TLR Signaling Pathway Gene Polymorphisms, Gene–Gene and Gene–Environment Interactions in Allergic Rhinitis

ABSTRACT

Background

Allergic rhinitis (AR) is a nasal inflammatory disease resulting from a complex interplay between genetic and environmental factors. The association between Toll-like receptor (TLR) signaling pathway and environmental factors in AR pathogenesis remains to be explored. This study aims to assess the genetic association of AR with single nucleotide polymorphisms (SNPs) in TLR signaling pathway, and investigate the roles of gene–gene and gene–environment interactions in AR.

Methods

A total of 452 AR patients and 495 healthy controls from eastern China were enrolled in this hospital-based case–control study. We evaluated putatively functional genetic polymorphisms in TLR2, TLR4 and CD14 genes for their association with susceptibility to AR and related clinical phenotypes. Interactions between environmental factors (such as traffic pollution, residence, pet keeping) and polymorphisms with AR were examined using logistic regression. Models were stratified by genotype and interaction terms, and tested for the significance of gene–gene and gene–environment interactions.

Results

In the single-locus analysis, two SNPs in CD14, rs2563298 (A/C) and rs2569191 (C/T) were associated with a significantly decreased risk of AR. Compared with the GG genotype, the GT and GT/TT genotypes of TLR2 rs7656411 (G/T) were associated with a significantly increased risk of AR. Gene–gene interactions (eg, TLR2 rs7656411, TLR4 rs1927914, and CD14 rs2563298) was associated with AR. Gene–environment interactions (eg, TLR4 or CD14 polymorphisms and certain environmental exposures) were found in AR cases, but they were not significant after Bonferroni correction.

Conclusion

The genetic polymorphisms of TLR2 and CD14 and gene–gene interactions in TLR signaling pathway were associated with susceptibility to AR in this Han Chinese population. However, the present results were limited to support the association between gene–environment interactions and AR.

Keywords:

• allergic rhinitis
• toll-like receptors
• CD14
• single nucleotide polymorphism
• gene–gene interaction
• gene–environment interaction

Introduction

Allergic rhinitis (AR) is a nasal inflammatory disease induced by immunoglobulin E (IgE)-mediated immune response, with main symptoms of itchy nose, rhinorrhea, sneezing and nasal congestion.¹ The prevalence of AR keeps increasing worldwide. In China, the AR prevalence in adults increased from 11.1% in 2005 to 17.6% in 2011.^2,3 AR and asthma often coexist, both triggered by genetic and environmental factors.⁴ The development of AR involves not only allergen exposure but also early-life factors, family history, ethnicity, and environmental factors, such as tobacco smoke, cooking fumes, living floors, lifestyle, air pollution and furniture pollution.

Toll-like receptors (TLRs) are type I transmembrane protein natural immune pattern recognition receptors, which can recognize pathogen-associated molecular patterns (PAMP) in nature, initiate intracellular signaling pathways, and activate innate immune response. In addition, TLRs also induce dendritic cell (DC) maturation and T cell activation, thus skewing the adaptive immune response towards Th1,⁵ and participate in the induction and perpetuation of asthma and atopy.⁶ Given the mediatory role of TLRs between innate and adaptive immunity, genetic variations of TLR signaling pathway genes may drive the progression of inflammatory and allergic diseases. A number of studies have demonstrated the association between single nucleotide polymorphisms (SNPs) of TLR signaling pathway genes and asthma in diverse populations;^7–14 however, little attention has been given to rhinitis.^15–18

According to the hygiene hypothesis,¹⁹ the allergic diseases may arise from bacterial or viral infections and exposure to non-infectious microbial agents (such as endotoxin) in the environment.²⁰ TLR signaling pathway genes may participate in the protective effect of microbial agents on allergy. Many studies that have examined the interplay between genetic susceptibility and environmental factors in allergic conditions only focus on asthma. A gene–environment interaction has been observed between CD14 rs2569190 and asthma, depending on the endotoxin exposure level in house dust.²¹ Moreover, the associations between polymorphisms of TLR2, TLR4, TLR6, TLR9 and CD14 genes in asthma were affected by environmental factors, such as house dust endotoxin level and living place.^10,22–24 Gene–environment interactions, which have rarely been analyzed in previous studies of AR, may provide new insight into AR pathogenesis. One study on AR children has shown no significant evidence of gene–environment interactions between traffic-related air pollution and GSTP1, TNF, TLR2, and TLR4 genes.¹⁷ At present, there is no literature reporting gene–gene and gene–environment interactions in AR based on a Chinese population. This study was the first to identify the associations of candidate genes and environmental factors with genetic predisposition to AR in the Chinese Han population.

According to the data about weather, air quality, transportation and lifestyle in East China, we selected several environmental factors that may be related to AR. We screened 10 SNPs of TLR2, TLR4 and CD14 genes in the TLR signaling pathway, and conducted a case–control study in East China Han population to explore the association between polymorphisms and AR, as well as gene–gene and gene–environment interactions.

Methods

Subjects

A total of 452 AR cases (302 males and 150 females) and 495 healthy controls (303 males and 192 females) were recruited from the First Affiliated Hospital of Nanjing Medical University, Nanjing, China. All subjects were Han Chinese in Jiangsu and Anhui provinces in eastern China. The diagnosis of AR was established according to the “Allergic Rhinitis and its Impact on Asthma (ARIA) 2008 update”¹ and the “Chinese Society of Allergy Guidelines for Diagnosis and Treatment of Allergic Rhinitis”.³ The medical history and clinical data were collected from outpatients’ medical records and face-to-face questionnaire surveys. The cases did not use glucocorticoids within 4 weeks, and antihistamines, leukotriene receptor antagonists and other anti-allergic drugs within 2 weeks before blood sampling. The controls were recruited from the hospital seeking health care or routine health examinations, and were frequency-matched with the cases in age (±5 years) and sex. The selection criteria for controls:^25,26 (1) no symptoms and medical history of AR and nasal diseases; (2) no symptoms and medical history of other allergic diseases, such as asthma, eczema and urticaria; (3) negative blood test for serum allergen-specific IgE; (4) no history of AR or other allergic diseases in the immediate family. All the study subjects were divided into < 18-year-old group and ≥ 18-year-old group for further stratification analysis.

