Human genetic variants of homologous recombination repair genes first found to be associated with Epstein–Barr virus antibody titers in healthy Cantonese

Abstract

Epstein–Barr virus (EBV) infection is a major risk factor for nasopharyngeal carcinoma (NPC). Despite high prevalence of infection among the general population worldwide, only a small proportion of infected individuals presents with seropositivity for EBV-specific IgA antibodies. This seropositive subgroup of EBV carriers has an elevated cumulative risk for NPC during their lifetime. Previous studies reported that the host homologous recombination repair (HRR) system participates in EBV lytic replication, suggesting a potential mechanism to influence EBV reactivation status and thus seropositivity. To investigate whether genetic variants of HRR genes are associated with the serostatus in a healthy population, we investigated the association between seropositivity for anti-VCA-IgA and 156 tagging SNPs in 35 genes connected with HRR in an observational study among 755 healthy Cantonese speakers in southern China. Six variant alleles of MDC1, RAD54L, TP53BP1, RPA1, LIG3 and RFC1 exhibited associations with seropositivity (p_trend from 0.0085 to 0.00027). Our study provides evidence that genetic variation within the HRR might affect an individual's propensity for EBV seropositive status of anti-VCA IgA antibody.

Epstein–Barr virus (EBV) is a member of the herpes virus family. More than 90% of humans worldwide have been infected with EBV.1, 2 Primary infection generally occurs in early childhood and latent infection persists for the rest of life. Yet, only a small proportion of adults in the general population3, 4 will present with seropositivity for EBV-specific IgA antibodies at any given time. Although it is possible for serostatus of an individual to fluctuate, a large-scale prospective study (N = 42,048) of healthy adults (age 30–59) in Zhongshan City of Southern China,5 involving seven tests over 16 years (December 1986 to December 2002), reported that ∼93% of the seronegative subjects maintained seronegativity, and 43% of seropositive subjects remained seropositive from test to test. Our study suggests that a subpopulation prone to EBV reactivation may exist, and that EBV seropositivity among adults may be restricted to this small subgroup of infected people.6

Nearly 100% of anaplastic or poorly differentiated nasopharyngeal carcinomas (NPCs) contain EBV genomic DNA and express EBV proteins, whereas normal nasopharyngeal epithelial cells are typically not infected.7 It is known that the majority of NPC patients exhibit elevated titers of IgA antibodies against EBV structural proteins, such as the viral capsid antigen (VCA). Therefore, EBV seropositivity has been regarded as a major risk factor for NPC.8

In a prospective cohort study conducted in Taiwan, anti-VCA IgA antibody was found to be present before the development of NPC, suggesting a temporal relationship consistent with viral causation. Study subjects positive for the anti-VCA IgA antibody also have an elevated risk of developing NPC during their lifetime.9 Further, the elevated risk is proportional to antibody titers—seropositive individuals with lower and higher anti-VCA IgA antibody levels had hazard ratios (HRs) for NPC of 9.5 and 21.4, respectively, using seronegative individuals as the referent group.10 Studies conducted in the Cantonese population have shown similar results.11 Therefore, VCA-IgA antibody has been widely used as a specific biomarker in the screening and early diagnosis of NPC in southern China.12, 13

NPC is rare in most parts of the world, but common in ethnic populations of southeastern Asia and southern China, particularly among the Cantonese.8 In contrast, EBV infection is worldwide. This difference suggests that, in addition to specific environmental risk factors in endemic areas, genetic factors may play a crucial role in NPC formation and development. Although previous studies have established a link between genetic factors and NPC risk, the potential association between genetic factors and elevated EBV antibody titer has never been previously tested.

Earlier studies have consistently reported that, during the lytic replication cycle, EBV can use the host DNA repair system to facilitate viral DNA synthesis, leading to enhanced fidelity of viral genomic replication and production of viral progeny.14–16 The host homologous recombination repair (HRR) system is the DNA repair pathway involved in EBV lytic replication.16–18 Several host DNA repair factors in HRR system, including P53, the cellular DNA damage-sensor Mre11/Rad50/Nbs1 complexes, ATM, ATR, RPA, RAD51, RAD52, TP53BP1, UNG and XPC,14–24 have been widely reported to be recruited during the lytic replication cycle.

