue toxicity and chronic inflammatory arthritis in transgenic mice.
J. Inflamm.
46
:
86
-97.
20
Redlich, K., S. Hayer, A. Maier, C. R. Dunstan, M. Tohidast Akrad, S. Lang, B. Turk, P. Pietschmann, W. Woloszczuk, S. Haralambous, et al
2002
. Tumor necrosis factor α-mediated joint destruction is inhibited by targeting osteoclasts with osteoprotegerin.
Arthritis Rheum.
46
:
785
-792.
21
Marcu, A., B. Bassit, R. Perez, S. Pinol-Roma.
2001
. Heterogeneous nuclear ribonucleoprotein complexes from Xenopus laevis oocytes and somatic cells.
Int. J. Dev. Biol.
45
:
743
-752.
22
Skriner, K., W. H. Sommergruber, V. Tremmel, I. Fischer, A. Barta, J. S. Smolen, G. Steiner.
1997
. Anti-A2/RA33 autoantibodies are directed to the RNA binding region of the A2 protein of the heterogeneous nuclear ribonucleoprotein complex: differential epitope recognition in rheumatoid arthritis, systemic lupus erythematosus, and mixed connective tissue disease.
J. Clin. Invest.
100
:
127
-135.
23
Dumortier, H., F. Monneaux, B. Jahn Schmid, J. P. Briand, K. Skriner, P. L. Cohen, J. S. Smolen, G. Steiner, S. Muller.
2000
. B and T cell responses to the spliceosomal heterogeneous nuclear ribonucleoproteins A2 and B1 in normal and lupus mice.
J. Immunol.
165
:
2297
-2305.
24
Schreiber, E., P. Matthias, M. M. Muller, W. Schaffner.
1989
. Rapid detection of octamer binding proteins with “mini-extracts,” prepared from a small number of cells.
Nucleic Acids Res.
17
:
6419
25
Swanson, M. S., G. Dreyfuss.
1988
. RNA binding specificity of hnRNP proteins: a subset binds to the 3′ end of introns.
EMBO J.
7
:
3519
-3529.
26
Habets, W. J., M. H. Hoet, B. A. De Jong, A. Van der Kemp, W. J. van Venrooij.
1989
. Mapping of B cell epitopes on small nuclear ribonucleoproteins that react with human autoantibodies as well as with experimentally-induced mouse monoclonal antibodies.
J. Immunol.
143
:
2560
-2566.
27
van Eekelen, C. A., M. H. Salden, W. J. Habets, L. B. van de Putte, W. J. van Venrooij.
1982
. On the existence of an internal nuclear protein structure in HeLa cells.
Exp. Cell Res.
141
:
181
-190.
28
Ma, A. S., K. Moran Jones, J. Shan, T. P. Munro, M. J. Snee, K. S. Hoek, R. Smith.
2002
. Heterogeneous nuclear ribonucleoprotein A3, a novel RNA trafficking response element-binding protein.
J. Biol. Chem.
277
:
18010
-18020.
29
Burd, C. G., M. S. Swanson, M. Gorlach, G. Dreyfuss.
1989
. Primary structures of the heterogeneous nuclear ribonucleoprotein A2, B1, and C2 proteins: a diversity of RNA binding proteins is generated by small peptide inserts.
Proc. Natl. Acad. Sci. USA
86
:
9788
-9792.
30
Cope, A. P., R. S. Liblau, X. D. Yang, M. Congia, C. Laudanna, R. D. Schreiber, L. Probert, G. Kollias, H. O. McDevitt.
1997
. Chronic tumor necrosis factor alters T cell responses by attenuating T cell receptor signaling.
J. Exp. Med.
185
:
1573
-1584.
31
Hatfield, J. T., J. A. Rothnagel, R. Smith.
2002
. Characterization of the mouse hnRNP A2/B1/B0 gene and identification of processed pseudogenes.
Gene
295
:
33
-42.
32
Krecic, A. M., M. S. Swanson.
