Interaction Between Susceptibility Loci in MAVS and TRAF3 Genes, and High-risk HPV Infection on the Risk of Cervical Precancerous Lesions in Chinese Population Free

With about 530,000 new cases annually, cervical cancer is a major malignant disease among women and accounts for 8% of the total cancer-related mortality (1, 2). Virtually all cervical cancers result from a persistent infection with certain high-risk types of the human papillomavirus (hrHPV) family (3). Cervical cancer develops through a multistep process, with three cervical intraepithelial neoplasia grades, 1 to 3 (CIN1–3; ref. 4). The transition from CIN to invasive cancer will take several years, even decades, which offers us many opportunities for intervention and only a minority of the infected individuals go on to develop invasive cervical cancer (5). Available data showed that an approximate 30% to 60% of sexually active women infected with genital HPV yet do not develop cancer, and the large majority of infections are cleared by the host immune system (2, 6). Therefore, HPV itself is not sufficient to cause cervical cancer and a variety of environmental and host factors are involved in the development of cervical cancer (7, 8).

Type I IFNs are critical components of the innate defense against viral infections (9), which can interfere with the viral transcription of HPV16 and HPV18 (10). However, the pathways by which HPV DNA virus triggers innate defenses are not well-established (11). Detection of viral nucleic acids within the cytoplasm via either the RIG-I–MAVS signaling axis or the cGAS–STING DNA–sensing pathway leads to the production of antiviral IFNs (12, 13), in which retinoic acid-inducible gene I (RIG-I) and mitochondrial antiviral signaling protein (MAVS) mediate IFN production in response to cytosolic double-stranded RNA or single-stranded RNA containing 5′-triphosphate (5′-ppp; ref. 14). Recently, there are indications that RNA-sensing pathways may modulate host responses against DNA infections (15). Chiu and colleagues (14) demonstrated RNA polymerase III (RNA Pol III) detects cytosolic DNA and converts cytosolic AT–rich DNA into RNA, then is sensed by RIG-I, and recruits the adaptor protein MAVS (16). MAVS further interacts with TNF receptor–associated factor 3 (TRAF3) to recruit downstream IRF3 and NFκB-activating kinases, resulting in IFNβ immune response (17), which can inhibit replication of HPV (18).

MAVS, also known as IPS-1, VISA, and Cardif (19, 20), plays a central role in regulating the complex events that lead to either antiviral or inflammatory responses (17). Melchjorsen and colleagues (21) found that the early IFN response was highly dependent on MAVS, because knockdown of MAVS in the macrophages resulted in a strong reduction in the levels of IFNI. Although, how MAVS functions or is regulated remains enigmatic. A recent study reported that the C-terminal TIM region of MAVS (455-PEENEY-460) was shown to bind TRAF3 and exclusively mediate the induction of IFN expression (22). Interestingly, all of the major viral recognition pathways characterized so far, require TRAF3 to initiate IFN production (23). Oganesyan and colleagues demonstrate that TRAF3 is a major regulator of type I IFN production and the innate antiviral response. Furthermore, gene deletion studies have identified TRAF3 as a critical mediator involved in the induction of the IFNs by the RIG-I pathway (24). Nevertheless, the true function of TRAF3 remained a mystery.

Genetic variability of the host has great impact on the outcome of an HPV infection, especially genetic factors that control the innate immunity (25). Moreover, the innate immune system plays an important role in cervical cancer progression (26). Thus, we hypothesized that individuals' genetic differences in the MAVS and TRAF3 gene could influence the expression of IFNI and the host antiviral response, and thus affect the individuals' susceptibility of developing cervical cancer.

In this study, eight polymorphisms in two candidate genes, including the MAVS gene (rs3746660, rs6052130, rs6116065, rs8116776, and rs914294), the TRAF3 gene (rs7156191, rs8022180, and rs12435483) were genotyped, and the possible associations of these SNPs with cervical precancerous lesions were investigated. Furthermore, we sought to study the potential interactions between these SNPs and environmental factors in the etiology of cervical precancerous lesions in the Chinese population.

In this study, all of the cervical specimens were collected with a broom-like device (Qiagen) and placed into a ThinPrep Pap test vial containing PreservCyt Solution. Referral Pap specimens were processed locally using the ThinPrep 2000 System (Hologic) and evaluated for routine screening cytology (8). A Pap smear was positive for squamous intraepithelial lesion (SIL) if low- (LSIL) and high-grade SIL (HSIL), as classified according to the Bethesda Classification System, was detected (27). This study included 164 SIL patients, 120 cases with LSIL (73.2%), and 44 cases with HSIL (26.8%). Four hundred and twenty eight control subjects were randomly selected from those without intraepithelial lesion or malignancy during the same period from the area of residence of the cases. All participants resided in southern China and were unrelated ethic Han Chinese. Participant written consent and ethical approval from the Ethics Committee of Medicine of Jinan University (Guangzhou, Guangdong, China) were obtained.

