HLA-G +3142 polymorphism as a susceptibility marker in two rheumatoid arthritis populations in Brazil

Tissue Antigens ISSN 0001-2815

HLA-G +3142 polymorphism as a susceptibility marker in tworheumatoid arthritis populations in BrazilT. D. Veit1, C. P. S. de Lima2, L. C. Cavalheiro3, S. M. Callegari-Jacques4, C. V. Brenol3, J. C. T. Brenol3,R. M. Xavier3, M. F. L. da Cunha Sauma2, E. J. M. dos Santos2 & J. A. B. Chies1

1 Laboratorio de Imunogenetica, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS Brazil2 Laboratorio de Genetica Humana e Medica, Universidade Federal do Para, Belem, Para Brazil3 Servico de Reumatologia, Hospital de Clınicas de Porto Alegre, Porto Alegre, RS Brazil4 Departamento de Estatıstica, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS Brazil

Key words

14 bp; 3′-untranslated region; Brazil; diseasesusceptibility; haplotype; human leukocyteantigen-G; polymorphism; rheumatoidarthritis

Correspondence

Jose Artur Bogo ChiesDepartment of GeneticsUniversidade Federal do Rio Grande do SulAv. Bento Goncalves, 9500Caixa Postal 1505391501-970 Porto Alegre, RSBrazilTel: +55 51 3308 6740Fax: +55 51 3308 7311e-mail: [email protected]

Received 21 June 2013; revised 9 January2014; accepted 20 January 2014

doi: 10.1111/tan.12311

Abstract

In this study, we sought to investigate the genetic influence of two HLA-G 3′-untranslated region (3′-UTR) polymorphisms – 14 bp (rs66554220) and +3142C>G(rs1063320) and their compounding haplotypes in susceptibility to rheumatoid arthritis(RA) in a two-region Brazilian study comprising of 539 patients and 489 controls.All subjects were polymerase chain reaction (PCR) genotyped for the referredpolymorphisms and logistic regression models controlling for sex, city and age wereperformed. Homozygozity for the +3142G allele was associated with an increasedrisk of RA [odds ratio (OR) = 1.45, 95% confidence interval (CI) = 1.075–1.959,PBonf = 0.030], whereas no association was observed for the 14 bp polymorphism.Haplotype comparisons between patients and controls showed a decreased frequencyof the delC haplotype in patients (OR = 0.70, 95% CI = 0.521–0.946, PBonf = 0.040),which remained significant in the rheumatoid factor (RF)-positive group (OR = 0.66,95% CI = 0.482–0.900, PBonf = 0.018), but not in the RF-negative group. Theseresults corroborate the hypothesis of an involvement of HLA-G in the susceptibility ofRA. The +3142G allele is associated with haplotype lineages that share high identityand are regarded as low producers. The presence of the G allele in homozygosis couldbe responsible for a low HLA-G expression profile that could favor the triggeringof RA.

Introduction

Rheumatoid arthritis (RA) is a chronic systemic inflamma-tory disease that can lead to joint deformities and permanentphysical disability. Onset is likely triggered by environmen-tal factors in susceptible individuals. Several genes have beenimplicated so far in the pathogenesis of RA. An importantgenetic region is the human leukocyte antigen (HLA) locus,which contributes to approximately half of the genetic sus-ceptibility for RA identified so far. Alleles from the HLA-DRB1 locus, notably those carrying the ‘shared epitope’sequence, are associated with an increased risk and a moresevere phenotype in RA (1). Recent evidences suggest thatsuch major histocompatibility complex (MHC) associationsare stronger or specific for anti-citrullinated protein antibod-ies (ACPA)-positive patients. However, this locus does notentirely explain MHC contribution to disease risk [reviewed inRef. (2)].