1999
. hnRNP complexes: composition, structure, and function.
Curr. Opin. Cell Biol.
11
:
363
-371.
33
Hietarinta, M., O. Lassila.
1996
. Clinical significance of antinuclear antibodies in systemic rheumatic diseases.
Ann. Med.
28
:
283
-291.
34
von Muhlen, C. A., E. M. Tan.
1995
. Autoantibodies in the diagnosis of systemic rheumatic diseases.
Semin. Arthritis Rheum.
24
:
323
-358.
35
Smolen, J. S., G. Steiner.
1998
. Are autoantibodies active players or epiphenomena?.
Curr. Opin. Rheumatol.
10
:
201
-206.
36
Masson-Bessiere, C., M. Sebbag, E. Girbal-Neuhauser, L. Nogueira, C. Vincent, T. Senshu, G. Serre.
2001
. The major synovial targets of the rheumatoid arthritis-specific antifilaggrin autoantibodies are deiminated forms of the α- and β-chains of fibrin.
J. Immunol.
166
:
4177
-4184.
37
Bas, S., S. Genevay, O. Meyer, C. Gabay.
2003
. Anti-cyclic citrullinated peptide antibodies, IgM and IgA rheumatoid factors in the diagnosis and prognosis of rheumatoid arthritis.
Rheumatology
42
:
677
-680.
38
La Cava, A., C. Fine, E. Rodriguez, A. Ilic, N. Sarvetnick, M. S. Horwitz.
2000
. Viruses, host responses, and autoimmunity.
Nat. Med.
6
:
693
-697.
39
Smolen, J. S., G. Steiner.
2001
. Rheumatoid arthritis is more than cytokines: autoimmunity and rheumatoid arthritis.
Arthritis Rheum.
44
:
2218
-2220.
40
Fournie, G. J., M. Mas, B. Cautain, M. Savignac, J. F. Subra, L. Pelletier, A. Saoudi, D. Lagrange, M. Calise, P. Druet.
2001
. Induction of autoimmunity through bystander effects: lessons from immunological disorders induced by heavy metals.
J. Autoimmun.
16
:
319
-326.
41
Mathis, D., L. Vence, C. Benoist.
2001
. β-Cell death during progression to diabetes.
Nature
414
:
792
-798.
42
Maier, B., L. Morawietz, V. Krenn, T. Kamradt, F. J. Descamps.
2004
. Remnant epitopes generate autoimmunity: from rheumatoid arthritis and multiple sclerosis to diabetes.
J. Immunol.
172
:
4503
-4509.
43
Green, E. A., R. A. Flavell.
2000
. The temporal importance of TNFα expression in the development of diabetes.
Immunity
12
:
459
-469.
44
Klein, L., M. Klugmann, K. A. Nave, V. K. Tuohy, B. Kyewski.
2000
. Shaping of the autoreactive T-cell repertoire by a splice variant of self protein expressed in thymic epithelial cells.
Nat. Med.
6
:
56
-61.
45
Gilkeson, G. S., P. Ruiz, A. J. Pritchard, D. S. Pisetsky.
1991
. Genetic control of inflammatory arthritis and glomerulonephritis in congenic lpr mice and their F1 hybrids.
J. Autoimmun.
4
:
595
-606.
46
Lanzavecchia, A..
1995
. How can cryptic epitopes trigger autoimmunity?.
J. Exp. Med.
181
:
1945
-1948.
47
Maki, M..
1996
. Coeliac disease and autoimmunity due to unmasking of cryptic epitopes?.
Lancet
348
:
1046
-1047.
48
Lan, M. S., N. K. Maclaren.
1998
. Cryptic epitope and autoimmunity.
Diabetes Metab. Rev.
14
:
333
-334.
49
Olson, J. K., J. L. Croxford, S. D. Miller.
2001
. Virus-induced autoimmunity: potential role of viruses in initiation, perpetuation, and progression of T-cell-mediated autoimmune disease.
Viral Immunol.
