Systemic Sclerosis (Scleroderma): Specific Autoantigen Are Selectively Overexpressed in Scleroderma Fibroblasts

This information is current as Xiaodong Zhou, Filemon K. Tan, Momiao Xiong, Dianna M. of October 2, 2021. Milewicz, Carol A. Feghali, Marvin J. Fritzler, John D. Reveille and Frank C. Arnett J Immunol 2001; 167:7126-7133; ; doi: 10.4049/jimmunol.167.12.7126 http://www.jimmunol.org/content/167/12/7126 Downloaded from

References This article cites 28 articles, 4 of which you can access for free at: http://www.jimmunol.org/content/167/12/7126.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on October 2, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2001 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Systemic Sclerosis (Scleroderma): Specific Autoantigen Genes Are Selectively Overexpressed in Scleroderma Fibroblasts1

Xiaodong Zhou,2* Filemon K. Tan,* Momiao Xiong,‡ Dianna M. Milewicz,† Carol A. Feghali,¤ Marvin J. Fritzler,¶ John D. Reveille,* and Frank C. Arnett*

The pathogenesis of systemic sclerosis (SSc) involves complex interactions between activated fibroblasts eventually leading to fibrosis, and impaired immune tolerance characterized by a variety of circulating SSc-specific autoantibodies. The expression of autoantigens in fibroblasts, a key target tissue in SSc, may play an important role in this process. To obtain a global view of this process, we examined expression profiles of SSc dermal fibroblasts using cDNA microarrays. The results show that dermal fibroblasts from SSc patients obtained from either affected or unaffected skin displayed a characteristic pattern of increased SSc centromeric ,(0.028 ؍ autoantigen compared with that from normal controls. In particular, fibrillarin (p -were significantly overex (0.02 ؍ and RNA polymerase II (220 kDa; p ,(0.042 ؍ centromeric autoantigen P27 (p ,(0.01 ؍ B(p Downloaded from pressed in SSc fibroblasts. Quantitative RT-PCR confirmed overexpression of these autoantigens and also revealed increased levels /The polymyositis .(0.0318 ؍ of DNA topoisomerase I transcripts in SSc fibroblasts compared with normal control fibroblasts (p To examine the specificity of these overexpressed .(0.09 ؍ scleroderma autoantigen gene was overexpressed in some SSc patients (p autoantigen genes for SSc and its tissue specificity for fibroblasts, cDNA microarrays of dermal fibroblasts from patients with eosinophilic fasciitis and scleromyxedema were studied as well as PBMC and muscle biopsies from SSc patients. None of these

tissues showed significant alterations in gene expression of SSc-specific autoantigens. Therefore, SSc-associated autoantigen genes http://www.jimmunol.org/ are selectively overexpressed in SSc dermal fibroblasts, a major tissue involved in disease pathogenesis. The Journal of Immu- nology, 2001, 167: 7126Ð7133.

ystemic sclerosis (SSc)3 is a multisystem disorder of con- anti-U3-ribonucleoprotein (RNP) (fibrillarin), polymyositis/sclero- nective tissue characterized by extensive cutaneous and derma autoantigen (PM-Scl), Th/To, and U1-RNP (3). Each SSc S visceral fibrosis and small vessel vasculopathy. While the patient typically produces only one of these autoantibodies, which, etiopathogenesis of SSc is unknown, pathological findings indicate in turn, correlates with certain disease manifestations and progno- that fibroblasts are activated to produce excessive amounts of col- sis. It is unknown, however, whether autoantibodies directly par- lagen and other extracellular matrix components. Moreover, endo- ticipate in any of the pathological manifestations. SSc-specific Abs by guest on October 2, 2021 thelial cell damage causing capillary loss and neo-intimal prolif- also are associated with specific class II alleles of the MHC. Con- eration in small vessels plays a prominent role in tissue damage siderable evidence suggests that such autoantibodies result from (1). Autoimmune mechanisms also are believed to be operative, autoantigen-driven Th cell-mediated immune responses (4). because patients with SSc spontaneously elaborate a variety of The present studies using a 4000-element cDNA microarray to highly disease-specific circulating autoantibodies to nuclear and profile gene expression in cultured dermal fibroblasts from SSc nucleolar Ags (2). The most common of these autoantibodies are patients demonstrated that several autoantigen genes specifically directed against DNA topoisomerase I; centromeric , es- targeted in SSc were overexpressed compared with dermal fibro- pecially centromeric protein B (CENP-B); and RNA polymerases blasts from normal controls. Therefore, we surveyed the expres- I, II, and/or III. Less frequent autoantibody specificities include sion of 72 known autoantigen genes from a variety of au- toimmune diseases that were represented on the array. Indeed, in SSc fibroblasts, several SSc-specific autoantigen genes appeared to *Division of Rheumatology and Clinical Immunogenetics and †Division of Medical be selectively overexpressed compared with those targeted in other , Department of Internal Medicine, University of Texas Medical School, autoimmune diseases. Transcript overexpression of these genes Houston, TX 77030; ‡Human Genetics Center, University of Texas School of Public Health, Houston, TX 77030; ¤Division of Rheumatology and Clinical Immunology, was confirmed using RT-PCR. Muscle tissue and PBMC from SSc Department of Medicine, University of Pittsburgh, Pittsburgh, PA 15261; and ¶Fac- patients did not show gene expression changes for the SSc-specific ulty of Medicine, University of Calgary, Calgary, Alberta, Canada autoantigens compared with their normal counterparts, nor did der- Received for publication July 27, 2001. Accepted for publication October 18, 2001. mal fibroblasts from patients with other fibrosing diseases. These The costs of publication of this article were defrayed in part by the payment of page findings may provide clues to the origins of autoimmune responses charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and pathogenetic mechanisms in SSc. 1 This work was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Specialized Center of Research in Scleroderma Grant IP50AR4488, National Materials and Methods Institutes of Health National Center for Research Resources Grant 3M01RR02558-12S1 Study subjects (to F.K.T.), and a grant from the RGK Foundation (Austin, TX). 2 Address correspondence and reprint requests to Dr. Xiaodong Zhou, Division of Rheu- For the microarray and quantitative-RT-PCR, 3-mm punch skin biopsies of matology and Clinical Immunogenetics, University of Texas Medical School, 6431 Fan- clinically affected and/or unaffected skin were obtained from 11 SSc pa- nin, MSB5.270, Houston, TX 77030. E-mail address: [email protected] tients with disease duration of Ͻ5 yr. Two patients were Choctaw Native 3 Abbreviations used in this paper: SSc, systemic sclerosis; CENP-B, centromeric Americans, three were African Americans, one was Mexican-American, protein B; sDNA, synthetic DNA; RNP, ribonucleoprotein; PM-Scl, polymyositis/ and five were Caucasians. Three patients had limited SSc, and the remain- scleroderma autoantigen. der had diffuse skin disease as defined by Leroy et al. (5). All SSc patients