HLA-G is a non-classical HLA molecule characterized bylimited tissue distribution under normal physiological condi-tions and low polymorphism at both DNA and protein levels.HLA-G presents unique characteristics as compared to its clas-sical counterparts, such as the expression of multiple isoformsgenerated by alternative splicing, which can be membrane-bound (G1–G4) or secreted (G5-G7). Moreover, it is the onlyHLA molecule capable of forming dimers that seem to bethe main contributors in its biological activity. It has beenshown that HLA-G plays a major role in immunosuppression,interacting with cells of the immune system and suppress-ing the immune response by different mechanisms. Thesemechanisms include the inhibition of the activity of cyto-toxic T lymphocytes and natural killer (NK) cells as well asprotection of class I-negative or allogeneic tumors from NK-mediated anti-tumor immunity (3). Also, the HLA-G moleculecan inhibit CD4+ T-cell alloproliferative responses (4), pro-liferation of T and peripheral blood NK cells (5–7) and can

260 © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons LtdTissue Antigens, 2014, 83, 260–266

T. D. Veit et al. HLA-G 3′-UTR haplotype in rheumatoid arthritis

also act on antigen-presenting cells (APCs) by inhibiting theirmaturation and function (8). In addition, HLA-G may exertlong-term immunotolerogenic effects through the generationof suppressor cells [reviewed in Ref. (9)]. Such features ren-der HLA-G an attractive candidate gene for susceptibilityto immune-mediated diseases. The 3′-untranslated region (3′-UTR) seems to play an important role in the regulation ofHLA-G expression, mainly through post-transcriptional regu-latory mechanisms. In silico analysis of this region has iden-tified numerous putative sites for microRNA binding, whichmay influence HLA-G expression, depending on the alleleand/or the biological context. According to the latest infor-mation, 11 polymorphic positions have been identified (10),many of which overlap putative microRNA binding sites.Among the 3′-UTR polymorphisms, a 14-bp insertion/deletion(INDEL) (rs66554220), which may be, among other things,involved in alternative splicing processes, seems to be impor-tant for post-transcriptional regulation. Two previous studieshave assessed the influence of the 14 bp INDEL in the sus-ceptibility to RA, but no association was observed (11, 12).However, considering other autoimmune diseases (ADs), ourgroup has observed a positive association of another 3′-UTRSNP, +3142C>G (rs1063320), with disease susceptibility inpatients with systemic lupus erythematosus (SLE) (13). In thisstudy, we sought to investigate the genetic influence of thesetwo HLA-G 3′-UTR polymorphisms and the combined 3′-UTR haplotype in the susceptibility to RA in a double-centerstudy comprising 546 RA patients from two independentBrazilian cohorts, a Southern-Brazilian cohort from Porto Ale-gre and a Northern-Brazilian cohort from the city of Belem.

Methods

The study sourced samples from two distinct areas in Brazil:the Southern sample comprised 339 RA European-derivedpatients and 299 healthy European-derived controls from theurban region of Porto Alegre, State of Rio Grande do Sul,Southern Brazil, situated at 30.02◦ S and 51.13◦ W. TheNorthern sample comprised 200 RA patients and 190 healthycontrols from the admixed population of Belem, NorthernBrazil, situated at 1.27◦ S and 48.30◦ W, totaling 539 RApatients and 489 controls. Patients from Porto Alegre wereunder the care of the Division of Rheumatology of the Hospi-tal de Clınicas de Porto Alegre (HCPA), whereas the patientsfrom Belem were collected from several medical centers. Allpatients were diagnosed according to the American Collegeof Rheumatology’s criteria for the classification of RA (14).

Patients having another connective tissue disease, otherthan secondary Sjogren syndrome, were excluded. The vastmajority of the patients (520 of 539) were evaluated for thepresence of rheumatoid factor (RF). The study was approvedby the Ethics Committee of all hospitals involved in thestudy, and informed consent was obtained from all patientsand controls.

Polymerase chain reaction amplification of the

+3142C>G and 14 bp polymorphism in the 3′-UTR

of the HLA-G gene and genotyping

Patient and control DNA were isolated from peripheral bloodusing a salting out method (15). The genotyping of 14 bppolymorphism of the HLA-G gene was performed as previ-ously described (16). Briefly, 100 ηg of genomic DNA wasamplified in a 25-ml reaction tube, with final concentrationsas follows: polymerase chain reaction [polymerase chainreaction (PCR) buffer] 1×, dNTP 0.2 mM, MgCl2 1.5 mM,Taq DNA polymerase 1.0 U and 10 pmol of each primer(GE14HLAG – 5′-GTGATGGGCTGTTTAAAGTGTCACC-3′, RGH4 – 5′-GGAAGGAATGCAGTTCAGCATGA-3′).The thermocycling conditions were as follows: 94◦C for2 min; 35 cycles of 94◦C for 30 s, 64◦C for 60 s and 72◦C for60 s; final extension of 72◦C for 10 min. The amplified PCRproducts were visualized in 6% polyacrylamide gel stainedwith ethidium bromide, with the amplicon sizes being 224 bpfor the 14 bp insertion allele and 210 bp for the 14 bp deletionallele.

