ARTHRITIS & RHEUMATISMVol. 52, No. 3, March 2005, pp 752–758DOI 10.1002/art.20877© 2005, American College of Rheumatology

Investigation of the SLC22A4 Gene (Associated WithRheumatoid Arthritis in a Japanese Population) in a

United Kingdom Population of Rheumatoid Arthritis Patients

Anne Barton, Stephen Eyre, John Bowes, Pauline Ho, Sally John, and Jane Worthington

Objective. Recent studies of 2 complex diseases,rheumatoid arthritis (RA) and Crohn’s disease (CD),showed associations with genes mapping to the cytokinegene cluster on 5q31. In particular, a functional single-nucleotide polymorphism (SNP) mapping to intron 1 ofthe organic cation transporter 1 (OCTN1; SLC22A4)gene was associated with RA in a Japanese population,and a haplotype of a different SNP in the same gene andone in an adjacent gene, OCTN2 (SLC22A5), was asso-ciated with CD. The purpose of this study was toinvestigate the association between the OCTN locus andRA in a Caucasian population.

Methods. Association with 11 SNPs spanning theSLC22A4 and SLC22A5 genes, including a putativeRA-causing functional polymorphism (rs3792876[slc2f2]) and a functional haplotype previously associ-ated with CD, was investigated in 909 RA cases and 594population controls in the UK. Genotyping was per-formed using 5�-allele discrimination assays. Estimatedhaplotype frequencies were generated using theexpectation-maximization algorithm and were com-pared between cases and controls.

Results. All SNPs were in Hardy-Weinberg equili-brium. We found no evidence for an association betweenRA and either the SNP (rs3792876 [slc2f2]) or thehaplotype previously reported to be associated with RAin a Japanese population. Similarly, no association

between RA and the haplotype associated with CD wasdetected.

Conclusion. Functional polymorphisms of theOCTN gene locus that have previously been associatedwith RA and CD were not found to be associated withRA in a UK population. The findings do not providesupport for a major role of these genes in the etiology ofRA in this population.

Rheumatoid arthritis (RA) is a chronic diseasecharacterized by inflammation of synovial joints, whichaffects up to 1% of the Caucasian population. Linkageand association with the major genetic susceptibilitylocus, HLA, is well established, but HLA is thought tocontribute less than one-third of the total genetic sus-ceptibility component (for review, see ref. 1). Identifyingnon-HLA susceptibility genes remains a challenge, but itwas recently reported that a novel RA susceptibilitygene, solute carrier 22 member 4 (SLC22A4;OMIM*604190), mapping to chromosome 5q31, hadbeen identified in a Japanese population (2). The 5q31region was targeted for investigation because it hadpreviously shown linkage in family studies of Crohn’sdisease (CD), atopic dermatitis, and bronchial asthma,and the region harbors a cytokine gene cluster.

In the Japanese study, single-marker and haplo-type analysis showed convincing association with a single-nucleotide polymorphism (SNP) mapping to intron 2 ofthe SLC22A4 gene, but the SNP was not thought to playa functional role (2). Studies of an intron 1 SNP inlinkage disequilibrium (LD) with it showed that thepolymorphism affected binding of a transcription regu-lator, RUNX-1. This was, therefore, thought to be thedisease-causing variant, particularly in light of severalother recent publications showing that polymorphismsaffecting RUNX-1 binding sites may be associated witha variety of autoimmune diseases (3,4).

Interestingly, a recent study found an association

Supported by the Arthritis Research Campaign. Dr. Barton isrecipient of a Wellcome Advanced Fellowship. Dr. Ho is recipient ofa Medical Research Council Clinical Training Fellowship. Dr. John’swork was supported by the Medical Research Council.

Anne Barton, MRCP, PhD, Stephen Eyre, MSc, John Bowes,BSc, Pauline Ho, MRCP, Sally John, PhD, Jane Worthington, PhD:University of Manchester, Manchester, UK.

Address correspondence and reprint requests to Anne Bar-ton, MRCP, PhD, ARC-EU, Stopford Building, University ofManchester, Manchester M13 9PT, UK. E-mail: ABarton@fs1.ser.man.ac.uk.

