A 3'-untranslated region polymorphism in the TBX21 gene encoding T-bet is a risk factor for genital herpes simplex virus type 2 infection in humans

Genital infection caused by herpes simplex virus type 2 (HSV-2) is the most common sexually transmitted ulcerative disease worldwide. Up to 25 % of the adult Swedish population is affected by genital HSV-2 infection (Forsgren et al., 1994; Persson et al., 1995). The clinical effects of HSV-2 infection range from asymptomatic infection to severe and recurrent episodes of genital and non-genital symptoms. The mechanism(s) underlying these different disease outcomes is not known, but high gamma interferon (IFN-γ) responses (Eriksson et al., 2004), previous HSV-1 infection (Langenberg et al., 1999), certain HLA alleles (Lekstrom-Himes et al., 1999), certain variants of the gene encoding Toll-like receptor 2 (TLR2; Bochud et al., 2007) and high levels of mannan-binding lectin (Gadjeva et al., 2004) increase the likelihood of an asymptomatic infection.

We recently showed that the Th1-inducing transcription factor T-bet is essential for both innate and the acquired immunity against genital infection with HSV-2 in mice (Svensson et al., 2005). T-bet was found to be especially important for natural killer cell and CD4⁺ T-cell-mediated responses to HSV-2 (Svensson et al., 2005), the immune responses of fundamental importance for the control of HSV-2 infection (Augenbraun et al., 1995; Biron et al., 1989). In essence, our study showed that natural killer cells from mice deficient in T-bet had severely reduced functional effects, including significantly impaired cytotoxic ability and significantly decreased IFN-γ production. In addition, the CD4⁺ T-cell-mediated IFN-γ responses to HSV-2 were impaired in T-bet-deficient animals, which prevented the development of a protective immune response to HSV-2 (Svensson et al., 2005).

Although the role of T-bet has been studied in a variety of diseases in mice, its relative importance in human infectious diseases is less well characterized. We therefore set out to evaluate the possible role of gene polymorphisms in the TBX21 gene encoding T-bet in susceptibility to, and severity of, genital HSV-2 infection in humans. We selected five of the 40 known human TBX21 polymorphisms based on them being (i) spread out over the whole gene, (ii) reasonably common among the Caucasian population, and (iii) not in 100 % disequilibrium according to their frequency reported by the National Center for Biotechnology Information (NCBI), and compared their frequency in 159 HSV-2-infected individuals and 186 HSV-2-negative control individuals. We showed that a single-nucleotide polymorphism (SNP) in the TBX21 gene is associated with the incidence, but not severity, of HSV-2 infection.

HSV-2-infected individuals and uninfected controls.
The HSV-2-infected group consisted of 159 individuals, 55 % males and 45 % females with a mean age of 38 years (range 20–70), recruited from the sexually transmitted disease (STD) clinics at Sahlgrenska University Hospital, Borås Hospital and Uddevalla Hospital, Sweden. Of these patients, 141 were Swedish Caucasians. 16 were non-Swedish Caucasians and two were non-Swedish and not Caucasians. Removal of the 18 individuals of non-Swedish ancestry from the analysis did not significantly change the P values and did not affect our conclusions.

HSV-2 infection was confirmed serologically by ELISA and Western blot (see below). The patients were divided into two groups based on their clinical status as follows.

The symptomatic HSV-2 infection group comprised 105 patients: 52 % males (mean age: 38 years, range: 23–70) and 48 % females (mean age: 39 years, range: 24–62), with a typical history of recurrent genital herpes. In addition to serology and clinical recurrences, symptomatic HSV-2 infection was also confirmed by PCR. The aim was to recruit individuals with more than six relapses year^–1 to ensure clinical disease; 81 of the 105 symptomatically infected individuals fulfilled this criterion, whilst the remaining 24 had fewer than six relapses year^–1. Thirty-nine of the symptomatic individuals were on antiviral treatment at the time of sampling and 63 had not received any treatment for HSV-2 infection (data regarding treatment were missing for three individuals) (Table 1).