Copyright © 2001 by The American Association of Immunologists 0022-1767/01/$02.00 The Journal of Immunology 7127

fulfilled the American College of Rheumatology Criteria for the classifi- Array images were analyzed with Pathways software (Research Genetics). cation of scleroderma (6). As controls with other fibrosing diseases, two Genes that gave a signal at or below background levels were discarded. patients each with scleromyxedema and eosinophilic fasciitis also were biopsied. These biopsies were obtained when the patients had active dis- Array analysis ease from lesional skin. Normal control skin samples were obtained from seven age- (Ϯ5 yr) and sex-matched patients with no history of autoim- To account for gene expression changes due to experimental variation, mune diseases who were undergoing routine dermatological surgery. gene expression data were normalized as follows. One control sample array Anti-coagulated blood was obtained from three SSc patients (these three was randomly selected as the reference. The reciprocal of the coefficient of patients also had fibroblasts explanted from skin biopsies) and from three the regression of the gene expression of subsequent arrays over the refer- age- and sex-matched normal controls. PBMC were isolated from the anti- ence array was taken as a normalizing factor. For each array, the normal- coagulated blood by Ficoll-Hypaque gradient centrifugation and were ization factor was derived in this manner, and the expression of each gene washed three times through centrifugation. in the array was multiplied by this factor to result in normalized gene Stored frozen skeletal muscle biopsies were obtained from three differ- expressions. The normalized data were transformed into logarithms and ent SSc patients who had under gone diagnostic biopsy and from three age- clustered using a hierarchical clustering algorithm (7). Student’s paired t and sex-matched controls. These SSc patients had symptoms of muscle tests were used to measure differences in the degree of gene expression in weakness and/or pain. Two of the biopsies were normal histologically, the normal vs disease tissues. The t statistic provides an estimate of the while the third showed changes compatible with a noninflammatory my- difference (and the precision of the difference) in mean expression ratios (or opathy. Three matched controls with stored frozen muscle biopsies who fold differences) of the two groups. Categorical differences were tested had muscle complaints but histopathologically normal muscle biopsies and using Fisher’s exact tests. The level of statistical significance was set at other normal diagnostic studies (creatine kinase levels and electromyogra- p Ͻ 0.05. phy) were similarly studied. Study subjects provided written informed con- sent, and the protocol was approved by the University of Texas committee Quantitative RT-PCR for the protection of human subjects. Downloaded from Quantitative real-time RT-PCR was performed using an ABI 7700 Se- Tissue culture quence Detector (Applied Biosystems, Foster City, CA) (8) for the tran- A 3-mm skin biopsy was stored in DMEM with 10% FCS supplemented scripts of DNA topoisomerase I, fibrillarin, CENP-B, and au- with an antibiotic and antimycotic. The tissue sample was subsequently toantigen p27 (CENP-p27). Specific quantitative assays were developed washed in 70% ethanol, PBS, and 10% FCS/DMEM. Cultured fibroblast using Primer Express software (PE Biosystems) following the recom- cell strains were established by mincing and placing tissues into 60-mm mended guidelines based on sequences from GenBank. The primers and culture dishes secured by glass coverslips. The primary cultures were main- probes for each gene are listed in Table I. Total RNA was extracted from tained in DMEM with 10% FCS and supplemented with antibiotic and study tissues as described previously, and they were the same RNA sam- http://www.jimmunol.org/ antimycotic. Low passage fibroblasts cell strains were plated at a density of ples as those used in the microarray assays. cDNA was synthesized in a 10-␮l total volume by the addition of 6 ␮l/well RT master mix, consisting 2.5 ϫ 105 cells in a 35-mm dish and incubated for RNA isolations. of 400 nM assay-specific reverse primer, 500 ␮M deoxynucleotides, Su- Microarrays perscript II buffer, DTT, and 10 U Superscript II reverse transcriptase (Life Technologies), to a 7700 96-well plate, followed by a 4-␮l volume of Total RNA was isolated from cultured fibroblast cell strains, muscle biop- sample (25 ng/␮l). Each sample was measured in triplicate plus a control sies, and PBMC using TRIzol reagent (Life Technologies, Gaithersburg, without reverse transcriptase. Each plate also contained an assay-specific MD) and treated with DNase I (Ambion, Austin, TX). Double-strand sDNA (synthetic amplicon oligo) standard spanning a 5-log range and a no cDNA probe with [33P]dCTP was generated from 1 ␮g total RNA using the template control. Samples were incubated in a thermocycler (MJ Research, Superscript cDNA system (Life Technologies). Labeled probe was purified Waltham, MA) for 30 min at 50¡C, followed by 72¡C for 10 min. Subse- by a chromatograph column (5 Prime33 Prime, Boulder, CO), and then quently, 40 ␮l PCR master mix (400 nM forward and reverse primers, 100 by guest on October 2, 2021 ␮ ␮ suspended in 100 l hybridization buffer. The sp. act. of the probe was nM fluorogenic probe, 3 mM MgCl2, 200 M deoxynucleotides, PCR determined in a scintillation counter. A total of 3 ϫ 106 cpm probe was buffer, and 1.25 U Taq polymerase (Life Technologies)) were added di- denatured at 95¡C for 8 min before hybridization. rectly to each well of the cDNA plate. Each assembled plate was then Microarray filters spotted with 4000 human cDNAs (GF-211, Research capped and run in the 7700 using the following cycling conditions: 95¡C Genetics, Huntsville, AL) were prehybridized in 5 ml hybridization buffer for 1 min, and 40 cycles of 95¡C for 12 s and 60¡C for 1 min. The resulting with 5 ␮l poly(A) (1 mg/ml) and 5 ␮l denatured Cot-1 DNA at 42¡C with data were analyzed using SDS software (Applied Biosystems, Foster City, gentle rotation for 3 h. The denatured probe was loaded into the same CA) with TAMRA as the reference dye. buffer, then the rotating hybridization was continued at 42¡C for 16 h. The Synthetic DNA oligos used as standards (sDNA) encompassed exactly hybridized filter was washed twice in 2ϫ SSC/1% SDS at 50¡C for 20 min, the entire 5ЈÐ3Ј amplicon for the assay (BioSource, Camarillo, CA). The followed by 15 min at room temperature. The hybridization signal was sDNA standards were quantified by absorbance at 260 nm. Standard curves detected by the Cyclone Phosphor System (Packard Bioscience, Downers were run from 2 to 200 pg sDNA on each plate. The number of molecules Grove, IL). Duplicate RNA samples were tested in all but four samples. per sDNA mass was then calculated from the length of the sDNA for each