Submitted for publication July 27, 2004; accepted in revisedform December 2, 2004.

752

between CD and a haplotype of 2 SNPs mapping to thesame genetic region (5). In that case, the haplotypeinvolved a SNP in the SLC22A4 gene (also known asorganic cation transporter 1 [OCTN1]) mapping to exon9, resulting in a nonconservative amino acid substitution,and a promoter polymorphism mapping to the adjacentSLC22A5 (also known as OCTN2) gene (OMIM*603377).The haplotype was shown to affect the transcription andtransporter functions of the genes.

The aim of the current study was to investigateassociations between SNPs mapping to the SLC22A4and A5 genes and RA in a UK population.

PATIENTS AND METHODS

Study design. A case–control (association) study wasundertaken to investigate associations between SNPs spanningthe SLC22A4 and SLC22A5 genes and RA. Unrelated patientswith RA (RA cases) were compared with healthy controlsubjects for their genotype at each locus. Pairwise LD wasmeasured between SNPs, and haplotypes were estimated usingthe expectation-maximization (EM) algorithm implemented inHelixTree (Golden Helix, Bozeman, MT).

RA cases. Cases with RA were obtained from theArthritis Research Campaign National Repository of patientsand families and from local clinics. For patients obtainedthrough the National Repository, 1 affected case per familywas selected for investigation. All RA patients satisfied theAmerican College of Rheumatology (formerly, the AmericanRheumatism Association) 1987 criteria for RA (6), as modifiedfor genetic studies of prevalent cases (7).

Control subjects with no history of inflammatory ar-thritis were recruited from blood donors and from generalpractice registers. All patients and controls were of UK

Caucasoid ethnic origin. All subjects were recruited withEthical Committee approval and provided informed consent.

Genotyping methods. Eleven SNPs spanning theSLC22A4 and SLC22A5 genes were selected for investigation(Figure 1). The SNPs were chosen as follows: first, to includethe SNP thought to cause RA in the Japanese population(rs3792876 [slc2f2]); second, to include the SNPs forming ahaplotype associated with RA in the Japanese population(rs3763112 [slc2–E1], rs1007602 [slc2–1], rs2073838 [slc2–F1],and rs2269822 [slc2–3]); third, to span the SLC22A4 gene withhigh density (rs3805673, rs272887, rs2304081, and rs2306772);and last, to include SNPs forming the haplotype associatedwith CD (rs1050152 [SLC22A4*L503F] and rs2631367[SLC22A5*G–207C]). All SNPs were genotyped using a Taq-Man 5�-allele discrimination assay (Applied Biosystems, War-rington, UK) according to the manufacturer’s instructions,except that a 5-�l, rather than a 25-�l, reaction volume wasused as described previously (8). Duplicate samples and neg-ative controls were included to ensure accuracy of genotyping.

Statistical analysis. Single-point analysis. Associationof RA with each of the 11 SNPs was tested using the chi-squaretest implemented in the Stata software (Stata, College Station,TX). The clinical phenotype of RA is heterogeneous, withvariation in the presence of such features as erosions andrheumatoid factor (RF), as well as age at onset. In addition,there is genotypic heterogeneity in terms of the carriage ofHLA–DR4 susceptibility alleles. Hence, stratification analysiswas undertaken to investigate whether associations werepresent in specific patient subsets based on sex, disease sever-ity, age at onset, and carriage of shared-epitope alleles.

Haplotype analysis. Pairwise LD measures (both the D�value and the correlation [r]) between individual SNPs wereinvestigated, and haplotypes were constructed using the EMalgorithm implemented in HelixTree software. Haplotype fre-quencies were compared between cases and controls using thechi-square test implemented in Stata software.

Figure 1. Diagrammatic representation of the region of chromosome 5q31 investigated, showing geneand single-nucleotide polymorphism marker positions. PDLIM4 � PDZ and LIM domain protein 4(OMIM*603422). Positional data obtained from National Center for Biotechnology Information, Build 34(online at www.ncbi.nlm.nih.gov).

