and Immunity (2001) 2, 373–380  2001 Nature Publishing Group All rights reserved 1466-4879/01 $15.00 www.nature.com/gene Six markers on the short arm of 6 in relation to HLA-DR3 and TNF−308A in systemic lupus erythematosus

MW van der Linden1,2, AR van der Slik3, E Zanelli3, MJ Giphart3, E Pieterman2, GMTh Schreuder3, RGJ Westendorp4 and TWJ Huizinga2 Departments of 1Clinical Epidemiology, 2Rheumatology, 3Immunohaematology and Blood Transfusion, 4General Internal Medicine, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands

Differences in allelic distribution at loci surrounding the human HLA-DRB1 and tumor necrosis factor (TNF) genes have been observed in association with systemic lupus erythematosus (SLE). We investigated whether the association of HLA- DRB1*0301 (HLA-DR3) and TNF−308A with SLE could be attributed to polymorphic markers in the chromosomal region encompassed by HLA-DRB1 and HLA-C. Ninety-one consecutive Caucasian patients with SLE and 253 controls (organ donors) were typed for HLA-DRB1, D6S1014, D6S273, TNFa, MIB, C1F2F5, and C1F3F2 and the single nucleotide at position −308 in the promoter of TNF. The independent contribution of alleles to disease susceptibility was estimated by cross-tabulation and multivariate logistic regression. Possession of TNF−308A was associated with susceptibility to SLE (odds ratio [95% confidence interval], 3.70 [2.24–6.11]). This remained present after stratification on possession of HLA-DR3 (pooled odds ratio, 2.53 [1.37–4.70]). Stratification revealed a possible association of possession of C1F2F5*192 with protection from SLE beyond the effects of HLA-DR3 and TNF−308A. A dosage effect was observed for −308A only (homozygotes, 7.75 [3.01–20.0], heterozygotes, 3.15 [1.85–5.37]). In multivariate analysis, possession of HLA-DR3, TNF−308A, and C1F2F5*192 remained independently associated with susceptibility to SLE (2.58 [1.29–5.18], 2.76 [1.43–5.31], and 0.26 [0.10–0.66], respectively). The association of possession of TNF−308A with susceptibility to SLE cannot be attributed to linkage to HLA-DR3 alone, nor to other polymorphic markers in the vicinity of the TNF gene. Further loci that are independently associated with SLE might be in the vicinity of marker C1F2F5. Genes and Immunity (2001) 2, 373–380.

