ARTHRITIS & RHEUMATISMVol. 56, No. 4, April 2007, pp 1286–1291DOI 10.1002/art.22444© 2007, American College of Rheumatology

Positive Association of SLC26A2 Gene Polymorphisms WithSusceptibility to Systemic-Onset Juvenile Idiopathic Arthritis

Rebecca Lamb,1 Wendy Thomson,1 the British Society of Paediatric and AdolescentRheumatology, Emma M. Ogilvie,2 and Rachelle Donn1

Objective. To investigate SLC26A2, the gene thatcauses diastrophic dysplasia, in juvenile idiopathic ar-thritis (JIA).

Methods. Nine polymorphisms across theSLC26A2 gene locus were investigated using MassArraygenotyping in 826 UK Caucasian JIA cases and 617ethnically matched healthy controls.

Results. Significant associations between multiplesingle-nucleotide polymorphisms (SNPs) acrossSLC26A2 and systemic-onset JIA were found. In eachcase, homozygosity for the minor allele conferred theincreased risk of disease susceptibility: rs1541915 (oddsratio [OR] 2.3, 95% confidence interval [95% CI] 1.4–3.7, P � 0.0003), rs245056 (OR 2.8, 95% CI 1.7–4.6, P �0.00002), rs245055 (OR 2.5, 95% CI 1.2–5.0, P � 0.004),rs245051 (OR 2.3, 95% CI 1.4–3.7, P � 0.0005),rs245076 (OR 2.7, 95% CI 1.3–5.4, P � 0.0015), andrs8073 (OR 2.3, 95% CI 0.9–5.6, P � 0.04).

Conclusion. These findings show the value ofusing monogenic disease loci as candidates for investi-gation in JIA. We identified a subgroup-specific associ-ation between SNPs within the SLC26A2 gene and

systemic-onset JIA. Our findings also highlightsystemic-onset JIA as being a distinctly different diseasefrom that in the other JIA subgroups.

Juvenile idiopathic arthritis (JIA) is the mostcommon chronic rheumatic disease of childhood and ischaracterized by persistent idiopathic inflammation of 1or more joints, with an onset before the age of 16 yearsand a duration of 6 weeks or more (1). JIA is clinicallyheterogeneous. Systemic-onset JIA accounts for 10–20% of JIA. Patients are often extremely unwell, withevidence of systemic inflammation, as manifested byfever, rash, serositis, and hyperferritinemia, in additionto arthritis, which may occur at presentation or later inthe disease course (1).

As part of a strategy of taking candidate genesfrom loci of monogenic syndromes with features incommon with JIA, we have previously shown WISP3, thegene that causes progressive pseudorheumatoid dyspla-sia, to confer susceptibility to polyarticular disease-course JIA (2). Continuing this approach in the presentstudy, we investigated the gene for diastrophic dysplasia(DTD), SLC26A2 (solute carrier family 26 [sulfate trans-porter] member 2; also called DTD sulfate transporter[DTDST]), in a well-defined cohort of UK JIA cases.

Mutations within the SLC26A2 gene (locus 5q32–q31) are known to cause a clinical spectrum of osteo-chondrodysplasias, with disease phenotypes rangingfrom severe/fatal disease to milder phenotypes (3). Thisrange of conditions includes achondrogenesis type IB, afatal skeletal dysplasia, atelosteogenesis type II (4),recessive multiple epiphyseal dysplasia (5,6), and themildest phenotype, DTD, an autosomal-recessive chon-drodysplasia (7). While many of the clinical aspects ofthe osteochondrodysplasias are evidently nonoverlap-ping with JIA, the associated early degenerative changesof joints occur in both (8). The osteochondrodysplasiasarise through loss-of-function mutations within the

Supported by the Epidemiology Unit Core Support programof the Arthritis Research Campaign, UK. Ms Lamb’s work wassupported by an MRC PhD studentship (G78/7554).

1Rebecca Lamb, PhD, Wendy Thomson, PhD, RachelleDonn, MB ChB, PhD: University of Manchester, Manchester, UK;2Emma M. Ogilvie, MMed: University College London, London, UK.