The PCR of the +3142C>G polymorphism was performedas described (17): 200 ng of genomic DNA was addedto a final volume of 25 μl, with final concentrations asfollows: PCR buffer 1×, 2.0 mM MgCl2; 0.2 mM of eachdNTP; 1.0 U of Taq polymerase and 10 pmol of each primer(GMIRNAF – 5′-CATGCTGAACTGCATTCCTTCC-3′,GMIRNAR – 5′-CTGGTGGGACAAGGTTCTACTG-3′).Thermocycling conditions were as follows: 94◦C for 5 min;32 cycles of 94◦C for 30 s, 65.5◦C for 30 s and 72◦C for 60 sfollowed by a final extension step at 72◦C for 5 min.

The amplified PCR products were cleaved with 3 U ofthe restriction enzyme BaeGI (New England Biolabs, Inc.,Ipswich, MA) according to the manufacturer’s instructions.RFLP products were analyzed by electrophoresis in a 1.5%agarose gel stained with ethidium bromide, with ampliconsizes of 406 bp for the C allele and 316 and 90 bp for theG allele.

Statistical analysis

The +3142C>G and 14 bp genotypic frequencies were com-pared for Hardy–Weinberg (HW) expectations using chi-squared tests. The allelic and genotypic frequencies of theHLA-G polymorphisms of controls and RA patients werecompared using chi-squared or Fisher exact test. When theresults for these tests were statistically significant, the residu-als were analyzed to identify local points of discrepancy.

We used binary and multinomial logistic regression tomodel the relationship between disease variables and the twopolymorphisms of interest. In this study, we used two outcomevariables, the first being dichotomous (RA patient or healthycontrol) and the second one with three possible values – RF-positive (RF+) RA patient, RF-negative (RF−) RA patientand healthy controls. When the latter tri-grouping was used,

© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd 261Tissue Antigens, 2014, 83, 260–266

HLA-G 3′-UTR haplotype in rheumatoid arthritis T. D. Veit et al.

the healthy control group was used as reference category.Polymorphisms were modeled each as two ‘dummy’ variables,indicating heterozygosis and homozygosis for allele 2 (14 bpinsertion for rs66554220 and +3142G for rs1063320), respec-tively. City of sampling, sex and age were used as covariables.For the specific case of age, three ‘dummy’ variables (38–45,46–54 and 55–85 years, with 20–37 years as the baseline cat-egory) were created for the analysis, as the effect of thisvariable on the outcome did not follow a linear pattern inthe logit.

Haplotype frequencies were estimated with the Mlocus soft-ware (18), which uses an expectation maximization algorithm(19). The estimated haplotype frequencies were comparedbetween patients and controls by means of chi-squared tests.Diplotypes were obtained by direct observation of the eightpossible non-ambiguous combinations of genotypes from thetwo polymorphisms, and by imputation in the double het-erozygotes, of haplotype frequencies obtained by the E-Malgorithm in the studied sample. Considering the allelicand diplotype frequencies of the study sample and there-fore the low frequency expected for rare haplotypes, doubleheterozygous individuals were imputed as delC/insG.

Bonferroni correction for multiple comparisons was appliedwhen P -value was significant. The significance level wasset at α = 0.05 (two tailed), and all statistical analyses wereperformed with spss 15.0 (SPSS, Inc., Chicago, IL) andwinPEPI (20).

Results

The mean patient age (±SD) and the mean age at diagnosiswere, respectively, 60.2 ± 12.2 years and 46.1 ± 13.4 years inPorto Alegre and 53.4 ± 14.1 years and 43.8 ± 14.1 years inBelem. The frequency of RA patients positive for RF (RF+)was 0.834 in Porto Alegre and 0.820 in Belem (Table 1).