Keywords: HLA-DRB1; TNF; genetics; susceptibility; systemic lupus erythematosus

Introduction be associated with susceptibility to SLE among African- Americans.12 Estimates of the allele frequency of the The chromosomal region encompassing the human leu- allele TNF−308A range from 17% to 28% in different Cau- kocyte antigen (HLA) Class II to Class I genes has been casian control populations.8,13–17 In our study among Eur- implicated in susceptibility to systemic lupus ery- opean Caucasians, the association between possession of thematosus (SLE) and other autoimmune diseases, such TNF−308A and SLE was independent of the existing as rheumatoid arthritis, insulin-dependent diabetes mel- association between possession of the HLA-DRB1*0301 litus, atopic astma, multiple sclerosis, and Crohn’s dis- (HLA-DR3) allele and SLE, in spite of linkage disequilib- 1–7 ease. Alleles of the HLA-Class II gene, HLA-DRB1, are rium between alleles of these loci.18 in linkage disequilibrium with genetic variations in the This previously detected association between presence 8,9 tumor necrosis factor (TNF) gene. of a TNF−308A allele and susceptibility to SLE might be Involvement of TNF in SLE was suggested by animal attributed to genetic variations in the region surrounding models, in which development of glomerulonephritis the TNF locus, apart from HLA-DR3. To investigate this 10 was delayed upon administration of recombinant TNF. possibility, we typed six microsatellite loci in a 1.8 million Disease activity in patients with SLE correlates with the base-pair region encompassed by HLA-DRB1 and HLA- 11 levels of circulating TNF. Recently, possession of the C in patients with SLE and controls. rare allele of a single nucleotide polymorphism located − at 308 base pairs from the transcription initiation site in Results the promoter of TNF (TNF2, TNF−308A), was shown to Demographic characteristics The mean age of the patients was similar to that of the Correspondence: Dr TWJ Huizinga, Department of Rheumatology, Build- controls (39.6 ± 1.5 vs 38.6 ± 1.1 years, mean ± standard ing 1, C4-R, Leiden University Medical Center, P.O. Box 9600, 2300 RC error). The patients were more often female than the con- Leiden, The Netherlands. E-mail: T.W.J.Huizinga.reumaȰlumc.nl MW van der Linden is supported by a grant of the Dutch Organiza- trols (92% and 46% females, respectively), but this did tion for Scientific Research (N.W.O.), grant no. 904–61–110. not influence genotype distributions among patients Received 8 February 2001; revised and accepted 6 August 2001 and controls. Microsatellite markers in SLE MW van der Linden et al 374 Allelic distribution of polymorphic markers contribution of each risk factor and thus to adjust for the Homozygosity of polymorphic markers was largest for effect of linkage disequilibrium, a stratified analysis was C1F3F2 with 39% in patients and 40% in controls and undertaken. The risk to develop SLE in the presence of lowest in MIB, with 15% and 11%. Four percent of all candidate alleles was studied in the presence and in the subjects were homozygous for HLA-DR3 (7% in patients, absence of HLA-DR3 separately. In Table 2 these odds 3% in controls) and 6% were homozygous for TNF−308A ratios are depicted separately for the HLA-DR3-positive (14% in the patients, 3% in the controls). and DR3-negative strata. To summarize the effect of each Individual alleles were studied. The relative risk to allele in both strata combined, the weighed estimate of develop disease, conferred by an extremely rare allele, the stratum-specificoddsratiosisgiven(Mantel–Haenszel may be larger than the relative risk conferred by a com- odds ratio). This pooled odds ratio is an estimate of the mon allele, merely as a consequence of small numbers. effect of each allele on susceptibility to (or protection from) To prevent the low background frequency of certain SLE, adjusted for its distribution over HLA-DR3-positive alleles from introducing spurious associations, the con- persons and HLA-DR3-negative persons. clusions were restricted to the effects of either relatively By this analysis of effects across the separate strata, the common alleles (more than 20% frequency of persons consequences of linkage disequilibrium between studied possessing at least of these alleles), or alleles with a clear alleles and HLA-DR3 became apparent. The odds ratios and distinctive impact on disease risk (unadjusted odds for presence of the alleles D6S1014*134 and MIB*350 ratio or unadjusted odds ratio−1 of 2.5 or greater). Also, became nonsignificant. This indicates that the effect of alleles that are bound to have such an impact merely on risk to develop SLE in the presence of these alleles is the basis of their low frequency, below 5%, are excluded. dependent on possession of HLA-DR3. For D6S273*139 Frequencies of persons possessing these 22 selected and TNFa*99, the stratum-specific odds ratios appeared alleles are depicted in Table 1, with odds ratios rep- to indicate opposite effects among HLA-DR3-positive resenting the risk to develop SLE among persons with and HLA-DR3-negative persons (test indicating hetero- the allele compared with that among persons without the geneity: P = 0.004 and P = 0.049, respectively). Therefore, allele. Frequencies of the presence of 10 alleles were sig- the pooled odds ratio is inconclusive for D6S237*139 and nificantly different among the patients compared with TNFa*99. Possession of the alleles D6S1014*143, controls (eight alleles present in increased frequencies, TNF−308A, C1F2F5*196 and C1F3F2*354 appeared to be two alleles present in decreased frequency). statistically significantly associated with susceptibility to SLE, independent of possession of HLA-DR3. The effect Stratified analysis was present and concordant in HLA-DR3-positives and Confounder effects arise, for example due to linkage dis- HLA-DR3-negatives. Possession of C1F2F5*192 appeared equilibrium with HLA-DR3. To estimate the individual protective, independent of possession of HLA-DR3.

Table 1 Frequencies of selected polymorphic marker alleles in patients with systemic lupus erythematosus and controls

Locus Allelea Patients (n = 91) Controls (n = 253) Odds ratiob (95% confidence interval) n (%) n (%)

HLA-DR 3 45 (49) 57 (23) 3.36 (2.03–5.58)* D6S1014 134 28 (31) 113 (45) 0.55 (0.33–0.92)* 137 63 (69) 178 (70) 0.95 (0.56–1.60) 143 42 (46) 47 (19) 3.76 (2.23–6.32)* D6S273 131 26 (29) 60 (24) 1.29 (0.75–2.21) 133 37 (41) 131 (52) 0.64 (0.39–1.04) 135 43 (47) 140 (55) 0.72 (0.45–1.17) 139 27 (29) 45 (18) 1.95 (1.12–3.39)* TNF −308 A 52 (57) 67 (26) 3.70 (2.24–6.11)* TNFa 99 56 (65) 125 (49) 1.64 (1.01–2.67)* 107 20 (22) 71 (28) 0.72 (0.41–1.27) 115 17 (19) 53 (21) 0.87 (0.47–1.59) 117 29 (32) 78 (31) 1.05 (0.57–1.55) MIB 326 17 (19) 61 (24) 0.72 (0.40–1.32) 336 31 (34) 90 (36) 0.94 (0.57–1.55) 350 49 (54) 81 (32) 2.48 (1.52–4.04)* C1F2F5 192 6 (7) 41 (16) 0.37 (0.15–0.89)* 196 41 (45) 55 (22) 2.95 (1.77–4.91)* 199 28 (31) 94 (37) 0.75 (0.45–1.26) C1F3F2 342 33 (36) 100 (40) 0.87 (0.53–1.43) 354 54 (59) 95 (38) 2.43 (1.49–3.96)* 360 23 (25) 82 (32) 0.71 (0.41–1.21)

aAlleles defined by fragment size (in base pairs), except HLA-DR3 and TNF-308A. bOdds ratio compares presence of at least one copy of the indicated allele to absence of that allele, among patients with SLE vs controls. *P value Ͻ 0.05.