Table 1. Clinical overview of the HSV-2-infected individuals

The asymptomatic HSV-2 infection group comprised 54 patients: 61 % males (mean age: 40 years, range: 21–66) and 39 % females (mean age: 34 years, range: 20–52). Asymptomatic patients were recruited from an ongoing screening study of HSV-2 infection in visitors to the STD clinics and among partners of HSV-2-infected patients. All had been given detailed information about the clinical spectrum of herpes and had been interviewed about genital symptoms. Presumed asymptomatic HSV-2-seropositive patients who, after this information, admitted to having genital symptoms were excluded from the study.

Of the 159 HSV-2-infected individuals, eight (six women and two men) were partners to HSV-2 individuals who were already included in the study; seven of these were asymptomatic and one was symptomatic. Furthermore, three individuals (all symptomatic) were on immunosuppressive treatment (against rheumatoid arthritis, Bechterew's disease and central nervous system vasculitis) at the time of sampling. Other diseases that were present among the study population were hepatitis B (one symptomatic individual), diabetes (two symptomatic), allergic asthma (one symptomatic), multiple sclerosis (one symptomatic) and lichen sclerosis (one asymptomatic). All individuals included in the study were screened for HSV-1 co-infection: 82 (52 symptomatic and 30 asymptomatic individuals) of the HSV-2-infected individuals were infected with HSV-1 and 75 (51 symptomatic and 24 asymptomatic individuals) were HSV-1 negative (data were missing for two individuals) (Table 1).

Permission for this study was granted by the Ethics Committee of the University of Gothenburg, Sweden, and all volunteers gave informed consent.

For the control group, 186 healthy HSV-2-negative adult blood donors, who were all self-reported Swedish Caucasians, were recruited from the Blood Bank at Sahlgrenska University Hospital, Sweden. This group consisted of 56 % males and 44 % females with a mean age of 39 years (range 19–65). All individuals were screened for HSV-1 and HSV-2 infection by ELISA; 94 of the healthy controls were positive for HSV-1 and 92 were HSV-1 negative. All individuals were routinely screened (and found to be negative) for blood-derived contaminating diseases, including hepatitis A and B, human immunodeficiency virus types 1 and 2, and human T-lymphotropic virus types 1 and 2.

Genotyping.
DNA used for genotyping was extracted from heparinized venous blood using a salting-out method (Aldener-Cannava & Olerup, 1996). Genotyping was performed for the 159 HSV-2-infected individuals and 186 healthy controls for five TBX21 reference SNPs: rs4794067 (T→C), rs2240017 (C→G), rs11652969 (G→A) (all Applera pre-made assays), rs11650354 (C→T) and rs17244587 (G→A) (both customized Applera assays), by TaqMan allelic discrimination at the Core Facility at the Sahlgrenska Academy, University of Gothenburg, Sweden.

ELISA for the detection of HSV-1- and HSV-2-specific antibodies.
Plasma from HSV-2-infected and uninfected individuals was screened for HSV-1 antibodies using an HSV-1 ELISA kit according to the manufacturer's instructions (HerpesSelect1 ELISA IgG; Focus Technologies). Plasma from HSV-2-uninfected individuals was screened for HSV-2 glycoprotein G (gG) antibodies using an HSV-2 ELISA kit according to the manufacturer's instructions (HerpesSelect2 ELISA IgG; Focus Technologies) and plasma from HSV-2-infected individuals was screened for membrane-anchored mature gG-2 (mgG-2)-specific antibodies using an ELISA as described previously (Tunback et al., 2003).

Western blotting.
Lysates of HSV-2 (strain B4327UR)-infected Hep-2 cells were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Membrane strips were washed in 0.3 % Tween 20 diluted in Tris-buffered saline (TBS-Tween, pH 7.5) for 30 min at 37 °C, followed by 1 ml blocking buffer (3 % powdered milk and 4 % fetal calf serum in TBS) for an additional 30 min at room temperature. Serum samples (10 µl) were added to each strip and incubated for 2 h. The strips were washed three times in TBS-Tween and horseradish peroxidase (HRP)-labelled anti-human IgG (Dako) diluted 1 : 100 in blocking buffer was added for 1 h at room temperature. After washing twice in TBS-Tween and once in TBS, the strips were developed. The substrate solution consisted of 30 µl H₂O₂ diluted in 50 ml TBS and mixed with 30 mg HRP substrate diluted in 20 ml methanol. Development was stopped in SuperQ water. A serum was considered to contain anti-HSV-2 antibodies if there was a reaction to one or more of the three bands representing mgG-2, the high-mannose precursor gG-2 or the C-terminal intermediate (Liljeqvist et al., 2002). All samples that were positive in the HSV-2-specific ELISA (see above) were also found to be positive in the Western blot.