Clinical Parameters

The questionnaire survey covered general information, demographic characteristics, disease symptoms and scores, physical signs, medical history, accompanying diseases, and environmental factors, including smoking, residence (floors), sunlight exposure, in-house chemicals, near-residence pollutants, near-residence traffic network, radiation exposure to computers, oil fume exposure and pet-keeping. A visual analogue scale (VAS) ranging from 0 cm (not bothersome at all) to 10 cm (extremely bothersome) was used to assess the patient’s subjective perception of nasal symptoms, including rhinorrhea, sneezing, nasal congestion, itchy nose and eyes, and total nasal symptoms. VAS scores < 5 were diagnosed as mild AR, and VAS scores ≥ 5 as moderate-to-severe AR.²⁷

About 5 mL of peripheral venous blood was collected from each subject for in vitro allergen detection. The levels of eosinophil cationic protein (ECP), serum total IgE and specific IgE were measured with ImmunoCAP assays (Phadia, Uppsala, Sweden). Total IgE and ECP were determined in all subjects. Specific IgE antibodies to common inhalant allergens including Dermatophagoides pteronyssinus (Der p; d1), Dermatophagoides farinae (Der f; d2), cat dander (e1), dog dander (e5), Blatella germanica (i6), Alternaria alternata (m6), Ambrosia elatior (w1), and Artemisia vulgaris (w6) were determined in AR cases. We collected patients allergic to dust mites (d1 and/or d2), one of the most common allergens in eastern China. The positive rates of other aeroallergens (e1, e5, i6, m6, w1 and w6) were low, and they were not the main allergens causing symptoms.

SNPs Selection and Genotyping

A total of 10 SNPs from TLR2, TLR4 and CD14 genes in the TLR signaling pathway were obtained: rs7656411 (G/T), rs76112010 (A/G) and rs7682814 (A/G) of TLR2 gene; rs10983755 (A/G), rs11536889 (G/C), rs1927914 (A/G) and rs7873784 (G/C) of TLR4 gene; rs2563298 (A/C), rs2569190 (A/G) and rs2569191 (C/T) of CD14 gene (Table 1). The data of 10 SNPs were selected by using genotype data of Han Chinese in Beijing (CHB) and Japanese in Tokyo (JPT) from the 1000 Genomes Project (March 2012) and based on previously published studies.^{12–17,28,29} Distributions of all genotypes in the control subjects were consistent with those estimated according to the minor allele frequency (MAF) > 0.05 and P-value of Hardy-Weinberg equilibrium (HWE) > 0.05. In addition, web-based tools were used to predict putative functions of genetic variants, including SNPinfo (https://snpinfo.niehs.nih.gov/), HaploReg (https://pubs.broadinstitute.org/mammals/haploreg/haploreg_v4.php), RegulomeDB (https://www.regulomedb.org/regulome-search/), MirSNP (http://bioinfo.bjm-u.ed-u.cn/mirsnp/search/), and RNAhybrid (https://bibiserv.cebitec.uni-bielefeld.de/rnahybrid/). Secondary structural changes and minimum free energy (MFE) changes caused by different SNP genotypes were predicted by using RNAfold (http://rna.tbi.univie.ac.at/).

Table 1 Gene Frequency Distribution of TLR2, TLR4 and CD14 Alleles

No.	SNPs	Gene	Location	Base Change	MAF			HWE P^b
					Database^a	Case	Control
1	rs7656411	TLR2	3ʹnear gene	G/T	0.419	0.482	0.451	0.168
2	rs76112010	TLR2	5ʹnear gene	A/G	0.136	0.203	0.169	0.661
3	rs7682814	TLR2	5ʹnear gene	A/G	0.167	0.202	0.172	0.286
4	rs10983755	TLR4	5’ near gene	A/G	0.267	0.127	0.154	0.457
5	rs11536889	TLR4	3ʹUTR	G/C	0.261	0.238	0.218	0.065
6	rs1927914	TLR4	5ʹnear gene	A/G	0.360	0.434	0.416	0.204
7	rs7873784	TLR4	3ʹUTR	G/C	0.122	0.274	0.259	0.537
8	rs2563298	CD14	3ʹUTR	A/C	0.151	0.123	0.184	0.067
9	rs2569190	CD14	5ʹUTR	A/G	0.500	0.380	0.412	0.988
10	rs2569191	CD14	5ʹnear gene	C/T	0.453	0.363	0.426	0.625

Notes: ^aMAF for CHB from the HapMap databases (http://www.hapmap.org) or PubMed (http://www.ncbi.nlm.nih.gov/snp). ^bTwo-sided χ² test for the allele frequencies of the controls.

Abbreviations: SNPs, single nucleotide polymorphisms; MAF, minor allele frequency; HWE, Hardy-Weinberg equilibrium.

Genomic DNA was purified from peripheral blood leukocytes using a commercial kit (Tiangen Biotech, Beijing, China) according to the manufacturer’s instructions, and stored at –70°C until usage. Genotyping was performed with the TaqMan SNP Genotyping Assay using the 384-well ABI 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). More than 15% of the samples were randomly selected for confirmation, and the discordance rate between genotypes was below 0.3%.

Statistical Analysis

The values of total IgE, specific IgE and ECP were transformed by log logarithm. The Student’s t-test was performed to analyze continuous variables, and the χ² test to analyze categorical variables and the genotype distributions of SNPs in two groups. HWE of the genotype distribution in controls was tested by a goodness-of-fit χ² test. The odds ratios (OR) and 95% confidence intervals (CI) were calculated by logistic regression analysis after adjustment for gender and age, to quantify the association between the polymorphisms and risk of AR. Gene–gene interactions were analyzed using multifactor dimensionality reduction (MDR) software (http://sourceforge.net/projects/mdr/). Logistic regression analyses were performed to explore gene–environment interactions, with adjustment for gender and age. Calculations were carried out using Statistical Analysis System software (version 9.1.3; SAS Institute, Cary, NC, USA). A P-value < 0.05 was considered statistically significant, and all statistical tests were two-sided. The Bonferroni correction (P < 0.05 divided by the number of SNPs analyzed, P < 0.005) was applied to adjust for multiple comparisons.