As HRR proteins play a role in viral replication, we hypothesized that HRR could modify individual propensity for EBV reactivation and possibly NPC risk. Therefore, the primary aim of our study was to investigate whether or not genetic variation in the HRR system is associated with anti-VCA IgA antibody serostatus in a healthy population from an NPC endemic area. As anti-VCA IgA antibody is the strongest predictor of NPC, it may be viewed as a biomarker of an intermediate phenotype in the pathway for development of NPC.

To test this hypothesis, we carried out an association study among 755 healthy Cantonese in an NPC high-risk area within Guangdong province. We genotyped 156 tagged single nucleotide polymorphisms (SNP) among 35 genes, representing known HRR genes and other DNA repair-related genes that may cooperate with HRR. The contribution of selected HRR polymorphisms to the anti-VCA-IgA antibody serostatus was evaluated. To our knowledge, this is the first comprehensive attempt to investigate the relationship between DNA repair pathway gene variants and EBV seropositivity.

Abbreviations

95% CI: 95% confidence interval; ATM: ataxia telangiectasia mutated; ATR: ataxia telangiectasia and Rad3 related; BRCA: breast cancer; EBV: Epstein–Barr virus; HRR: homologous recombination repair; LIG: DNA ligase; MDC1: mediator of DNA-damage checkpoint 1; Mre11: meiotic recombination 11; Nbs1: nibrin; NPC: nasopharyngeal carcinoma; OR: odds ratio; RAD: Escherichia coli RecA homolog; RFC: replication factor C; RPA: replication protein; SNP: single nucleotide polymorphism; TP53: tumor protein p53; TP53BP1: tumor protein p53-binding protein 1; UNG: uracil-DNA glycosylases; VCA: viral capsid antigen; XPC: xeroderma pigmentosum C; XRCC: X-ray repair cross-complementing group

Material and Methods

Study population

Healthy individuals from rural and urban localities were enrolled between October 2005 and October 2007. Fourteen rural localities were included in our study. In each locality, two or three villages were randomly selected, and a total of 35 villages were included. All of the rural subjects were randomly selected from the registry of permanent residents with no more than one member from each family selected. Urban subjects were randomly selected from people who attended local hospitals for health examinations. All subjects were native residents without a history of malignancy or immunological disease.

A standardized personal interview based on a structured questionnaire was used to collect information on sociodemographic characteristics, cigarette smoking, alcohol consumption, diet, personal history of sinusitis or other nasal diseases and family history of NPC. Blood samples were collected at the time of enrollment. Each participant was provided free services, including serologic testing for the anti-VCA-IgA antibody, otorhinolaryngologic examinations, and medical consultations performed by physicians of the Sun Yat-Sen University Cancer Center (SYSUCC). Those who were positive for the anti-VCA-IgA antibody were informed and then referred to the NPC department at SYSUCC, where they underwent fiberoptic endoscopy performed by an otorhinolaryngologist. A total of 755 healthy Cantonese were included in our study, and the response rates were similar between rural and urban localities (64 and 67%).

All study subjects had signed informed consent before epidemiological data, and blood samples were collected by trained SYSUCC staff interviewers.

EB-VCA-IgA testing

The measurements of EB-VCA-IgA antibody titers were conducted from peripheral blood samples. A commercial immunoenzymatic assay4, 5, 25, 26 was used in accordance with the kit instructions (Guangdong Zhongshan Biotechnology Company, China). This kit is routinely used to diagnose NPC in the hospital setting. Detailed information is described in Supporting Information (EBV_antiboby_test).

Serial dilutions of quality control sera were applied to each assay for evaluation of intraset variability. To minimize experimental error, all of the tests were conducted in the same laboratory by the same technician. The coefficients of variation for titers 1:10, 1:20, 1:40, 1:80, 1:160 and 1:320 were 8.44, 8.92, 8.59, 6.74, 7.11 and 6.59%, respectively.