Statistical analysis.
Genotype and allele frequencies were compared using a χ² test. Differences were considered to be significant for values of P<0.05. Calculations were done using StatView, version 5.0 (SAS Institute). Odds ratios were calculated using an online device (). A Hardy–Weinberg test and haplotype reconstruction with an association test were performed using Haploview. To test the probability of finding false positives, a permutation test was performed for the whole set of markers and haplotypes with 10 000 permutations.

A polymorphism at rs17244587 in TBX21 is associated with the incidence of genital HSV-2 infection
To assess gene polymorphism association with susceptibility to genital HSV-2 infection, we compared the frequency of genotypes and haplotypes for five TBX21 gene polymorphisms, rs4794067, rs2240017, rs11652969, rs11650354 and rs17244587 (Fig. 1), in 159 HSV-2-infected individuals and 186 healthy control individuals. All genotypes analysed were found to be in Hardy–Weinberg equilibrium. In a few cases, we were unable to determine the exact genotype due to technical problems relating to inadequate quality of the DNA.

(36K): Fig. 1. Linkage disequilibrium map showing the five SNPs in the TBX21 gene that were analysed for their association with HSV-2 infection (r2 format). D' is the normalized linkage disequilibrium coefficient; LOD is the logarithm of odds.

For four of the five polymorphisms, no differences in frequency could be seen between HSV-2-infected and healthy individuals (Table 2). However, rs17244587 yielded a strong correlation between HSV-2 infection and the A allele variant (P=9.3x10^–8). This polymorphism was located in the 3'-untranslated region (3'UTR) of TBX21. Six per cent of the HSV-2-infected individuals were homozygous for this variant compared with none of the healthy individuals, and 27 % of the HSV-2-infected group were heterozygous compared with only 11 % of the healthy control group (P=3x10^–6 for genotype frequency; P=3x10^–6 for allelic frequency before and after permutation test; Table 2).

Table 2. Genotype frequencies in HSV-2-infected individuals and healthy controls for the five SNPs investigated in the TBX21 gene

Seven different haplotypes with a frequency of >1 % were found for the five SNPs analysed out of 32 expected, and five of these comprised 95 % of the chromosomes. This is probably due to relatively high linkage disequilibrium (LD) at this genetic locus. However, a more careful analysis of the LD map showed LD only for three SNPs (1, 3 and 4), whilst SNPs 2 and 5 were in perfect equilibrium with all of the other SNPs (Fig. 1). One of the seven combinations, haplotype 4 (TCGCA), with a relatively low frequency in HSV-2-infected individuals (estimated frequency 14 %), was found to be significantly overrepresented among HSV-2-infected individuals compared with healthy controls (2.9 %; P=3.1x10^–7) before and after the permutation test (Table 3). This haplotype contains the A allele of SNP 5, which obviously explains the difference. On the other hand, the most common haplotype, haplotype 1 (TCGCG), was more common among the healthy control individuals (P=0.03; Table 3). The latter difference was, however, not significant after the permutation test (empiric P=0.16), which implies that this most likely represented a false-positive effect.

Table 3. Haplotype frequencies in HSV-2-infected individuals and healthy controls for the five SNPs in the TBX21 gene

No correlation between TBX21 polymorphisms and HSV-2 disease severity
To evaluate whether any of the polymorphisms had a role in the severity of genital HSV-2 infection, we compared the five polymorphisms in symptomatic-infected patients and in asymptomatic-infected individuals. No significant difference was found between these groups in any of the polymorphic sites investigated, nor was any haplotype difference found between these two groups (data not shown).