Results

Characteristics of Subjects

The demographic characteristics of the study subjects are summarized in Table 2. There was no significant difference in the distribution of age (P = 0.597) and sex (P = 0.073). Apparently, serum levels of total IgE (264.0 [120.6–593.5] kU/L) and ECP (12.7 [5.2–28.7] μg/L) in AR cases were significantly higher (P < 0.001) than those in controls (26.8 [11.1–52.2] kU/L and 4.6 [3.1–7.5] μg/L, respectively). In AR cases, the serum levels of allergen-specific IgE against Der p and Der f were 29.0 (4.6–72.5) kUA/L and 24.1 (5.0–69.9) kUA/L, respectively. A total of 247 (54.6%) cases presented mild (VAS < 5) and 194 (42.9%) presented moderate-to-severe (VAS ≥ 5) AR, with a mean VAS score of 5.25 ± 2.39; 118 (26.1%) cases reported to have concomitant asthma. According to the questionnaire, asthma information missed in 63 (13.9%) patients, for the related questions in their questionnaires were not answered.

Table 2 Distribution of Selected Variables Among Cases and Controls

Variables	Case (n = 452)		Control (n = 495)		P^a
	N	%	N	%
Age (years), mean ± SD	19.70 ± 12.60		20.10 ± 12.61		0.597
Gender
Male	302	66.8	303	61.2	0.073
Female	150	33.2	192	38.8
Duration of rhinitis (year), mean ± SD	6.62 ± 5.87
Total nasal symptoms (VAS), mean ± SD	5.25 ± 2.39
Serum total IgE (kU/L), median (IQR)^b	264.0 (120.6–593.5)		26.8 (11.1–52.2)		< 0.001*
Serum specific IgE (kUA/L), median (IQR)^b
Dermatophagoides pteronyssinus	29.0 (4.6–72.5)
Dermatophagoides farinae	24.1 (5.0–69.9)
Serum ECP (μg/L), median (IQR)^b	12.7 (5.2–28.7)		4.6 (3.1–7.5)		< 0.001*
With asthma^c
Yes	118	26.1
No	271	60.0

Notes: ^aDerived from two-sided χ² test for comparison of discrete variables and unpaired Student’s t-test for continuous variables. ^bSelective variables were transformed into logarithmic model before unpaired Student’s t-test between cases and controls. ^cSome information of concomitant asthma of allergic diseases was not available in cases. *Statistically significant (P < 0.05).

Abbreviations: IQR, interquartile range; ECP, eosinophil cationic protein.

Association Between SNPs and AR Risk in Single-Locus Analyses

The genotype and allele distributions of the 10 SNPs and their associations with AR risk are presented in Table 3. The single-locus analyses revealed that the genotype frequencies of two SNPs rs2563298 (A/C) and rs2569191 (C/T) in CD14 were significantly different between the cases and the controls (rs2563298: P = 0.003; rs2569191: P = 0.008). Multivariate logistic regression analysis indicated that the variant CA and AA genotypes of CD14 rs2563298 were associated with a significantly decreased risk of AR, compared with the wild-type CC genotype (adjusted OR = 0.65, 95% CI = 0.48–0.89 for CA and adjusted OR = 0.40, 95% CI = 0.18–0.89 for AA). For the CD14 rs2569191, compared with the CC genotype, the CT and TT genotypes were associated with a significantly decreased risk of AR (adjusted OR = 0.65, 95% CI = 0.48–0.87 for CT and adjusted OR = 0.64, 95% CI = 0.44–0.95 for TT). We also found that the dominant models of CD14 rs2563298 (CA/AA vs CC) and rs2569191 (CT/TT vs CC) were significantly associated with AR risk (adjusted OR = 0.62, 95% CI = 0.46–0.83 for rs2563298 and 0.65, 0.49–0.85 for rs2569191). The genotype and allele frequencies of TLR2 rs7656411 showed no significant difference between cases and controls; however, compared with the GG genotype, the GT and GT/TT genotypes were associated with a significantly increased risk of AR (adjusted OR = 1.41, 95% CI = 1.04–1.91 for GT and adjusted OR = 1.35, 95% CI = 1.01–1.79 for GT/TT). No significant difference was found between the associations of genotype and allele frequencies with AR susceptibility in the other seven SNPs (P > 0.05).

Table 3 Genotype and Allele Frequencies in TLR2, TLR4 and CD14 Polymorphisms Among Cases and Controls