Candidate gene selection, tagged SNP selection and genotyping

Genomic DNA was isolated from peripheral blood lymphocytes using the QIAamp DNA blood midi kit (Qiagen, Germany) according to the manufacturer's protocol. A list of genes involved in DNA repair pathways were compiled from the inventory of human DNA repair genes published by Wood et al.27 and from the National Center for Biotechnology Information (NCBI) databases. The genes in the HRR pathway or HRR-related pathways were specifically selected as the candidate genes for our study, including ATM, ATR, BRCA1, BRCA2, LIG1, LIG3, LIG4, MDC1, MRE11, NBS1, RAD1, RAD17, RAD50, RAD51A, RAD51C, RAD51L1, RAD51L3, RAD52, RAD54B, RAD54L, RAD9A, RFC1, RFC2, RFC3, RFC4, RFC5, RPA1, RPA2, RPA3, TP53, TP53BP1, UNG, XPC, XRCC2 and XRCC3. Han Chinese SNP data for the candidate genes, including sequences 5–10 kb upstream and downstream of each gene, were obtained from the NCBI dbSNP database and the International HapMap Project database. Then, according to the selection criteria of r² > 0.8 and minor allele frequency ≥ 0.05 in the Han Chinese population, we selected haplotype tagging SNPs (htSNPs) to capture the common haplotypes (i.e., frequencies ≥ 0.05) associated with each fragment.28, 29 After eliminating SNPs that were not suitable for analysis on the Illumina GoldenGate genotyping platform (design score < 0.60), one tagged SNP was selected per bin. [A bin was defined as a set of markers in strong linkage disequilibrium (LD) with each other (i.e., r² ≥ 0.80).] A total of 156 tagged SNPs across the 35 genes were evaluated in our study.

Genotyping was performed using the Illumina High-Multiplex BeadArray genotyping system.30, 31 An Illumina oligonucleotide pool assay including the selected loci on candidate genes was designed. The assay used allele-specific extension methods and universal PCR amplification reactions. DNA samples were processed through the highly multiplexed GoldenGate protocol using barcoded microwell plates and robust automation systems. Genotype calls were made using the Genotyping module of BeadStudio version 3.2. Out of 156 SNPs, 150 SNPs had call rates above 98%. Three SNPs had call rates of 95%, 92% and 86%, respectively. The remaining three SNPs failed genotyping and were removed from further analyses.

Statistical analysis

Statistical analyses were conducted using the Stata 10.0 statistical software package (Stata Corp.; College Station, TX) and HelixTree Genetics Analysis software (version 6.4.1; Golden Helix, Bozeman, MT). Demographic variables are presented as the mean ± 1 standard deviation for continuous variables and as percentages for categorical variables.

Before conducting the association analysis of genetic variants with a positive anti-VCA-IgA serostatus, we used X² tests to determine whether the genotype distributions for each SNP followed Hardy–Weinberg equilibrium under Mendelian biallelic expectations. Titers of anti-VCA-IgA antibody were dichotomized into binary variables (positive or negative). A titer of greater than or equal to 1:10 was defined as positivity for the IgA antibodies.4, 11 Univariate genotype associations with anti-VCA-IgA antibody serostatus were assessed using the Cochran–Armitage trend test. The threshold for significance of p values was defined as 0.01.

The values of false-positive report probability (FPRP) were assessed according to the method described by Wacholder et al.32

Results

A total of 755 healthy Cantonese people, including both seropositive (n = 128) and seronegative (n = 627) individuals, were included in an observational study of HRR genotype association with seropositivity risk. A cutoff value of 1:10 for the anti-VCA IgA antibody was used to differentiate between seropositive and seronegative subjects. No differences were observed between the seropositive and seronegative groups in sex, age, smoking history, Cantonese salted fish exposure or family history of NPC (Table 1).

Table 1. Characteristics of the study population

Six variant alleles among six genes (MDC1, RAD54L, TP53BP1, RPA1, LIG3 and RFC1) were significantly associated with seropositivity based on Cochran–Armitage trend test of allele-associated risk using individuals homologous for the common allele as the referent (p_trend from 0.0085 to 0.00027) (Table 2). ORs increased proportionately with the number of risk alleles for each significant gene, even after adjusting for age, sex, smoking status, family history of NPC and salted fish exposure. For example, the ORs of TT, TC and CC genotypes for rs17102086 in RAD54L were 1.00 (reference), 1.90 (1.13–3.20) and 2.67 (1.51–4.69), respectively, and the p_trend was 0.00087. (The genotypes, frequencies and p_trend values of total 156 SNPs are shown in Supporting Information Table 1).