Contributors to the British Society of Paediatric and Adoles-cent Rheumatology are as follows: Dr. M. Abinun, Dr. M. Becker, Dr.A. Bell, Professor A. Craft, Dr. E. Crawley, Dr. J. David, Dr. H. Foster,Dr. J. Gardener-Medwin, Dr. J. Griffin, Dr. A. Hall, Dr. M. Hall, Dr.A. Herrick, Dr. P. Hollingworth, Dr. L. Holt, Dr. S. Jones, Dr. G.Pountain, Dr. C. Ryder, Professor T. Southwood, Dr. I. Stewart, Dr. H.Venning, Dr. L. Wedderburn, Professor P. Woo, and Dr. S. Wyatt.

Address correspondence and reprint requests to RachelleDonn, MB ChB, PhD, Arthritis Research Campaign EpidemiologyUnit and Centre for Molecular Medicine, University of Manchester,Stopford Building, Oxford Road, Manchester M13 9PT, UK. E-mail:Rachelle.Donn@manchester.ac.uk.

Submitted for publication September 12, 2006; accepted inrevised form December 6, 2006.

1286

SLC26A2 gene that result in absent or severely defectiveprotein synthesis. It was our hypothesis that mildergenetic changes (single-nucleotide polymorphisms[SNPs]) give rise to an altered, but less dramatic, quan-tity and/or quality of the SLC26A2 protein, and that thiscontributes to the oligogenetic basis of JIA.

PATIENTS AND METHODS

Patients and controls. Blood samples were obtainedfollowing informed consent. Ethics approval was obtainedfrom the Multi-centre Research Ethics Committee (MREC99/8/84) and the University of Manchester Committee on theEthics of Research on Human Beings (8/92/[i] [b]). DNAsamples from JIA patients were collected as part of the BritishSociety of Paediatric and Adolescent Rheumatology nationalrepository for JIA. All cases were classified according to theInternational League of Associations for Rheumatology clas-sification criteria (1).

DNA samples from 826 UK Caucasian JIA cases wereavailable for genotyping. The total number of JIA casesavailable for genotyping per subgroup was as follows: 133 withsystemic-onset JIA, 215 with persistent oligoarticular JIA, 127with extended oligoarticular JIA, 176 with rheumatoid factor(RF)–negative polyarticular JIA, 53 with RF-positive polyar-ticular JIA, 58 with enthesitis-related JIA, 57 with psoriaticJIA, and 7 with unclassifiable JIA. In addition, a controlpopulation consisting of 617 unrelated and ethnically matchedhealthy individuals was used. The controls were blood donorsor had been recruited via general practitioners’ registers.

Genotyping. Nine SNPs were genotyped across theSLC26A2 locus, giving an average spacing between SNPs of�300 bp (Figure 1). Genotyping was performed using high-throughput MassArray DNA analysis with matrix-assisted laser

desorption ionization–time-of-flight mass spectrometry (Se-quenom, San Diego, CA). Amplifications were conductedaccording to the manufacturer’s protocol. Polymerase chainreaction primers and probe sequences are available from theauthors on request.

Statistical analysis. Prior to analysis, the raw data were“cleaned” by the removal of data for DNA samples that failedto genotype for �20% of the SNPs assayed, as well as byremoval of data for SNPs that failed to genotype for �20% ofthe samples.

Hardy-Weinberg equilibrium was determined sepa-rately in the cases and controls with the use of Stata version 8software (Stata, College Station, TX). SNPs were removedfrom analysis if deviation from Hardy-Weinberg equilibrium(P � 0.05) was observed in the controls.

Single-point analysis. Association of the SLC26A2 SNPswas investigated using the chi-square test, and odds ratios(ORs) with 95% confidence intervals (95% CIs) were calcu-lated using Stata version 8 software. P values less than or equalto 0.05 were considered significant. Initially, genotype frequen-cies for each SNP were compared between the JIA cases (as awhole) and the controls. Then, subgroup analysis was per-formed in which each JIA subgroup was individually comparedwith the controls.