The observed genotype distributions agreed withHardy–Weinberg equilibrium (HWE) expectations forboth polymorphisms in both cohorts. The polymorphisms

in the 3′-UTR showed linkage disequilibrium (P < 0.001).The values of D ′ between the 14 bp polymorphism andthe +3142G>C in controls and RA patients were 0.948and 0.924 and the values for r2 were 0.511 and 0.479,respectively.

Genotype and allelic frequencies in the total sample, as wellas in each city, are shown in Table 2. A preliminary analysisdid not show differences between the allelic frequencies ofcontrol individuals from Porto Alegre and Belem, whereasthe difference in the +3142G allele frequency almost reachedthe significance level (0.504 vs 0.576, P = 0.057) with Belempresenting the higher value.

We then performed logistic regression models to evaluatethe influence of HLA-G polymorphisms on susceptibilityto RA adjusted for city of sampling, sex and age, ascovariables, in that order. In order to identify a possiblelinear correlation among the genotypes’ odds ratio (OR), thedel/del genotype was used as reference and was comparedto the other genotypes’ OR individually. After controllingfor the covariables, no significant OR was observed, whichsignified an influence of the 14 bp polymorphism in RAsusceptibility (heterozygote genotype: OR = 1.17, P = 0.299;homozygote for ins: 1.23, P = 0.312, compared with thedel/del genotype). When performing the same analysis withthe +3142 polymorphism, however, we observed that beinghomozygous for the G allele was associated with greater RArisk compared with that for the CC genotype (OR = 1.53, 95%CI = 1.03–2.28, P = 0.035), while heterozygous individualsfor the G allele presented a similar risk as CC individuals(OR = 1.08, 95% CI = 0.753–1.545, P = 0.680). A simplifiedmodel including only the GG genotype as a risk factor for RAyielded an OR of 1.45 (95% CI = 1.075–1.959, PBonf = 0.030;Table 3, ‘All patients’). We also evaluated whether theinfluence of this genotype was restricted to a subgroupof patients. For this, we performed a multinomial logisticregression model subdividing the patient groups according tothe positivity to RF, using healthy controls as the referencecategory. In this analysis, the influence of the GG genotype

Table 1 Rheumatoid arthritis (RA) patient and control features

Porto Alegre (South) Belem (North)

Freq. (N) Mean ± SD N Freq. (N) Mean ± SD N

RAFemales 0.811 (275) 339 0.925 (185) 200Age 60.2 ± 12.2 339 53.4 ± 14.1 200Age at diagnosis 46.1 ± 13.4 339 43.8 ± 14.1 200Disease time (years) 14.1 ± 7.9 339 8.5 ± 7.6 200RF positivity 0.834 (281) 337a 0.820 (150) 183a

ControlsFemales 0.418 (125) 299 0.584 (111) 190Age 47.7 ± 11.7 299 43.0 ± 12.9 190

RF, rheumatoid factor.aSome patients had not been tested for RF.



Table 2 Human leukocyte antigen (HLA)-G +3142 genotype and allele frequencies in patients and controls

Controls RA Controls POA RA POA Controls Belem RA BelemN (frequency) N (frequency) N (frequency) N (frequency) N (frequency) N (frequency)

14 bp Genotypedel/del 171 (0.355) 179 (0.333) 105 (0.357) 110 (0.324) 66 (0.351) 69 (0.348)ins/del 233 (0.483) 269 (0.501) 136 (0.463) 170 (0.501) 97 (0.516) 99 (0.500)ins/ins 78 (0.162) 89 (0.166) 53 (0.180) 59 (0.174) 25 (0.133) 30 (0.152)P-valuea 0.772 0.605 0.865

Alleledel 575 (0.596) 627 (0.584) 346 (0.588) 390 (0.575) 229 (0.609) 237 (0.598)ins 389 (0.404) 447 (0.416) 242 (0.412) 288 (0.425) 147 (0.391) 159 (0.402)P-valuea 0.592 0.628 0.741

+3142 GenotypeCC 97 (0.204) 106 (0.199) 67 (0.229) 76 (0.226) 30 (0.163) 30 (0.154)CG 246 (0.517) 251 (0.473) 150 (0.514) 159 (0.473) 96 (0.522) 92 (0.472)GG 133 (0.279) 174 (0.328) 75 (0.257) 101 (0.301) 58 (0.315) 73 (0.374)P-valuea 0.234 0.451 0.477