Table I. Primers used in quantitative RT-PCR and quality control data for each assay a

Correlation Genes Accession No. Primers Slope y-Intercept Coefficient

Topo I AA232856 Forward (1511ϩ) TCGTGTGGAGCACATCAATC Reverse (1586Ϫ) CTTCCCGAGGAAGTCAAACTC Ϫ3.338 38.175 0.995 Probe (1536ϩ) CCAGAGTTGGATGGTCAGGAATATGTGGT

Fibrillarin AA663986 Forward (700ϩ) CATGCTCATCGCAATGGTG Reverse (765Ϫ) ACAATCCGGGTCTGGTCTG Ϫ3.248 38.299 0.993 Probe (721ϩ) TGTGATCTTTGCTGATGTGGCCCAG

CENP-B AA455481 Forward (2587ϩ) TTGCCATGGTCAAGAGGTACC Reverse (2670Ϫ) CCAGATCGTGTTCCAAGTGG Ϫ3.323 38.157 0.998 Probe (2610ϩ) ACCTCCTTCCCCATTGATGACCGC

CENP-p27 AA456077 Forward (143ϩ) TCCCAAGCTCGGGAGCACCA Reverse (224Ϫ) CGTCCTGAGTTCCGTCGGGTC Ϫ3.085 29.894 0.998 Probe (179Ϫ) TGGCCTCAGCCTCAGAGCTCCC

a Probes contain a reporter dye (6-FAM) at its 5Ј end and a quencher dye (TAMRA) at its 3Ј end. 7128 AUTOANTIGENS IN SCLERODERMA FIBROBLASTS assay. Each set of standards yielded a linear curve (r2 Ͼ 0.992), and effi- fibroblasts (4 of 16 SSc autoantigens, excluding PM-Scl (100 ciencies calculated from the slope of each standard curve were Ͼ95%. kDa), vs 3 of 56 non-SSc autoantigens) resulted in a p value of Due to the inherent inaccuracies in quantitating total RNA by absor- 0.039 and an odds ratio of 5.9, while inclusion of PM-Scl yielded bance, the amount of RNA added to an RT-PCR from each sample was more accurately determined by measuring the ␤-actin transcript levels in a p value of 0.01 and an odds ratio of 8.0. If vimentin and laminin each sample. The final data were normalized to ␤-actin and are presented S were considered as SSc-related autoantigens, then the likelihood as the molecules of transcript/molecules of ␤-actin ϫ 100 (percentage of that these observations were occurring by chance alone is 7 of 16 ␤-actin). vs1of56(p ϭ 0.00005; odds ratio, 42.8). A semiquantitative RT-PCR method was used for confirmation of mi- croarray data for the RNA polymerase II and PM-Scl transcripts. The prim- The altered expression of autoantigen genes in SSc skin fibro- ers and probe for RNA polymerase II are: forward primer (2534ϩ), CAT blast strains did not show significant differences in expression CGAGAAGGCACACAACA; reverse primer (2615Ϫ), GCGGTTC when fibroblasts from clinically affected and unaffected skin were ACCTGATTCTCAA; and probe (2573Ϫ), CGTCTGCCGCAGAGTGT studied in the same patients (data not show). Increased expression ϩ TCCCT. The primers and probe for PM-Scl are: forward primer (1852 ), levels (ratios ϭ 1.4 or higher) of two or more of the five autoan- TCTCTTTGGACCTCACGACTG; reverse primer (1916Ϫ), CACTGGT TGGGATGATTGGA; and probe (1874ϩ), TCCCATGCCCCTCCGG tigen genes was found in 10 of the 11 SSc patients (Table IV). ATG. RNA samples were run as described above without a standard However, the pattern of SSc autoantigen gene expression did not (sDNA). Molecules of transcript for each sample were determined math- correlate with the type of SSc-specific autoantibody expressed by ematically in Microsoft Excel from the slope and y-intercept of a transcript- each patient (Table IV). specific standard curve (sDNA) run on a separate plate. Although this method is not preferred for absolute quantitation, it will give correct rel- The cluster analysis of all SSc and control fibroblast strains ative values for comparison of the samples to one another. based on the expressions of the 72 autoantigen genes resulted in classification of the fibroblast strains into two major groups (Fig. Downloaded from Autoantibody testing 2). The first group included 8 of the 11 SSc fibroblast strains whose All patient sera were tested for antinuclear Abs by indirect immunofluo- gene expression patterns were marked by overexpression of SSc rescence using HEp-2 cells as Ag substrate (Antibodies, Davis, CA). Anti- autoantigen genes. The other three SSc fibroblast strains and all centromere was determined visually by their distinctive indirect immuno- seven control fibroblast strains were classified into a second major fluorescence patterns on HEp-2 cells. Anti-topoisomerase I (Scl-70) Abs group, but two of the three SSc fibroblast strains were classified were detected by passive immunodiffusion against calf thymus extracts under different subgroups (Fig. 2).

