SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

Autoimmunity to proteins, such as type II collagen and cartilage intermediate layer protein, that are produced by chondrocytes has been reported in patients with osteoarthritis (OA) as well as in patients with rheumatoid arthritis (RA). However, it remains to be determined whether the overall specificities of the autoimmunity differ between OA and RA patients. This study sought to clarify the differences by applying proteomic surveillance for the detection of autoantigens comprehensively.

Methods

Serum samples were obtained from 20 patients with OA, 20 patients with RA, and 20 healthy volunteers. Human chondrocyte proteins were separated from the sera by 2-dimensional electrophoresis, and antigenic protein spots were detected by Western blotting. The antigenic proteins were then identified by mass fingerprinting. The antigenicity of the identified proteins was confirmed and the prevalence of the autoantibodies in the OA, RA, and other disease groups was determined with the use of recombinant proteins. In addition, autoepitopes were mapped on the antigens.

Results

Nineteen protein spots were recognized only by the OA sera, but not by the RA sera. One of these proteins was identified as triosephosphate isomerase (TPI). IgG-type anti-TPI autoantibodies were detected in 24.7% of the serum samples and 24.1% of the synovial fluid samples from the patients with OA, whereas <6% of the RA and systemic lupus erythematosus samples were positive for anti-TPI. In addition, multiple autoepitopes were identified on TPI.

Conclusion

The overall profile of autoimmunity in OA differs from that in RA, which may reflect the OA-specific pathologic role of autoimmunity. The autoantibody to TPI, detected predominantly in the OA samples and produced by the antigen-driven mechanism, has the potential to be used as a diagnostic marker for OA.

Osteoarthritis (OA) is generally accepted as a slowly progressive degenerative disease, mainly caused by physical stresses and aging factors. However, involvement of immunologic pathways has also been recently implicated in the pathophysiology of OA. Specifically, autoimmunity to cartilage-related components, such as anti–chondrocyte surface antibodies (1), as well as cellular immune responses to chondrocyte membranes (2), the cartilage link protein, and cartilage aggrecans (3) have been described. In this context, we have also recently reported the existence of autoimmunity against cartilage-related proteins such as YKL-39 (4), cartilage intermediate layer protein (5), and osteopontin (6). However, most of the autoimmune responses to such antigens are detected both in patients with OA and in patients with rheumatoid arthritis (RA).

Considering that chronic synovitis is observed in both OA and RA (7–9), the autoimmune responses could be a part of epiphenomena that are associated with destruction/degradation of the cartilage. In this respect, if OA-specific autoimmunity is demonstrated, the pathologic role of this autoimmunity in OA would be strongly implicated. To investigate the OA-specific autoimmunity, it would be of great help to compare the overall profiles of autoimmunity between OA and RA patients.

To this end, we carried out this study to detect the proteins comprehensively recognized by autoantibodies in patients with OA as compared with those in patients with RA, using a proteomic approach. Specifically, with serum samples from patients with OA and RA, we separated the human chondrocyte proteins by 2-dimensional (2-D) electrophoresis, detected antigenic protein spots by Western blotting, and identified the antigenic proteins by mass fingerprinting. We next prepared some of the identified proteins as recombinant proteins, by which we confirmed their antigenicity and determined the prevalence of the autoantibodies in OA, RA, and other disease categories. Furthermore, we mapped the autoepitopes.

As a result, we found that the overall profiles of the chondrocyte-related autoantigens overlapped only partially between OA and RA. Interestingly, some proteins were recognized as an autoantigen much more frequently by the OA samples than by the RA samples. One such protein was identified as triosephosphate isomerase (TPI). In fact, TPI was recognized by 24.7% of the tested serum samples from the OA patients, but by only 5.6% of the tested serum samples from the RA patients, in an enzyme-linked immunosorbent assay (ELISA) using recombinant TPI.

TPI is a glycolytic enzyme that interconverts D-glyceraldehyde-3-phosphate (GDP) and dihydroxyacetone phosphate (DHAP). Since another glycolytic enzyme of glucose-6-phosphate isomerase (GPI) was reported to be a disease-causing autoantigen in RA-like murine arthritis, we hypothesized that the autoimmunity to TPI may be involved in the pathology of OA. Taken together, our results could promote the understanding of the role of autoimmunity in OA.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients and articular cartilage tissues.