AlleleC 440 (0.462) 463 (0.436) 284 (0.486) 311 (0.463) 156 (0.424) 152 (0.390)G 512 (0.538) 599 (0.564) 300 (0.504) 361 (0.537) 212 (0.576) 238 (0.610)P-valuea 0.256 0.412 0.356

del, 14 bp deletion; ins, 14 bp insertion; POA, Porto Alegre; RA, rheumatoid arthritis.aFisher exact test comparing controls and respective patient groups.

Table 3 Positive association between rheumatoid arthritis (RA) and RAsubtypes according to RF and +3142 genotypes

OR 95% CI P PBonf

All patients (n = 531)a

CC+CG 1 —GG 1.451 1.075–1.959 0.015 0.030

RF+ RA (n = 425)b

CC+CG 1 —GG 1.542 1.127–2.110 0.007 0.014

RF− RA (n = 88)b

CC+CG 1 —GG 1.442 0.862–2.414 0.163 —

CI, confidence interval; OR, odds ratio; PBonf, P after Bonferronicorrection; RF, rheumatoid factor.aBinary logistic regression, controlled by sex, city and age.bMultinomial logistic regression, controlled by sex, city and age.

remained statistically significant for the RF-positive group(OR = 1.54, 95% CI = 1.127–2.110, PBonf = 0.014), whereasfor the RF-negative group, the OR was similar (1.4, 95%CI = 0.862–2.414; Table 3).

The second step of the analysis involved the investigationof a potential association of HLA-G haplotypes and RA.A significant difference in the haplotype frequencies wasobserved between healthy controls from Porto Alegre andfrom Belem, with an increased frequency of the deletion-G (delG) haplotype in the control population from Belem(0.180 vs 0.112), whereas the frequency of the deletion-C(delC) haplotype was increased in the control population fromPorto Alegre (0.477 vs 0.426) (Table 4). This pattern wasalso observed among RA patients, with significant differences

in the frequencies of the delC and delG haplotypes betweenpatients from the two cities (Table 4). Further comparisonsbetween patients and controls showed some sex differences:in women, the delG haplotype was more frequent in patientsboth in Porto Alegre (0.124 vs 0.073, residual P = 0.029) andBelem: (0.233 vs 0.160, residual P = 0.025), and a higherfrequency of the insG haplotype was observed among malepatients overall compared with all male controls (0.466 vs0.397, residual P = 0.045).

We also analyzed the protection associated with the pres-ence of the delC haplotype by logistic regression, adjustingfor sex, age and city, as we found a decreased frequencyof the delC haplotype in patients compared with controls inboth cities. Double heterozygotes were imputed as delC/insG,as the chance of a delG/insC double heterozygote was verylow, with percentages ranging from 0.3% in controls fromPorto Alegre to 2.1% in cases from Belem (data not shown).The result of this analysis is presented in Table 5: the pres-ence of a delC haplotype yielded a significant OR (0.702,95% CI = 0.521–0.946, PBonf = 0.040), suggesting protectionagainst RA. After stratifying the patients by positivity to RF,only the comparison between controls and RF+ RA patientspresented a significant OR (0.659, 95% CI = 0.482–0.900,PBonf = 0.018), while the OR for RF− RA patients yielded avalue close to 1 (0.962, 95% CI = 0.568–1.631, P = 0.886).

Discussion

In this work, we assessed the influence of two polymorphismslocated at the 3′-UTR of the HLA-G gene in the susceptibilityto RA. Due to the inflammatory nature of this disease and the



Table 4 The estimated human leukocyte antigen (HLA)-G haplotype frequencies in patients and controls

Haplotype Controls RA Controls POA RA POA Controls Belem RA BelemN (frequency N (frequency) N (frequency N (frequency) N (frequency) N (frequency)