using commercial kits (INOVA, Diagnostics, Inc., San Diego, CA). Anti- http://www.jimmunol.org/ fibrillarin (9) and anti-RNA polymerases I, II, and III (10) were detected by To assess the specificity of SSc autoantigen overexpression, we immunoprecipitation as described previously. Anti-PM-Scl Abs were de- examined muscle biopsies and PBMC from three SSc patients and tected by immunoprecipitation and confirmed by Ouchterlony double im- three controls using the same assays. We found no such altered munodiffusion (11). expression of these genes in either tissue of any of these patients (Fig. 3). Of potential interest, however, the 70-kDa U1-RNP gene Results showed 2-fold increased expression levels in all three of the mus- Microarrays cle biopsies from SSc patients. Thirty-two genes of approximately 4000 genes on the array To further determine the disease specificity of autoantigen gene

showed altered expression levels that were statistically signifi- overexpression for SSc, we examined gene profiles of cultured by guest on October 2, 2021 cantly different ( p Յ 0.05) from those of controls and showed an dermal fibroblasts from skin biopsies of two eosinophilic fasciitis adequate signal above background. These included seven autoan- and two scleromyxedema patients. None of them showed increased tigen genes and 25 nonautoantigen genes (Tables 2 and 3). The gene expressions of SSc autoantigens, but the No55 gene showed cluster analysis of all SSc and control fibroblast strains based on increased expression in one of each of these cases (Fig. 4). the expressions of these genes resulted in a clear distinction be- Finally, an additional 25 genes also displayed statistically sig- tween patients and controls (Fig. 1). nificantly ( p Ͻ 0.05) altered expression levels in SSc fibroblasts There were 72 cDNAs on the microarray that have been re- compared with control fibroblasts (Table III). Although some ra- ported as autoantigen genes in various autoimmune diseases. Six- tios between SSc patients and controls are less than 2-fold, these teen of these autoantigens were SSc specific or occurred frequently genes demonstrated consistent patterns and adequate signal above in SSc patients (Table II). The remaining 56 autoantigens occurred background levels. The functions of these genes included extra- specifically in other connective tissue diseases, such as systemic cellular matrix composition and regulation, apoptosis, gene tran- lupus erythematosus, Sjo¬gren’s syndrome, poly- or dermatomyo- scription, inflammation, metabolic regulation, and others. The ex- sitis, and certain organ-specific autoimmune disorders, or as non- pression of some collagen genes, e.g., collagen type I ␣-2 and specific autoantigens in a variety of disease states. collagen type III ␣-1 genes, were increased in some SSc patients. Among the 16 SSc-specific or -associated autoantigen genes, Over 2-fold increased gene expression of HLA-C and HLA-DRA1 five demonstrated an average increased expression of 1.44-fold or also was observed in four SSc patients. higher in the SSc patients than in the controls. The average ex- pression levels of four of these five autoantigen genes were sta- tistically significantly different between patients and controls using Quantitative RT-PCR t statistics, including CENP-B, CENP-p27, U3-RNP (fibrillarin), Quantitative RT-PCR was used to confirm the gene expression and the 220-kDa subunit of RNA polymerase II. The PM-Scl (100 levels for SSc-specific autoantigens in skin fibroblast strains. Be- kDa) autoantigen gene also showed an average 1.44-fold increased cause topoisomerase I is a major autoantigen in SSc (but was not expression, but in fewer patients ( p ϭ 0.091). Only 3 of the 56 detected in our microarray assays as its signal was not much higher non-SSc-specific autoantigen genes showed statistically signifi- than background), it also was specifically assayed using RT-PCR. cantly increased expression levels. Two of these, laminin S (ratio, The overall gene expression levels for CENP-B, CENP-p27, fibril- 1.58; p ϭ 0.008) and vimentin (ratio, 1.52; p ϭ 0.008), have been larin, RNA polymerase II, and PM-Scl (100 kDa) in this assay reported as nonspecific autoantigens in SSc previously (12, 13). were reasonably concordant with those seen on the cDNA microar- The third, nucleolar autoantigen No55 (ratio, 1.67; p ϭ 0.053) has rays, although several genes showed 2- to 3-fold higher levels by not been studied in SSc to our knowledge. A statistical comparison RT-PCR. The topoisomerase I gene displayed an average 4.5-fold of the number of overexpressed SSc autoantigen genes in SSc increase in expression level using RT-PCR (Table IV). The Journal of Immunology 7129

Table II. Comparisons of average gene expression ratios of 72 autoantigens in 11 SSc vs 7 normal control skin cultured fibroblasts a

Gene Disease Accession No. Av. Ratio p

Centromere protein B (80 kDa) SSc AA455481 1.56 0.010 Centromere autoantigen C SSc AA461496 1.03 0.726 (312 kD) SSc AA411850 0.97 0.706 Centromeric autoantigen (27 kDa) SSc, Sjo¬gren’s syndrome AA456077 1.65 0.042 DNA topoisomerase I SSc AA232856 1.00 0.955 Fibrillarin SSc AA663986 1.72 0.028 Autoantigen PM-SCL (100 kD) SSc, myositis AA487064 1.44 0.091 Pericentriole material 1 (PCM-1) SSc AA164440 0.96 0.359 Fibrillin 1 SSc AA418674 1.14 0.142 Gu protein SSc, NS AA465386 0.97 0.500 RNA polymerase II (220 kDa) SSc, SLE AA479052 1.44 0.020 RNA polymerase II (140 kDa) SSc, SLE N74956 0.95 0.797 RNA polymerase I subunit (hRPA39) SSc AA733038 0.98 0.997 RNA polymerase III subunit (RPC62) SSc AA282063 1.08 0.857 U1 snRNP (70 kDa) MCTD R02346 1.10 0.200 U1 snRNP-specific C protein SLE, SSc AA253448 1.06 0.261 U1 snRNP-A SLE R70488 1.03 0.711 U2 snRNP-A SLE AA122272 1.24 0.179 U5 snRNP (100 kDa) SLE AA598470 1.09 0.844 Downloaded from snRNP polypeptides D1 SLE H16255 1.10 0.866 U4/U6 snRNP SLE AA703250 1.01 0.364 snRNP polypeptides B SLE AA599116 1.13 0.645 snRNP polypeptide N SLE T54926 1.03 0.102 snRNP polypeptide E SLE AA678021 0.92 0.875 Sm protein G SLE AA133577 0.86 0.726 Sm protein F SLE AA668189 1.04 0.706