Multipoint analysis. Linkage disequilibrium was calcu-lated, and the association of varying-length haplotypes wastested using the expectation-maximization algorithm. This wasimplemented using HelixTree software (http://www.goldenhelix.com/index.jsp).

Correction for multiple testing. To limit Type II error, 2methods were used. First, permutation testing was used, andempirical P values were generated. This was instigated inClump (http://linkage.rockefeller.edu/soft/clump.html), aMonte Carlo–based method for assessing significance in case–control association studies. One hundred thousand simulationswere performed. Empirical P (Pe) values less than or equal to

Figure 1. Schematic representation of the SLC26A2 gene locus on chromosome 5q32–q33.1, showing exon locations andsingle-nucleotide polymorphisms (SNPs; denoted by their rs numbers). The length of the gene is 27 kb. SNP annotation wasbased on the National Center for Biotechnology Information dbSNP database refSNP and using genomic build 34.

SLC26A2 SNPs AND SUSCEPTIBILITY TO SYSTEMIC-ONSET JIA 1287

0.05 were considered significant. Second, the Bonferroni cor-rection was applied. Correction for the 8 SNPs analyzed andthe 7 subgroups of JIA investigated (n � 56) was used.Corrected P (Pcorr) values less than or equal to 0.05 wereconsidered significant.

RESULTS

Nine SNPs (rs1541915, rs245056, rs245055,rs245051, rs245076, rs3756307, rs30832, rs3776070, andrs8073) were genotyped across the SLC26A2 locus.SLC26A2 SNP rs3776070 deviated from Hardy-Weinberg equilibrium in both the JIA cases (P � 0.05)and the controls (P � 0.03). Additionally, the successrate for this SNP was �80% in both the JIA cases (76%)and the controls (79%), and this SNP was thereforeexcluded from further analysis.

Genotype frequencies of the remaining SNPswere compared between the JIA cases and the controls.No significant associations (P � 0.05) were observed forany of the SNPs (Table 1). To investigate the possibilitythat SLC26A2 SNPs confer susceptibility to specificsubgroups of JIA, genotype frequencies of SNPs incontrols were compared with those in each JIA subgroupindividually (Table 2).

Highly significant associations between 5 SNPsacross the SLC26A2 gene and systemic-onset JIA wereobserved: rs1541915 (P � 0.0015), rs245056 (P �0.0001), rs245055 (P � 0.0052), rs245051 (P � 0.0015),and rs245076 (P � 0.0018). When a recessive mode ofinheritance was applied to the data, where 2 copies ofthe minor allele confer susceptibility, the level of signif-

Table 1. Genotype frequencies in JIA cases versus controls for SLC26A2 SNPs*

SNP,genotype

JIA cases ControlsP for

genotypeassociation

Samplesize

P forHWE

Genotypefrequency, no. (%)

Samplesize

P forHWE

Genotypefrequency, no. (%)

rs1541915AA 664 0.63 215 (32.4) 563 0.73 195 (34.6)AC 332 (50.0) 277 (49.2)CC 117 (17.6) 91 (16.2) 0.64

rs245056AA 682 0.48 228 (33.4) 559 0.42 213 (38.1)AT 341 (50.0) 272 (48.7)TT 113 (16.6) 74 (13.2) 0.12

rs245055TT 724 0.57 396 (54.7) 585 0.91 342 (58.5)TC 275 (38) 212 (36.2)CC 53 (7.3) 31 (5.3) 0.21

rs245051TT 746 0.76 276 (37.0) 595 0.6 234 (39.3)TC 352 (47.2) 284 (47.8)CC 118 (15.8) 77 (12.9) 0.31

rs245076AA 648 0.19 355 (54.8) 557 0.81 332 (59.6)AG 240 (37.0) 195 (35.0)GG 53 (8.2) 30 (5.4) 0.08

rs3756307GG 697 1 509 (73.0) 574 0.48 428 (74.6)GA 174 (25.0) 138 (24.0)AA 14 (2.0) 8 (1.4) 0.64

rs30832GG 659 1 644 (97.7) 563 0.06 552 (98.0)GA 15 (2.3) 10 (1.8)AA 0 (0.0) 1 (0.2) 0.46

rs3776070AA 591 0.05 516 (87.3) 490 0.03 425 (86.8)AT 69 (11.7) 59 (12.0)TT 6 (1.0) 6 (1.2) –

rs8073AA 716 0.06 465 (64.9) 586 0.89 396 (67.6)AC 214 (29.9) 171 (29.2)CC 37 (5.2) 19 (3.2) 0.21