Serum samples were obtained from 93 patients with OA (83 women, 10 men; ages 42–96 years, mean 74.9 years), 54 patients with RA (46 women, 8 men; ages 34–80 years, mean 62.5 years), and 43 patients with systemic lupus erythematosus (SLE) (41 women, 2 men; ages 20–57 years, mean 41.3 years). All patients were diagnosed according to the respective classification criteria for each disease (10–12). Age- and sex-matched healthy serum samples were selected as a control for each group of OA, RA, and SLE patients (44 women and 8 men ages 42–89 years [mean 73 years] were matched with the OA group, 31 women and 8 men ages 37–80 years [mean 60.7 years] were matched with the RA group, and 26 women and 2 men ages 20–63 years [mean 40.3 years] were matched with the SLE group).

Synovial fluid samples were obtained from 29 patients with OA and from 19 patients with RA. A sample of articular cartilage tissue was obtained from a 23-year-old man during hip joint replacement surgery for the treatment of femoral fracture. The radiographic grading of OA joints was carried out in accordance with the criteria of Kellgren and Lawrence (13). All samples were obtained after patients gave their informed consent, and this study was approved by the local institutional ethics committee.

Chondrocyte culture, 2-D electrophoresis, Western blotting, and purification of autoantibodies.

Chondrocytes were prepared from cartilage of the femoral head as described previously (14). Whole-cell lysates were prepared from the cultured chondrocytes by the freeze–thaw method in a lysis buffer (7M urea, 2M thiourea, 4% CHAPS) and stored at −80°C until used (15). The 2-D electrophoresis was performed as described elsewhere (15, 16). Briefly, the extracted protein samples were loaded onto 7-cm Immobiline drystrips (pH range 3–10; Amersham Biosciences, Uppsala, Sweden) at room temperature overnight. Up to 400 μg of the extracted proteins was applied onto the drystrip gels for Western blotting and up to 1,000 μg for the analysis by mass spectrometry. Isoelectric focusing was performed using MultiPhor II (Amersham Biosciences). The second electrophoresis was then performed in 12.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) slab gels. After the electrophoresis, the gels were stained with Coomassie brilliant blue or used for protein transfer onto nitrocellulose membranes.

Western blotting was performed as described previously (5). The serum samples for incubating the membranes were diluted to 1:500. The bound antibodies were reacted with horseradish peroxidase–conjugated goat anti-human IgG (Zymed, San Francisco, CA) and visualized by diaminobenzidine. In addition, the autoantibodies to TPI were purified from the recombinant proteins immobilized on the nitrocellulose membrane as described previously (17).

Protein identification.

Protein spots on the gel stained with Coomassie brilliant blue, which corresponded to positive spots on Western blot membranes, were recovered and then in-gel digested with trypsin, and finally identified by matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometry as described previously (18, 19). A list of the determined peptide masses was subjected to mass fingerprinting using the Mascot software program (Matrix Science, London, UK) (20), in which the National Center for Biotechnology Information (NCBI) protein databases were searched.

Preparation of recombinant fusion proteins.

In accordance with the nucleotide sequence of the human TPI complementary DNA (cDNA) (21), 2 DNA primers were prepared using reverse transcription–polymerase chain reaction, to amplify the entire protein-coding regions of the TPI cDNA from messenger RNA (mRNA) that was extracted from the human articular chondrocytes. To map the autoepitopes on TPI, 5 cDNA fragments that covered and overlapped the entire protein-coding region (TPI-1 to TPI-5), and 3 more cDNA fragments that covered and overlapped the TPI-1 region (TPI-1a to TPI-1c) (shown in Figure 2a) were amplified similarly. The nucleotide sequences of the primers are shown in Table 1.

thumbnail image

Figure 2. Preparation of fusion proteins with triosephosphate isomerase (TPI). a, The construction of the TPI fusion proteins was achieved with DNA fragments that encoded the entire protein coding region and 8 small regions of cDNA for TPI, amplified by reverse transcription–polymerase chain reaction: TPI (747 bp), TPI-1 (165 bp), TPI-2 (180 bp), TPI-3 (180 bp), TPI-4 (180 bp), TPI-5 (162 bp), TPI-1a (69 bp), TPI-1b (72 bp), and TPI-1c (69 bp). b, Each of the DNA fragments, cloned into a pMAL-eHis vector, was expressed as a maltose binding protein (MBP) fusion protein in Escherichia coli. The purified fusion proteins, as well as MBP as a fusion partner, were separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and stained with Coomassie brilliant blue. Lane 1, MBP; lane 2, TPI-MBP; lanes 3–7, TPI-1-MBP to TPI-5-MBP, respectively; lanes 8–10, TPI-1a-MBP to TPI-1c-MBP, respectively. M = molecular weight markers.