snRNP core protein Sm D2 SLE T62529 1.05 0.496 http://www.jimmunol.org/ Sm-like protein CaSm SLE AA628430 1.08 0.955 H2A SLE AA426352 1.19 0.715 .X SLE H95392 0.99 0.910 Histone H1D SLE T66815 0.99 0.359 .1 SLE AA456298 1.26 0.142 Histone H1X SLE W81318 1.00 0.900 Ki nuclear autoantigen SLE AA486324 1.05 0.422 Proliferating cell nuclear Ag SLE AA450265 1.05 0.666 Annexin V SLE AA451895 1.14 0.064 RNP autoantigen SS-A/Ro (60 kDa) SLE, Sjo¬gren’s syndrome AA010352 0.91 0.060 RNP autoantigen SS-A/Ro (52 kDa) SLE, Sjo¬gren’s syndrome N45131 1.12 0.310 by guest on October 2, 2021 Autoantigen La (SS-B/La) SLE, Sjo¬gren’s syndrome H29485 0.99 0.258 Alanyl-tRNA synthetase Myositis AA156571 0.99 0.881 Isoleucine-tRNA synthetase Myositis AA410636 0.88 0.060 Histidyl-tRNA synthetase Myositis H61209 0.91 0.310 Glycyl-tRNA synthetase Myositis AA629909 0.89 0.258 Threonyl-tRNA synthetase Myositis AA630628 0.90 0.098 Signal recognition particle (19 kDa) Myositis AA411407 1.00 0.997 Signal recognition particle (54 kDa) Myositis AA599078 1.01 0.857 Mi-2 protein (218 kDa) Myositis N34372 0.99 0.569 Bullous pemphigoid Ag 1 (230 kDa) Bullous pemphigoid H44785 1.02 0.809 Bullous pemphigoid Ag 2 (180 kDa) Bullous pemphigoid H87536 0.95 0.470 Amphiphysin (128-kDa autoantigen) Stiff-man syndrome H06483 1.02 0.814 Insulin receptor Diabetes AA001614 0.95 0.434 Thyroid autoantigen 70 kDa (Ku Ag) Organ specific AA486207 1.00 0.960 Myasthenic syndrome Ag B Organ specific R92452 0.99 0.910 Leiomodin 1 Organ specific AA441933 1.06 0.310 Thyroid autoantigen Organ specific AA486239 1.13 0.385 Nuclear autoantigenic sperm protein Organ specific AA644128 1.34 0.267 Nucleolar autoantigen No55 NS W81191 1.67 0.053 Laminin S NS AA156802 1.58 0.008 Laminin ␥1 NS H24650 1.09 0.242 Laminin, ␣4 NS R43734 0.99 0.915 Nuclear autoantigen GS2NA NS AA418821 1.00 0.908 Cytoplasmic autoantigen NS AA481276 1.08 0.107 Autoantigen (Ngp-1) NS AA446557 1.09 0.260 Myelin basic protein NS H17080 0.71 0.199 Annexin XI (56-kDa autoantigen) NS AA464982 0.91 0.061 Vimentin NS AA486321 1.52 0.008 Keratin 7 NS AA485959 1.24 0.258 Keratin 14 NS H44051 1.07 0.106 Golgin-95 (gm 130) NS AA424786 1.12 0.098 Golgi Ag gcp372 (giantin) NS AA485974 0.87 0.074 Golgin-245 (gcp 230) NS AA460981 1.08 0.179 Golgin-97 NS R44140 1.31 0.056

a SLE, systemic lupus erythematosus; MCTD, mixed connective tissue disease; NS, nonspecific autoantibodies in various diseases; snRNP, small nuclear RNP. In boldface are those genes with altered expression levels and statistically significant p values. 7130 AUTOANTIGENS IN SCLERODERMA FIBROBLASTS

Table III. Twenty-five nonautoantigen genes show altered expression levels between 11 SSc and 7 normal control fibroblasts a

Gene Title Function p Accession No. Ratio

Matrix metalloproteinase 1 (MMP1) Extracellular matrix 0.04153 AA143331 3.8 1 Stromelysin (MMP3) Extracellular matrix 0.02461 W51794 2.4 1 SPARC/osteonectin Extracellular matrix 0.045 H95960 2.0 1 Cysteine protease Extracellular matrix 0.03644 AA425938 1.6 1 Metallopeptidase 1 Extracellular matrix 0.0455 H98666 1.5 1 Cystatin C Extracellular matrix 0.04277 AA599177 1.6 1 Amyloid precursor protein-binding family B Apoptosis 0.00449 AA488247 1.6 1 Amyloid ␤A4 precursor protein Apoptosis 0.03191 W42849 1.6 1 Amyloid precursor protein-binding family A Apoptosis 0.001 H19687 1.6 1 Zinc finger protein 10 (KOX 1) Transcription 0.02146 AA877082 1.8 1 cAMP response element-binding protein 1 Transcription 0.04247 AA161486 1.8 1 Transportin for hnRNP Transcription 0.0042 R08897 1.7 1 Colony-stimulating factor 1 (M-CSF) Inflammation 0.02796 T55558 1.4 1 Cystein-rich, angiogenic inducer 61 Inflammation 0.02181 AA777187 1.6 1 Phosphoglycerate mutase 1 Glycolytic pathway 0.01019 AA676970 1.6 1 Triosephosphate isomerase 1 Glycolytic enzyme 0.01964 AA663983 1.6 1 Human cell adhesion protein (SQM1) Cell adhesion 0.04282 AA428058 1.6 1 Epidermal surface Ag 1 Cell adhesion 0.03833 R73545 1.5 1 1 Eph family protein Cellular receptor 0.03158 AA609284 1.6 Downloaded from Myoglobin Metabolic regulation 0.00335 AA176581 2.0 1 Mitochondrial RNA polymerase Mitochondrial gene expression 0.01946 R31115 1.5 1

CCR1 Inflammatory 0.0008 AA036881 3.8 2 Insulin-like growth factor-binding protein 3 Inflammatory 0.04461 AA598601 1.9 2 Human hemopoietic progenitor kinase Growth regulation 0.0056 T50313 2.5 2 Splicing factor Sip1 Pre-mRNA splicing 0.0016 H78241 2.4 2 http://www.jimmunol.org/ a Average ratios are labeled as increased (1) or decreased (2) in SSc patients. hnRNP, Heterogeneous RNP; SPARC, secreted protein, acidic, and rick in cysteine.