AlldelC 447 (0.457) 460 (0.424)b 285 (0.477) 309 (0.451)b 162 (0.426) 151 (0.377)b

delG 136 (0.139)a 176 (0.162)c 67 (0.112)a 87 (0.126)c 69 (0.180)a 89 (0.223)c

insC 8 (0.008) 13 (0.012) 7 (0.012) 7 (0.010) 1 (0.003) 6 (0.014)insG 387 (0.396) 437 (0.402) 239 (0.399) 283 (0.412) 148 (0.391) 154 (0.386)Comparison with Belem P = 0.008 P = 0.0004

FemaledelC 226 (0.479) 402 (0.465) 127 (0.509) 256 (0.465) 99 (0.445) 146 (0.383)delG 53 (0.112)d 157 (0.124)d 18 (0.073)d 68 (0.124)d 35 (0.160)e 89 (0.233)e

insC 6 (0.013) 10 (0.008) 5 (0.018) 5 (0.008) 1 (0.006) 5 (0.014)insG 187 (0.396) 361 (0.403) 100 (0.401) 221 (0.403) 87 (0.390) 140 (0.369)Comparison with Belem P = 0.011 P = 0.0001

MaledelC 221 (0.437) 60 (0.402) 158 (0.454) 51 (0.402) 63 (0.398) 9 (0.300)delG 82 (0.162) 19 (0.113) 49 (0.139) 15 (0.113) 33 (0.209) 4 (0.133)insC 2 (0.004) 2 (0.018) 2 (0.006) 2 (0.018) 0 (0.000) 0 (0.000)insG 201 (0.397)f 77 (0.466)f 139 (0.400) 60 (0.466) 62 (0.392) 17 (0.567)Comparison with Belem P = 0.21 P = 0.69

POA, Porto Alegre; RA, rheumatoid arthritis.aResidual P = 0.002.bResidual P = 0.019.cResidual P > 0.001.dResidual P = 0.029.eResidual P = 0.025.fResidual P = 0.045.

Table 5 delC haplotype carrier frequencies and association with rheumatoid arthritis (RA) and RA subtype susceptibility

All individuals Porto Alegre Belem

delC carriers Non-carriersa delC carriers Non-carriersa delC carriers Non-carriersa OR (95% CI) P PBonfb

Controls 335 (0.711) 136 (0.289) 209 (0.728) 78 (0.272) 126 (0.685) 58 (0.315) — — —RA (all) 352 (0.665) 177 (0.335) 233 (0.693) 103 (0.307) 119 (0.617) 74 (0.383) 0.702 (0.521–0.946)c 0.020c 0.040Controls 335 (0.711) 136 (0.289) 209 (0.728) 78 (0.272) 126 (0.685) 58 (0.315) — — —RF+ RA 276 (0.650) 148 (0.350) 189 (0.680) 89 (0.320) 87 (0.596) 59 (0.404) 0.659 (0.482–0.900)d 0.009d 0.018RF− RA 63 (0.716) 25 (0.284) 42 (0.750) 14 (0.250) 21 (0.677) 11 (0.323) 0.962 (0.568–1.631)d 0.886d —

CI, confidence interval; OR, odds ratio; RA, rheumatoid arthritis; RF, rheumatoid factor.aPossible genotypes: delG/delG, delG/insG, insC/insC, insC/insG, insG/insG.bP after Bonferroni correction (n = 2).cBinary logistic regression taking controls and RA (all patients) as outcomes and adjusting for age, sex and city.dMultinomial logistic regression taking controls, RF+ RA and RF− RA as outcomes and adjusting for age, sex and city.

immunoregulatory features attributed to the HLA-G molecule,the HLA-G gene is a good candidate for association with RA.Although previous studies have assessed the influence of a3′-UTR polymorphism in the susceptibility to RA (14 bp),this is the first one to analyze the influence of the +3142polymorphism and 3′-UTR haplotypes in this disease.