NO ASSOCIATION OF SLC22A4 WITH RA IN A UK POPULATION 753

Power. Sample sizes were calculated based on pub-lished allele frequencies (minor allele frequencies of 28–49%)so that each of the association studies had at least 80% powerto detect a gene conferring the same odds ratio of 1.5 asdetected in the Japanese study for associated SNPs (oddsratios of 1.5–2.0) at the 5% significance level assuming adominant inheritance model (2,5).

Multiple testing. Association of RA with 11 SNPs and 1haplotype were investigated in this study. The data werestratified as described above to fully explore whether theSLC22A4/A5 gene locus is associated with RA. Although it isdifficult to know the appropriate correction factor that shouldbe applied (because the SNPs are tightly linked, but theBonferroni test assumes independence of loci), we used a

Table 1. Comparison of genotype frequencies for SLC22A4/A5 SNPs in RA cases and controls*

SNP Genotype

No. (%) of allRA cases(n � 909)

No. (%) ofall controls(n � 594)

No. (%) of RA cases

WithRF

Witherosion

With onset ageabove the median

With onset agebelow the median

rs3763112 (slc2–E1) G/G 173 (20.3) 108 (19.0) 125 (19.5) 114 (20.2) 92 (19.0) 81 (22.1)A/G 433 (50.9) 283 (49.6) 334 (52.0) 282 (50.0) 253 (52.2) 180 (49.2)A/A 245 (28.8) 179 (31.4) 183 (28.5) 168 (29.8) 140 (28.8) 105 (28.7)

P 0.55 0.54 0.79 0.65 0.44rs1007602 (slc2–1) C/C 92 (11.1) 62 (11.0) 72 60 45 47

T/C 360 (43.4) 248 (44.0) 270 235 206 154T/T 378 (45.5) 254 (45.0) 285 258 221 157

P 0.98 0.94 0.85 0.70 0.62rs3792876† (slc2f2) T/T 2 (0.2) 6 (1.0) 1 1 0 2

T/C 112 (12.7) 87 (14.7) 88 75 69 43C/C 766 (87.1) 501 (84.3) 578 506 432 334

P 0.07 0.09 0.11 0.06 0.23rs2073838 (slc2–F1) A/A 3 (0.3) 6 (1.0) 1 2 1 2

A/G 110 (12.1) 86 (14.8) 87 74 68 42G/G 796 (87.6) 490 (84.2) 596 530 451 345

P 0.08 0.06 0.15 0.16 0.14rs3805673 A/A 3 (0.3) 4 (0.7) 1 2 1 2

A/G 97 (11.1) 72 (12.5) 75 63 58 39G/G 777 (88.6) 500 (86.8) 580 519 436 341

P 0.42 0.29 0.45 0.50 0.54rs272887 A/A 75 (8.6) 48 (8.2) 61 48 37 38

A/G 356 (40.7) 241 (41.3) 270 231 199 157G/G 444 (50.7) 294 (50.5) 331 294 260 184

P 0.96 0.84 0.94 0.78 0.60rs2304081 T/T 2 (0.2) 6 (1.0) 1 1 0 2

T/C 110 (12.4) 84 (14.4) 87 74 67 43C/C 774 (87.4) 493 (84.6) 578 513 438 336

P 0.07 0.09 0.03‡ 0.06 0.26rs2306772 A/A 3 (0.3) 6 (1.1) 2 2 1 2

A/G 108 (12.4) 82 (14.5) 85 73 66 42G/G 764 (87.3) 477 (84.4) 567 512 433 331

P 0.11 0.20 0.19 0.18 0.22rs1050152§

(SLC22A4*L503F)T/T 156 (19.3) 112 (20.5) 120 104 95 61T/C 383 (47.3) 260 (47.6) 292 254 217 166C/C 271 (33.4) 174 (31.9) 198 183 148 123