According to the manufacturer's instructions, total DNA from cervical cells was extracted using a commercial magnetic beads kit (Chemagen; PerkinElmer). Then, the MassARRAY (Sequenom) technique based on matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) was used to detect 16 HPV types including 2 low-risk (6, 11) and 14 high-risk (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68; ref. 28). We performed all of the procedures in the clinical standard laboratory of Beijing Genomics Institute (Shenzhen, China).

Genomic DNA from peripheral blood was extracted with phenol–chloroform DNA extraction methods. The samples were stored at −20°C until they were used. The DNA concentration was determined using a spectrophotometer (NanoDrop ND-1000, PerkinElmer). Samples with a mean OD260 nm/OD280 nm of 1.8 to 2.0 and DNA concentration > 20 ng/μL were considered to be free of contamination.

Tag SNPs (tagSNP) across the MAVS and TRAF3 genes were determined through a search of the HapMap database. Furthermore, the tagSNPs were selected using the following criteria: (i). the population of the HapMap selected Chinese Han Beijing (CHB); (ii). using Haploview by Tagger function; (iii). SNPs with a minor allele frequency (MAF) greater than 5%; (iv). the pairwise tagging with r² ≥ 0.8 (29, 30). Eventually, 8 tagSNPs were selected, including rs3746660, rs6052130, rs6116065, rs8116776, rs914294 in MAVS and rs7156191, rs8022180, rs12435483 in TRAF3. Supplementary Table S3 shows these SNP sites and locations.

MALDI-TOF MS method was used to genotype these 8 SNPs. Detailed primer and amplified length for the determination of the 8 SNP genotypes is provided in Supplementary Table S2. Sequenom MassARRAY RS1000 was used for genotyping and the related data were performed using Typer software version 4.0. The method of this part was described in our previous study (8).

With Ficoll–Hypaque density-gradient centrifugation, the peripheral blood mononuclear cells (PBMC) were isolated from EDTA-anticoagulated blood. TRIzol (Invitrogen) was used to extract total RNA from PBMCs. Then, we used a transcriptase cDNA kit (Takara-PrimeScript RT Master Mix Kit) to finish reverse transcription. And qRT-PCR analysis was further performed to quantify the mRNA expression of MAVS and TRAF3 genes with the SYBR PrimerScript RT-PCR Kit (Takara). On a Bio-Rad's CFX96 RT System (Bio-Rad Laboratories), the assays were performed. Cycle conditions were 95°C for 30 seconds followed by 45 cycles at 95°C for 5 seconds and 60°C for 30 seconds. The expression of individual MAVS and TRAF3 measurements was calculated relative to expression of β-actin using 2^−ΔΔC_t method (8). Mann–Whitney test was used to test the differences of relative mRNA levels between the cases and controls.

The dependence of the allele frequencies between the cases and controls was calculated by the χ² test. Moreover, we used a t test or χ² to evaluate differences in the demographic characteristics and high-risk HPV infection status between the cases and controls. Logistic regression was used to calculate ORs and their relative 95% confidence intervals (CI) for risk estimation. The haplotypes of MAVS and TRAF3 gene were analyzed using SNPStats online program (https://www.snpstats.net). Haplotypes with a frequency less than 0.03 would be excluded from analysis. All MAVS and TRAF3 gene polymorphisms identified in the best models of gene–environment interactions were calculated using MDR software (version 1.0.0) and MDR permutation testing software (version 1.0 beta 2; refs. 30, 31). The best model for each order of interaction was selected by cross-validation consistency (CVC) and testing-balanced accuracy (TBA). Interaction models showing highest TBA and CVC were further tested by 1,000 folds permutation tests and a sign test P value of 0.05 (32). All other statistical analyses were performed using SPSS software v.16.0 (SPSS, Inc.).

The interaction effects were determined on the multiplicative and additive scale. Interaction on the multiplicative scale was calculated by logistic regression model. On the additive scale, three measures to examine biological interaction were as follows: (i) attributable proportion due to interaction (AP), (ii) relative excess risk due to interaction (RERI), and (iii) synergy index (S; ref. 33). RERI > 0, AP > 0, or S > 1 indicates biological interaction (34, 35). Furthermore, S > 1 for synergetic effects and S < 1 for antagonistic effects (36, 37).