We observed an association between the +3142 GG geno-type and the risk for RA and, conversely, a protective effect ofthe presence of the delC haplotype (especially, for the risk ofRF+ RA). These genotypes were associated with SLE suscep-tibility in the same way in a previous study of our group (13).The results of the analysis are consistent with a recessive effect

of the G allele in RA susceptibility. This makes sense if weconsider that the +3142 SNP is located within a microRNAbinding site and the G allele is associated with a highermicroRNA affinity and thus to a decreased HLA-G expres-sion. Under this assumption, G would be the ‘low producingallele’ and C the ‘high producing allele’ and the presence ofat least one copy of the later allele would be protective againstthe risk of RA. HLA-G expression has been described as apotential prognostic marker in RA, as patients who respondedwell to anti-RA treatment had increased levels of sHLA-Gin blood (21). Also, a higher frequency of the 14 bp deletionallele, considered a high HLA-G producer and preferentially



linked to the +3142C allele, was observed in patients whichreached remission after 12 months of DMARD therapy (21).These observations are consistent with the hypothesis of anincreased susceptibility to RA associated with ‘low producer’alleles.

Although we were unable to associate the risk of RA witha single haplotype, as the delG haplotype is more prevalent infemale patients and insG has a higher frequency among malepatients, these two haplotypes are associated with extendedhaplotype clusters that encompass alleles that are very similarto each other: the HG0104 (delG) and the HG010102 andHG010103 (insG) extended haplotype clusters share 88%identity in the SNPs localized on the HLA-G promoterregion (23/26 SNPs). On the other hand, the delC haplotype,which was associated with protection to RA, is associatedwith a haplotype cluster (HG010101) that shares at mostonly 42% (11/26) identity with the former three haplotypeclusters among the known promoter polymorphic sites (22),characterizing a distinct haplotype lineage. Notably, the HLA-G gene has a low rate of recombination according to Refs (22)and (23) and that UTR-1 (Lineage HG010101a) is associatedwith high sHLA-G expression according to Martelli-Palominoet al. (24). In this perspective, the +3142 SNP could beregarded as a remarkable tag SNP for two different HLA-G haplotype lineages with different expression patterns anddifferential influence in disease.

Despite the known correlation between the 14 bp polymor-phism and the HLA-G expression levels (mainly in blood),and the observed correlations between high HLA-G levelsand lower severity disease parameters, until now no associ-ation between the 14 bp polymorphism and susceptibility toRA was reported. In this sense, we suggest that the +3142SNP could be a better marker for disease susceptibility thanthe 14 bp polymorphism.

Concluding, we have described an association between aHLA-G 3′-UTR SNP (+3142) with susceptibility to RA.The GG genotype was shown to be a risk factor for thedisease. The G allele is associated with haplotype lineagesthat share high identity and are regarded as low producers. Thepresence of the G allele in homozygosis could be responsiblefor a low HLA-G expression profile which could favor thetriggering of RA. Under a broader view, it has been previouslyhypothesized that many ADs might share a common geneticorigin, with some polymorphisms being common risk factorsfor the development of ADs, while others being disease-specific (25). The fact that the same +3142G genotype (GG)is associated with the susceptibility to both SLE and RA, tworelated ADs, suggests an important involvement of the HLA-Gmolecule in the pathogenesis of this group of diseases.

Acknowledgments

This work was supported by the Conselho Nacional de Desen-volvimento Cientifico e Tecnologico (CNPQ) and Fundacao

de Amparo a Pesquisa do Estado do Rio Grande do Sul(FAPERGS).

Conflict of interest

The authors have declared no conflicting interests.

References

1. MacGregor A, Ollier W, Thomson W, Jawaheer D, Silman A.HLA-DRB1*0401/0404 genotype and rheumatoid arthritis:increased association in men, young age at onset, and diseaseseverity. J Rheumatol 1995: 22: 1032–6.

2. Perricone C, Ceccarelli F, Valesini G. An overview on thegenetic of rheumatoid arthritis: a never-ending story.Autoimmun Rev 2011: 10: 599–608.

3. Riteau B, Menier C, Khalil-Daher I et al. HLA-G1co-expression boosts the HLA class I-mediated NK lysisinhibition. Int Immunol 2001: 13: 193–201.

4. Riteau B, Menier C, Khalil-Daher I et al. HLA-G inhibits theallogeneic proliferative response. J Reprod Immunol 1999: 43:203–11.

5. Bahri R, Hirsch F, Josse A et al. Soluble HLA-G inhibits cellcycle progression in human alloreactive T lymphocytes.J Immunol 2006: 176: 1331–9.

6. LeMaoult J, Caumartin J, Daouya M et al. Immune regulationby pretenders: cell-to-cell transfers of HLA-G make effector Tcells act as regulatory cells. Blood 2007: 109: 2040–8.

