Variability at the 3′ untranslated region of the HLA-G gene: a study on patients with AIDS and cytomegalovirus retinochoroiditis

Cytomegalovirus retinochoroiditis (CMV-R) is the primary cause of blindness among AIDS patients. Since HLA-G is associated with the modulation of the immune response, we hypothesized that variability at the 3′ untraslated region (3′UTR) of the gene could be implicated on the predisposition to CMV-R. We evaluated whether HLA-G 3′UTR influences CMV-R development in Brazilian AIDS patients. Peripheral blood DNA was obtained from two groups of patients: (1) AIDS exhibiting CMV-R (n = 40) and (2) AIDS without CMV-R (n = 147). HLA-G 3′UTR typing was performed using sequencing analysis. Allele, genotype and haplotype frequencies were evaluated using Fisher’s exact test accompanied by the calculation of the odds ratio and its 95% confidence interval (95% CI). The etiologic (EF) and preventive fractions were also estimated. Compared to AIDS patients without CMV-R, AIDS patients with CMV-R showed increased frequencies of the: (1) + 3001T allele, (2) the + 3001C/T genotype and (3) the UTR-17 (InsTTCCGTGACG) haplotype (EFs = 0.02–0.04). The UTR-3 (DelTCCGCGACG) haplotype was associated with protection against CMV-R development. Although the risk for developing CMR-V at the population level was relatively low (EF), the identification of HLA-G 3′UTR variation sites may help to further evaluate the role of post-transcriptional factors that may contribute to the existent immunosuppresion caused by HIV per se.

Although studies regarding the classical HLA class I and class II profiles of patients with AIDS exhibiting ocular manifestations have been reported so far [17][18][19] , there is no information regarding the study of the nonclassical HLA-G gene in patients exhibiting both AIDS and CMV-R. The present study was undertaken based on the hypothesis that variation sites at the HLA-G 3′UTR in patients with AIDS exhibiting or not CMV-R may discriminate variation sites associated with the regulation of the immune checkpoint gene.
The frequency of the HLA-G 3′UTR alleles is shown in Table 1. Only the + 3001T allele was overrepresented in CMV-R AIDS patients compared to AIDS without CMV-R (P = 0.046), conferring an OR 18.631 (95% CI 0.885-392.043) and an Etiologic Fraction = 0.024.

Discussion
Due to the systemic immunodepression, patients with AIDS are susceptible to opportunistic pathogens, and one of the most frequent is CMV that, in approximately 85% of cases, induces CMV-R, particularly in patients presenting other risk factors (e.g. CD 4 + cells count < 50). Despite the broad availability of cART in many countries, 4.9-7.6% of the patients are believed to develop CMV-R during some evolution stages of AIDS [20][21][22][23][24] .
The individual variability regarding the time of onset and the rate of progression of CMV-R suggest that genetic factors such as histocompatibility alleles and HLA specific haplotypes may be involved in disease susceptibility. The study of genetic susceptibility in infectious diseases usually poses a challenge because most studies compare infected patients with healthy controls, for whom one cannot predict whether or not the individual would be susceptible to or resistant to the infectious agent. In this study, both groups were infected by HIV and developed AIDS, and just one developed CMV-R. Noteworthy, the CMV infection is present in 90% of the healthy Brazilian individuals 24 and in about 98% of the patients with AIDS 25 . Because the previous contact with CMV in Brazilian patients with AIDS is extremely frequent, the development of CMV-R was the dependable variable.
Considering that low CD4 cell count itself propitiate an immunosuppressive condition to develop CMV-R 17 , and also considering that HLA-G has immunosuppressive properties, the study of the regulatory region of the HLA-G gene is of relevance to predict the differential chance of ocular involvement in patients with AIDS. No literature data have been reported regarding the variability at the 3′UTR of the HLA-G gene in patients with AIDS, exhibiting or not CMV-R. Noteworthy, in the present series, just one variable site was associated with CMV-R when this group of AIDS patients was compared with AIDS patients without CMV-R. Accordingly, the + 3001T allele, the + 3001C/T genotype, and the UTR-17 (InsTTC CGT GACG) haplotype contains the + 3001T Table 2. Frequency and absolute number of the HLA-G 3′UTR genotypes observed in patients with AIDS exhibiting CMV retinitis (CMV-R: Group I) and in patients with AIDS without CMV-R (Group II).  www.nature.com/scientificreports/ allele, and they were all associated with susceptibility to CMV-R. The + 3001C/T variable site is uncommon in healthy Brazilian individuals, but has been encountered in African populations 24 . This allele was also associated with infectious diseases, including leprosy 4 . Although some polymorphic sites at the HLA-G 3′UTR (14 bp Ins/Del, + 3142C/G and + 3187A/G) have been functionally studied 11,13 , the + 3001C/T has not, and in analogy to other variation sites it may be a differential target for microRNA binding. In addition to the low frequency of + 3001C/T, the etiologic fractions conferred by the + 3001T allele, + 3001C/T genotype and UTR-17 haplotype were also low, varying from 0.02 to 0.04, thus contributing to 2-4% of the CMV-R susceptibility at the population level. This result indicates that the + 3001C/T variation may act in concert with other variable sites at HLA-G 3′UTR or surrounding variants. Indeed, the + 3001T allele is in linkage disequilibrium with the 14 bp Insertion, + 3142G and the + 3187A alleles, which are included at the HLA-G UTR-17 haplotype, and these variable sites have individually been associated with decreased production of HLA-G 7,11,13 .
Considering the 5′UTR segment, UTR-17 is associated with the HLA-G promoter PROMO-0103e (0103 group), which contains variation sites that diverge from the most common promoter haplotypes (Promo-010101 and PROMO-010102 groups), particularly at positions − 725 (a T nucleotide compared to C or G nucleotides in other haplotypes), − 509 (a G nucleotide compared to a C nucleotide in other haplotypes) and − 56 (a T nucleotide compared to a C nucleotide in other haplotypes) 5 . In contrast to susceptibility alleles, the + 3001C/C and the UTR-3 haplotype (exhibiting the + 3001C allele) were more robustly associated with protection against CMV-R, conferring a preventive fraction of 0.945 and 0.116, respectively. Previous studies from our group have reported that UTR-3 has been associated with median levels of HLA-G production 26 . Concluding, considering that the rare + 3001T allele appeared in few patients with AIDS and CMV-R, and considering the large variation of the confidence interval including the number 1, only the study of a large sample of patients would ascertain the role of the + 3001T allele on the susceptibility to CMV-R in patients with AIDS.
Besides HLA-G, other genes of the Major Histocompatibility Complex have been previously associated with CMV-R, including: (1) the overrepresentation of TNF microsatellites (TNFc2 alleles, as well as the 4-1-G-2-2-1 haplotype, which contains the TNFc2 allele) in patients with AIDS, inducing a rapid progression of the disease as well as the development of accelerated CMV-R 27 , (2) the HLA-A31 antigen was associated with the development of AIDS, irrespective of the presence or not of CMV-R 17 , (3) the HLA-C*07 allele group was assoaciated with protection against the development of CMV-R when compared to patients with AIDS without CMV-R. In addition, the HLA-C*05 allele group was associated with susceptibility to and the HLA-C*16 allele group with protection against AIDS development, irrespective of the presence or not of CMV-R 19 , and (4) the HLA-B35 antigen was associated with the rapid progression of AIDS and development of retinochoroidites induced by Toxoplasma gondii 18 .
The classical HLA class I molecules (HLA-A, -B and -C) play a major role on the defense against viruses by presenting virus-derived peptides to cytotoxic TCD8 cells, and the lack of classical class I molecules on the surface of virus infected cells provides a signal for the cytotoxic action of NK cells. In this aspect, HLA-G may inhibit the cytotoxic activity of both CD8 and NK cells, facilitating virus escape from the host immune response and permitting virus spread 6,28 . Although HLA-A, HLA-B and HLA-C form common extended haplotypes due to the linkage disequilibrium among these genes, the most striking linkage disequilibrium of the HLA-G gene is with Table 3. Frequency and absolute number of the HLA-G 3′UTR haplotypes observed in patients with AIDS exhibiting CMV retinitis (CMV-R: Group I) and in patients with AIDS without CMV-R (Group II). ) 3001 3003 3010 3027 3032 3035 3142 3187 3196 3227 Group 1 Group II Odds ratio, etiologic fraction (EF) or preventive fraction (PF) P value  (14) UTR-10  (2) UTR-50 www.nature.com/scientificreports/ the HLA-A gene 29,30 , and the typing of both classical and non-clasical class I genes is important to understand the association of HLA genes with virus disorders.