SNPs	Genotype	Case		Control		Crude OR (95% CI)	Adjusted OR (95% CI)^a	P^b
		N	%	N	%
TLR2
rs7656411		n = 441		n = 493
	GG	112	25.4	156	31.6	1.00 (reference)	1.00 (reference)	0.079
	GT	233	52.8	229	46.5	1.42 (1.05–1.92)	1.41 (1.04–1.91)
	TT	96	21.8	108	21.9	1.23 (0.86–1.79)	1.22 (0.84–1.76)
	GT/TT	329	74.6	337	68.4	1.36 (1.02–1.81)	1.35 (1.01–1.79)	0.035*
	T allele	0.482		0.451				0.186
rs76112010		n = 446		n = 485
	GG	287	64.4	331	68.3	1.00 (reference)	1.00 (reference)	0.144
	GA	137	30.7	141	29.1	1.12 (0.84–1.49)	1.15 (0.86–1.53)
	AA	22	4.96	13	2.7	1.95 (0.97–3.95)	1.98 (0.98–4.02)
	GA/AA	159	35.7	154	31.2	1.19 (0.91–1.53)	1.22 (0.93–1.60)	0.209
	A allele	0.203		0.172				0.089
rs7682814		n = 441		n = 486
	GG	285	64.6	330	67.9	1.00 (reference)	1.00 (reference)	0.074
	GA	134	30.4	145	29.8	1.07 (0.81–1.42)	1.09 (0.82–1.45)
	AA	22	5.00	11	2.26	2.32 (1.10–4.86)	2.29 (1.09–4.81)
	GA/AA	42	15.9	38	13.9	1.15 (0.88–1.52)	1.17 (0.90–1.55)	0.292
	A allele	0.202		0.172				0.097
TLR4
rs10983755		n = 423		n = 455
	GG	327	77.3	327	71.9	1.00 (reference)	1.00 (reference)	0.157
	GA	84	19.9	115	25.3	0.73 (0.53–1.01)	0.72 (0.52–0.98)
	AA	12	2.8	13	2.9	0.92 (0.42–2.05)	0.98 (0.44–2.18)
	GA/AA	96	22.7	128	28.1	0.75 (0.55–1.02)	0.74 (0.54–1.01)	0.065
	A allele	0.128		0.155				0.101
rs11536889		n = 427		n = 482
	GG	254	59.5	298	61.8	1.00 (reference)	1.00 (reference)	0.765
	GC	143	33.5	153	31.7	1.10 (0.83–1.46)	1.09 (0.82–1.45)
	CC	30	7.0	31	6.4	1.13 (0.67–1.93)	1.18 (0.69–2.01)
	GC/CC	173	40.5	184	38.1	1.10 (0.85–1.44)	1.10 (0.84–1.44)	0.471
	C allele	0.238		0.223				0.458
rs1927914		n = 411		n = 453
	AA	128	31.1	161	35.5	1.00 (reference)	1.00 (reference)	0.286
	AG	209	50.9	207	45.7	1.27 (0.94–1.72)	1.25 (0.92–1.69)
	GG	74	18.0	85	18.8	1.10 (0.74–1.62)	1.09 (0.73–1.63)
	AG/GG	283	68.9	292	64.5	1.22 (0.92–1.62)	1.20 (0.90–1.60)	0.171
	G allele	0.434		0.416				0.445
rs7873784		n = 433		n = 453
	GG	235	54.3	251	55.4	1.00 (reference)	1.00 (reference)	0.644
	GC	159	36.7	169	37.3	1.01 (0.76–1.33)	1.00 (0.76–1.33)
	CC	39	9.0	33	7.28	1.26 (0.77–2.07)	1.27 (0.77–2.01)
	GC/CC	198	45.7	202	44.6	1.05 (0.80–1.36)	1.05 (0.80–1.37)	0.734
	C allele	0.274		0.259				0.496
CD14
rs2563298		n = 434		n = 479
	CC	336	77.4	326	68.1	1.00 (reference)	1.00 (reference)	0.003*
	CA	89	20.5	131	27.3	0.66 (0.48–0.90)	0.65 (0.48–0.89)
	AA	9	2.1	22	4.60	0.40 (0.18–0.88)	0.40 (0.18–0.89)
	CA/AA	98	68.5	153	31.9	0.62(0.46–0.84)	0.62 (0.46–0.83)	0.002*
	A allele	0.123		0.183				<0.001*
rs2569190		n = 450		n = 481
	AA	174	38.7	166	34.5	1.00 (reference)	1.00 (reference)	0.351
	AG	210	46.7	233	48.4	0.86 (0.65–1.14)	0.86 (0.64–1.15)
	GG	66	14.7	82	17.1	0.78 (0.88–1.92)	0.76 (0.52–1.94)
	AG/GG	276	61.3	315	65.5	0.84 (0.52–1.31)	0.84 (0.64–1.09)	0.188
	G allele	0.380		0.413				0.150
rs2569191		n = 441		n = 472
	CC	187	42.4	153	32.4	1.00 (reference)	1.00 (reference)	0.008*
	CT	188	42.6	236	50.0	0.65 (0.49–0.87)	0.65 (0.48–0.87)
	TT	66	15.0	83	17.6	0.65 (0.44–0.96)	0.64 (0.44–0.95)
	CT/TT	254	57.6	319	67.6	0.65 (0.50–0.85)	0.65 (0.49–0.85)	0.002*
	T allele	0.363		0.426				0.006*

Notes: ^aAdjusted for age and sex in logistic regression model. ^bTwo-sided χ² test for the distributions of genotype and allele frequencies. *Statistically significant (P < 0.05).

Abbreviations: SNPs, single nucleotide polymorphisms; OR, odds ratio; CI, confidence interval.

Association of Stratification Analysis Between the SNPs and AR

In the stratification analysis (Table 4), compared with the controls, the dominant model of TLR2 rs7656411 (GT/TT vs GG) exhibited a significantly increased risk of AR in the subgroups of males (adjusted OR = 1.60, 95% CI = 1.12–2.30) and concomitant asthma (adjusted OR = 1.61, 95% CI = 1.13–2.28). Furthermore, a significantly decreased risk of AR was found in the dominant model of CD14 rs2563298 (CA/AA vs CC) in the subgroups of < 18-year-old, females, lower total IgE, mild symptoms, moderate-to-severe symptoms, and asthma (with/without). This decreased risk was also more pronounced in the dominant model of CD14 rs2569191 (CT/TT vs CC) among all the subgroups of age, gender, asthma, VAS, and total IgE, compared with that in the controls. However, only the associative significance of CD14 rs2563298 in the subgroup of < 18-year-old, females, without asthma and lower total IgE, and that of CD14 rs2569191 in the subgroup of without asthma, moderate-to-severe symptoms and lower total IgE remained statistically evident after Bonferroni correction.