Table 2. Genotype frequency¹ and variant association with VCA IgA antibody status

FPRP values for associations between MDC1, RAD54L, TP53BP1, RPA1, LIG3 and RFC1 with anti-VCA-IgA antibody seropositivity were 0.008, 0.033, 0.214, 0.490, 0.088 and 0.217, respectively, when the prior probability was estimated as 11/156 = 0.07. [A total of 11 genes out of a total of 35 candidate HRR genes (156 SNPs) have been shown to have a role in EBV lytic replication. Assuming only one tagging SNP per gene would be informative with regard to EBV antibody titer, the prior probability was estimated to be 11/156 = 0.07].

Discussion

In our article, we report an association between seropositivity for EBV anti-VCA-IgA antibody and specific variant alleles of genes in the HRR pathway. Our results provide suggestive evidence that host genetic variants could influence EBV serostatus.

Although previous studies have not focused on the relationship between genetic variation and production of EBV antibodies against specific structure proteins such as VCA, there have been several indications that such a correlation might exist. Ho et al. reported that apparently healthy individuals showing detectable levels of anti-VCA IgA tend to aggregate in families of NPC patients. These authors argued that an autosomal recessive gene may be involved in the anti-VCA IgA response.33 In 2004, Pickard et al. reported that healthy individuals in multiplex NPC families in Taiwan also had elevated IgA antibody titers against VCA compared to the general population. They further demonstrated a genetic contribution, although shared environmental factors could not be excluded.34 A genetic study of 34 short tandem repeat markers provided the first evidence indicating that genetic factors may lead to chronic EBV infection in Guangxi of southern China, although the results were not robust.35 These previous reports prompted us to undertake a systematic approach to explore the possible contribution of genetic factors to EBV seropositivity.

During the EBV lytic replication cycle, EBV may use the host DNA repair system to facilitate viral DNA synthesis, enhance the fidelity of genomic replication and produce progeny virus. Specifically, several host DNA repair factors are recruited, including P53, the cellular DNA damage-sensor Mre11/Rad50/Nbs1 complexes, ATM, ATR, RPA, RAD51, RAD52, TP53BP1, UNG and XPC,14–24 most of which are components of the HRR pathway. The EBV lytic program results in inhibition of host cellular DNA replication and disruption of DNA repair, leading to accumulation of unrepaired DNA and consequent genomic instability.36 Cytogenetic studies have suggested that genomic instability may be a common feature in the development of NPC.37–40 As HHR pathway proteins appear to play an important role in lytic reactivation of EBV virus, we hypothesized that the genes coding proteins involved in lytic reactivation of virus may influence seropositivity rates. We, therefore, chose genes of the HRR pathway as potential candidates.

We found that variant alleles of MDC1, RAD54L, TP53BP1, RPA1, LIG3 and RFC1 were significantly associated with elevated anti-VCA IgA titers among healthy individuals from an NPC endemic area. Among these six genes, the products of the MDC1 and TP53BP1 genes have been shown to interact intracellularly. A direct biophysical interaction between MDC1 and TP53BP1 has been reported, and MDC1 acts as an upstream regulator of TP53BP1 in the DNA damage response pathway and in tumor suppression.41–44 Furthermore, a functional interaction between EBV replication protein Zta and host protein TP53BP1 has been recently reported by Bailey et al.17 They have shown that knockdown of TP53BP1 expression reduces EBV lytic replication, and the association between Zta and TP53BP1 is involved in the EBV lytic replication cycle. Thus, there exist potential biochemical roles in viral lytic reactivation for these two genes that could implicate their two putative risk alleles in EBV seropositivity. It is noteworthy that the identified risk allele of MDC1 (rs10947087) is a rare polymorphism, prevalent only in Asian and Sub-Saharan African populations, according to the NCBI dbSNP database (http://www.ncbi.nlm.nih. gov/SNP/snp_ref.cgi?type=rs&rs=10947087). The allele frequencies of the genes identified in our study in general Chinese population and European population are listed in Supporting Information Table 2.

RPA1 is a subunit of the RPA complex (RPA1, RPA2 and RPA3) that can be recruited and retained with RPA2 and RPA3 subunits during EBV replication.16 Hyperphosphorylated RPA is involved in EBV lytic genome synthesis and maintenance of viral genome integrity, so a biochemical mechanism for RPA1's role in viral reactivation is evident. Although genetic variants of RAD54L, LIG3 and RFC1 were also observed to be associated with seropositivity, no functional relationship between lytic replication or viral reactivation has yet been reported.