P 0.77 0.93 0.75 0.99 0.42rs2631367§

(SLC22A5*G–207C)G/G 198 (23.4) 124 (23.0) 146 130 120 78G/C 409 (48.3) 256 (47.4) 324 283 226 183C/C 240 (28.3) 160 (29.6) 178 160 136 104

P 0.87 0.63 0.77 0.75 0.72rs2269822 (slc2–3) T/T 9 (1.1) 10 (1.9) 5 5 5 4

T/C 181 (22.8) 116 (22.3) 137 126 100 81C/C 605 (76.1) 395 (75.8) 460 400 342 263

P 0.50 0.30 0.37 0.66 0.69

* Rheumatoid arthritis (RA) cases were stratified by the presence of rheumatoid factor (RF) and erosions, and by age at onset above or below themedian. SNPs � single-nucleotide polymorphisms.† SNPs thought to cause RA in a Japanese population.‡ Corrected P � 0.33.§ SNPs forming functional haplotype associated with Crohn’s disease in a Canadian population.

754 BARTON ET AL

Bonferroni correction factor of 11. Uncorrected P values arepresented, and for those showing statistical significance, thecorrected P values are also given.

RESULTS

DNA was available for 909 RA cases and 594population controls. In the RA case cohort, 656 (72.2%)

patients were female, 607 (67.8%) had erosions, 689(75.8%) were seropositive for RF, and the median age atarthritis onset was 43.5 years (interquartile range 32–55years). Carriage of shared-epitope alleles was as follows:181 (19.9%) patients carried 0 copies, 425 (46.8%)carried 1 copy, and 284 (31.2%) carried 2 copies.

Among the controls, information about sex was

Table 2. Comparison of genotype frequencies for SLC22A4/A5 SNPs in RA cases and controls stratified by sex and carriage of SE alleles*

SNP Genotype

No. (%) female No. (%) maleNo. (%) with 0 copies

of SE allelesNo. (%) with 1/2 copies

of SE alleles

Cases Controls Cases Controls Cases Controls Cases Controls

rs3763112 (slc2–E1) G/G 54 (20.3) 116 (19.9) 51 (22.7) 36 (18.9) 36 (20.7) 70 (20.3) 137 (20.2) 38 (16.9)A/G 132 (49.6) 314 (51.2) 114 (50.7) 91 (47.9) 86 (49.4) 170 (49.3) 347 (51.3) 113 (50.2)A/A 80 (30.1) 183 (29.9) 60 (26.6) 63 (33.2) 52 (29.9) 105 (30.4) 193 (28.5) 74 (32.9)

P 0.87 0.32 0.99 0.36rs1007602 (slc2–1) C/C 63 (10.6) 31 (11.6) 27 (12.1) 18 (9.6) 20 (11.8) 39 (11.3) 72 (10.9) 23 (10.5)

T/C 252 (42.5) 120 (44.9) 102 (45.5) 78 (41.7) 69 (40.8) 154 (44.8) 291 (44.0) 94 (42.7)T/T 278 (46.9) 116 (43.5) 95 (42.4) 91 (48.7) 80 (47.4) 151 (43.9) 298 (45.1) 103 (46.8)

P 0.64 0.42 0.69 0.91rs3792876† (slc2f2) T/T 2 (0.3) 3 (1.1) 0 (0.0) 2 (1.0) 1 (0.6) 4 (1.1) 1 (0.1) 2 (0.9)

T/C 89 (14.0) 46 (16.5) 22 (9.5) 25 (12.6) 27 (15.2) 54 (15.0) 85 (12.1) 33 (14.1)C/C 544 (85.7) 230 (82.4) 210 (90.5) 171 (86.4) 150 (84.2) 302 (83.9) 616 (87.8) 199 (85.0)

P 0.17 0.17 1.0 0.14rs2073838 (slc2–F1) A/A 2 (0.3) 3 (1.1) 1 (0.4) 2 (1.0) 1 (0.6) 4 (1.1) 2 (0.3) 2 (0.9)