Genes and Immunity Microsatellite markers in SLE MW van der Linden et al 375 Table 2 Odds ratios to systemic lupus erythematosus for presence of alleles, stratified on presence of HLA-DR3

Locus Allele Unadjusted Stratified Pooleda (n = 344) (n = 344) DR3-positive DR3-negative (n = 102) (n = 242)

D6S1014 134 0.55 0.57 0.70 0.65 (0.33–0.92) (0.23–1.40) (0.36–1.34) (0.38–1.10) 143 3.76 2.12 3.35 2.39* (2.23–6.32) (0.74–6.05) (0.72–15.5) (1.01–5.68) D6S273 139 1.95 0.52 2.57 0.79b (1.12–3.39) (0.23–1.17) (0.72–9.19) (0.40–1.56) TNF −308 A 3.70 2.32 2.70 2.53* (2.24–6.11) (0.82–6.58) (1.25–5.80) (1.37–4.70) TNFa 99 1.64 2.73 0.70 1.00b (1.01–2.67) (0.82–9.15) (0.36–1.38) (0.02–1.81) MIB 350 2.48 1.73 1.74 1.74 (1.52–4.04) (0.74–4.04) (0.87–3.47) (1.02–2.97) C1F2F5 192 0.37 0.23 0.33 0.27* (0.15–0.89) (0.07–0.74) (0.08–1.43) (0.11–0.68) 196 2.95 2.00 1.67 1.84* (1.77–4.91) (0.86–4.65) (0.69–4.02) (1.00–3.39) C1F3F2 354 2.43 1.27 1.27 1.73* (1.49–3.96) (0.53–3.02) (1.09–4.05) (1.03–2.92) aPooled odds ratio is a weighed average of the stratum-specific odds ratios (Mantel–Haenszel odds ratio). Intervals beween parentheses are 95% confidence intervals. bLack of homogeneity across the strata. *Pooled odds ratio with P Ͻ 0.05.

Alleles of polymorphic markers in the vicinity of TNF, TNF−308A. Analysis in this particular group of patients such as TNFa, might be in linkage disequilibrium to and controls suggests that possession of TNFa*99 does TNF−308 (Figure 1). To adjust for this, risk to develop not contribute significantly to susceptibility to SLE SLE dependent on various alleles was also analyzed after beyond possession of TNF−308A. There was heterogen- stratification on TNF−308A (Table 3). Linkage disequilib- eity across the strata in the comparison of D6S1014*134 rium between the TNFa*99 allele and TNF−308A became and D6S273*139 (P Ͻ 0.05) and this prohibited further apparent when comparing the unadjusted odds ratio pooled analysis for these two alleles, stratified on pos- with the pooled odds ratio that adjusts for possession of session of TNF−308A.

Figure 1 Chromosomal location of polymorphic markers on the short arm of . Chromosomal location of HLA-DR, HLA-C, HLA-B, polymorphic microsatellite markers D6S1014, D6S273, TNFa, MIB, C1F2F5, and C1F3F2, and a single nucleotide polymorphism in the promoter of TNF. TNF G−308A. The relative positions of LTA, LTB and LST1 in the TNF gene cluster are indicated, as well as the relative positions of the genes encoding complement factor 4B, complement factor 4A, factor B, complement factor 2, heat-shock A1, and small nuclear ribonucleoprotein (C4B, C4A, Bf, C2, HSPA1 and snRNP) (not to scale).

Genes and Immunity Microsatellite markers in SLE MW van der Linden et al 376 Table 3 Odds ratios to systemic lupus erythematosus for presence of alleles, stratified on presence of TNF−308A

Locus Allele Unadjusted Stratified Pooleda (n = 344) (n = 344) −308A-positive −308A-negative (n = 119) (n = 225)

HLA-DR 3 3.36 1.79 2.07 1.88* (2.03–5.58) (0.80–3.97) (0.75–5.73) (1.00–3.54) D6S1014 134 0.55 0.31 1.04 0.63b (0.33–0.92) (0.13–0.72) (0.52–2.08) (0.37–1.06) 143 3.76 1.47 5.32 1.95* (2.23–6.32) (0.68–3.20) (1.46–19.39) (1.00–3.82) D6S273 139 1.95 0.38 16.73 0.80b (1.12–3.39) (0.18–0.80) (3.24–86.46) (0.43–1.47) TNFa 99 1.64 2.02 0.64 1.01 (1.01–2.67) (0.80–5.15) (0.31–1.35) (0.58–1.76) MIB 350 2.48 1.56 1.62 1.59 (1.52–4.04) (0.72–3.40) (0.76–3.49) (0.94–2.76) C1F2F5 192 0.37 0.20 0.54 0.32* (0.15–0.89) (0.05–0.71) (0.15–1.88) (0.13–0.80) 196 2.95 1.47 1.52 1.48 (1.77–4.91) (0.68–3.20) (0.47–4.94) (0.77–2.84) C1F3F2 354 2.43 1.13 2.16 1.59 (1.49–3.96) (0.51–2.49) (1.06–4.40) (0.94–2.71)

aPooled odds ratio is a weighted average of the stratum-specific odds ratios (Mantel–Haenszel odds ratio). Intervals beween parentheses are 95% confidence intervals. bLack of homogeneity across strata. *Pooled odds ratio with P Ͻ 0.05.