HLA-G haplotype 2960 (14 bp
In this study, we reported that susceptibility to or protection against CMV-R in patients with AIDS is associated with specific variation sites at the HLA-G 3′UTR, and these findings deserve further studies to evaluate specific post-transcriptional factors that may target these gene segments, and consequently the differential production of HLA-G. The genetic susceptibility to produce differential levels of immunomodulatory molecules such as HLA-G, in concert with the immunosuppression caused by the HIV infection, may further influence host immunosurveillance, propitiating the concomitant appearance of other viral infections, including CMVinduced complications.

Methods
All investigation was performed following the guidelines of the National Committee of Ethics in Research (CONEP) Brazil. The study was submitted to the Ethics Committee of the University Hospital of the Ribeirão Preto Medical School, demanding the use of samples stored at a Bank of Biological Samples (protocols # 1525/1998, #8992/2001 and #7581/2007). The Ethics Committee approved the use of the stored samples, waiving the informed consent (protocol #4084/2009). All additional patients who were further included in the study signed informed consent.
The samples were obtained from two groups of individuals: Group I (n = 40) patients with AIDS and CMV-R diagnosed according to current criteria; and Group II (n = 147), patients with AIDS diagnosed according to current criteria and without CMV-R diagnosis, regardless of the evolutionary phase of the latter or of whether or not they had started antiviral treatment.
DNA was obtained using a salting-out procedure. The 3′UTR of the HLA-G gene was amplified by the polymerase chain reaction (PCR) using with two primers: HLA-G 3′UTR alleles and genotypes were estimated by direct counting using the GENEPOP 3.4 software 31 . Linkage disequilibrium was evaluated using the ARLEQUIN v.3.1 32 software and adherence to Hardy-Weinberg equilibrium was determined by the Guo and Thompson test 33 using the GENEPOP 3.4 software 31 . Since the base sequencing in the gamete phase was unknown, the haplotypes were inferred using the PHASE 34 and Expectation-Maximization 35 methods.
The allele, genotype and haplotype frequencies were compared between groups by Fisher's exact test with the calculation of the Odds Ratio (OR) and 95% confidence interval (95% CI). The level of significance was set at P ≤ 0.05. We also estimated the etiologic fraction that indicates the contribution of each allele, genotype, or haplotype to the susceptibility to CMV-R, and the preventive fraction, that indicates how much these factors contribute to protection against the development of CMV-R.