Download figure to PowerPoint

Table 1. Nucleotide sequences of primers*
Primers for amplification of cDNA sequence
  • *

    TPI = triosephosphate isomerase.

Encoding entire protein of TPI
 Sense 5′-TTTGGATCCATGGCGCCCTCCAGGAAGTTCTTC-3′
 Antisense 5′-TTTGTCGACTTGTTTGGCATTGATGATGTCCAC-3′
Encoding fragments of TPI
 TPI-1
  Sense 5′-TTTGGATCCATGGCGCCCTCCAGGAAGTCCTCC-3′
  Antisense 5′-TTTGTCGACCTTCTGCCGGGCGAAGTCGATATA-3′
 TPI-2
  Sense 5′-TTTGGATCCACTGCCTATATCGACTTCGCCCGG-3′
  Antisense 5′-TTTGTCGACCTCCCCAAAGACATGCCTTCTCTC-3′
 TPI-3
  Sense 5′-TTTGGATCCCACTCAGAGAGAAGGCATGTCTTT-3′
  Antisense 5′-TTTGTCGACCGCGTTATCTGCGATGACCTTTGT-3′
 TPI-4
  Sense 5′-TTTGGATCCGAGCAGACAAAGGTCATCGCAGAT-3′
  Antisense 5′-TTTGTCGACGGTGCTCTGAGCCACCGCATCAGA-3′
 TPI-5
  Sense 5′-TTTGGATCCAACGTCTCTGATGCGGTGGCTCAG-3′
  Antisense 5′-TTTGTCGACTTGTTTGGCATTGATGATGTCCAC-3′
Encoding fragments of TPI-1
 TPI-1a
  Sense 5′-TTTGGATCCATGGCGCCCTCCAGGAAGTTCTTC-3′
  Antisense 5′-TTTGTCGACCCCCAGACTCTGCTTCCGCCCGTT-3′
 TPI-1b
  Sense 5′-TTTGGATCCGGGCGGAAGCAGAGTCTGGGGGAG-3′
  Antisense 5′-TTTGTCGACCACCTCGGTGTCGGCCGGCACCTT-3′
 TPI-1c
  Sense 5′-TTTGGATCCAAGGTGCCGGCCGACACCGAGGTG-3′
  Antisense 5′-TTTGTCGACCTTCTGCCGGGCGAAGTCGATATA-3′

The amplified cDNA fragments of TPI (747 bp), TPI-1–5, and TPI-1a–1c were subcloned into a plasmid expression vector of pMAL-eHis, a derivative from pMAL-c2 (New England Biolabs, Beverly, MA). Thus, recombinant TPI proteins were produced in Escherichia coli as a fusion protein with the maltose binding protein (MBP). The recombinant proteins were purified by the histidine-Ni+ affinity purification system as described previously (22).

Analysis by ELISA.

The ELISA was performed as described previously (22). Briefly, each well of a multititer plate (Cook; Dynatech, Alexandria, VA) was coated with 10 μg/ml of the individual purified fusion protein or MBP (as a background) in a carbonate buffer (50 mM sodium carbonate, pH 9.6). To adsorb the reactivity of the serum samples with MBP, the serum samples were diluted and incubated with 40 μg/ml of MBP in phosphate buffered saline–Tween containing 1% bovine serum albumin for 2 hours before incubation with the coated recombinant proteins. The serum samples diluted at 1:500 were used for the screening. The bound antibodies were reacted with the same secondary antibodies as used in Western blotting. The reactivity of the serum samples in response to the TPI fusion proteins was expressed as the optical density or arbitrary binding units, as described previously (5), and 100 binding units was defined as the cutoff point for reactivity.

Statistical analysis.

Differences in the prevalence of the anti-TPI antibodies among the disease categories were compared by using the chi-square test. The differences in mean binding units among the disease categories and the differences in the mean radiographic grades and ages between the anti-TPI–negative and –positive patients were compared by Student's t-test.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Detection of chondrocyte-producing autoantigens by 2-D electrophoresis and Western blotting.