Discussion expressions were statistically significant. Whether these changes While using cDNA microarrays containing 4000 known human ultimately prove to be clinically or pathogenetically relevant re- genes to investigate transcript expression in cultured dermal fibro- mains to be seen. Another comparison examining the number of blasts of patients with SSc, we noted that several genes known to SSc-related autoantigens showing altered expression (4 of 16) in be autoantigens in SSc were overexpressed compared with normal SSc fibroblasts as opposed to non-SSc-related autoantigens (3 of fibroblasts. The average increased folds of expressions of four au- 56) indicated that the likelihood that these observations were due by guest on October 2, 2021 toantigen genes (CENP-B, CENP-p27, fibrillarin, and the 220-kDa to chance alone was small ( p ϭ 0.039). In addition, another SSc subunit of RNA polymerase II) were Ն1.44 in the SSc fibroblasts autoantigen gene, PM-Scl (100 kDa), also appeared to show compared with normal fibroblasts. In some cases, these ratios were increased expression levels averaging 1.44-fold in six of the 11 Ͼ2-fold. Although it is difficult to predict which ratio of these SSc patients, although the overall comparison of expression increases may be biologically important, comparisons between all levels between patients and controls was not statistically sig- SSc patients and controls showed that these alterations in gene nificant ( p ϭ 0.09).

FIGURE 1. Cluster analysis of gene ex- pression levels of 32 genes that showed sta- tistically significant alterations and adequate signals above background between SSc pa- tients and controls. Gene names and acces- sion numbers are listed on the right of the cluster. Colors from green to red assignment represent gene expression levels from lower to higher, respectively, based on individual data. The genes expressed in similar patterns are clustered on the left of the diagram. Based on the individual gene expression pro- file, SSc patients and controls are clustered on the top of the diagram. hnRNP, Hetero- geneous RNP. The Journal of Immunology 7131

Table IV. Autoantibody and autoantigen gene profiles for 11 SSc patientsa

Patient

1 2 3 4 5 6789 1011

Autoantibody Topo I RNAP CENP PM-Scl Unknown CENP CENP CENP Topo I PM-Scl RNAP I, III I, II, III

cDNA microarray Topo I ND ND ND ND ND ND ND ND ND ND ND Fibrillarin 1.7 0.9 2 1.2 1.8 1.5 2.1 3.1 1.2 1.9 1.5 CENP-B 2 1.3 2.5 1 1.6 1.3 1.6 1.8 1.7 0.9 1.3 CENP-p27 1.3 2.6 1.2 1.2 1.8 1.5 1.2 1.8 1.4 2.9 1.3 RNAP II 1 1.8 1.2 1.2 1.1 1.8 2.5 1.6 1.2 0.9 1.4 PM-Scl (100 kDa) 1.8 1.8 0.9 1.6 1.2 1.4 2.1 1.2 0.9 1.7 1.2

Quantitative RT-PCR Topo I 2 5 8 15 10 2.5 2.5 2 1 2 1.7 Fibrillarin 2.1 1 6.1 2.7 1.7 1 2.1 3.1 1 1.5 1.7 CENP-B 2.5 4.2 5.4 1.5 2 1 1.6 2.1 3.7 1 1 CENP p27 0.8 2.3 1.2 1.4 1.9 1 1.1 1 1 3.4 1.8 RNAP II 1.1 2.6 1.4 1.1 1.1 NA NA NA NA NA NA Downloaded from PM-Scl (100 kDa) 1.8 2.3 1.1 1.7 1 NA NA NA NA NA NA

a Each patient shows specific serum autoantibody and altered gene expression levels of autoantigens as compared with age/sex matched normal controls. All numbers indicate fold increase in gene expression level. Abs to: Topo I, DNA topoisomerase I; RNAP I, II, III, RNA polymerase I, II, III; CENP, centromeric protein. NA, not assayed.

The majority of SSc patients demonstrated increased expression SSc patients and are associated with diffuse skin involvement and http://www.jimmunol.org/ of two or more autoantigen genes, but there was no correlation pulmonary fibrosis (16, 17). between the patterns of autoantigen gene expression and autoan- DNA-dependent RNA polymerase II, a complex multisubunit tibody profiles of individual patients. RT-PCR confirmed increased enzyme, is responsible for the transcription of protein-coding transcripts of these autoantigens in the majority of SSc patients’ genes. Autoantibodies to RNA polymerases I, II, and III occur fibroblasts and, in addition, showed increased levels of DNA to- frequently in SSc, often in various combinations; however, anti- poisomerase I transcripts in SSc fibroblasts (Ն1.5-fold in 10 of 11, RNA polymerase III Abs are most common and can be used to 4.5-fold on the average). The reason for our inability to detect predict a high likelihood of diffuse skin disease and severe visceral