Table 4 Stratification Analyses of TLR2, TLR4 and CD14 Polymorphisms in the Dominant Model in Cases and Controls

Variables	Subgroups	TLR2 rs7656411		P	CD14 rs2563298		P	CD14 rs2569191		P
		N (Case/Control)^a	Adjusted OR (95% CI)^b		N (Case/Control)^a	Adjusted OR (95% CI)^b		N (Case/Control)^a	Adjusted OR (95% CI)^b
Age (years)	< 18	247/239	1.33 (0.88–1.99)	0.173	244/239	0.54 (0.36–0.80)	0.003**	247/236	0.65 (0.44–0.94)	0.022*
	≥ 18	194/254	1.36 (0.91–2.06)	0.163	191/240	0.73 (0.47–1.13)	0.164	194/236	0.66 (0.44–0.98)	0.037*
Gender	Male	295/303	1.60 (1.12–2.30)	0.011*	289/296	0.75 (0.52–1.07)	0.115	293/289	0.66 (0.47–0.93)	0.018*
	Female	146/190	0.97 (0.60–1.57)	0.894	146/183	0.42 (0.25–0.72)	0.002**	148/183	0.62 (0.40–0.98)	0.038*
Asthma	No	263/493	1.28 (0.81–2.03)	0.292	261/479	0.45 (0.27–0.75)	0.002**	262/472	0.53 (0.35–0.81)	0.003**
	Yes	116/493	1.61 (1.13–2.28)	0.007*	115/479	0.60 (0.42–0.85)	0.006*	116/472	0.70 (0.51–0.96)	0.027*
VAS	< 5	247/493	1.25 (0.88–1.77)	0.207	241/479	0.65 (0.45–0.92)	0.017*	246/472	0.72 (0.52–0.99)	0.048*
	≥ 5	194/493	1.45 (0.99–2.12)	0.055	193/479	0.60 (0.40–0.88)	0.013*	195/472	0.57 (0.41–0.80)	0.001**
Total IgE^c	Lower	135/493	1.41 (0.99–2.00)	0.059	234/479	0.45 (0.31–0.67)	<0.0001**	237/472	0.61 (0.44–0.84)	0.002**
	Higher	206/493	1.31 (0.90–1.89)	0.154	200/479	0.82 (0.57–1.19)	0.339	204/472	0.71 (0.50–0.99)	0.047*

Notes: ^aControls were stratified accordingly when dividing cases into age and gender subgroups, while they were kept as a whole in the situation of the other subgroups. ^bAdjusted for age and gender in logistic regression model. ^cLower: below the 90th percentile of logarithmic total IgE level; Higher: above the 90th percentile of logarithmic total IgE level. *Nominally significant (P < 0.05) but did not withstand Bonferroni correction. **Significant after Bonferroni correction. Dominant model: MW/MM vs WW; MW: heterozygotes; MM: mutation homozygotes; WW: wild homozygotes.

Abbreviations: OR, odds ratio; CI, confidence interval; VAS, visual analogue scale.

Locus–Locus Interactions of SNPs in AR

We investigated the locus–locus interactions of 10 SNPs in TLR2, TLR4 and CD14 genes using MDR analysis. As shown in Table 5, CD14 rs2569191 polymorphism was a significant single-locus model, with a cross-validation consistency (CVC) of 8/10 and a test accuracy of 53.52% (P = 0.0036), and the best interaction model was the three-factor model (TLR2 rs7656411, TLR4 rs1927914 and CD14 rs2563298), with a testing accuracy to 53.08% and a CVC of 6/10 (P < 0.0001), as determined empirically by the permutation testing.

Table 5 Multifactor Dimensionality Reduction Models for Locus–Locus Interactions

Model^a	CV Training	CV Testing	CV Consistency	OR (95% CI)	P
A10	0.5544	0.5352	8/10	1.5555 (1.1468–2.1099)	0.0036*
A3, A10	0.5759	0.5298	6/10	1.8436 (1.3591–2.5009)	< 0.0001*
A1, A6, A8	0.6074	0.5308	6/10	2.4043 (1.7642–3.2768)	< 0.0001*

Notes: ^aA1: rs7656411; A3; rs7682814; A6: rs1927914; A8: rs2563298; A10: rs2569191. *Statistically significant (P < 0.05).

Abbreviations: CV, cross-validation; OR, odds ratio; CI, confidence interval.

Besides, we applied the interaction dendrogram to determine whether there was a synergistic relationship among these polymorphisms in the best model (Figure 1). The closer distance between TLR2 rs7656411 and TLR4 rs1927914 in the diagram indicated a stronger synergistic interaction, which showed that there may be a gene–gene synergistic interaction between TLR2 and TLR4.

Figure 1 Tree diagram of the best genotype models. The distance between single nucleotide polymorphisms (SNPs) indicates the intensity of the interactions. The color indicates the type of interactions. Red, orange, and green denote strong, moderate and weak interactions, respectively. A1: rs7656411; A3: rs7682814; A6: rs1927914; A8: rs2563298; A10: rs2569191.

Gene–Environment Interactions in AR Cases

As shown in Table 6, a total of 237 patients in the case group were asked to answer questionnaires covering environmental factors, including smoking, living or working floors, sunlight exposure, recently renovated or purchased furniture, polluting enterprises nearby, distance to main road, time of using computer, cooking fumes and pet keeping. Since these environmental data are difficult to be obtained from the controls, we only questionnaire-surveyed the cases, and explored gene–environment interactions through the logistic regression in a case-only study.³⁰

Table 6 Statistics of Environmental Factors in AR Case Group

Variables	N	%
Smoke
No	212	89.5
Yes	25	10.5
Living or working floors
< 4 floor	125	59.2
≥ 4 floor	86	40.8
Daytime sunshine
< 4 hours	30	13.4
≥ 4 hours	194	86.6
Recently renovated or purchased furniture
No	110	46.8
Yes	125	53.2
Polluting enterprises nearby
No	176	81.5
Yes	40	18.5
Distance to main road
< 1000 meters	170	78.3
≥ 1000 meters	47	21.7
Time of using computer per day
< 3 hours	143	72.2
≥ 3 hours	55	27.8
Cooking fumes
No	125	54.3
Yes	105	45.7
Pet-keeping
No	219	93.6
Yes	15	6.4

Abbreviation: AR, allergic rhinitis.