For the role of TP53 in the EBV lytic cycle progression, several studies have indicated that p53 is recruited to viral replication compartments22, 24; in addition, immediate-early viral Zta gene can interact with TP53 and that both proteins can regulate the other's biological functions.20, 21 Tsai and coworkers19 first observed that EBV lytic cycle can be induced significantly in TP53-positive NPC cell lines but not in p53-null cell, TP53 contributed to the expression of Zta gene, our study provided the first evidence that TP53 is involved in the regulation of EBV lytic cycle initiation. However, in our study, we did not find the polymorphism in TP53 associated with seropositive for anti-VCA IgA antibody.

Regarding the potential functions of the six associated SNPs, there is little evidence that these SNPs have certain direct functions themselves. The LIG3 SNP was located in the 3′ UTR region, which might indicate that it has a role in message stability or protein expression levels, and this same SNP allele has previously been reported to be associated with early-onset lung cancer among Europeans.45 The MDC1 SNP is located in exon 10 but is nonsynonymous. All of the remaining four SNPs are located in introns. It thus seems most likely that these SNPs are in LD with functional unidentified functional sequences, rather than being functional determinants themselves (Table 3). This is to be expected, because the SNPs were chosen as tags to inform on all other SNPs that are in high LD, rather than for any putative functional features they might have.

Table 3. The characteristics of six SNPs identified in our study

On the basis of the results of our study, we speculate that a host–virus interaction between the products of DNA repair genes and EBV lytic replication might be an important aspect of NPC development. Host DNA repair genes, particularly in the HRR pathway, could participate in and enhance productive viral replication. The notion is supported by the observation that inhibited expression of HRR genes results in suppression of EBV lytic replication.14

A strength of our study is that we comprehensively and systematically investigated the association between genetic polymorphisms of the 35 most important genes in the HRR pathway and the EBV antibody titer in a healthy population using the htSNP approach. Using a tagging strategy, a subset of SNPs was selected to represent the majority of genetic variants in a defined genomic region. This approach is considered to be a powerful method for identification of the association between genetic loci and a disease state.46 Any given set of tagging SNPs may or may not be functionally related to a particular disease state, but may indirectly identify the true causal variants related to that disease state.

We need to point out that, as we collected blood samples only once per subject, it was not possible to document the fluctuations in antibody titers. This is a limitation of our study. Nevertheless, a large-scale cohort study in Cantonese of Zhongshan City, where individuals were tested for seropositivity seven times over 16 years,5 suggests that the seronegative population is quite stable, because only 7% seroconverted to positive from test to test. Although the level of antibody titer in seropositive population could wane over time, less than 4% of individuals ever testing seronegative go on to develop NPC,5 suggesting that even one-time serologic screening is valuable as a risk marker for the subpopulation prone to NPC.

While using a more stringent FPRP value of 0.2, only MDC1, RAD54L and LIG3 were preserved as being associated with anti-VCA-IgA antibody seropositivity, suggesting many of these results will likely fail to reproduce in additional populations. Noteworthy, our study is just a descriptive study, the interpretation of the results is speculative and the mechanisms involved in the interplay between cellular DNA repair and EBV reactivation merit further investigation.

Another limitation of the study is that we could only measure the associations of allelic variants with seropositivity and not with NPC itself. However, all seropositive individuals do no progress to NPC, and the unknown environmental and host factors that modify progression to NPC can obscure the ability to measure a direct association between the SNP and NPC, even if the association with seropositivity is true. Regardless, in the current study design, we could only measure the associations of allelic variants with seropositivity and not with NPC itself. Potential associations of these SNPs with NPC warrant further investigation.

In conclusion, our study provides evidence that genetic variation among the HRR genes might affect an individual's propensity for EBV seropositive status of anti-VCA IgA antibody. These findings are in agreement with previous reports suggesting a heritable component to EBV seropositivity and are informed by recent cellular findings demonstrating that HRR proteins are host factors that cooperate in lytic viral reactivation of EBV. Although direct associations between the risk alleles and NPC could not be ascertained here, the strength of seropositivity in predicting NPC risk suggests that genetic modifiers of seropositivity would be expected to concomitantly modify NPC risk as well. Studies of HRR gene associations directly with NPC risk are needed.

Acknowledgements

The authors thank all participants who were keen on supporting their studies. They also thank their staff members who have made great efforts for data collection, input and validation.