changes in topoisomerase I on the arrays may be due to the fact involvement in SSc (18). Anti-RNA polymerase II Abs have been by guest on October 2, 2021 that the sensitivity of array-based assays is lower than that of quan- reported most often in systemic lupus erythematosus; however, titative RT-PCR. Abs to phosphorylated RNA polymerase II occur frequently in Moreover, autoantigen transcripts were not altered in the fibro- Japanese SSc patients who also have anti-topoisomerase I Abs (19) blasts of four patients with other fibrosing diseases (two each with as well as in Choctaw Native Americans, who have a high prev- scleromyxedema and eosinophilic fasciitis). Finally, microarray alence of SSc and anti-topoisomerase I Abs (20) (F. C. Arnett, analyses of muscle tissue and PBMC from SSc patients also failed unpublished observations). to detect any alterations in autoantigen gene expression as seen in Finally, PM-Scl is another nucleolar particle. It consists of sev- the fibroblasts, thus suggesting that this phenomenon was selective eral polypeptides, of which two proteins of 75 and 100 kDa have for SSc dermal fibroblasts (of those tissues studied). been identified as the major antigenic components. Anti-PM-Scl All the autoantigens detected here are targeted in SSc. Anti- Abs occur almost exclusively in a small percentage of Caucasians centromere Abs are highly specific for scleroderma, most typically with SSc, myositis, or both (21). the limited form of the disease or the calcinosis, Raynaud phe- Thus, the SSc autoantigens represent a variety of different pro- nomenon, esophageal dysmotility, sclerodactyly, and telangiecta- teins with differing cellular locations and functions. The reason sia syndrome. Although centromeric autoantigens include that they should be targeted by an autoimmune response primarily CENP-A, CENP-B, CENP-C, and others, CENP-B is considered in scleroderma and singly in individual patients is unclear, al- the major one, since Abs in high titer to CENP-B are consistently though several hypotheses recently have been proposed. Tan (22) found in anti-centromere-positive sera (14). More recently using has pointed out that the majority of SSc-specific autoantigens lo- autoimmune sera from a patient with anti-centromere Abs, Muro et calize to the nucleolus during some phase of their cell cycles. al. (15) identified a novel 27-kDa protein (p27) to which approx- Building upon this observation, Rosen and colleagues (23) have imately 2% of anti-centromere Abs reacted. All cases with anti- noted that heavy metals also are concentrated in the nucleolus and p27 had SSc and/or Sjo¬gren’s syndrome. may catalyze oxidative reactions or in other ways fragment these Fibrillarin is a component of a small nucleolar RNP particle proteins and reveal cryptic epitopes vulnerable to initiating an au- thought to participate in the first step in processing preribosomal toimmune response. It is likely that CD4-positive T lymphocytes RNA. In , fibrillarin is associated with U3, U8, and U13 are involved, because each of these autoantibody responses is as- small nuclear RNAs. Anti-fibrillarin Abs are highly specific for sociated with specific and largely different MHC class II alleles (2, SSc, typically occurring in a small subset with diffuse skin thick- 24, 25). ening and telangiectasia, and visceral involvement in men and Af- The findings in the present study that several of these SSc-re- rican Americans (9). lated autoantigens are specifically and selectively overexpressed in DNA topoisomerase I catalyzes the breaking and joining of SSc fibroblasts are provocative and raise the possibility that DNA strands. Autoantibodies to topoisomerase I are specific for activated fibroblasts may represent the cellular source of the 7132 AUTOANTIGENS IN SCLERODERMA FIBROBLASTS

FIGURE 2. Cluster analysis of gene expression levels of 72 autoan- tigens in 11 SSc vs 7 control skin fi- broblasts. Gene names and accession numbers are listed on the right of the cluster. Colors from green to red as- signment represent gene expression Downloaded from levels from lower to higher, respec- tively, based on individual data. The genes expressed in the similar pat- terns are clustered on the left of the diagram. Based on the individual gene expression profile, patients and con- http://www.jimmunol.org/ trols are clustered on the top of the di- agram. snRNP, Small nuclear RNP by guest on October 2, 2021

autoantigens driving the immune responses leading to the char- not seen, it is possible that multiple Ags can be associated with acteristic SSc-specific autoantibodies. Although the clear rela- a particular autoimmune disease. For example, a primary au- tionships between expression levels of specific autoantigen toantigen may be responsible for the initiation of the autoim- genes and corresponding autoantibodies in patients’ sera were mune process and others become involved later in the course of

FIGURE 3. Transcript levels of autoantigens in three different tissues of FIGURE 4. Transcript levels of autoantigens in skin fibroblasts from SSc patients compared with normal controls. Error bars indicate SD. RNA- three fibrotic diseases compared with seven normal controls. Scl-Myx, PII, RNA polymerase II. Scleromyxedema; EF, eosinophilic fasciitis; RNAPII, RNA polymerase II. The Journal of Immunology 7133 the disease, i.e., epitope spreading (26). It is also important to 7. Eisen, M. B., P. T. Spellman, P. O. Brown, and D. Botstein. 1998. Cluster anal- know that autoantibody responses to specific autoantigen are ysis and display of -wide expression patterns. Proc. Natl. Acad. Sci. USA 95:14863. associated with certain HLA class II alleles (9), and many of the 8. Heid, C. A., J. Stevens, K. J. Livak, and P. M. Williams. 1996. Real time quan- epitopes are influenced by modification in various ways, such as titative PCR. Genome Res. 6:986. metal-catalyzed oxidation (23, 27). This hypothesis would also 9. Arnett, F. C., J. D. Reveille, R. Goldstein, K. M. Pollard, K. Leaird, E. A. Smith, E. C. Leroy, and M. J. Fritzler. 1996. Autoantibodies to fibrillarin in systemic suppose that these autoantigens would of necessity need to be sclerosis (scleroderma). An immunogenetic, serologic, and clinical analysis. Ar- externalized via cellular destruction or apoptosis or expressed thritis Rheum. 39:1151. on the cell surface, where they could be presented to the im- 10. Okano, Y., V. D. Steen, and T. A. Medsger, Jr. 1993. Autoantibody reactive with mune system. Alternatively, the activated fibroblast itself might RNA polymerase III in systemic sclerosis. Ann. Intern. Med. 119:1005. 11. Oddis, C. V., Y. Okano, W. A. Rudert, M. Trucco, R. J. Duquesnoy, and serve as an APC. In the microarrays of SSc fibroblasts studied T. A. Medsger, Jr. 1992. Serum autoantibody to the nucleolar antigen PM-Scl. here, significantly increased expression of both class I and class Clinical and immunogenetic associations. Arthritis Rheum. 35:1211. II MHC genes was observed in some patients; however, it is 12. Gabrielli, A., M. Montroni, S. Rupoli, M. L. Caniglia, F. DeLustro, and G. Danieli. 1988. A retrospective study of antibodies against basement membrane unknown whether actual MHC molecules appear on the fibro- antigens (type IV collagen and laminin) in patients with primary and secondary blast cell surface in SSc and function normally. Raynaud’s phenomenon. Arthritis Rheum. 31:1432. The present results represent a first attempt at using multiple 13. Tsuneda, Y., Y. Kitajima, and S. Mori. 1988. Autoantibodies to vimentin-type intermediate filament in patients with progressive systemic sclerosis. J. Dermatol. gene expression assays to understand molecular pathogenesis in 15:128. SSc fibroblasts. It provides important preliminary evidence that the 14. Earnshaw, W., B. Bordwell, C. Marino, and N. Rothfield. 1986. Three human primary target tissue in SSc may play potential roles in initiation or chromosomal autoantigens are recognized by sera from patients with anti- centromere antibodies. J. Clin. Invest. 77:426. perpetuation of autoimmunity by the potential overexpression of