The correlations between environmental factors and polymorphisms of TLR4 and CD14 were evident in AR patients (Table 7). Compared with homozygous wild type GG, the proportion of AR patients with TLR4 rs10983755 GA/AA genotype living or working on the ≥ 4 floor was significantly less than that on the < 4 floor (OR = 0.51, 95% CI = 0.27–0.99), indicating a negative correlation between TLR4 rs10983755 and floors (P = 0.048). TLR4 rs11536889 had also a negative correlation with pet keeping (P = 0.043), while AR cases in the pet group were significantly less than those in the non-pet group (OR = 0.12, 95% CI = 0.02–0.93) under the dominant model (GC/CC vs GG). In addition, sunlight exposure exhibited significantly positive correlation with three SNPs (OR = 2.66, 95% CI = 1.12–6.34 for TLR4 rs7873784; 2.22, 1.00–4.90 for CD14 rs2569190 and 2.56, 1.15–5.73 for CD14 rs2569191). However, none of these interactions withstood Bonferroni correction. No significant interactions were observed between genotypes of other SNPs and environmental factors (data not shown).

Table 7 Association Between Polymorphisms and Environmental Factors in the Dominant Model Among AR Case Group

SNPs	Genotype	Floors		Adjusted OR (95% CI)^a	P	Sunlight Hours		Adjusted OR (95% CI)^a	P	Pet-Keeping		Adjusted OR (95% CI)^a	P
		< 4F	≥ 4F			< 4h	≥ 4h			No	Yes
TLR4 rs10983755	GG	111	62	1.00 (reference)	0.048*	20	131	1.00 (reference)	0.730	147	11	1.00 (reference)	0.998
	GA/AA	41	17	0.51 (0.27–0.99)		7	51	1.18 (0.47–2.99)		57	4	1.00 (0.30–3.33)
TLR4 rs11536889	GG	76	49	1.00 (reference)	0.515	16	121	1.00 (reference)	0.268	128	14	1.00 (reference)	0.043*
	GC/CC	44	34	1.21 (0.68–2.15)		14	65	0.64 (0.29–1.41)		82	1	0.12 (0.02–0.93)
TLR4 rs7873784	GG	67	48	1.00 (reference)	0.725	22	95	1.00 (reference)	0.027*	118	5	1.00 (reference)	0.128
	GC/CC	55	36	0.90 (0.52–1.59)		8	94	2.66 (1.12–6.34)		96	10	2.38 (0.78–7.27)
CD14 rs2569190	AA	39	33	1.00 (reference)	0.352	15	64	1.00 (reference)	0.049*	79	5	1.00 (reference)	0.812
	AG/GG	85	53	0.76 (0.42–1.36)		15	129	2.22 (1.00–4.90)		139	10	1.15 (0.37–3.52)
CD14 rs2569191	CC	45	39	1.00 (reference)	0.186	18	76	1.00 (reference)	0.022*	94	6	1.00 (reference)	0.777
	CT/TT	79	46	0.68 (0.39–1.20)		12	116	2.56 (1.15–5.73)		123	9	1.17 (0.40–3.44)

Notes: ^aAdjusted for age and gender in logistic regression model. *Nominally significant (P < 0.05) but did not withstand Bonferroni correction. Dominant model: MW/MM vs WW; MW: heterozygotes; MM: mutation homozygotes; WW: wild homozygotes.

Abbreviations: AR, allergic rhinitis; SNPs, single nucleotide polymorphisms; OR, odds ratio; CI, confidence interval.

Functional Assessment of SNPs

As shown in Table 8, using SNPinfo, HaploReg, RegulomeDB and MirSNP, we performed a functional annotation analysis of these 10 SNPs and estimated that TLR2 rs7656411, CD14 rs2563298 and CD14 rs2569191 all possessed promoter histone marks, enhancer histone marks, changed motifs, and selected expression quantitative trait locus (eQTL) hits. Moreover, for CD14 rs2563298 located in the 3ʹUTR region, using MirSNP and RNAhybrid, we predicted that miR-451 can be bound to rs2563298 A-allele but not to rs2563298 C-allele. RNAfold predicted that rs2563298 A to C substitution led to alteration of CD14 secondary structure, with the MFE decreasing from −6.8 kcal/mol to −8.0 kcal/mol (Figure 2).

Table 8 Functional Annotation Information of 10 SNPs

SNPs	Promoter Histone Marks	Enhancer Histone Marks	DNAse	Proteins Bound	eQTL Results	Motifs Changed	MicroRNA^a	Function Annotation	Score^b
rs7656411	BLD	BLD			12 eQTL results	PU.1, Rad21			5
rs76112010		BLD							4
rs7682814						4 altered motifs			6
rs10983755			BLD		2 eQTL results	AIRE, DMRT4, Tgif1			7
rs11536889		12 tissues	4 tissues	STAT3	4 eQTL results	Ik-1	hsa-miR-1208, hsa-miR-1236	3ʹUTR	4
rs1927914		BLD			4 eQTL results	6 altered motifs			6
rs7873784		BLD			3 eQTL results		hsa-let-7d-3p, hsa-miR-16-1-3p, hsa-miR-4528	3ʹUTR	5
rs2563298	13 tissues	7 tissues			41 eQTL results	Myb, SIX5	hsa-miR-451a	3ʹUTR	6
rs2569190	20 tissues	11 tissues	25 tissues	8 bound proteins	49 eQTL results	5 altered motifs		5ʹUTR	1b
rs2569191	BLD, LIV	4 tissues			2 eQTL results	Mef2, Pou2f2, TATA			5

Notes: ^aPredicted possible binding microRNA by SNPinfo and MirSNP. ^bBased on RegulomeDB.

Abbreviations: SNPs, single nucleotide polymorphisms; BLD, blood; LIV, liver; eQTL, expression quantitative trait locus; UTR, untranslated region.

Figure 2 In silico prediction of CD14 folding structures and minimum free energy (MFE) changes corresponding to rs2563298 A to C allele. The arrows indicated the changes in structure caused by rs2563298.

Discussion

TLRs modulate Th1/Th2 immune balance via a variety of cells closely related to AR, such as dendritic cells, mast cells and regulatory T-cells (Treg). In this study, we explored the associations of genetic variations and environmental factors with the susceptibility to AR in a Chinese Han population. The selected genes, TLR2, TLR4 and CD14, are implicated in the pathogenesis of allergy. We found that polymorphisms of rs2563298 (A/C) and rs2569191 (C/T) in CD14, and rs7656411 (G/T) in TLR2 were significantly associated with AR. In addition, gene–gene and gene–environment interactions may contribute to the development of AR.