The discordant effects of D6S273*139 among D6S1014*134 were not homogeneous over HLA-DR3- TNF−308A-positive and -negative persons resulted from positive and HLA-DR3-negative persons, nor over an imbalanced distribution of D6S273*139 alleles over TNF−308A-positive and -negative persons, a second TNF−308A-positive and -negative strata. Forty-three out regression model was analysed in which D6S273*139 and of 67 (64%) TNF−308A-carrying controls also had D6S1014*134 were not considered. This analysis indicated D6S273*139, and 21 out of 52 (40%) TNF−308A-carrying that possession of three alleles contributed independently patients also had D6S273*139. These proportions were to risk of, or protection from SLE (HLA-DR3 2.58 [1.29– inverted among TNF−308-negative persons: two out of 5.18], TNF−308A 2.76 [1.43–5.31], and C1F2F5*192 0.26 186 (1%) TNF−308A-negative controls had D6S273*139, [0.10–0.66]). Possession of the alleles D6S1014*137 and whereas six out of 39 (15%) TNF−308A-negative patients D6S273*131 remained in the model at the last regression had D6S273*139. step (step 16), but their effects on SLE did not reach stat- Of all nine alleles tested, HLA-DR3, D6S1014*143, and istical significance (1.81 [0.98–3.35] and 1.65 [0.91–2.96], C1F2F5 196 were statistically significantly associated with respectively). susceptibility to SLE, independent of the effect of TNF−308A. C1F2F5*192 was statistically significantly Dose-response and co-dominant effects associated with protection from SLE, after adjustment for To investigate the possibility of a dose-response relation possession of TNF−308A (Table 3). of the alleles suggested above, frequencies of hetero- zygotes and homozygtes with these alleles were com- Multivariate analysis pared with frequencies of persons without the allele. The preliminary analyses above suggested that some Odds ratios for these heterozygotes and homozygtes in alleles are involved in susceptibility to SLE. These alleles patients and controls are depicted in Table 4. Both hetero- were further investigated. To estimate the independent zygous and homozygous possession of the alleles HLA- contribution of possession of these alleles, multivariate DR3, TNF–308A and C1F2F5*192 was statistically signifi- logistic regression was used. cantly associated with risk to develop SLE. This was not In a first model, all 22 alleles from Table 1 were the case for D6S273*131. The odds ratio increased in a entered. Results indicated that D6S1014*143 was associa- proportional manner for persons who possessed either ted with risk to develop SLE (multivariate odds ratio zero, one or two TNF−308A alleles. No such proportional [95% confidence interval] 4.37, [1.81–10.56]), at the cost of increase was observed for possession of the other alleles. HLA-DR3, which was removed from the regression equ- ation at backward regession step 13. Presence of Discussion TNF−308A was associated with increased susceptibility to SLE, and D6S273*139 and C1F2F5*192 appeared protec- Six microsatellite markers in the region surrounding the tive (odds ratios 3.14 [1.57–6.28], 0.36 [0.15–0.87] and 0.27 TNF locus, and HLA-DR3, did not explain a previously [0.10–0.70], respectively). However, since the analyses described association between possession of the A allele above showed that the distribution of D6S273*139 and of a single nucleotide polymorphism in the promoter of

Genes and Immunity Microsatellite markers in SLE MW van der Linden et al 377 Table 4 Gene–dose effect of polymorphic markers that are independently associated with susceptibility to systemic lupus erythematosus

Locus Genotype Patients (n = 91) Controls (n = 253) Odds ratioa (95% confidence interval) n(%)n(%)

HLA-DR x/x 46 (51) 196 (77) 1b 3/x 39 (43) 50 (20) 3.32* (1.96–5.63) 3/3 6 (7) 7 (3) 3.65* (1.71–11.4) −308 G/G 39 (43) 186 (74) 1b — A/G 39 (43) 59 (23) 3.15* (1.85–5.37) A/A 13 (14) 8 (3) 7.75* (3.01–20.0) C1F2F5 x/x 85 (93) 212 (84) 1b — 192/x 6 (7) 38 (15) 0.39* (0.16–0.97) 192/192 0 (0) 3 (1) 0.71* (0.66–0.77) aOdds ratio compares possession of the indicated genotype to the reference category, among patients with SLE vs controls. bReference category. *Genotype with P-value Ͻ 0.05 compared with reference category.