Among cases and controls, a binary classification was used both for high-risk HPV infection (high-risk infection vs. low-risk HPV infection or non-HPV infection) and for genotypes (homozygous for the major allele vs. one or two copies of the minor allele). The risk for cervical precancerous lesions for a given SNP and high-risk HPV infection status was expressed by OR_i,j where the first index (i) indicated the high-risk HPV infection status coded as 0 for low-risk HPV-infected or non-HPV–infected and 1 for high-risk HPV-infected subjects, and the second index (j) indicated the SNP genotype, coded as 0 for subjects homozygous for the major allele and 1 for subjects bearing one or two copies of the minor allele. Subjects who were low-risk HPV-infected and homozygous for the major allele were considered the reference group, thus coding their cervical precancerous lesions risk as OR₀₀=1. The relative ORs were obtained by logistic regression. The CIs were calculated by the regression coefficients and the corresponding covariance matrix (34). Excel spreadsheet (www.epinet.se) was used to calculate additive interaction: S, RERI, and AP and their corresponding CIs (34).

Hardy–Weinberg equilibrium was reassessed for the 8 SNPs comparing the observed genotype frequencies with the expected frequencies using χ² test with significance level at P < 0.05 (38). The Akaike Information Criterion (AIC) was calculated for each genetic model and the genetic model with the smallest AIC was selected (39).

Statistical significance was set at P < 0.05 for each analysis.

The demographic characteristics are shown in Supplementary Table S1. All observed genotype frequencies in both SIL cases and controls conform to Hardy–Weinberg equilibrium (P > 0.05). As shown in Table 1, no SNP was found to be associated with the risk of cervical precancerous lesions under the allelic model.

The genotype association of 8 SNPs in the MAVS and TRAF3 genes with cervical precancerous lesion risk was shown in Table 2, in which rs6052130 in MAVS gene and rs12435483 in TRAF3 gene were significantly associated with cervical precancerous lesion. After adjusting for age, passive smoking, BMI and initial pregnancy of age, individuals who carried CA and CA+AA genotype in MAVS rs6052130 were at 1.48 and 1.47 times higher risk of developing cervical precancerous lesion than individuals with CC genotype (CA vs. CC: OR_adjusted = 1.48, 95% CI, 1.02–2.16, P = 0.041; CA+AA vs. CC: OR_adjusted = 1.47, 95% CI, 1.02–2.12, P = 0.038). In addition, TC and TC+TT genotypes in TRAF3 rs12435483 were found to have a significant protective effect (OR_adjusted = 0.67, 95% CI, 0.45–0.98, P = 0.041 and OR_adjusted = 0.69, 95% CI, 0.48–0.99, P = 0.044, respectively) compared with CC genotype. Moreover, the AIC values in the dominant model were the smallest, 698.1 and 698.3 for rs6052130 and rs12465483, respectively (Supplementary Fig. S1) while the other 6 SNPs were not observed to be relevant to the risk of cervical precancerous lesions.

No haplotype in MAVS and TRAF3 genes was found to be associated with cervical precancerous lesions (Supplementary Table S4).

The interactive effects of high-risk HPV infection and each SNP were calculated on the basis of an additive scale and multiplicative scale. On the additive scale, we examined the RERI, AP, and S between MAVS rs6052130 (wild or mutant) and high-risk HPV infection (positive or negative) in their association with cervical precancerous lesion. The RERI_adjusted, AP_adjusted, and S_adjusted were 3.10 (95% CI, 0.02–6.18), 0.42 (95% CI, 0.13–0.71), and 1.93 (95% CI, 1.03–3.61), respectively, suggesting that there was statistically significant synergistic interaction (Table 3). And the OR of MAVS rs6052130 mutant and high-risk HPV infection on cervical precancerous lesions was the highest (OR_adjusted, 7.46; 95% CI, 4.19–13.26; Fig. 3). The multiplicative interaction analysis revealed no interaction between high-risk HPV infection and each SNP (Table 3).

The results of MDR analysis for one- to four-factor models were shown (Table 4). The best interaction model of predicting cervical precancerous lesions risk was the three-factor model, including high-risk HPV infection, TRAF3 rs12435483, and number of full-term pregnancies, which yield the highest TBA (69.67%) and the maximum CVC (10/10) and P for permutation test = 0.0000–0.0010.