TLR2 has a wide recognition spectrum, which can recognize bacterial peptidoglycans, lipoproteins, viral envelope proteins and other microbial components, activate signal transduction pathways, and induce adaptive immunity. In this study, the GT/TT genotype of rs7656411 in TLR2 was associated with AR risk (P = 0.035). The rs7656411 was located in the 3ʹnear gene region, which may participate in mRNA transcription and affect gene expression. TLR2 rs7656411 was first reported to be associated with AR, while this rs7656411 has been confirmed to be associated with asthma in Chinese population.⁷ Few studies have investigated the role of TLR2 SNPs in the etiology of AR.^15,17 The results of studies on TLR2 SNPs and the risk of asthma in different populations are contradictory.^22,31–35 Moreover, meta-analyses analyzed the correlation of four SNPs (rs5743708, rs3804099, rs3804100 and rs4696480) of TLR2 to susceptibility of asthma, suggesting that rs4696480 and rs3804099 were associated with asthma.^9,11

TLR4 acts as a surface receptor on macrophages for bacterial endotoxin or lipopolysaccharide (LPS) in a dose-dependent manner, and activates macrophages to produce cytokines that affect Th1/Th2 balance. A recent study reported that polymorphisms of rs4986790 and rs4986791 in TLR4 were associated with COVID-19 severity, cytokine storm, and mortality,³⁶ while the association between these two SNPs with asthma was almost negative result.^10,31,32 In this study, four SNPs of TLR4 (rs10983755, rs11536889, rs1927914 and rs7873784) were not significantly associated with AR risk. A lack of association between TLR4 SNPs and asthma risk was also reported in a Chinese population with two SNPs (rs10983755 and rs1927914),¹² and a European population with rs11536889;¹³ whereas rs10759930 in TLR4 was associated with AR risk.³⁷ However, more negative results have demonstrated that TLR4 gene polymorphisms may not be strongly associated with allergic diseases; this may be explained by the effect of endotoxin level, the race of the population, and the complexity of the allergic mechanism.

In this study, three SNPs of CD14 were located in the functional region: rs2563298 was located in the 3ʹUTR region, which may participate in the translation of mRNA regulated by microRNA; rs2569190 and rs2569191 were located in the 5ʹUTR region and 5ʹnear gene region, which may be involved in the expression of promoter region. Our study first explored that the polymorphisms of rs2563298 and rs2569191 in CD14 significantly decreased the risk of AR in Chinese population. Studies in Egypt¹⁴ and Pakistan²⁸ confirmed that rs2569191 polymorphism was associated with asthma. No association between rs2569190 polymorphism and AR was also reported in a meta-analysis.²⁹ However, a study from northern China¹⁶ found that the TT genotype of rs2569190 was associated with AR. This contradiction may arise from the low reliability of the results in that study with a small sample size (92 cases and 72 controls) and the environmental differences in regions between northern and eastern China.

In this study, 118 patients (26.1%) with AR had concomitant asthma. AR and asthma are interrelated in many aspects, including epidemiology, physiology, histology, immunopathology and therapeutic principles. Since upper and lower respiratory tract diseases usually coexist in one patient in an interdependent manner,³⁸ the concept of “one airway, one disease” has been proposed.¹ In the stratification analysis, the mutation heterozygous/homozygous genotypes of rs2563298 and rs2569191 in CD14 gene could reduce the risk of AR with and without asthma, and the AR without asthma demonstrated higher associative significance to withstand the Bonferroni correction. We speculate that AR combined with asthma may bring about more serious allergic symptoms than simple AR, suggesting that the genetic polymorphisms do play a protective role in the allergic diseases. A study of French population has suggested that the heterozygous/homozygous genotype of CD14 rs2563298 can reduce the risk of allergic asthma.²² Another study in a Chinese population has shown that the frequency of CD14 rs2569191 A-allele in allergic asthma is significantly lower than that in non-allergic asthma.³⁹ Moreover, we believe that although AR and asthma share many similarities, the inhibitory effects of genetic variations in TLR pathway on AR, asthma or other allergic diseases may differ to some extent, as Micheal et al²⁸ confirmed that in Pakistani adults, the CD14 rs2569191 is significantly associated with atopic asthma, but not AR.

In the stratification analysis, the mutation heterozygous/homozygous of rs2563298 (CA/AA) and rs2569191 (CT/TT) in CD14 reduced the risk of AR in patients with lower IgE levels, while rs2569190 was not associated with total IgE levels. However, the association between CD14 rs2569190 and atopy has been confirmed in studies of diverse populations including French,⁴⁰ Australian⁴¹ and Chinese,⁴² but not in German⁴³ and Japanese⁴⁴ populations. The controversial results may be explained by the various alleles frequencies among races. Furthermore, Martinez et al²¹ have reported that the CD14 rs2569190 C-allele is a risk factor for allergic phenotypes at low levels of endotoxin exposure, whereas the T-allele is a risk factor at high levels of exposure, suggesting CD14 polymorphisms may be associated with allergic sensitization and environmental endotoxin exposure. After Bonferroni correction, this study still showed that CD14 rs2569191 was associated with AR severity evaluated by VAS score. The associations of SNPs with allergic disease severity vary across studies.^33,45 VAS is a tool to evaluate AR severity using subjective indexes which may depend on individual cognition and sensitivity. Therefore, the correlation with disease severity cannot be fully explained from the perspective of SNPs alone, since the severity of AR is related to many other factors, such as quality of life, sleep status, daily activities, workload, course of disease, and so on.