To evaluate the overall profiles of autoimmunity in patients with OA and those with RA, we surveyed autoantibodies/autoantigens by 2-D electrophoresis and subsequently by Western blotting in a total of 60 tests, with the use of serum samples from 20 patients with OA, 20 patients with RA, and 20 healthy donors. We used proteins extracted from human chondrocytes for the 2-D electrophoresis/Western blotting, since one of the main targets both in RA and in OA is cartilage.

We found that comparable numbers of proteins were recognized as autoantigens by Western blotting of the OA and RA samples (representative results are shown in Figures 1b and c, respectively, versus healthy controls in Figure 1d). As summarized in Table 2, we found 62 positive spots from the 60 Western blots, by nonoverlapping counting. Interestingly, 19 protein spots were detected in the OA group only, 11 spots in the RA patient group only, and 22 spots in both the OA and RA groups. These spots on the membranes were identified on the 2-D electrophoresis gel by staining with Coomassie brilliant blue, as shown in Figure 1e. The detection of the apparently OA-specific and RA-specific autoantigens indicates that the profiles of autoimmunity are different between OA and RA.

thumbnail image

Figure 1. Two-dimensional (2-D) electrophoresis and Western blotting of chondrocyte-derived proteins. Proteins extracted from human chondrocytes were separated by 2-D electrophoresis and then were stained with Coomassie brilliant blue (a), or were transferred onto nitrocellulose membranes and reacted with serum samples, diluted at 1:500, from 20 patients with osteoarthritis, 20 patients with rheumatoid arthritis, and 20 healthy donors (representative results in b, c, and d, respectively). All positive spots by Western blotting are indicated with identification numbers 1–62 on a Coomassie brilliant blue–stained 2D electrophoresis gel (e).

Download figure to PowerPoint

Table 2. Detection of autoantigenic chondrocyte protein spots with 2-dimensional electrophoresis/Western blotting*
Protein spot identification numberMolecular weightpIOA (n = 20)RA (n = 20)HC (n = 20)
  • *

    Except where indicated otherwise, values are the number of positive samples. OA = osteoarthritis; RA = rheumatoid arthritis; HC = healthy control.

1265 1 
2267.2 1 
3274.81  
42874  
528.55.4 1 
6305.71  
7309.611 
8324.62  
9347.63  
10355.4 2 
11359.912 
12366.42  
13366.62  
14369.655 
1536.55.51  
16376.51  
17384.73  
18394.811 
19391083 
20414.8 1 
21427.121 
22439.2211
234310123
24456.51  
25459.521 
264752  
27484.7 1 
28485.411 
29504.264 
30514.511 
31515.311 
32516.52  
33516.72  
34516.9431
35517.281110
36524.531 
37526.9111
38527.1232
39546.5 1 
40624.81 1
41635.422 
42646 2 
43675.32 1
44704.91  
45716.622 
46716.722 
47716.822 
4871722 
49726.454 
50726.554 
51785.912 
5278612 
53805.3 1 
54817.2  1
55825.7 2 
56835.211 
57944.71  
58945.12  
591105.42  
601184.9  1
611305.41  
621505.6 1 

Identification of TPI by mass fingerprinting.

We next focused on the apparently OA-specific autoantigens, in particular, the protein spot 4 (see Table 2), which had a molecular weight of ∼28 kd and a pI value of ∼7.0; protein spot 4 was detected only in the OA group, at a high frequency of 20% (4 of 20 serum samples) (Figure 1b and Table 2). The mass of trypsin-digested peptides from protein spot 4 was determined by MALDI-TOF mass spectrometry. By mass fingerprinting using the Mascot program and NCBI protein databases, it was found that the mass data covered 28% of the amino acid sequence of TPI (data not shown). This indicated that spot 4 was most likely TPI.

Analysis of the antigenicity of TPI, using recombinant proteins.