14
:
227
-250.
50
Visan, L., I. A. Visan, A. Weishaupt, H. H. Hofstetter, K. V. Toyka, T. Hunig, R. Gold.
2004
. Tolerance induction by intrathymic expression of P0.
J. Immunol.
172
:
1364
-1370.
51
Nichols, R. C., X. W. Wang, J. Tang, B. J. Hamilton, F. A. High, H. R. Herschman, W. F. Rigby.
2000
. The RGG domain in hnRNP A2 affects subcellular localization.
Exp. Cell Res.
256
:
522
-532.
52
Prakken, B. J., S. Roord, A. Ronaghy, M. Wauben, S. Albani, W. van Eden.
2003
. Heat shock protein 60 and adjuvant arthritis: a model for T cell regulation in human arthritis.
Springer Semin. Immunopathol.
25
:
47
-63.
53
van Eden, W., R. van der Zee, B. Prakken.
2005
. Heat-shock proteins induce T-cell regulation of chronic inflammation.
Nat. Rev. Immunol.
5
:
318
-330.
54
Panayi, G., V. Corrigall.
2004
. BiP: a new biologic immunomodulator for the treatment of rheumatoid arthritis.
Autoimmun. Rev.
3
: (Suppl 1):
S16
-S17.
55
de Graeff-Meeder, E. R., M. Voorhorst, W. van Eden, H. J. Schuurman, J. Huber, D. Barkley, R. N. Maini, W. Kuis, G. T. Rijkers, B. J. Zegers.
1990
. Antibodies to the mycobacterial 65-kd heat-shock protein are reactive with synovial tissue of adjuvant arthritic rats and patients with rheumatoid arthritis and osteoarthritis.
Am. J. Pathol.
137
:
1013
-1017.
56
Barker, R. N., A. D. Wells, M. Ghoraishian, A. J. Easterfield, Y. Hitsumoto, C. J. Elson, S. J. Thompson.
1996
. Expression of mammalian 60-kD heat shock protein in the joints of mice with pristane-induced arthritis.
Clin. Exp. Immunol.
103
:
83
-88.
57
Schett, G., K. Redlich, Q. Xu, P. Bizan, M. Groger, M. Tohidast-Akrad, H. Kiener, J. Smolen, G. Steiner.
1998
. Enhanced expression of heat shock protein 70 (hsp70) and heat shock factor 1 (HSF1) activation in rheumatoid arthritis synovial tissue: differential regulation of hsp70 expression and hsf1 activation in synovial fibroblasts by proinflammatory cytokines, shear stress, and antiinflammatory drugs.
J. Clin. Invest.
102
:
302
-311.
58
Kindas Mugge, I., G. Steiner, J. S. Smolen.
1993
. Similar frequency of autoantibodies against 70-kD class heat-shock proteins in healthy subjects and systemic lupus erythematosus patients.
Clin. Exp. Immunol.
92
:
46
-50.
59
Zugel, U., S. H. Kaufmann.
1999
. Role of heat shock proteins in protection from and pathogenesis of infectious diseases.
Clin. Microbiol. Rev.
12
:
19
-39.
60
Kim, H..
2003
. Diagnostic significance of antibodies to heat shock proteins.
Clin. Chim. Acta
337
:
1
-10.
61
Prakken, B., W. Kuis, W. van Eden, S. Albani.
2002
. Heat shock proteins in juvenile idiopathic arthritis: keys for understanding remitting arthritis and candidate antigens for immune therapy.
Curr. Rheumatol. Rep.
4
:
466
-473.
62
Vossenaar, E. R., S. Nijenhuis, M. M. Helsen, A. van der Heijden, T. Senshu, W. B. van den Berg, W. J. van Venrooij, L. A. Joosten.
2003
. Citrullination of synovial proteins in murine models of rheumatoid arthritis.
Arthritis Rheum.
48
:
2489
-2500.
63
Vossenaar, E. R., M. A. van Boekel, W. J. van Venrooij, M. Lopez-Hoyoz, J. Merino, R. Merino, L. A. Joosten.