* The single-nucleotide polymorphisms (SNPs) are listed as they occur in the SLC26A2 gene, 5� to 3�. The sample size represents the number ofindividuals with a genotyping result for each SNP. JIA � juvenile idiopathic arthritis; HWE � Hardy-Weinberg equilibrium.

1288 LAMB ET AL

icance increased, conferring more than a 2-fold in-creased risk of disease: rs1541915 (odds ratio [OR] 2.3,95% confidence interval [95% CI] 1.4–3.7, P � 0.0003),rs245056 (OR 2.8, 95% CI 1.7–4.6, P � 0.00002),rs245055 (OR 2.5, 95% CI 1.2–5.0, P � 0.004), rs245051(OR 2.3, 95% CI 1.4–3.7, P � 0.0005), rs245076 (OR2.7, 95% CI 1.3–5.4, P � 0.0015), and rs8073 (OR 2.3,95% CI 0.9–5.6, P � 0.04).

A weak association-by-genotype of 2 of the SNPsand psoriatic JIA was also seen: rs1541915 (P � 0.03)and rs30832 (P � 0.05).

The positive associations with systemic-onset JIAwere maintained following permutation testing:rs1541915 (Pe � 0.00052), rs245056 (Pe � 0.00005),

rs245055 (Pe � 0.00588), rs245051 (Pe � 0.00069),rs245076 (Pe � 0.00296), and rs8073 (Pe � 0.044). Forpsoriatic JIA significant association was maintained withrs1541915: Pe � 0.037.

Furthermore, when the Bonferroni correction, ahighly conservative test which assumes independence ofall the parameters tested, was applied, the associationswith psoriatic JIA were lost. The associations withsystemic-onset JIA, however, remained significant for 3of the SNPs tested: rs1541915 (Pcorr � 0.0168), rs245056(Pcorr � 0.0011), and rs245051 (Pcorr � 0.028).

Significant linkage disequilibrium was observedacross the SLC26A2 SNPs examined (i.e., between thestudied SNPs located most 5� [rs1541915] and most 3�

Table 2. Genotype frequencies in JIA subgroups versus controls for SLC26A2 SNPs*

SNP,genotype

Systemic-onset JIApatients

PersistentoligoarticularJIA patients

ExtendedoligoarticularJIA patients

RF-negativepolyarticularJIA patients

RF-positivepolyarticularJIA patients

Enthesitis-related JIA

patientsPsoriatic JIA

patients No. (%)of

controlsNo. (%) P No. (%) P No. (%) P No. (%) P No. (%) P No. (%) P No. (%) P

rs1541915AA 30 (27.0) 65 (34.2) 30 (31.9) 45 (34.1) 16 (39.5) 15 (33.3) 13 (26.5) 195 (34.6)AC 47 (42.4) 95 (50.0) 43 (45.8) 69 (52.3) 16 (42.1) 27 (60.0) 33 (67.4) 277 (49.2)CC 34 (30.6) 0.0015† 30 (15.8) 0.98 21 (22.3) 0.34 18 (13.6) 0.72 7 (18.4) 0.61 3 (6.7) 0.18 3 (6.1) 0.03† 91 (16.2)

rs245056AA 28 (26.2) 69 (37.3) 31 (30.4) 53 (36.3) 12 (32.4) 18 (37.5) 16 (30.2) 213 (38.1)AT 47 (43.9) 93 (50.3) 50 (49.0) 73 (50.0) 18 (48.7) 24 (50.0) 34 (64.1) 272 (48.7)TT 32 (29.9) 0.0001† 23 (12.4) 0.92 21 (20.6) 0.1 20 (13.7) 0.92 7 (18.9) 0.57 6 (12.5) 0.98 3 (5.7) 0.08 74 (13.2)