Furthermore, the polymorphism at rs17244587 could not be correlated with the severity of disease with respect to the number of relapses for symptomatically infected individuals. The AA genotype was found in 7 % of individuals with more than six relapses year^–1, in 5 % of individuals with less than six relapses year^–1 and in 5 % of the asymptomatic individuals (corresponding to a frequency of 6 % for the whole HSV-2-infected group). In addition, the allelic distribution among the heterozygous AG individuals and the homozygous GG individuals was comparable to the whole group (Table 4).

Table 4. Genotype frequencies in rs17244587 in HSV-2-infected individuals depending on severity of infection

No correlation between the TBX21 rs17244587 polymorphism and HSV-1 co-infection
To evaluate whether any of the genotypes at the rs17244587 site correlated with HSV-1 infection, all individuals (both HSV-2 patients and healthy controls) were assessed for antibodies against HSV-1. We found that the distribution of genotypes at the TBX21 rs17244587 position among HSV-1-infected and -uninfected individuals was comparable to the genotype distribution between HSV-2-infected individuals and healthy control individuals (data not shown). Therefore, no correlation could be found between HSV-1 co-infection and rs17244587. In this study, we showed that genetic variation in the TBX21 gene encoding the transcription factor T-bet, specifically the SNP rs17244587, is strongly associated with the incidence of genital HSV-2 infection. A relatively rare allele with an A instead of a G in this position was significantly more common among HSV-2-infected individuals compared with uninfected age- and sex-matched controls, and a homozygous AA genotype was found only among HSV-2-infected subjects. However, the rs17244587 polymorphism was not associated with disease severity, as no differences were observed between patients with recurrent disease and those with an asymptomatic, silent infection.

We have, for the first time, been able to link a polymorphism in the TBX21 gene to the incidence of an infectious disease, in this case HSV-2. Several studies have implied genetic risk factors in the susceptibility to and pathogenesis of HSV-2 infection. The variations observed affect the host immune response through (i) variations in the TBX21 gene (as shown in this study), which are associated with the incidence of HSV-2 infection, (ii) variations in the gene encoding TLR2 (Bochud et al., 2007), which affects the degree of HSV-2 shedding and lesion formation, and (iii) the specificity of certain HLA alleles (Lekstrom-Himes et al., 1999), which affects the clinical status of infected individuals. Thus, the individual composition of variations in immune-related genes can influence both the susceptibility to HSV-2 infection and the outcome of the concomitant disease. When discussing the genetic influence of HSV-2 susceptibility, differences in virus exposure among the study population should be taken into consideration, as we do not know whether (and to what extent) the volunteers in the uninfected control population have been exposed to HSV-2. We can, however, conclude that variations in the TBX21 rs17244587 site are associated with the incidence of HSV-2 infection.

Variations in the TBX21 gene have previously been associated with inflammatory diseases of presumed non-infectious origin. Raby et al. (2006) showed, in a family study, an association between asthma and a SNP upstream of the TBX21 gene, but no association with rs17244587 (designated c.2122 in that paper) was found. An upstream polymorphism in TBX21, as well as the rs17244587 polymorphism, is also associated with airway hyper-responsiveness in asthmatic children (Raby et al., 2006). Yet another TBX21 variation has been implicated in the incidence of type 1 diabetes (Sasaki et al., 2004). Interestingly, all previously found associations between a disease and rs17244587 or other TBX21 SNPs have been fairly moderate in comparison with that found for rs17244587 and HSV-2 infection.

There was no difference in allelic frequency of the rs17244587 polymorphism between HSV-1-infected and HSV-1-seronegative individuals. This is surprising as HSV-1 and HSV-2 are very similar and require a robust Th1 immune response for their containment. However, HSV-1 infection is often established during the first year of life (Tunback et al., 2003) when the child's immune system is still immature and Th2 biased (Wilson & Kollmann, 2008), whereas HSV-2, being sexually transmitted, is usually acquired after puberty when the immune system is fully mature and more prone to Th1 responses. Thus, variations in the TBX21 gene might not influence HSV-1 acquisition in young children due to their inherent Th2 bias.