A/G 87 (13.3) 45 (16.7) 22 (9.1) 25 (12.8) 27 (15.1) 53 (15.1) 83 (11.4) 33 (14.3)G/G 565 (86.4) 221 (82.2) 219 (90.5) 168 (86.2) 151 (84.3) 295 (83.8) 645 (88.3) 195 (84.8)

P 0.10 0.34 0.97 0.16rs3805673 A/A 2 (0.3) 2 (0.7) 1 (0.4) 1 (0.5) 1 (0.6) 3 (0.9) 2 (0.3) 1 (0.4)

A/G 77 (12.2) 35 (12.9) 19 (8.1) 21 (11.2) 26 (14.8) 44 (12.6) 71 (10.1) 28 (12.3)G/G 550 (87.5) 235 (86.4) 216 (91.5) 166 (88.3) 149 (84.6) 302 (86.5) 628 (89.6) 198 (87.3)

P 0.68 0.54 0.80 0.45rs272887 A/A 47 (7.5) 22 (8.1) 26 (11.2) 15 (7.7) 18 (10.2) 30 (8.4) 57 (8.2) 18 (7.9)

A/G 258 (41.0) 110 (40.4) 92 (39.5) 77 (39.7) 75 (42.4) 146 (41.0) 281 (40.3) 95 (41.9)G/G 325 (51.5) 140 (51.5) 115 (49.3) 102 (52.6) 84 (47.4) 180 (50.6) 360 (51.5) 114 (50.2)

P 0.94 0.48 0.70 0.91rs2304081 T/T 2 (0.3) 3 (1.1) 0 (0.0) 2 (1.0) 1 (0.6) 4 (1.1) 1 (0.2) 2 (0.9)

T/C 87 (13.7) 45 (16.5) 22 (9.3) 23 (11.9) 27 (15.1) 52 (14.7) 83 (11.7) 32 (14.0)C/C 548 (86.0) 224 (82.4) 214 (90.7) 168 (87.1) 151 (84.3) 298 (84.2) 623 (88.1) 195 (85.1)

P 0.13 0.21 0.93 0.12rs2306772 A/A 3 (0.5) 3 (1.1) 0 (0.0) 2 (1.1) 1 (0.6) 4 (1.2) 2 (0.3) 2 (0.9)

A/G 85 (13.4) 43 (16.2) 22 (9.5) 23 (0.13) 27 (15.1) 51 (14.9) 81 (11.6) 31 (13.9)G/G 544 (86.1) 220 (82.7) 209 (90.5) 159 (86.4) 151 (84.3) 287 (83.9) 613 (88.1) 190 (85.2)

P 0.25 0.93 0.22rs1050152‡

(SLC22A4*L503F)T/T 117 (20.0) 47 (18.4) 38 (17.8) 39 (21.4) 33 (20.0) 67 (20.0) 123 (19.1) 45 (21.3)T/C 280 (47.9) 124 (48.6) 99 (46.5) 88 (48.4) 69 (41.8) 163 (48.7) 314 (48.7) 97 (46.0)C/C 188 (32.1) 84 (33.0) 76 (35.7) 55 (30.2) 63 (38.2) 105 (31.3) 208 (32.2) 69 (32.7)

P 0.88 0.45 0.27 0.71rs2631367‡

(SLC22A5*G–207C)G/G 140 (23.1) 52 (20.5) 56 (24.7) 44 (24.9) 36 (21.7) 76 (23.3) 162 (23.8) 48 (22.4)G/C 298 (49.1) 124 (48.8) 107 (47.1) 79 (44.6) 75 (45.2) 153 (46.9) 334 (49.0) 103 (48.1)C/C 169 (27.8) 78 (30.7) 64 (28.2) 54 (30.5) 55 (33.1) 97 (29.8) 185 (27.2) 63 (29.4)

P 0.59 0.85 0.73 0.80rs2269822 (slc2–3) T/T 7 (1.2) 7 (2.9) 2 (1.0) 2 (1.2) 2 (1.2) 8 (2.5) 7 (1.1) 2 (1.0)

T/C 134 (23.1) 62 (25.6) 45 (22.0) 35 (20.2) 42 (26.1) 64 (20.2) 139 (21.9) 52 (25.5)C/C 438 (75.7) 173 (71.5) 158 (77.0) 136 (78.6) 117 (72.7) 245 (77.3) 488 (77.0) 150 (73.5)

P 0.15 0.89 0.26 0.53

* SNPs � single-nucleotide polymorphisms; RA � rheumatoid arthritis; SE � shared epitope.† SNP thought to cause RA in a Japanese population.‡ SNPs forming functional haplotype associated with Crohn’s disease in a Canadian population.