TNF, TNF−308A, and susceptibility to SLE. Therefore we Centromeric from TNF, and located 275 kilo base pairs interpret this association as independent of the HLA-DR3 centromeric from D6S273, is one of the genes encoding allele. The frequently reported association between pos- the classical-pathway complement activator C4, C4A.25 session of HLA-DR3 and risk to develop SLE was con- Complement component gene variations, particularly firmed using a new control group.18 The data also sup- C4A gene deletions and silent alleles, have been impli- port an independent association of the allele C1F2F5*192 cated in susceptibility to SLE in various ethnic groups with protection from SLE. This allele might be in linkage since the 1970s.1,26–31 Indeed, a central European multi- disequilibrium with alleles from other polymorphic loci center study of 300 patients with SLE and 453 of their such as HLA-B. first-degree relatives demonstrated that increased pos- These findings do not exclude the possibility that the session frequency of C4AQ0 among patients with SLE TNF gene itself is causally related to the disease. Alterna- was dependent on possession of HLA-B8-DR3.32 Strong tively, alleles in linkage disequilibrium with alleles at the linkage disequilibrium between these loci may not fully TNF gene or its promoter might be responsible for the have been adjusted for in the present analysis, since C4A association. For example, D6S273 alleles are in linkage alleles were not typed in this population. disequilibrium with alleles at loci within the TNF gene It has been suggested that inherited differences in TNF cluster, in spite of the physical distance in-between of production are related to different alleles of the HLA- 150000 base pairs. We found divergent effects at the DRB1 locus.24,33 HLA-DR3 presence was suggested to be D6S1014 and D6S273 loci. This might indicate the possi- associated with high, and HLA-DR2 possession with low bility of additional susceptibility loci located between TNF production in mitogen-stimulated peripheral blood HLA-DRB1 and TNF. This possibility is in agreement lymphocytes.23,34 Others could not confirm this.16 Since with the finding of altered allelic distribution at the the TNF promoter polymorphism at −308 base pairs from D6S273 locus among patients with rheumatoid arthritis the transcription initiation site is present on the extended who did not possess the QKRAA/QRRAA epitope.19 haplotype HLA-A1, B8, DR3, it has been attempted to However, the divergence of the effects of the polymor- attribute these associations between TNF production and phic markers under study does not directly support a HLA-genotypes to genetic variations in the TNF pro- role for such additional loci. The number of D6S273*139 moter. No association between TNF−308A or DR3 and alleles in the absence of TNF−308 A was very small both endotoxin-induced production of TNF could however be in our patients and our controls. Thus, small fluctuations demonstrated in peripheral blood samples of family in allelic frequencies between D6S273 and TNF−308 may members of patients with meningococcal disease, nor in lead to large shifts in the odds ratio of that particular stra- broncho-alveolar lavage samples of patients with sarcoid- tum. osis.17,35 Others described increased production of TNF in A possible association between alleles of marker loci the presence of both −308A and HLA-DR3.13 This lack within the TNF gene cluster and susceptibility to SLE has of functionality of the TNF−308 polymorphism might be previously been suggested.20–23 Despite linkage disequi- explained by differences in stimulus or cell type used.36 librium between TNF−308A and TNFa*99, no additional Since increased odds ratios both for heterozygous and or independent effect of TNFa*99 over HLA-DR3 was homozygous presence of HLA-DR3, TNF−308A and observed. This is in agreement with the previously C1F2F5*192 were observed, recessive effects of these reported linkage disequilibrium between TNFa*99 and alleles on susceptibility to SLE seem unlikely, whereas HLA-DR3.24 Stratified analysis suggested that possession dominant and co-dominant effects might be possible. In of TNFa*99 could not predict susceptibility to SLE better the case of TNF−308A, the odds ratio for homozygous than TNF−308A. This makes the interpretation less likely, possession was higher than that for heterozygous pos- that TNFa*99 or a gene in its immediate vicinity, such as session. This is compatible with a gene–dose relation of the gene-encoding lymphotoxin-A (LTA), confers TNF−308A, which strengthens the interpretation that this additional or independent susceptibility to SLE. TNFa is locus has causal relevance for the disease. For HLA-DR3 located 3500 base pairs telomeric from LTA, which is a gene–dose relation was not observed in the present itself ca. 3000 base pairs telomeric from the TNF gene.25 sample.