To confirm and analyze the antigenicity of TPI, we prepared full-length and variously truncated recombinant TPI proteins as a fusion protein with MBP (Figure 2a) and purified them as shown in Figure 2b. First, we separated the purified full-length TPI with MBP (TPI-MBP), as well as MBP alone as a control, by SDS-PAGE and tested its reactivity with the 4 serum samples that had shown reactivity with protein spot 4 by Western blotting in the first screening. As a result, all 4 serum samples reacted positively in response to TPI-MBP, but not to MBP alone (Figure 3a). This confirmed that spot 4 was TPI.

thumbnail image

Figure 3. The antigenicity of TPI. a, The purified TPI-MBP and MBP as a control were separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and stained with Coomassie brilliant blue (left panel of a: Lane 1, TPI-MBP; lane 2, MBP), or transferred onto membrane and stained with serum samples from 4 patients with osteoarthritis (OA) that reacted with protein spot 4 (OA38, OA39, OA50, and OA58) and a serum from a healthy control (HC). b and c, The chondrocyte-derived proteins, separated by 2-dimensional electrophoresis, were reacted with the whole sera of anti-TPI (spot 4) antibody–positive patients (b) and with the autoantibodies to TPI-MBP purified from the same patients (c) (a representative case, OA38, is shown). d, The anti–TPI-MBP autoantibodies were measured in serially diluted serum samples by enzyme-linked immunosorbent assay (ELISA). Representative cases, OA50 with positive reactivity to TPI and OAn3 with negative reactivity, are shown. e, The anti–TPI-MBP autoantibody titers were measured in serum samples before and after adsorption with TPI-MBP by ELISA. Representative cases, OA50 with positive reactivity to TPI and OAn3 with negative reactivity, are shown. OD = optical density; TPI/MBP = anti-TPI titers in the serum samples adsorbed with MBP; TPI/TPI = anti-TPI titers in the serum samples adsorbed with TPI-MBP (see Figure 2 for other definitions).

Download figure to PowerPoint

To confirm this in a different way, we purified the autoantibodies to TPI-MBP from the positive serum samples by affinity purification. With the purified autoantibodies, we then stained the chondrocyte proteins separated by 2-D electrophoresis and transferred them onto membranes. As a result, only protein spot 4 was stained, as shown in Figures 3b and c. To further confirm specific binding of the autoantibodies to TPI, we measured anti–TPI-MBP titers in variously diluted serum samples by ELISA. The anti–TPI-MBP titers increased in a serum concentration–dependent manner, but the anti-MBP antibodies did not (Figures 3d and e). Furthermore, the serum reactivity with TPI-MBP was completely cancelled by the addition of TPI-MBP, but not by MBP alone (Figure 3d). These data provide evidence that autoantibodies specific to TPI were produced in these patients.

Prevalence of autoantibodies to TPI in patients with OA, RA, and SLE.

We next tried to determine the frequency of the anti-TPI autoantibodies in patients with OA as well as in patients with RA and patients with SLE, using ELISA. Specifically, we tested 93 serum samples from patients with OA as well as 54 from patients with RA and 43 from patients with SLE; as a negative control, we tested 52 serum samples from healthy donors. As shown in Figure 4a, the mean binding units were much higher in the OA group than in the other groups (P < 0.01 for each). The prevalence of the autoantibodies to TPI-MBP was 24.7% in the OA patients (23 of 93), 5.6% in the RA patients (3 of 54), 4.7% in the SLE patients (2 of 43), and 2.5% in the healthy donors (3 of 119). The frequency of the anti-TPI autoantibodies was ∼5 times greater in the OA patients as compared with the RA or SLE patients, and the differences were statistically significant (χ2 = 8.5995 and χ2 = 7.8744, respectively; P < 0.01).

thumbnail image

Figure 4. Detection of the autoantibodies to TPI by enzyme-linked immunosorbent assay (ELISA) and Western blotting. a, The autoantibodies to TPI in serum samples from patients with osteoarthritis (OA), rheumatoid arthritis (RA), and systemic lupus erythematosus (SLE) were detected by ELISA. The dotted line indicates the defined cutoff for positivity of 100 binding units. Serum samples diluted at 1:500 were used. b, The autoantibodies to TPI in synovial fluid (SF) samples were measured by ELISA as in a. SF samples diluted at 1:500 were used. The dotted line indicates the defined cutoff for positivity of 100 binding units. c, The autoantibodies to TPI in serum and SF samples in identical patients were measured by ELISA as in a. SF and serum samples were diluted at 1:500. d, Three paired samples in which the SF samples reacted with TPI by ELISA were further tested by Western blotting. A pair of the anti-TPI–negative RA samples was used as a negative control. Both the serum and the SF dilution was 1:500. Lane 1, TPI-MBP; lane 2, MBP. HC = healthy control; Ponceau S = the membrane was stained with Ponceau S (see Figure 2 for other definitions).