7. Caumartin J, Favier B, Daouya M et al. Trogocytosis-basedgeneration of suppressive NK cells. EMBO J 2007: 26:1423–33.

8. Horuzsko A, Lenfant F, Munn DH, Mellor AL. Maturation ofantigen-presenting cells is compromised in HLA-G transgenicmice. Int Immunol 2001: 13: 385–94.

9. Carosella ED, HoWangYin K-Y, Favier B, LeMaoult J.HLA-G-dependent suppressor cells: diverse by nature, function,and significance. Hum Immunol 2008: 69: 700–7.

10. Castelli EC, Moreau P, Oya e Chiromatzo A et al. In silicoanalysis of microRNAS targeting the HLA-G 3′ untranslatedregion alleles and haplotypes. Hum Immunol 2009: 70: 1020–5.

11. Veit TD, Vianna P, Scheibel I et al. Association of the HLA-G14-bp insertion/deletion polymorphism with juvenile idiopathicarthritis and rheumatoid arthritis. Tissue Antigens 2008: 71:440–6.

12. Rizzo R, Rubini M, Govoni M et al. HLA-G 14-bppolymorphism regulates the methotrexate response inrheumatoid arthritis. Pharmacogenet Genomics 2006: 16:615–23.

13. Consiglio CR, Veit TD, Monticielo OA et al. Association of theHLA-G gene +3142C>G polymorphism with systemic lupuserythematosus. Tissue Antigens 2011: 77: 540–5.

14. Arnett FC, Edworthy SM, Bloch DA et al. The AmericanRheumatism Association 1987 revised criteria for theclassification of rheumatoid arthritis. Arthritis Rheum 1988: 31:315–24.

15. Lahiri DK, Nurnberger JI. A rapid non-enzymatic method forthe preparation of HMW DNA from blood for RFLP studies.Nucleic Acids Res 1991: 19: 5444.



16. Hviid TV, Hylenius S, Hoegh AM, Kruse C, Christiansen OB.HLA-G polymorphisms in couples with recurrent spontaneousabortions. Tissue Antigens 2002: 60: 122–32.

17. Cordero EA, Veit TD, da Silva MA, Jacques SM, Silla LM,Chies JA. HLA-G polymorphism influences the susceptibilityto HCV infection in sickle cell disease patients. Tissue Antigens2009: 74: 308–13.

18. Long JC. Multiple Locus Haplotype Analysis , 3.0 edn. Softwareand documentation distributed by the author. Bethesda, MD,Section on Population Genetics and Linkage, Laboratory ofNeurogenetics, NIAAA, National Institutes of Health, 1999.

19. Long JC, Williams RC, Urbanek M. An E-M algorithm andtesting strategy for multiple-locus haplotypes. Am J Hum Genet1995: 56: 799–810.

20. Abramson JH. WINPEPI (PEPI-for-Windows): computerprograms for epidemiologists. Epidemiol Perspect Innov 2004:1: 6.

21. Rizzo R, Farina I, Bortolotti D et al. HLA-G may predict thedisease course in patients with early rheumatoid arthritis. HumImmunol 2013: 74: 425–32.

22. Castelli EC, Mendes-Junior CT, Veiga-Castelli LC, Roger M,Moreau P, Donadi EA. A comprehensive study of polymorphicsites along the HLA-G gene: implication for gene regulationand evolution. Mol Biol Evol 2011: 28: 3069–86.

23. Santos KE, Lima TH, Felıcio LP et al. Insights on the HLA-GEvolutionary history provided by a nearby Alu insertion. MolBiol Evol 2013: 30: 2423–34.

24. Martelli-Palomino G, Pancotto JA, Muniz YC et al. Polymor-phic sites at the 3′ untranslated region of the HLA-G gene areassociated with differential HLA-G soluble levels in theBrazilian and French population. PLoS One 2013: 8: e71742.

25. Anaya J-M, Gomez L, Castiblanco J. Is there a commongenetic basis for autoimmune diseases? Clin Dev Immunol2006: 13: 185–9.


Documents

HLA-G +3142 polymorphism as a susceptibility marker in two rheumatoid arthritis populations in Brazil