Genes and Immunity Microsatellite markers in SLE MW van der Linden et al 378 The influence of confounding genetic determinants, TNFa, C1F2F5 and C1F3F2, encompass the HLA class III due to linkage disequilibrium, was adjusted for by strati- /TNF/lymphotoxin region and extend into HLA class I fied analysis. The influence of a linked locus on the effects between HLA-C and HLA-B. Detailed information on of alleles was analyzed by stratification according to pos- these microsatellites have been published elsewhere.38–40 session or non-possession of the alleles at the linked Amplification was carried out by fluorescent-primer locus. Since the microsatellite markers show a high polymerase chain reaction (PCR) in a total reaction vol- degree of polymorphism, the chance of bearing an allele ume of 25 ml containing approximately 50 ng of genomic is affected by the chance of bearing any other allele at the DNA, 12.5 pmol (D6S1014, MIB, C1F2F5 and C1F3F2) or locus. Among heterozygotes, the results of the stratified 25 pmol (D6S73 and TNFa) of each primer, 200 mM of analysis may therefore have been influenced by pos- dNTPs (Pharmacia Biotech, Uppsala, Sweden), 5% (v/v) session of this second allele. Among homozygotes, there glycerol, and 2.5 units of Taq polymerase (Amplitaqaˆ, is no such influence. If, among the heterozygotes the Perkin Elmer, Foster City, CA, USA). Amplification second allele is responsible for the association, one might buffer contained 50 mM KCl, 1.5 mM MgCl2, 10 mM Tris- expect that the odds ratio (or odds ratio−1) of hetero- HCl, at pH 8.4 (room temperature) and 0.06 mg/ml bov- zygous possession of the alleles HLA-DR3, TNF−308A, ine serum albumin. Touchdown PCR was utilised to and C1F2F5*192 is substantially higher than the odds increase sensitivity of annealing during the first cycles ratios of homozygous possession. For HLA-DR3 and (annealing temperature 58 to 68°C) using a Peltier Ther- C1F2F5*192, no difference was observed between the mal Cycler (PTC-200, MJ Research, MA, USA). odds ratios of heterozygotes and homozygotes, whereas For the patients’ samples, PCR products were loaded TNF−308A homozygotes had an increased risk to onto a denaturing acrylamide sequencing gel (Reprogel develop SLE compared with heterozygotes. High Resolution, Amersham Pharmacia Biotech) in Auto- The allele C1F2F5*192 might point to a locus that is mated Laser Fluorescence-DNA sequencer (ALF-express possibly involved in protection from SLE. Given the high II, Amersham Pharmacia Biotech). Electrophoresis (1500 degree of polymorphism of C1F2F5, the number of per- V, 60 mA, 30 W at 55°C) and data collection were perfor- sons with the allele is inevitably small. The finding that a med using ALF Manager software (Amersham Pharma- protective effect was nevertheless statistically significant, cia Biotech). Data were analyzed using ALFWin Frag- might indicate that the locus C1F2F5 points to a relevant ment Analyzer 1.01 software (Amersham Pharmacia genetic determinant for the disease. The protective effect Biotech). Allele assignment was performed by using CY-5 of this allele might therefore be a reflection of a total labeled internal size markers and running external allelic effect of the other alleles at the locus. Since the locus ladders over the size range of the involved microsatellite C1F2F5 is located 75 kilo base pairs telomeric from HLA- locus. Allelic ladders were composed of PCR products B and 20 kilo base pairs centromeric from HLA-C, this from Epstein–Barr virus-transformed lymphoblastoid cell effect might be due to linkage to HLA class I loci. lines, which had been sequenced previously.41 Analysis In conclusion, linkage disequilibrium to alleles from six of the control group was performed using a 96-lane ABI polymorphic microsatellite markers in the region, and 377-DNA sequencer (Applied Biosystems, Foster City, HLA-DR3, could not explain an independent association CA, USA). Data were analyzed using Genescan 3.1 and between TNF−308A and susceptibility to SLE. Genotyper 2.0 software (Applied Biosystems). Accurate allele sizing was achieved using Genescan-ROX 400HD internal size marker (Applied Biosystems) and external Patients and methods allelic ladders were run as described above.

Study population HLA-DRB1 typing One hundred and seven consecutive Caucasian patients HLA-DRB1 typing was performed by sequence specific with SLE at the Rheumatology Department of Leiden oligonucleotide (SSO) typing on PCR amplified genomic University Medical Center (The Netherlands) were DNA as previously described.42 The presence of the included. The recruitment of 99 of these patients has been HLA-DRB1*0301 allele will be referred to as HLA-DR3 described elsewhere.18 All patients fulfilled the 1982 possession. American College of Rheumatology (ACR) criteria for the classification of SLE as assessed by review of the medical Statistical analysis records. For the present analyis, 278 Dutch Caucasian Sixteen patients and 16 controls had incomplete typing organ donors from the Department of Immunohaematol- data for all six microsatellites. Nine controls had incom- ogy and Blood Transfusion of the Leiden University plete data for the TNF−308 promoter gene polymorphism Medical Center served as a control population. due to lack of DNA. This resulted in 91 patients and 253 controls for analysis. Amplification and determination of microsatellite Differences in allelic distribution of polymorphic mark- markers ers between patients and controls were assessed by chi- DNA was isolated by phenol-chloroform extraction and square analysis. To estimate the relative risk of specific ethanol precipitation of SDS-lysed and proteinase-K genetic risk factors, odds ratios with 95% confidence treated whole blood of the patients according to intervals were calculated according to Woolf.43 In strati- described methods.37 Six polymorphic microsatellite mar- fied analyses, Mantel–Haenszel’s odds ratio was used to kers, D6S1014, D6S273, TNFa, MIB, C1F2F5, and C1F3F2 estimate the odds ratio pooled over strata. Discordance were studied. Their chromosomal location is indicated in of stratum-specific effects, opposite effects across strata, Figure 1. The most centromeric of these microsatellites, was tested using Breslow and Days’s homogeneity test. D6S1014, is located in the vicinity of HLA-DR, whereas In logistic regression models, presence of alleles was sim- the other, more telomeric microsatellites, D6S273, MIB, ultaneously entered as categorical variables. The number