NO ASSOCIATION OF SLC22A4 WITH RA IN A UK POPULATION 755

available for 470 subjects, 273 (58.1%) of whom werefemale. Of the 570 controls with HLA–DRB1 genotypeinformation available, 359 (63%) carried 0 copies ofshared-epitope alleles, 168 (29.5%) carried 1, and 43(7.5%) carried 2.

Genotype frequencies for all SNPs were inHardy-Weinberg equilibrium both in the cases and inthe controls. The genotype error rate was �0.05%, asassessed by concordance of duplicate samples acrossdifferent plates. Allele frequencies determined in theUK population differed from those reported previouslyin the Japanese population. For example, the slc2–F1SNP (rs2073838), which was most strongly associatedwith RA in the Japanese population, was reported ashaving a control minor allele frequency of 31% (2)compared with 8.4% in the current study in a UKpopulation.

Findings of single-point analysis. No significantassociation between RA and any of the SNPs tested wasdetected in this UK population. A trend toward anassociation was detected with the SNP thought to causeRA in the Japanese population (rs3792876) and with allSNPs that showed high levels of correlation with it(rs2073838, rs2304081, and rs2306772) (Table 1) (2).However, this trend did not remain after correction formultiple testing. Furthermore, the direction of the asso-ciation was in the direction opposite to that reported inthe Japanese study with the rare slc2f2 (rs3792876)*Tallele being less common in patients than in controls in

the UK cohort (2). Stratification by the presence of RFand erosions, by sex, and by carriage of shared-epitopealleles revealed no evidence of association (Tables 1 and2).

Findings of haplotype analysis. A plot of pairwiseLD measures between the 11 SNPs spanning theSLC22A4/A5 locus is shown in Figure 2. There wasstrong LD across the region, with D� measurements of�0.9. Haplotype analysis of 5 SNPs common to bothstudies and defining the haplotype associated with RA inthe Japanese study (rs3763112, rs1007602, rs3792876,rs2073838, and rs2269822) revealed that 5 haplotypesexisted at a frequency of �1% in the UK population.Together, these haplotypes captured �98% of the vari-ation. No difference in the frequency of these haplotypeswas observed between cases and controls (Table 3). Thesusceptibility haplotype in the Japanese study is charac-terized by G, A, and T at positions slc2–E1 (rs3763112),slc2-F1 (rs2073838), and slc2–3 (rs2269822), respectively(2). This haplotype was extremely rare in our population(Table 3). No difference was seen in the frequency of thehaplotype associated with CD in this cohort (P � 0.11)(data not shown) (5).

DISCUSSION

A recent Japanese study demonstrated an asso-ciation between RA and a functional polymorphismmapping to the SLC22A4 gene (2). Interestingly, afunctional haplotype at the same locus has also recentlybeen associated with Crohn’s disease (5). We sought toreplicate these findings, but we found no evidence ofassociation between either the SLC22A4 gene or theSLC22A4–SLC22A5 CD-associated haplotype with RAin our UK population.

The failure to replicate the finding of a geneassociation in different populations of RA patients could

Figure 2. Graphic representation of D� (top left) and correlation(bottom right) values for 11 single-nucleotide polymorphisms (SNPs)investigated across the SLC22A4/A5 locus. SNPs are represented inthe correct order across the locus.