Download figure to PowerPoint

Similarly, we used ELISA to detect the autoantibodies to TPI in synovial fluid samples from 29 OA patients and 19 RA patients. As a result, 7 (24.1%) of 29 OA synovial fluid samples were found to react with TPI-MBP, but none of the 19 RA synovial fluid samples was found to be positive (Figure 4b) (χ2 = 6.47; P < 0.02). In addition, in the ELISA of paired samples of serum and synovial fluid from 9 OA patients, both the serum and synovial fluid samples from 1 patient (OAp2) were found to be anti-TPI antibody positive, whereas only the synovial fluid samples from 2 patients (OAp1 and OAp3) were found to be positive (Figure 4c). The samples from these 3 patients with OA showed similar results on Western blotting, although the serum sample from patient OAp1 was found weakly positive on Western blot (Figure 4d). In contrast, all of the paired samples from the 9 patients with RA were negative for the anti-TPI autoantibody (Figure 4c). Thus, the anti-TPI autoantibodies in synovial fluid appeared to be specific for OA.

Analysis of autoepitopes on TPI in OA.

We next mapped the autoepitopes on TPI. Using all of the OA serum samples that were positive by ELISA and Western blotting, we tested their immune reactivity in response to the 5 fragments of TPI described in Figure 2. As a result, TPI was found to carry multiple autoepitopes. The results from ELISA and Western blotting are summarized in Table 3 and representative results are shown in Figure 5a. One OA serum sample recognized, on average, 3.5 of the 5 regions. Interestingly, TPI-1-MBP was recognized by 91.3% of the positive serum samples, whereas TPI-2-MBP, TPI-3-MBP, TPI-4-MBP, and TPI-5-MBP were recognized by 86.9%, 43.5%, 78.3%, and 60.9% of the positive serum samples, respectively.

Table 3. The results of TPI epitope mapping*
SampleTPI-1-MBPTPI-2-MBPTPI-3-MBPTPI-4-MBPTPI-5-MBP
EAWBE + WEAWBE + WEAWBE + WEAWBE + WEAWBE + W
  • *

    TPI = triosephosphate isomerase; MBP = maltose binding protein; EA = enzyme-linked immunosorbent assay; WB = Western blotting; E + W = EA + WB; OA = osteoarthritis; RA = rheumatoid arthritis; SLE = systemic lupus erythematosus; HC = healthy control.

OA               
 OA02+++++++++++
 OA07++++++++++++
 OA10++++++++++
 OA11+++++++++
 OA17++++++++++
 OA21+++++++++
 OA23+++++++++++
 OA32++++++++++++
 OA34++++++
 OA38++++++++++++++
 OA39+++++++++
 OA42+++
 OA50+++++
 OA58+++++
 OA68+++++
 OA71+++++++++++
 OA72++++++++++++++
 OA75++++++++++
 OA80++++++++++
 OA81+++++++
 OA86++++++++++++
 OA98++++++++++
 OA149+++++
 Total positive19/2319/2321/2315/2314/2320/232/238/2310/2313/2313/2318/239/2314/2314/23
Other               
 RA24++++++++
 RA28++++++++++++++
 RA35++++++++++++
 SLE72++++++++++++++
 SLE77+++++++++++
 HC110+++++
 HC146+++++++
 Total positive6/76/77/76/76/77/71/72/72/75/74/75/74/75/75/7
All, total positive25/3025/3028/3021/3020/3027/303/3010/3012/3018/3017/3022/3013/3019/3019/30
thumbnail image

Figure 5. The results of Western blotting to map the autoepitopes on TPI. a, The fusion proteins of TPI, separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), were reacted with the anti-TPI–positive serum samples from patients with osteoarthritis (OA). Lane 1, MBP; lane 2, TPI-MBP; lanes 3–7, TPI-1-MBP to TPI-5-MBP, respectively. I = Ponceau S staining. Representative OA cases are shown in II–VIII. b, MBP alone and the fusion proteins of TPI-1a-MBP, TPI-1b-MBP, and TPI-1c-MBP (lanes 1–4, respectively) were separated in 10% SDS-PAGE, and then were reacted with anti–TPI-1-MBP–positive serum samples. I = Ponceau S staining. Representative OA cases are shown in II–VII. See Figure 3 for other definitions.