Genes and Immunity Microsatellite markers in SLE MW van der Linden et al 379 of alleles in the model was stepwise reduced by back- moter polymorphisms, production and susceptibility to multiple ward regression, which removes each allele variable from sclerosis in different groups of patients. J Neuroimmunol 1997; the model equation as soon as its conditional likelihood 72: 149–153. ratio exceeds the threshold of 0.10. The remaining allele 15 Rudwaleit M, Tikly M, Khamashta M et al. Interethnic differ- variables, all below the threshold, were considered allele- ences in the association of tumor necrosis factor promoter poly- morphisms with systemic lupus erythematosus. J Rheumatol presence variables that independently predicted risk to 1996; 23: 1725–1728. develop SLE. Multivariate logistic regression analysis 16 Stuber F, Udalova IA, Book M et al. −308 tumor necrosis factor adjusts for multiple testing, as the hypotheses concerning (TNF) polymorphism is not associated with survival in severe individual alleles are all tested at the same time. sepsis and is unrelated to lipopolysaccharide inducibility of the human TNF promoter. JInflamm 1995;46:42–50. 17 Westendorp RG, Langermans JA, Huizinga TW et al. Genetic Acknowledgements influence on cytokine production and fatal meningococcal dis- G Jones, W Verduyn (Immunohaematology and Blood ease [published erratum appears in Lancet 1997; 349: 656]. Lancet 1997; 349: 170–173. Transfusion) and BA de Jong (Clinical Epidemiology) are 18 Rood MJ, van Krugten MV, van der Linden MW et al.TNF−308A gratefully acknowledged for their contribution. The and HLA-DR3 alleles contribute independently to susceptibility authors are indebted to Dr BPC Koeleman to systemic lupus erythematosus. Arthritis Rheum 2000; 43: (Immunohaematology and Blood Transfusion) and Dr 129–134. FW Dekker (Clinical Epidemiology) for critical dis- 19 Singal DP, Li J, Ye M, Lei K. D6S273 microsatellite polymor- cussions. phism and susceptibility to rheumatoid arthritis. Tissue Antigens 1998; 52:353–358. 20 d’Alfonso S, Colombo G, Della BS, Scorza R, Momigliano-Richi- References ardi P. Association between polymorphisms in the TNF region and systemic lupus erythematosus in the Italian population. 1 Arnett FC, Goldstein R, Duvic M, Reveille JD. Major histocom- Tissue Antigens 1996; 47: 551–555. patibility complex genes in systemic lupus erythematosus, 21 Hajeer AH, Worthington J, Davies EJ et al. TNF microsatellite Sjogren’s syndrome, and polymyositis. (Review). Am J Med 1988; a2, b3 and d2 alleles are associated with systemic lupus ery- 85:38–41. thematosus. Tissue Antigens 1997; 49: 222–227. 2 Hajeer AH, Dababneh A, Makki RF et al. Different gene loci 22 Tarassi K, Carthy D, Papasteriades C et al. HLA-TNF haplotype within the HLA-DR and TNF regions are independently associa- heterogeneity in Greek SLE patients. Clin Exp Rheumatol 1998; ted with susceptibility and severity in Spanish rheumatoid 16:66–68. arthritis patients. Tissue Antigens 2000; 55: 319–325. 23 Jacob CO, Fronek Z, Lewis G, Hansen JA, McDevitt HO. Heri- 3 Bali D, Gourley S, Kostyu DD et al. Genetic analysis of multiplex table major complex class II-associated differ- rheumatoid arthritis families. Genes Immun 1999; 1:28–36. ences in production of tumor necrosis factor ␣: relevance to gen- 4 Obayashi H, Nakamura N, Fukui M et al.Influence of TNF etic predisposition to systemic lupus erythematosus. Proc Natl microsatellite polymorphisms (TNFa) on age-at-onset of insulin- Acad Sci USA 1990; 87: 1233–1237. dependent diabetes mellitus. Hum Immunol 1999; 60: 974–978. 24 Pociot F, Briant L, Jongeneel CV et al. Association of tumor 5 Albuquerque RV, Hayden CM, Palmer LJ et al. Association of necrosis factor (TNF) and class II major histocompatibility com- polymorphisms within the tumour necrosis factor (TNF) genes plex alleles with the secretion of TNF-alpha and TNF-beta by and childhood asthma. Clin Exp Allergy 1998; 28: 578–584. human mononuclear cells: a possible link to insulin-dependent 6 Sandberg-Wollheim M, Ciusani E, Salmaggi A, Pociot F. An Eur J Immunol 23 evaluation of tumor necrosis factor microsatellite alleles in gen- diabetes mellitus. 1993; : 224–231. etic susceptibility to multiple sclerosis. Mult Scler 1995; 1: 181– 25 The MHC sequencing consortium. Complete sequence and gene 185. map of a human major histocompatibility complex. Nature 1999; 7 Louis E, Peeters M, Franchimont D et al. Tumour necrosis factor 401: 921–923. (TNF) gene polymorphism in Crohn’s disease (CD): influence 26 Dunckley H, Gatenby PA, Hawkins B, Naito S, Serjeantson SW. on disease behaviour? Clin Exp Immunol 2000; 119:64–68. Deficiency of C4A is a genetic determinant of systemic lupus 8 Wilson AG, de Vries N, Pociot F et al. An allelic polymorphism erythematosus in three ethnic groups. J Immunogenet 1987; 14: within the human TNF-alpha promoter region is strongly asso- 209–218. ciated with HLA A1, B8 and DR3 alleles. J Exp Med 1993; 177: 27 Ochs HD, Rosenfeld SI, Thomas ED et al. Linkage between the 557–560. gene (or genes) controlling synthesis of the fourth component of 9 Sanchez-Mazas A, Djoulah S, Busson M et al. A linkage disequi- complement and the major histocompatibility complex. N Engl J librium map of the MHC region based on the analysis of 14 loci Med 1977; 296: 470–475. haplotypes in 50 French families. Eur J Hum Genet 2000; 8:33–41. 28 Partanen J, Koskimies S, Johansson E. C4 null phenotypes 10 Gordon C, Ranges GE, Greenspan JS, Wofsy D . Chronic therapy among lupus erythematosus patients are predominantly the with recombinant tumor necrosis factor-alpha in autoimmune result of deletions covering C4 and closely linked 21- NZB/NZW F1 mice. Clin Immunol Immunopathol 1989; 52: 421– hydroxylase A genes. J Med Genet 1988; 25: 387–391. 434. 29 Schur PH, Marcus-Bagley D, Awdeh Z, Yunis EJ, Alper CA. The 11 Studnicka-Bencke A, Steiner G, Petera P, Smolen JS. Tumour effect of ethnicity on major histocompatibility complex comp- necrosis factor alpha and its soluble receptors parallel clinical lement allotypes and extended haplotypes in patients with sys- disease activity in systemic lupus erythematosus. Br J Rheum temic lupus erythematosus. Arthritis Rheum 1990; 33: 985–992. 1996; 35: 1067–1074. 30 Tappeiner G, Hintner H, Scholz S et al. Systemic lupus ery- 12 Sullivan KE, Wooten C, Schmeckpeper BJ, Goldman D, Petri thematosus in hereditary deficiency of the fourth component of MA. A promoter polymorphism of tumor necrosis factor alpha complement. J Am Acad Dermatol 1982; 7:66–79. associated with systemic lupus erythematosus in African- 31 Ulgiati D, Abraham LJ. Comparative analysis of the disease- Americans. Arthritis Rheum 1997; 40: 2207–2211. associated complement C4 gene from the HLA-A1, B8, DR3 13 Bouma G, Crusius JB, Oudkerk Pool M et al. Secretion of tumour haplotype. Exp Clin Immunogenet 1996; 13:43–54. necrosis factor ␣ and lymphotoxin ␣ in relation to polymor- 32 Hartung K, Baur MP, Coldewey R et al. Major histocompatibility phisms in the TNF genes and HLA-DR alleles. Relevance for complex haplotypes and complement C4 alleles in systemic inflammatory bowel disease. Scand J Immunol 1996; 43: 456–463. lupus erythematosus. Results of a multicenter study. J Clin 14 Huizinga TW, Westendorp RGJ, Bollen EL et al. TNF-alpha pro- Invest 1992; 90: 1346–1351.