We further performed a risk group analysis of the combinations of three factors, hrHPV infection, mutant-type of TRAF3 rs12435483, and higher number of full-term pregnancies (Fig. 4). Individuals with negative high-risk HPV infection, wild-type of TRAF3 rs12435483, and term births ≤2 times were used as reference group. Independent of the status of rs12435483, those with high-risk HPV exposure and low number of full-term pregnancies was decreased about 2 times risk of cervical precancerous lesions compared with those with more than 2 times full-term pregnancies.

To explore the effects of different genotypes on gene expression, the MAVS and TRAF3 mRNA expression levels in healthy individuals were detected and no significant difference was observed between the CA+AA genotype and CC homozygotes in MAVS mRNA expression (P = 0.438; Fig. 1A). The TRAF3 mRNA level in TC/TT individuals of SNP rs12435483 was significantly higher than in CC individuals (P = 0.028; Fig. 2A).

To further study the functional role of MAVS/rs6052130 and TRAF3/rs12435483 in the expression of IFN, we carried out a RT-PCR assay to evaluate the mRNA levels of IFNβ expression from healthy individuals.

As shown in Fig.1B, the IFNβ mRNA level in CA/AA individuals of SNP MAVS/rs6052130 was significantly lower compared with CC individuals (P = 0.048).

We also investigated whether the expression of IFNβ was affected by different TRAF3/rs12435483 genotypes. The IFNβ mRNA level in CT/TT individuals of SNP rs12435483 was significantly higher than in CC individuals (P = 0.039; Fig. 2B).

Cervical precancerous lesions is a complex disease with a multifactorial etiology (40). The role of RIG-I pathway in host defense to multiple RNA viruses is well established; however, no association among the population about polymorphisms in the RIG-I pathway involved in the host defense to HPV DNA viruses in cervical precancerous lesions has been reported. And the roles of the key gene in RIG-I pathway in the pathogenesis of cervical precancerous lesions are still not well known. Here, we found that MAVS and TRAF3 genes influence the risk of cervical precancerous lesions, which might provide new insights into the modulation of innate immune response against HPV.

A recent study indicated that the RIG-I–MAVS signaling pathway plays an important role in the innate immune response to HPV infection (41). rs6052130 resides in the intron 5–6 region of the MAVS gene. We found that the heterozygous genotype (CA) in MAVS rs6052130 was significantly associated with an increased risk of cervical precancerous lesions. Previous studies have confirmed that SNPs in MAVS exerted inhibitory effects on antiviral signaling in response to dsRNA (42). In addition, Lad and colleagues (43) found the splicing variants of MAVS, MAVS 1a, inhibits RIG-I and MAVS activity. Interestingly, our study also observed a synergetic interaction between high-risk HPV infection and MAVS rs6052130, which indicates that the rs6052130 variant in the MAVS may inhibit antiviral signaling in response to HPV. Our functional analysis in PBMCs showed that the MAVS rs6052130 CA or AA genotype might downregulate the production of IFNβ, although a significant difference of MAVS mRNA expression in different rs6052130 genotypes was not observed. Therefore, we speculated that the rs6052130 variant in the MAVS gene might contribute to dislocation of MAVS from the outer mitochondria membrane (42), which may lead to loss of RLR antiviral signaling in response to HPV, thereby promoting evasion of HPV, leading to a more robust infection and ultimately contributing to an increased risk of cervical precancerous lesions. More studies are warranted to elucidate these speculations.

Furthermore, MAVS was shown to interact with TRAF3 through TRAF-binding consensus sequences (13). TRAF3 bridged the upstream MAVS and downstream kinase TBK1 and assembles the active MAVS–TRAF3–TBK1 signaling complex (44). The expression of type I IFN elicited by the infection of RNA viruses was severely reduced in TRAF3-deficient mouse embryonic fibroblast cells, indicating a crucial role for TRAF3 in RIG-I signaling. Although, the molecular mechanisms by which TRAF3 is involved in DNA viral response pathways remain unclear. rs12435483 is located in intron 2–3 region of TRAF3 gene. This polymorphism was suggested to increase susceptibility to type II diabetes (45). In our study, the TC heterozygote in rs12435483 was observed to decrease the risk of cervical precancerous lesions by a 0.67-fold. Moreover, we found that the minor allele of rs12435483 was associated with an increased TRAF3 expression phenotype at mRNA levels, and a significant upregulation of IFNβ mRNA expression by PBMCs in TC or TT genotype of rs12435483 compared with CC carriers was also observed. As our previous study showed, an intronic polymorphism also played an important role in mRNA or protein expression (8). We, therefore, propose that carriers of the rs12435483 variant might enhance the TRAF3 transcriptional activity, resulting in a higher level of TRAF3 expression and thereby enhancing IFNβ induction, then contributing to clearance of the HPV virus (8, 46) and ultimately contributing to a decreased risk of cervical precancerous lesions. Modulating TRAF3 activity may be a productive approach to host-directed cervical precancerous lesions therapy. These findings need to be further confirmed in a larger population. We have also found an interaction between high-risk HPV infection and TRAF3 rs12435483, which was especially strong among women with more than 2 times full-term pregnancies (Fig. 4). It is therefore likely that a better understanding of the interplay between different risk factors may shed light on the pathogenesis of cervical precancerous lesions (47).