Table 3. Haplotypes formed by the 5 SNPs rs3763112 (slc2–E1),rs1007602 (slc2–1), rs3792876 (slc2f2), rs2073838 (slc2–F1), andrs2269822 (slc2–3)*

HaplotypeNo. (%) of RA cases

(n � 909)No. (%) of controls

(n � 594)

A, T, C, G, C 784 (43.5) 450 (45.0)†G, C, C, G, C 584 (32.4) 324 (32.5)G, T, C, G, C 216 (12.0) 98 (9.8)A, C, T, A, T 112 (6.2) 76 (7.6)A, T, C, G, T 80 (4.4) 36 (3.6)

* SNPs � single-nucleotide polymorphisms; RA � rheumatoid arthri-tis.† P � 0.20.

756 BARTON ET AL

be due to a number of factors. First, it could have arisendue to a Type II error (false-negative). There weredifferent SNP allele frequencies between the 2 studypopulations, but this is unlikely to have arisen as a resultof genotyping error. In our study, duplicate sampleswere included, and the genotyping error rate was foundto be low (�0.05%); the genotype frequencies did notdiffer from Hardy-Weinberg expectations; and the allelefrequencies closely reflected the frequencies previouslyreported in Caucasian populations (online at www.hapmap.org) (5). Despite the lower allele frequency ofthe RA-associated SNP in the UK population than inthe Japanese population (8.4% and 31% minor allelefrequency for the slc2–F1 [rs2073838] SNP, respec-tively), the sample size used in the current study wasnevertheless sufficiently large that we had 80% power todetect a genotypic relative risk of 1.5 at the 5% signifi-cance level, which is similar to that reported in theJapanese study (2).

A second possibility is that LD patterns maydiffer in the Japanese and UK populations and that theassociated SNPs reported in the Japanese study may beacting as markers for another gene in the region. Itshould be noted that in the Japanese study, a commonhaplotype (haplotype 1) was found at increased fre-quency in RA cases compared with controls (23% versus18%, respectively). Haplotype 1 was distinguished fromthe other haplotypes by polymorphism at 4 sites: slc2–E1(rs3763112), slc2–1 (rs1007602), slc2–F1 (rs2073838),and slc2–3 (rs2269822). These SNPs spanned 4 genes,but the data were interpreted as providing support onlyfor the slc2–F1 (rs2073838) polymorphism, whichmapped to intron 2 of the SLC22A4 gene. Association tothe other genes, however, was not excluded, and it maybe that another gene in the region is actually thedisease-causing gene. We did investigate the associationof RA with all the SNPs defining the same haplotype,but we found no evidence for association in our UK RApatients. It is noteworthy that the Japanese susceptibilityhaplotype was found at very low frequency in our UKpopulation.

A third explanation for the discrepant findingsmay be that the gene only plays a role in susceptibility toRA in the presence of environmental factors to whichthe Japanese population, but not the UK population, isexposed. Differences in diet and other lifestyle factorsare known to exist between the 2 ethnic groups, but wedo not have a detailed history of environmental expo-sures in the current cohort that would allow explorationof possible gene–environment interactions.

Interestingly, the proposed functional SNP

(slc2f2 [rs3792876]) does not appear to have been di-rectly tested for association with RA in the Japanesestudy. It was 1 of 9 additional SNPs that mapped tointron 1 of the SLC22A4 gene and which were in tightLD with the slc2–F1 (rs2073838) SNP. The slc2f2(rs3792876) SNP was selected as the most likely func-tional mutation because polymorphism was predicted toaffect binding of a transcription factor, RUNX-1. Sev-eral recent reports have suggested that RUNX-1binding-site polymorphisms may be associated with anumber of autoimmune diseases (3,4). There is no doubtthat the functional data provide support that the puta-tive RUNX-1 binding site does indeed affect RUNX-1binding. Since the SLC22A4 gene product was found tobe increased in RA joints and lymph nodes, it might bepredicted that the RA-associated slc2f2 (rs3792876)*Tallele should increase RUNX-1 binding and gene tran-scription. Surprisingly, the opposite effect was found inthe Japanese study, since the slc2f2 (rs3792876)*T allelereduced RUNX-1 binding, resulting in decreased tran-scription in the cell types studied. In contrast, in ourstudy, the T allele was underrepresented in RA casescompared with controls, although this did not achievestatistical significance (uncorrected P � 0.07). A numberof other SNPs also showed a trend toward an associationwith RA. This is because they were in LD with, hadsimilar allele frequencies to, and showed high correla-tion values with, the slc2f2 (rs3792876) SNP.