The allelic frequency of the A allele in rs17244587 in Caucasians in two previous studies was 7.1 and 9.5 % (Raby et al., 2006; Ylikoski et al., 2004). According to the NCBI database, the frequency of this variation in 120 Caucasians of European descent is around 12 %. This is considerably lower than in our HSV-2-infected group (19 %), but somewhat higher than in our control group (5.3 %), which may well reflect differences in allelic frequencies in different populations. Together, these observations suggest that the TBX21 gene product T-bet is an important factor affecting the incidence of several different human diseases. Furthermore, the accumulated data suggest that distinct TBX21 variations are associated with either anti-infectious or overtly inflammatory diseases. The strong correlation between the A allele at the rs17244587 site and HSV-2 infection, together with our observation that a haplotype in TBX21 containing the A allele variant also showed a strong association with HSV-2 infection, further indicate that this variation in TBX21 rs17244587 increases the incidence of HSV-2 infection and possibly also susceptibility to HSV-2. Due to the lack of LD between rs17244587 and other SNPs at this locus (r²≤0.02), it is not surprising that only one haplotype with the rare A allele of rs17244587 showed such an association. However, we cannot exclude the influence of unknown variations linked to the rs17244587 SNP.

An interesting question is, of course, the functional consequence of the rs17244587 SNP, in particular in relation to the increased frequency of the A allele at this position among HSV-2-infected individuals. Although it is premature to imply any function without proper experimental proof, one may speculate that the location of this SNP in the non-translated 3'UTR could affect the stability of the TBX21 mRNA and subsequently influence downstream activity related to T-bet function. Such effects on mRNA stability by single-nucleotide changes in the 3'UTR were recently demonstrated for several other genes, including thymidylate synthase (Pullmann et al., 2006), RET (Griseri et al., 2007), CD24 (Wang et al., 2007), NCALD (Kamiyama et al., 2007) and CPB2 (Boffa et al., 2008), and such variations are considered to be one of the possible mechanisms whereby non-coding sequences may affect downstream events. In the case of the rs17244587 SNP, it is therefore tempting to speculate that an A allele in this position reduces the TBX21 mRNA stability and thus the translation of T-bet, leading to reduced IFN-γ-driven cell-mediated immune responses and thereby an increased susceptibility to HSV-2 infection.

Our study investigating the association between SNPs in the TBX21 gene and the incidence of HSV-2 infection was based on a comparison of a well-defined group of patients with gender- and age-matched HSV-2-seronegative healthy controls. This matching is reasonable when designing an association study such as this. Furthermore, it reflects the requirement for stringent selection criteria for a control group, in particular when analysing a common disease with a low SNP allele penetrance, where the general population represents a mixture of carriers. Nevertheless, our results must be taken with caution due to the obvious limitations in the study design using DNA samples from a single population, and a validation of our finding in different populations and ethnical groups is warranted.

In summary, we have shown that genetic variation in the TBX21 gene encoding the transcription factor T-bet, specifically the SNP rs17244587, is strongly associated with the incidence of genital HSV-2 infection. A relatively rare allele with an A instead of a G in this position was significantly more common among HSV-2-infected individuals compared with uninfected age- and sex-matched controls, and a homozygous AA genotype was found only among HSV-2-infected subjects. However, the rs17244587 polymorphism was not associated with disease severity, as no differences were observed between patients with recurrent disease and those with an asymptomatic, silent infection.

This work was supported by the Swedish Medical Research Council (including a Senior Research position for K. E.), the Torsten and Ragnar Söderberg Foundation, Västra Götalandsregionen (FoU-bidrag) and the Swedish state under the ALF Agreement. Permission for this study was granted by the Ethics Committee of Gothenburg University and all volunteers gave informed consent.

Footnotes

^,†, Gun-Britt Löwhagen³, Petra Tunbäck^3,4, Lars Bellner¹ †Present address: Queen Mary University of London, Institute of Cell and Molecular Science, 4 Newark Street, London E1 2AT, UK. ‡Present address: Department of Pharmacology, New York Medical College, Grassland Reservation, Valhalla, NY 10595, USA.