2004
. Absence of citrulline-specific autoantibodies in animal models of autoimmunity.
Arthritis Rheum.
50
:
2370
-2372.
64
van Gaalen, F. A., J. van Aken, T. W. Huizinga, G. M. Schreuder, F. C. Breedveld, E. Zanelli, W. J. van Venrooij, C. L. Verweij, R. E. Toes, R. R. de Vries.
2004
. Association between HLA class II genes and autoantibodies to cyclic citrullinated peptides (CCPs) influences the severity of rheumatoid arthritis.
Arthritis Rheum.
50
:
2113
-2121.
65
Hoffman, I. E., I. Peene, H. Pottel, A. Union, F. Hulstaert, L. Meheus, K. Lebeer, L. De Clercq, L. Schatteman, S. Poriau, et al
2005
. Diagnostic performance and predictive value of rheumatoid factor, anti-citrullinated peptide antibodies, and the HLA shared epitope for diagnosis of rheumatoid arthritis.
Clin. Chem.
51
:
261
-263.
66
Hassfeld, W., G. Steiner, W. Graninger, G. Witzmann, H. Schweitzer, J. S. Smolen.
1993
. Autoantibody to the nuclear antigen RA33: a marker for early rheumatoid arthritis.
Br. J. Rheumatol.
32
:
199
-203.
67
Nell, V., K. P. Machold, T. A. Stamm, G. Eberl, H. Heinzl, M. Uffmann, J. S. Smolen, G. Steiner.
2005
. Autoantibody profiling as early diagnostic and prognostic tool for rheumatoid arthritis.
Ann. Rheum. Dis.
64
:
1731
-1736.
68
Mayeda, A., S. H. Munroe, J. F. Caceres, A. R. Krainer.
1994
. Function of conserved domains of hnRNP A1 and other hnRNP A/B proteins.
EMBO J.
13
:
5483
-5495.
69
Kamma, H., M. Fujimoto, M. Fujiwara, M. Matsui, H. Horiguchi, M. Hamasaki, H. Satoh.
2001
. Interaction of hnRNP A2/B1 isoforms with telomeric ssDNA and the in vitro function.
Biochem. Biophys. Res. Commun.
280
:
625
-630.
70
Brumwell, C., C. Antolik, J. H. Carson, E. Barbarese.
2002
. Intracellular trafficking of hnRNP A2 in oligodendrocytes.
Exp. Cell Res.
279
:
310
-320.
71
Matsui, M., H. Horiguchi, H. Kamma, M. Fujiwara, R. Ohtsubo, T. Ogata.
2000
. Testis- and developmental stage-specific expression of hnRNP A2/B1 splicing isoforms, B0a/b.
Biochim. Biophys. Acta
1493
:
33
-40.
72
Zhou, J., J. L. Mulshine, E. J. Unsworth, F. M. Scott, I. M. Avis, M. D. Vos, A. M. Treston.
1996
. Purification and characterization of a protein that permits early detection of lung cancer: identification of heterogeneous nuclear ribonucleoprotein-A2/B1 as the antigen for monoclonal antibody 703D4.
J. Biol. Chem.
271
:
10760
-10766.
73
Tani, H., K. Ohshima, S. Haraoka, M. Hamasaki, H. Kamma, S. Ikeda, M. Kikuchi.
2002
. Overexpression of heterogeneous nuclear ribonucleoprotein B1 in lymphoproliferative disorders: high expression in cells of follicular center origin.
Int. J. Oncol.
21
:
957
-963.
74
Pino, I., R. Pio, G. Toledo, N. Zabalegui, S. Vicent, N. Rey, M. D. Lozano, W. Torre, J. Garcia-Foncillas, L. M. Montuenga.
2003
. Altered patterns of expression of members of the heterogeneous nuclear ribonucleoprotein (hnRNP) family in lung cancer.
Lung Cancer
41
:
131
-143.