Download figure to PowerPoint

Since TPI-1-MBP was recognized most frequently, we further investigated the epitopes in the TPI-1 region. The result from Western blotting revealed that TPI-1a-MBP, TPI-1b-MBP, and TPI-1c-MBP were recognized by 42.8%, 19.0%, and 95.2% of the anti–TPI-1-MBP–positive serum samples (Figure 5b). This indicates that the 23 amino acid residues from positions 33 to 55 of TPI include one of the main epitopes recognized by the anti-TPI IgG.

Correlations between clinical parameters and the anti-TPI autoantibody.

We next investigated clinical parameters between the anti-TPI autoantibody–positive and –negative patients. Since there were only 3 anti-TPI–positive serum samples from the patients with RA and there were only 2 from the patients with SLE, we concentrated on the OA patients (Table 4). The mean age of the anti-TPI–positive group was slightly older than that of the anti-TPI–negative group, but this difference was not statistically significant (P > 0.05). The distribution ratio of females to males in the anti-TPI–positive group was higher than that in the anti-TPI–negative group, but again the difference was of no statistical significance (P > 0.05). However, the mean radiographic grade in the anti-TPI–positive group was significantly lower than that in the anti-TPI–negative group (P < 0.05).

Table 4. Comparison of clinical data between anti-TPI–positive and –negative patients with osteoarthritis*
GroupNo.Age, mean ± SD yearsSex, no. female/no. maleRadiographic grade, mean ± SD
  • *

    P < 0.05 versus patients negative for autoantibodies to triosephosphate isomerase (TPI).

Antibody positive2377.4 ± 9.167/31.36 ± 0.81*
Antibody negative7073.5 ± 13.720/31.8 ± 1.18

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In the present study, we sought to detect the autoantibodies produced by chondrocytes in patients with OA and RA, using a proteomic approach to compare autoimmune profiles between the two diseases. In the first experiment involving 2-D electrophoresis/Western blotting of 20 serum samples each from the OA, RA, and healthy control groups, we detected 62 immunogenic protein spots, of which 52 were recognized by the serum samples from patients with OA or RA. It is of importance that 41 of the 52 protein spots were recognized by the OA serum samples, whereas 33 of the 52 protein spots were recognized by the RA serum samples. Thus, the range of autoimmunity in OA is at least comparable with that in RA.

Since inflammation is much more severe, generally, in RA than in OA, the wide range of autoimmunity would not simply reflect the severity of the joint inflammation. Furthermore, 19 protein spots reacted with the OA serum samples only, and 11 protein spots reacted with the RA serum samples only. This demonstrated that RA and OA possess unique autoimmune profiles. The difference in the profiles may reflect different involvement of the autoimmunity in the pathology between OA and RA.

We concentrated on protein spot 4, one of the apparently OA-specific autoantigens in the 2-D electrophoresis/Western blot screening. Mass fingerprinting revealed that protein spot 4 was most likely TPI, which was confirmed by ELISA and Western blotting. We confirmed a higher frequency of detection (24.7%) of the anti-TPI autoantibodies in the patients with OA, but these were infrequent in RA patients as well as SLE patients and healthy donors. The frequency of 24.7% determined by ELISA corresponded well with the result of the first screening by 2-D electrophoresis/Western blotting (autoantibody frequency of 20%).

Predominant detection of the anti-TPI autoantibodies in OA was more distinct in the measurement of synovial fluid. Furthermore, in the paired samples of sera and synovial fluid from individuals, the anti-TPI antibody titers were higher in the synovial fluid samples than in the sera (Figure 4c). This finding may reflect the pathologic importance of the anti-TPI autoantibodies in the joints.