This study has important clinical implications. Because of hereditary nature of the rs6052130 genotype, effective intervention and prevention measures can be implemented by monitoring high-risk HPV infection status. Moreover, the interactions between rs6052130-mutant and high-risk HPV exposure are important at the population level because among rs6052130-positive women, once changing infection status from high-risk HPV to low-risk HPV would reduce the incidence rate of cervical precancerous lesions by about 40.9% (Fig. 3). Consequently, we suggest that high-risk HPV infection should be prevented in healthy individuals, especially in those women with rs6052130 variant. It is a priority for healthy rs6052130-mutant individuals receiving the HPV vaccine and regular screening to measure status of high-risk HPV infection. Moreover, our data provide novel insight into how MAVS and TRAF3 functions in antiviral innate immune signaling. A further investigation of the RIG-I signaling pathway and a better understanding of how innate immune system works would be a novel target for anti-HPV strategies in the prevention and treatment of cervical precancerous lesions (48).

Several limitations of our study merit further discussion. First, the limited sample size reduced the study power to detect the association of SNP with cervical precancerous lesions. Thus, larger studies and an independent cohort should be taken to validate the association between MAVS rs6052130, TRAF3 rs12435483 polymorphisms, and cervical precancerous lesions. And we would like to mention that we did not include the lesion stratification results because Pap smear might not be reliable enough in grading the severity of SIL (49). Therefore, SIL was only divided as present or absent. Although the methods of calculating interactions based on the MDR could apply to multiple factors at multiple levels, the mechanism how the three factors, including high-risk HPV infection, TRAF3 rs12435483, and number of full-term pregnancies, contribute to cervical precancerous lesions is not well understood. Taking into account that no report has been studied that the expression level of MAVS/TRAF3 in cervical cells, it remains to be elucidated whether an increase of MAVS/TRAF3 expression in cervix correlates with an increase of their expression in PBMCs. It should be noted that the mRNA levels were only tested in healthy control individuals, because the expression of various genes related to the inflammatory response may change during the clinical course of cervical precancerous lesions (50).

In conclusion, our study provides evidence that the MAVS rs6052130 and TRAF3 rs12435483 polymorphism influence the risk of cervical precancerous lesions. In addition, this study also observes that two-way and three-way gene–environment interactions exist in the etiology of cervical precancerous lesions. Our data support the mechanism research that RIG-I signaling pathway plays a crucial role in the antiviral host response to HPV in Chinese. This study provides new insights into the mechanism of HPV infection and pathogenesis, and suggests that MAVS and TRAF3 are new potential antiviral targets in the prevention and treatment of HPV-associated cervical precancerous lesions.

No potential conflicts of interest were disclosed.

Conception and design: G. Yang, C. Jing

Development of methodology: D. Liu, Y. Han, J. Wu, S. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Liu, Z. Wen, X. Huang, X. Ye, C. Huang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.Xiao, C. Zeng, Z. Zhou, C. Guo, G. Yang, C. Jing

Writing, review, and/or revision of the manuscript: D.Xiao, C. Jing

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.Xiao, Y. Wang, M. Ou, G. Yang, C. Jing

Study supervision: X. Wei, G. Yang, C. Jing

This work was supported in part by the Major Research Plan of the National Natural Science Foundation of China (grant no. 91543132); National Natural Science Foundation of China (grant no. 81541070, 30901249, and 81101267); the Guangdong Natural Science Foundation (grant no. 2018A030313601, 10151063201000036, S2011010002526, and 2016A030313089); Guangdong Province Medical Research Foundation (grant no. A2014374, A2015310) and Project from Jinan University (grant no. 21612426, 21615426, JNUPHPM2016001, and JNUPHPM2016002).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.