A haplotype of 2 SNPs mapping to the samechromosome 5q31 locus has recently been shown to beassociated with another chronic inflammatory disease,CD (5). Given that both RA and CD respond toanti–tumor necrosis factor � therapy, we speculated thatcommon inflammatory pathways exist and that theremay be some overlap in the genetic control of thesepathways. However, we found no evidence of an associ-ation between the CD haplotype and RA. Althoughpatients with CD are at increased risk of developing aninflammatory arthritis, this is much more likely to beseronegative for RF, and there is no evidence that theincidence of RA is increased in these patients (9).Similarly, polymorphisms of the CARD15 gene, whichhave been shown to be disease-causing in CD, also showno association in RA populations (10,11).

In summary, in this study of a UK population, wefound no evidence of an association between RA andeither a potential functional disease-causing SLC22A4SNP previously reported to be associated with RA in aJapanese population or a functional haplotype of SNPsmapping to SLC22A4 and SLC22A5 previously found tobe with CD. The findings do not provide support for a

NO ASSOCIATION OF SLC22A4 WITH RA IN A UK POPULATION 757

major role of these genes in the etiology of RA in ourpopulation.

REFERENCES

1. Barton A, Ollier W. Genetic approaches to the investigation ofrheumatoid arthritis. Curr Opin Rheumatol 2002;14:260–9.

2. Tokuhiro S, Yamada R, Chang X, Suzuki A, Kochi Y, Sawada T,et al. An intronic SNP in a RUNX1 binding site of SLC22A4,encoding an organic cation transporter, is associated with rheu-matoid arthritis. Nat Genet 2003;35:341–8.

3. Helms C, Cao L, Krueger JG, Wijsman EM, Chamian F, GordonD, et al. A putative RUNX1 binding site variant betweenSLC9A3R1 and NAT9 is associated with susceptibility to psoriasis.Nat Genet 2003;35:349–56.

4. Prokunina L, Castillejo-Lopez C, Oberg F, Gunnarsson I, Berg L,Magnusson V, et al. A regulatory polymorphism in PDCD1 isassociated with susceptibility to systemic lupus erythematosus inhumans. Nat Genet 2002;32:666–9.

5. Peltekova VD, Wintle RF, Rubin LA, Amos CI, Huang Q, Gu X,et al. Functional variants of OCTN cation transporter genes areassociated with Crohn disease. Nat Genet 2004;36:471–5.

6. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS, et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988;31:315–24.

7. MacGregor AJ, Bamder S, Silman AJ. A comparison of theperformance of different methods of disease classification forrheumatoid arthritis: results from an analysis from a nationwidetwin study. J Rheumatol 1994;21:1420–6.

8. Barton A, Bowes J, Eyre S, Spreckley K, Hinks A, John S, et al. Afunctional haplotype of the PADI4 gene associated with rheuma-toid arthritis in a Japanese population is not associated in a UnitedKingdom population. Arthritis Rheum 2004;50:1117–21.

9. Palm O, Moum B, Ongre A, Gran JT. Prevalence of ankylosingspondylitis and other spondyloarthropathies among patients withinflammatory bowel disease: a population study (the IBSENstudy). J Rheumatol 2002;29:511–5.

10. Steer S, Fisher SA, Fife M, Cuthbert A, Newton J, Wordsworth P,et al. Development of rheumatoid arthritis is not associated withtwo polymorphisms in the Crohn’s disease gene CARD15. Rheu-matology (Oxford) 2003;42:304–7.

11. Ferreiros-Vidal I, Barros F, Pablos JL, Carracedo A, Gomez-Reino JJ, Gonzalez A. CARD15/NOD2 analysis in rheumatoidarthritis susceptibility. Rheumatology (Oxford) 2003;42:1380–2.

758 BARTON ET AL