rs245055TT 57 (47.1) 122 (61.3) 61 (55.4) 89 (60.1) 16 (43.2) 24 (47.1) 25 (46.3) 342 (58.5)TC 49 (40.5) 64 (32.2) 40 (36.4) 51 (34.5) 19 (51.4) 25 (49.0) 26 (48.1) 212 (36.2)CC 15 (12.4) 0.0052† 13 (6.5) 0.52 9 (8.2) 0.48 8 (5.4) 0.92 2 (5.4) 0.14 2 (3.9) 0.21 3 (5.6) 0.22 31 (5.3)

rs245051TT 38 (29.9) 78 (38.6) 38 (35.2) 69 (43.1) 15 (37.5) 20 (38.5) 17 (32.7) 234 (39.3)TC 57 (44.9) 94 (46.5) 49 (45.4) 74 (46.3) 21 (50.0) 25 (48.0) 30 (57.7) 284 (47.8)CC 32 (25.2) 0.0015† 30 (14.9) 0.79 21 (19.4) 0.19 17 (10.6) 0.59 5 (12.5) 0.91 7 (13.5) 0.99 5 (9.6) 0.41 77 (12.9)

rs245076AA 56 (47.1) 112 (60.5) 45 (51.7) 79 (58.9) 13 (44.8) 23 (53.5) 25 (53.2) 332 (59.6)AG 47 (39.5) 59 (31.9) 34 (39.1) 47 (35.1) 14 (48.3) 19 (44.2) 19 (40.4) 195 (35.0)GG 16 (13.4) 0.0018† 14 (7.6) 0.47 8 (9.2) 0.22 8 (6.0) 0.96 2 (6.9) 0.22 1 (2.3) 0.48 3 (6.4) 0.62 30 (5.4)

rs3756307GG 84 (71.8) 132 (69.1) 75 (72.8) 106 (73.6) 29 (82.9) 41 (80.4) 39 (75.0) 428 (74.6)GA 28 (23.9) 58 (30.4) 23 (22.3) 36 (25.0) 6 (17.1) 9 (17.6) 13 (25.0) 138 (24.0)AA 5 (4.3) 0.1281 1 (0.5) 0.17 5 (4.9) 0.06 2 (1.4) 0.95 0 (0.0) 0.64 1 (2.0) 0.41 0 (0.0) 0.93 8 (1.4)

rs30832GG 108 (98.2) 182 (98.9) 85 (98.8) 137 (97.9) 35 (97.2) 43 (95.6) 50 (92.6) 552 (98.0)GA 2 (1.8) 2 (1.1) 1 (1.2) 3 (2.1) 1 (2.8) 2 (4.4) 4 (7.4) 10 (1.8)AA 0 (0.0) 1 0 (0.0) 0.8 0 (0.0) 1 0 (0.0) 0.78 0 (0.0) 0.53 0 (0.0) 0.28 0 (0.0) 0.05† 1 (0.2)

rs8073AA 77 (62.1) 139 (70.9) 60 (56.1) 98 (67.1) 25 (65.8) 32 (62.7) 32 (64.0) 396 (67.6)AC 38 (30.6) 45 (23.0) 41 (38.3) 0.06 42 (28.8) 11 (28.9) 18 (35.3) 17 (34.0) 171 (29.2)CC 9 (7.3) 0.0944 12 (6.1) 0.07 6 (5.6) 6 (4.1) 0.87 2 (5.3) 0.69 1 (2.0) 0.68 1 (2.0) 0.77 19 (3.2)

* Persistent oligoarticular juvenile idiopathic arthritis (JIA) affects �4 joints; extended oligoarticular JIA affects �5 joints after 6 months.Rheumatoid factor (RF)–negative and RF-positive polyarticular JIA affects �5 joints within the first 6 months. All JIA patients were classifiedaccording to the criteria of the International League of Associations for Rheumatology (1). SNPs � single-nucleotide polymorphisms.† Statistically significant P value versus controls.