Genes and Immunity Microsatellite markers in SLE MW van der Linden et al 380 33 Wilson AG, Symons JA, McDowell TL, McDevitt HO, Duff GW. 38 Udalova IA, Nedospasov SA, Webb GC, Chaplin DD, Turets- Effects of a polymorphism in the human tumor necrosis factor kaya RL. Highly informative typing of the human TNF locus alpha promoter on transcriptional activation. Proc Natl Acad Sci using six adjacent polymorphic markers. Genomics 1993; 16: USA 1997; 94: 3195–3199. 180–186. 34 Bendtzen K, Morling N, Fomsgaard A et al. Association between 39 Foissac A, Salhi M, Cambon-Thomsen A. Microsatellites in the HLA-DR2 and production of tumour necrosis factor alpha and HLA region: 1999 update. Tissue Antigens 2000; 55: 477–509. interleukin 1 by mononuclear cells activated by lipopolysaccha- 40 Zanelli E, Jones G, Pascual M et al. The telomeric part of the ride. Scand J Immunol 1988; 28: 599–606. HLA region predisposes to rheumatoid erthritis independent of 35 Somoskovi A, Zissel G, Seitzer U et al. Polymorphisms at pos- the class II loci. Hum Immunol 2001; 62:75–84. ition −308 in the promoter region of the TNF-alpha and in the 41 Degli-Esposti MA, Griffiths MS, Daly LN et al. Characterization first intron of the TNF-beta genes and spontaneous and lipo- of 4AOHW cell line panel including new data for the 10IHW polysaccharide-induced TNF-alpha release in sarcoidosis. Cyto- panel. Hum Immunol 1993; 38:3–16. kine 1999; 11: 882–887. 42 Verduyn W, Doxiadis II, Anholts J et al. Biotinylated DRB 36 Kroeger KM, Steer JH, Joyce DA, Abraham LJ. Effects of stimu- sequence-specific oligonucleotides. Comparison to serologic lus and cell type on the expression of the −308 tumour necrosis HLA-DR typing of organ donors in eurotransplant. Hum Immu- factor promoter polymorphism. Cytokine 2000; 12: 110–119. nol 1993; 37:59–67. 37 Sambrook J., Fritsch EF, Maniatis T. Analysis and cloning of 43 Woolf B. On estimating the relation between blood group and eukaryotic genomic DNA. In: Nolan C, Ford N, Ferguson M disease. Ann Hum Genet 1995; 19: 251. (eds). Molecular Cloning: A Laboratory Manual. Second edn. Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 1989, 9, pp 9.16–9.17.

Genes and Immunity