15. Muro, Y., T. Yamada, M. Himeno, and K. Sugimoto. 1998. cDNA cloning of a Downloaded from SSc-related autoantigen genes. Further studies of protein levels of novel autoantigen targeted by a minor subset of anti-centromere antibodies. Clin. these autoantigens will be necessary in the future. In addition, the Exp. Immunol. 111:372. 16. Kuwana, M., J. Kaburaki, Y. Okano, T. Tojo, and M. Homma. 1994. Clinical and possibility that the SSc-specific autoantibodies might target au- prognostic associations based on serum antinuclear antibodies in Japanese pa- toantigens expressed on the fibroblast cell surface or released into tients with systemic sclerosis. Arthritis Rheum. 37:75. the extracellular matrix should be explored. The endogenous 17. Diot, E., B. Giraudeau, P. Diot, D. Degenne, L. Ritz, J. L. Guilmot, and E. Lemarie. 1999. Is anti-topoisomerase I a serum marker of pulmonary involve- and/or exogenous stimuli leading to fibroblast activation and fi- ment in systemic sclerosis? Chest 116:715. http://www.jimmunol.org/ brosis are currently unknown. Moreover, similar studies of endo- 18. Harvey, G. R., S. Butts, A. L. Rands, Y. Patel, and N. J. McHugh. 1999. Clinical thelial cells in SSc, another major cell type involved in pathogen- and serological associations with anti-RNA polymerase antibodies in systemic esis, are needed and are in progress. sclerosis. Clin. Exp. Immunol. 117:395. 19. Satoh, M., M. Kuwana, T. Ogasawara, A. K. Ajmani, J. J. Langdon, D. Kimpel, As a hypothesis, the overexpression of disease-specific autoan- J. Wang, and W. H. Reeves. 1994. Association of autoantibodies to topoisom- tigens in the target tissues of other autoimmune diseases also erase I and the phosphorylated (IIO) form of RNA polymerase II in Japanese should be explored. In recent microarray studies of affected muscle scleroderma patients. J. Immunol. 153:5838. 20. Arnett, F. C., R. F. Howard, F. Tan, J. M. Moulds, W. B. Bias, E. Durban, biopsies in poly- and dermatomyositis patients, several were found H. D. Cameron, G. Paxton, T. J. Hodge, P. E. Weathers, et al. 1996. Increased to overexpress certain aminoacyl tRNA synthetase genes (28). Au- prevalence of systemic sclerosis in a Native American tribe in Oklahoma: asso- ciation with an Amerindian HLA haplotype. Arthritis Rheum. 39:1362.

toantibodies to tRNA synthetases are characteristic of these auto- by guest on October 2, 2021 21. Kuwana, M., J. Kaburaki, F. C. Arnett, R. F. Howard, T. A. Medsger, Jr., and immune muscle diseases (29). The application of genomic tech- T. M. Wright. 1999. Influence of ethnic background on clinical and serologic nology to complex and poorly understood diseases, such as SSc, features in patients with systemic sclerosis and anti-DNA topoisomerase I anti- by virtue of a more comprehensive view of gene expression and body. Arthritis Rheum. 42:465. 22. Tan, E. M. 1991. Autoantibodies in pathology and cell biology. Cell 67:841. molecular pathways may help unite seemingly disparate 23. Casciola-Rosen, L., F. Wigley, and A. Rosen. 1997. Scleroderma autoantigens observations. are uniquely fragmented by metal-catalyzed oxidation reactions: implications for pathogenesis. J. Exp. Med. 185:71. References 24. Fanning, G. C., K. I. Welsh, C. Bunn, R. Du Bois, and C. M. Black. 1998. HLA associations in three mutually exclusive autoantibody subgroups in UK systemic 1. Medsger, T. A., Jr. 2000. Systemic sclerosis (scleroderma): clinical aspects. In sclerosis patients. Br. J. Rheumatol. 37:201. Arthritis and Allied Conditions, 14th Ed. W. J. Koopman, ed. Lippincott Williams 25. Kuwana, M., H. Inoko, H. Kameda, T. Nojima, S. Sato, K. Nakamura, & Wilkins, Philadelphia, p. 1590. T. Ogasawara, M. Hirakata, Y. Ohosone, J. Kaburaki, et al. 1999. Association of 2. Arnett, F. C. 1995. HLA and autoimmunity in scleroderma (systemic sclerosis). human leukocyte antigen class II genes with autoantibody profiles, but not with Int. Rev. Immunol. 12:107. disease susceptibility in Japanese patients with systemic sclerosis. Intern. Med. 3. Bunn, C. C., and C. M. Black. 1999. Systemic sclerosis: an autoantibody mosaic. 38:336. Clin. Exp. Immunol. 117:207. 4. Kuwana, M., T. A. Medsger, Jr., and T. M. Wright. 1995. T cell proliferative 26. Mocci, S., K. Lafferty, and M. Howard. 2000. The role of autoantigens in auto- response induced by DNA topoisomerase I in patients with systemic sclerosis and immune disease. Curr. Opin. Immunol. 12:725. healthy donors. J. Clin. Invest. 96:586. 27. Rodenburg, R. J., J. M. Raats, G. J. Pruijn, and W. J. van Venrooij. 2000. Cell 5. LeRoy, E. C., C. Black, R. Fleischmajer, S. Jablonska, T. Krieg, T. A. Medsger, death: a trigger of autoimmunity? BioEssays 22:627. Jr., N. Rowell, and F. Wollheim. 1988. Scleroderma (systemic sclerosis): clas- 28. Zhou, X. D., F. K. Tan, M. Xiong, M. Dimachke, and F. C. Arnett. 2000. Gene sification, subsets and pathogenesis. J. Rheumatol. 15:202. expression profile of muscle biopsies from patients with inflammatory myopa- 6. Subcommittee for Scleroderma. 1980. Criteria of the American Rheumatism As- thies. Arthritis Rheum. 43:S275 (Abstr.). sociation Diagnostic and Therapeutic Criteria Committee: preliminary criteria for 29. Targoff, I. N. 2000. Update on myositis-specific and myositis-associated autoan- the classification of systemic sclerosis (scleroderma). Arthritis Rheum. 23:581. tibodies. Curr. Opin. Rheumatol. 12:475.