SLC26A2 SNPs AND SUSCEPTIBILITY TO SYSTEMIC-ONSET JIA 1289

[rs8073], with D� � 0.87 and r2 � 0.49). However, nohaplotype association of an order of magnitude greaterthan the single-point association was observed.

DISCUSSION

SLC26A2 is the gene that causes a family ofosteochondrodysplasias. All of these are characterizedby abnormal growth and remodeling of cartilage andbone (9). Because of the clinical overlap between theseosteochondrodysplasias and JIA, we hypothesized thatSNPs within the same gene, SLC26A2, and not thepreviously described mutations, may confer susceptibil-ity to JIA. Significant associations within the 5� region ofthe SLC26A2 gene and systemic-onset JIA were identi-fied, which remained after correction for multiple test-ing. The most significant association observed was withSNP rs245056 within the first intron of the SLC26A2gene, which confers an almost 3-fold increased risk ofdisease susceptibility, with an OR of 2.8 (95% CI1.7–4.6) at Pe � 0.00005 and Pcorr � 0.0011.

SLC26A2 encodes an anion exchanger that trans-ports extracellular sulfate into the cytoplasm for thesulfation of proteoglycans in the extracellular matrix ofcartilage (10). Within cartilage, the proteoglycan mole-cules form a highly hydrated, gel-like substance in whichthe fibrous proteins are embedded; this resists compres-sive forces on the matrix while allowing the rapiddiffusion of nutrients, metabolites, and hormones be-tween the blood and the tissue cells (11). Despite theexpression of SLC26A2 in numerous different tissues(12), mutations in this gene are only known to function-ally affect cartilage.

Our finding of significant associations with just 1subgroup of JIA, the systemic-onset JIA subgroup, wassomewhat unexpected. In the osteochondrodysplasias,some residual activity of the mutated SLC26A2 protein,combined with the use of alternative sources of sulfate,appears to be sufficient to supply the needs of cell typesother than chondrocytes. This may be how cartilage-specific defects arise (13,14). In systemic-onset JIA, acomplex genetic disease, alternative sources of sulfate orsulfate transporters may also be defective. The combi-nation of subtle changes in more than one gene involvedin sulfate transportation may contribute to the systemicmanifestations that are characteristic of systemic-onsetJIA. One way to determine whether multiple such locicontribute to systemic-onset JIA would be to examinedata generated from whole-genome studies. Recently,Thompson et al (15) performed a genome-wide scan ofsibling pairs with juvenile rheumatoid arthritis. How-

ever, no systemic-onset JIA–specific linkage peaks,which would be indicative of potential candidate loci,were identified due to the very limited patient samplesize (n � 11) studied (15).

Alternatively, our findings could suggest hetero-geneity in the pathogenesis of the cartilage involvementin JIA, with SLC26A2 contributing to the cartilagedamage in systemic-onset JIA only. Currently, there areno histopathologic or imaging studies to address thispoint.

While JIA is the most common chronic rheu-matic disease of childhood, it is a rare condition, affect-ing 1 in 10,000 children under the age of 16 years. Theclinical subgrouping means that the numbers of patientswith any 1 clinical presentation are relatively small forconducting association-based genetic studies. With thisin mind, we tried to overcome the potential problem thatthe observed positive associations between SLC26A2and systemic-onset JIA might be false-positive ones byusing both permutation testing and Bonferroni correc-tion. Additional support for the importance of theseobservations is required, and functional studies to ad-dress the consequence of the SNP polymorphisms withinrelevant cell types should now be conducted.

In summary, we studied a very large group of UKCaucasian JIA cases and ethnically matched healthycontrols and found significant associations betweenSLC26A2 and systemic-onset disease. These findingsreiterate the value of investigating loci that cause mo-nogenic syndromes as JIA susceptibility genes. Furtherstudies in different populations of systemic-onset JIAshould now be conducted in an attempt to replicate thefindings of SLC26A2 as a susceptibility gene forsystemic-onset JIA.