Our experiments also sought to detect the IgG types of the anti-TPI autoantibodies. Although we also checked the IgA, IgM, and IgD types, they were rarely detected (data not shown). In the epitope mapping, each of the 8 regions (TPI-1 to TPI-5 and TPI-1a to TPI-1c) were recognized by a part of the OA serum samples. This indicates that there are at least 7 epitopes on TPI. Since the combination of the reactive regions was different among the serum samples, the antigenicity of the 7 epitopes would be different from each other. Furthermore, 1 serum sample recognized an average of 3.5 regions. The recognition of the multiple epitopes would not be produced by cross-reaction, but rather, would be more likely to be produced by the antigen-driven mechanism; that is, B cells specific for each of the epitopes were activated by the help of the antigen-specific T cell. The dominance of the IgG-type anti-TPI autoantibodies, which indicated class-switching, would also support the antigen-driven mechanism. In this respect, the generating mechanisms and pathologic roles of the anti-TPI autoantibodies in OA would be different from those of the IgM-type anti-TPI autoantibodies previously reported in patients with acute hepatitis A virus infection, infectious mononucleosis, or malaria (23–25).

Clinically, the mean radiographic grade of the OA joints was significantly lower in the anti-TPI–positive patients than in the anti-TPI–negative patients. This suggests that production of the anti-TPI autoantibodies would be related to the early phases of OA.

It remains unclear whether the anti-TPI autoantibodies play a substantial role in the development of OA. Informatively, TPI is a housekeeping glycolytic enzyme that interconverts GDP and DHAP. The autosomal recessive disease of TPI deficiency is characterized by chronic hemolytic anemia, neurologic disturbances, susceptibility to bacterial infection, and cardiomyopathy (26, 27). However, decreased production of ATP and accumulation of DHAP failed to explain the occurrence of hemolytic anemia (28).

In the case of chondrocytes, energy production in the chondrocytes has been demonstrated to be dependent mainly on the anaerobic metabolism (29). Since the anaerobic condition needs more amounts of glucoses than does the aerobic condition to produce the same level of energy, interconversion from DHAP to GDP by TPI would be of critical importance in chondrocytes. In this context, conditions involving stress on the joints, such as OA, may increase the need for TPI. Accordingly, we observed that the mRNA of TPI in cultured chondrocytes was increased by stimulation with interleukin-1β, which has a pathologic role in OA (data not shown). Considering these data, if the anti-TPI autoimmunity inhibits the enzymatic activity of TPI, it can directly lead to damage of the chondrocytes. Further studies are needed to elucidate whether such a pathway is possible.

TPI can be important as an antigen to generate chronic inflammation in OA, even though it is milder than in RA. Recently, GPI, a ubiquitously expressed protein and also one of the glycolytic enzymes, was demonstrated to be a pathogenic autoantigen in a K/B×N murine arthritis model, and the anti-GPI autoantibodies were detected in a subset of patients with RA (30–32). The ways in which the antibodies to a ubiquitous cytoplasmic enzyme provoke joint-specific autoimmune disease are further explained in some investigations.

Maccioni et al generated anti-GPI monoclonal antibodies (mAb) from spontaneously activated B cells in lymphoid organs of arthritic K/B×N mice and injected these mAb into healthy recipients individually (33). They found that the pathogenicity of the mAb depended on the ability to form mAb/GPI multimers by simultaneous recognition of different epitopes. Matsumoto and colleagues observed the deposits of GPI/anti-GPI and C3 on the lining of the articular cavity and cartilage surface in both arthritic mice and human arthritic joints (34). In contrast with the findings in the joints, C3 was not located with GPI–anti-GPI IgG complexes in kidney glomeruli. Based on these phenomena, the authors speculated that GPI could be passively transferred from the serum into synovial fluid and deposited on the surface of cartilage, which started inflammation (34). The deposition of immune complexes and complement components on the surface of articular cartilage was found in OA as well as in RA (35). In the present study, anti-TPI IgG from OA was demonstrated to recognize multiple epitopes on TPI, with high titers in both the serum and synovial fluid from the OA patients. Taking these facts together, the potential of the alternative complement pathway to play a pathophysiologic role in OA may be related to TPI and anti-TPI IgG. However, further studies are needed to elucidate this concept.

In summary, we demonstrated different autoimmune profiles in patients with OA as compared with patients with RA, by a proteomic approach. One of the OA-specific autoantibodies was the IgG-type anti-TPI autoantibody, which we have now described for the first time, to our knowledge. The anti-TPI autoantibody would be a potent diagnostic marker for OA and may play pathogenic roles in the development of OA.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors would like to thank Miss Mie Kanke and Miss Yumiko Ajiri for their technical assistance, and Miss Ishitani for secretarial assistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES