Polymorphisms in TLR2 Are Associated with Increased Viral Shedding and Lesional Rate in Patients with Genital Herpes Simplex Virus Type 2 Infection

Herpes simplex virus type 2 (HSV-2) is the most frequent cause of genital ulceration worldwide [1]. The infection is characterized by a large array of clinical manifestation, ranging from asymptomatic infection to severe ulcerations, as well as by wide variability in viral shedding [2]. Whether this variability can be explained by viral or host factors remains unknown.

Recent studies have illustrated the importance of Toll-like receptor (TLRs) in the innate immune recognition of viruses [3]. TLRs are transmembrane receptors expressed on the surface or the endosomal lumen of immune and epithelial cells that can detect microbe-associated molecular patterns from a variety of organisms, including viral envelope proteins and nucleic acids. On activation, TLRs activate transcription factors, thereby leading to the production of inflammatory cytokines, activation of innate effector mechanisms, and modulation of adaptive immunity. Several studies using mice deficient in TLRs or cellular complementation systems with human TLRs showed that herpesviruses (including HSV, cytomegalovirus, and varicella-zoster virus) are detected by TLR2, TLR9, and possibly TLR3 [3, 4].

Common polymorphisms in TLR genes influence susceptibility to respiratory syncytial virus (TLR4) [5] and modify the course of HIV-1 infection (TLR9) [6]. Mutations in innate pathways downstream of TLRs have been associated with fatal HSV infections [7]. We hypothesized that polymorphisms in TLR genes can modify the frequency of viral shedding and lesions in genital HSV-2 infection. To test these hypotheses, we analyzed the frequencies of single-nucleotide polymorphisms (SNPs) in 4 TLR genes in a cohort of 128 HSV-2-seropositive persons prospectively assessed for their frequency of genital viral shedding and lesions.

Patients and methods. Study participants were HSV-2-infected persons who were assessed prospectively for disease severity by daily sampling of genital secretions for HSV and who kept a diary of genital signs and symptoms for a period of at least 30 days at the Virology Research Clinic of the University of Washington between November 1988 and April 2005. Age, sex, ethnicity, HSV-1 serostatus, HIV-1 serostatus, and clinical history of genital herpes were extracted from standardized case report forms. Swabs obtained from participants seen at the clinic in earlier years were evaluated for shedding frequency using viral culture (n = 25), whereas more recent samples were assessed using HSV DNA polymerase chain reaction (PCR) (n = 103), as described elsewhere [8]. Subjects gave written informed consent using protocols approved by the Institutional Review Board of the University of Washington and the Western Institutional Review Board.

Minimal haplotype tagging SNPs for each candidate gene (4 in TLR2, 4 in TLR3, 5 in TLR4, and 2 in TLR9) were selected from the Innate Immunity Program in Genomic Applications database (available at: http://innateimmunity.net/). Five additional SNPs that have been previously reported were also genotyped, for a total of 20 SNPs. SNP detection used a mass spectrometry genotyping platform. SNPs in noncoding regions were located with a number referring to their position in the gDNA sequence upstream (indicated by a minus sign [-]) or downstream (indicated by a plus sign [+]) from the translational start site (bp = 1). SNPs situated in the coding region were located using a number referring to their position on the mRNA sequence relative to the “A” of the translational start site (ATG; “A” = bp 1).

Hardy-Weinberg equilibrium and pairwise linkage disequilibrium (LD) were calculated with the genhw and the pwld programs developed in Stata (version 9; StataCorp). We considered r²> 0.80 to represent strong LD. Haplotypes were inferred using Hplus [9]. Analysis was done using SAS for Windows (version 9.1; SAS Institute). Descriptive statistics included graphical analysis, frequencies for categorical, and median and ranges for continuous measures. Shedding rate was defined as number of days on which HSV was detected out of days observed. Lesional rate was the number of days on which lesions were reported on participant diaries out of valid diary recordings. Differences in median person-level shedding frequency and lesional frequency were compared by Kruskal-Wallis test.

Poisson regression was used to test for differences in viral shedding rate and in lesional rate by SNP. Extra-Poisson variation, or overdispersion, was adjusted by incorporating a scale parameter. Alleles present in at least 5 subjects were tested univariately, although each analysis was adjusted for demographic factors known previously to be associated with shedding rate: time since infection and sex [2]. We performed an overall Wald test for a difference in outcome between either the heterozygous or the homozygous genotype for the rare allele and the homozygous genotype for the prevalent allele. A multiple comparisons test was performed using the method of Li and Ji [10], a P value correction method that computes the effective number on independent tests based on the correlations between the pairs of SNPs. The false discovery rate was then controlled at 5%. Risk ratios for significant SNP-level associations were computed assuming either the additive (additional risk for each copy of the minor allele), the dominant (risk for having 1 or 2 copies of the minor allele), or the recessive (risk for having 2 copies of the minor allele) model.

Haplotype analysis was performed using Hplus, a regression method that incorporates haplotype uncertainty into precision estimates for associations with the outcome and incorporates associations with outcome variables into haplotype construction. We computed shedding and lesional rates on the natural log scale. Analysis was again adjusted for sex and time since HSV-2 acquisition. Once haplotypes were computed, estimates of average differences in the outcome of each haplotype were tested versus the most common haplotype, the control. Because all estimates were computed at once, and all haplotypes were simultaneously compared with a control, no multiple comparison adjustment was needed.

Assuming a median shedding study period of 60 days and a correlation of shedding within subjects of 0.2, the power to show a 2-fold and 3-fold increase in a baseline rate of 5% shedding (or lesions) was 64% and 89%, respectively, for genetic polymorphisms present in 10% of individuals.

Results. The cohort included 128 HSV-2-seropositive white participants who obtained at least 30 days of genital swabs for PCR (n = 103) or viral culture (n = 25). The median age was 40 years (interquartile range, 31–49 years; range, 22–76 years), and 74 (58%) were women. All participants were HIV seronegative and 53 (41%) were seropositive for both HSV-1 and HSV-2. The time since primary infection was available in 109 patients (median, 8 years; IQR, 3–16 years; range, <1 month to 33 years).

Shedding and lesional rates among patients were highly correlated (Spearman's correlation coefficient ρ = 0.64, P < .001). The median shedding rates by PCR were not found to differ between sexes (13% [range, 0%–76%] in men and 15% [range, 0%–78%] in women), and neither did the lesional rates (6% [range, 0%–67%] in men and 8% [range, 0%–80%] in women). Reported median rates of lesions did depend on the time since infection (8% in persons infected for ⩾10 years, 6% in those infected for 2–9 years, 10% in those infected for ⩾1 year, and 0% in those with unknown history of HSV infection; P = .008). The median shedding rate by PCR was not significantly influenced by the time since infection.

Differences in viral shedding and lesional rates by TLR alleles are shown in table 1. Among the 20 SNPs tested, 2 pairs of SNPs were in strong LD (TLR4 1063 A/G and TLR4 1363 C/ T [r² = 1]; and TLR9+1174 G/A and TLR9 1635 A/G [r² = 0.76], consistent with previous observations [6]. A SNP located in intron 1 of TLR2 (-15607 A/G) was significantly associated with both shedding and lesional rates. The median shedding rate was 16.7% when G was present versus 12.0% when it was absent (risk ratio for dominant effect, 1.60; P = .02). The median lesional rate was 10.4% when G was present and 5.3% when it was absent (risk ratio for dominant effect, 2.16; P ! .001). The associations were not stronger when the analysis was performed using the additive model (risk ratio for shedding rate, 1.16 [P = .31 ]; risk ratio for lesional rate, 1.47 [P = .01 ]).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1.

Genetic organization of the TLR2 gene. The figure and legend are available in their entirety in the online edition of the Journal of Infectious Diseases.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1.

Association of Toll-like receptor (TLR) alleles with lesional and shedding rates in patients with herpes simplex virus type 2 infection.

Differences in shedding and lesional rates by TLR haplotypes are shown in table 2. Two TLR2 haplotypes (haplotypes 2 and 4) were significantly associated with increased shedding and lesional rates. The median shedding rate was 16.7% versus 12.5% for the presence versus the absence of haplotype 2 (risk ratio for dominant effect, 1.54; P = .03) and 27.3% versus 12.6% for the presence versus the absence of haplotype 4 (risk ratio for dominant effect, 1.80; P = .01). The lesional rate was 10.0% versus 5.6% for the presence versus the absence of haplotype 2 (risk ratio for dominant effect, 2.00; P = .002) and 8.2% versus 6.4% for the presence versus the absence of haplotype 4 (risk ratio for dominant effect, 1.35; P = .30). The associations were not stronger when the analysis was performed using the additive model (risk ratio for shedding rate 1.24 [for haplotype 2, P = .16] and 1.62 [for haplotype 4, P = .02]; risk ratio for lesional rate 1.48 [for haplotype 2, P = .01] and 1.36 [for haplotype 4, P = .27]). Haplotype 2 but not 4 contains the TLR2 -15607G allele that was associated with both end points in the SNP analyses (figure 1)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 2.

Association of Toll-like receptor (TLR) haplotypes with lesional and shedding rates in patients with herpes simplex virus type 2 infection.

Discussion. In the present study, we showed that 2 haplotypes in TLR2 (haplotypes 2 and 4) are associated with increased shedding and lesional rate in patients with HSV-2 infection. Haplotype 2 contained an SNP located in intron 1, which was significantly associated with both end points.

TLR2 has been initially identified as the detector of Gram-positive bacteria. Further investigations showed that TLR2 forms a heterodimer with TLR1 or TLR6 to detect triacyl or diacyl lipopeptides, respectively, present in bacterial cell walls [3]. Subsequently, several studies implied TLR2 in the detection of herpesviruses [4, 11], but the precise viral molecules interacting with TLR2 (presumably envelope proteins) have not been identified yet [3].

Polymorphisms in TLR2 have been previously associated with increased susceptibility to several infectious diseases. A Turkish study revealed an association between susceptibility to tuberculosis disease and a TLR2 SNP associated with impaired function (2258 G/A, R753Q) [12]. Cells transfected with 2258A TLR2 plasmids had decreased responsiveness to bacterial lipopeptides, compared with those transfected with the wild-type plasmids [13]. Allelic variants of a microsatellite polymorphism (GT repeat) in the exon 2 of TLR2 have been associated with increased susceptibility to tuberculosis disease in Korean patients, a finding that might be due to weaker promoter activities and lower TLR2 expression in CD14⁺peripheral blood monocytes [14]. Another TLR2 SNP, -16934 A/T, was associated with increased prevalence of sepsis due to Gram-positive bacteria among critically ill patients with systemic inflammatory response syndrome [15].

Although several studies addressed the role of TLR2 SNPs and microsatellite markers with human disease, few studies have performed systematic genotyping of TLR2 haplotypes using tagging SNPs. A limited number of TLR2 haplotypes that are in LD with functional polymorphisms may contain most of the SNPs reported so far and explain most phenotypes. Further studies using systematic genotyping of TLRs haplotypes with tagging SNPs and TLR2 microsatellites in various cohorts of different ethnicities will be necessary to clarify this issue.

Similarly to other genetic association studies, the present study may be limited by population stratification. To minimize this risk, we limited the analyses to white persons. Another possible limitation is that TLR2 SNPs and haplotypes may be in LD with mutations in other gene(s) located nearby that could be responsible for the infection phenotype. However, the fact that TLR2 has been clearly involved in the innate immune response to herpesviruses, including HSV-2, suggests that the TLR2 SNPs may be responsible for the observed phenotypes. Finally, this study may be limited by its relatively small sample size, so that associations with SNPs or haplotypes that have a smaller effect on the end points might not have been detected. Overall, our data show that polymorphisms in TLR gene can, at least in part, explain different clinical manifestations due to HSV-2 infections and thereby strongly suggest a role of TLR2 in the innate immune response to HSVs.

• Received January 11, 2007.
• Accepted February 6, 2007.

• © 2007 by the Infectious Diseases Society of America