AUTHOR CONTRIBUTIONS

Drs. Lamb and Donn had full access to all of the data in thestudy and take responsibility for the integrity of the data and theaccuracy of the data analysis.Study design. Thomson, Donn.Acquisition of data. Lamb, Thomson, Ogilvie, Donn.Analysis and interpretation of data. Lamb, Donn.Manuscript preparation. Lamb, Thomson, Ogilvie, Donn.Statistical analysis. Lamb, Donn.

REFERENCES

1. Petty RE, Southwood TR, Baum J, Bhettay E, Glass DN, MannersP, et al. Revision of the proposed classification criteria for juvenileidiopathic arthritis: Durban, 1997. J Rheumatol 1998;25:1991–4.

2. Lamb R, Thomson W, Ogilvie E, Donn R. Wnt-1–induciblesignaling pathway protein 3 and susceptibility to juvenile idio-pathic arthritis. Arthritis Rheum 2005;52:3548–53.

1290 LAMB ET AL

3. Rossi A, Superti-Furga A. Mutations in the diastrophic dysplasiasulfate transporter (DTDST) gene (SLC26A2): 22 novel muta-tions, mutation review, associated skeletal phenotypes, and diag-nostic relevance. Hum Mutat 2001;17:159–71.

4. Hastbacka J, Superti-Furga A, Wilcox WR, Rimoin DL, CohnDH, Lander ES. Atelosteogenesis type II is caused by mutations inthe diastrophic dysplasia sulfate-transporter gene (DTDST): evi-dence for a phenotypic series involving three chondrodysplasias.Am J Hum Genet 1996;58:255–62.

5. Sheffield EG. Double-layered patella in multiple epiphyseal dys-plasia: a valuable clue in the diagnosis. J Pediatr Orthop 1998;18:123–8.

6. Ballhausen D, Bonafe L, Terhal P, Unger SL, Bellus G, Classen M,et al. Recessive multiple epiphyseal dysplasia (rMED): phenotypedelineation in eighteen homozygotes for DTDST mutationR279W. J Med Genet 2003;40:65–71.

7. Walker BA, Scott CI, Hall JG, Murdoch JL, McKusick VA.Diastrophic dwarfism. Medicine (Baltimore) 1972;51:41–59.

8. Petty RE. Classification of childhood arthritis: a work in progress.Baillieres Clin Rheumatol 1998;12:181–90.

9. Rimoin DL. The chondrodystrophies. Adv Hum Genet 1975;5:1–118.

10. Habuchi O. Diversity and functions of glycosaminoglycan sulfo-transferases. Biochim Biophys Acta 2000;1474:115–27.

11. Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P. Celljunctions, cell adhesion, and the extracellular matrix. In: Molecu-lar biology of the cell. New York: Garland Science; 2002. p.949–1010.

12. Haila S, Hastbacka J, Bohling T, Karjalainen-Lindsberg ML, KereJ, Saarialho-Kere U. SLC26A2 (diastrophic dysplasia sulfate trans-porter) is expressed in developing and mature cartilage but also inother tissues and cell types. J Histochem Cytochem 2001;49:973–82.

13. Superti-Furga A, Rossi A, Steinmann B, Gitzelmann R. A chon-drodysplasia family produced by mutations in the diastrophicdysplasia sulfate transporter gene: genotype/phenotype correla-tions. Am J Med Genet 1996;63:144–7.

14. Esko JD, Elgavish A, Prasthofer T, Taylor WH, Weinke JL.Sulfate transport-deficient mutants of Chinese hamster ovary cells:sulfation of glycosaminoglycans dependent on cysteine. J BiolChem 1986;261:15725–33.

15. Thompson SD, Moroldo MB, Guyer L, Ryan M, Tombragel EM,Shear ES, et al. A genome-wide scan for juvenile rheumatoidarthritis in affected sibpair families provides evidence of linkage.Arthritis Rheum 2004;50:2920–30.

SLC26A2 SNPs AND SUSCEPTIBILITY TO SYSTEMIC-ONSET JIA 1291