Using MDR model, we discovered that the locus–locus interactions of TLR2, TLR4 and CD14 genes might be associated with the susceptibility to AR. In the presence of CD14, LPS can activate TLR2 or TLR4 signaling pathways to enhance Th2 type immune response.⁴⁶ The interactions between CD14 and allergy-related genes were also found in the Philippine allergic population (CD14, IL4, FCER1B, IL4RA and ADRB2)⁴⁷ and in Korean children with asthma (CD14, IL4RA, IL13, IL13RA1 and CTLA4).⁴⁸

A large number of studies^49–52 have confirmed that environmental factors, including rural farm environment, tobacco smoke, traffic pollution, climate change, nutrition and occupational factors, are related to IgE levels, atopic and allergic diseases. At present, there are many studies on the role of gene–environment interaction in the pathogenesis of asthma in TLR pathway,^10,22–24 but few on TLR pathway with AR.¹⁷ In this study, 237 patients providing data about environmental factors were collected in a case-only study.³⁰ The premise of the application is that genotype and environmental exposure should occur as independent factors in a general population. In this study, we can only evaluate the factor–factor associations in cases but not the main effect between cases and controls. Although none of these interactions withstood Bonferroni correction, the results of the case-only analysis still suggest a possible interaction between gene and environment during the development of AR.

Most of the AR patients were collected from two provinces with a humid climate in East China. Asthma and allergic symptoms are associated with mold and humidity.⁵³ A study of 400 Swedish children showed that low home ventilation rate in combination with moldy odor from the building structure increased the risk of allergic symptoms in children.⁵⁴ Compared with the higher floors, the lower floors are more humid, moldy and ventilated, which may increase the susceptibility to AR. We found that the patients carrying TLR4 rs10983755 GA/AA genotype living on higher floors (≥ 4 F) were less than those living on lower floors (< 4 F), but this result did not pass the Bonferroni correction. According to the function annotation query (Table 8), TLR4 rs10983755 is located in the 5ʹnear gene region, which may enrich DNase hypersensitivity sites, eQTL hits and changed motifs and affect the expression of TLR4 protein. In the TLR signaling pathway, the binding of TLR4 to bacterial endotoxin or LPS was affected by the environmental humidity and moldiness in a dose-dependent manner. Therefore, we suspected that the mutation of TLR4 rs10983755 may have a synergistic effect with environmental humidity and moldiness in the pathogenesis of AR. Similarly, Chen et al⁵⁵ have found gene–environment interactions between the polymorphism rs769214 and moldy odor in AR children. However, more functional research is needed to verify this interaction.

Moreover, we deduced a positive interaction between gene polymorphisms and sunlight exposure in AR. We found that AR patients carrying CD14 rs2569191 CT/TT genotype were more prone to sunshine exposure, whereas CD14 rs2569191 T-allele decreased the risk of AR in this study. The lack of environmental data in the controls may cause data deviation. Therefore, the interaction between CD14 rs2569191 and sunlight exposure in AR, and related TLR pathways, needs to be further studied. Ultraviolet radiation (UVR) from sunlight stimulates anti-inflammatory and immunosuppressive pathways in the skin that modulate psoriasis, atopic dermatitis and some systemic diseases (such as multiple sclerosis, type 1 diabetes and asthma) through the actions of UVR-induced regulatory cells and mediators, including 1,25-dihydroxy vitamin D3, IL-10, and nitric oxide.⁵⁶ The potential beneficial effects of UVR in controlling allergic airways disease have been evaluated in murine asthma models showing that UVR on the skin can repress Th2 response to asthma.⁵⁷ Rueter et al⁵⁸ have reported that exposure to UVR can reduce the risk of eczema in early childhood but has undefined associations with other allergic disease outcomes. Furthermore, it has been discovered that vitamin D, from either sunshine exposure or oral intake, may function in the process of allergic diseases.⁵⁹ However, the association between vitamin D level and AR risk is still controversial.⁶⁰

In addition, pet-keeping seems to protect against AR in this study, though this result failed to challenge Bonferroni correction. Opinions differ a lot about whether pet-keeping can prevent allergic diseases.^61,62 It has been established that the CD14 rs2569190 polymorphism is associated with increased total and specific serum IgE levels in children with frequent contact with pets.⁶³ Interestingly, another study has found that the sensibility to asthma is different between adults keeping cats and dogs.⁶⁴ Thereby, the effects of pet-keeping are influenced by various aspects, like pet species, individual allergic sensitivity and wide environmental exposure to allergen.

Several limitations exist in our study. First, this study only collected AR patients sensitive to dust mites in eastern China, but did not include patients allergic to other allergens (such as pollen, cockroach, and molds), for dust mites were the major allergens in this area. However, other allergens may also arouse various gene–environment interactions. Second, only analysis of gene–environment interactions in AR cases may not be powerful enough to cement the present evidence. A larger sample size and environmental data for healthy controls were required. Finally, functional experiments, especially combined with environmental factors, are required to further explain the correlation between genetic polymorphisms in TLR signaling pathway and AR pathogenesis from the perspective of the mechanism.

Conclusions

The polymorphisms of TLR2 and CD14 and gene–gene interactions in TLR signaling pathway were associated with susceptibility to AR in the Han Chinese population. However, the present results were limited to support the association between gene-environment interactions and AR. Moreover, more environmental data in the general population and functional studies are needed to warrant these findings.

Abbreviations

AR, allergic rhinitis; ARIA, allergic rhinitis and its impact on asthma; CHB, Han Chinese in Beijing; CI, confidence intervals; CVC, cross-validation consistency; DC, dendritic cell; ECP, eosinophil cationic protein; eQTL, expression quantitative trait locus; HWE, Hardy-Weinberg equilibrium; JPT, Japanese in Tokyo; LPS, lipopolysaccharide; MAF, minor allele frequency; MDR, multifactor dimensionality reduction; MFE, minimum free energy; OR, odds ratios; PAMP, pathogen-associated molecular patterns; SNPs, single nucleotide polymorphisms; TLR, toll-like receptor; Treg, regulatory T-cells; UTR, untranslated region; UVR, ultraviolet radiation; VAS, visual analogue scale.

Ethics Approval and Informed Consent

The research protocol complying with the Declaration of Helsinki was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (2015-SRFA-122) and written informed consent was obtained from all participants.

Consent for Publication

We have obtained the informed consent from all patients or their legal guardians.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis, and interpretation, or in all these areas. All authors took part in drafting, revising, or critically reviewing the article; have agreed on the journal to which the article has been submitted; gave final approval for the version to be published; and agreed to be accountable for the contents of the article.

Disclosure

The authors report no conflicts of interest in this work.

Acknowledgments

We are grateful to the participants in this study and all staff involved in the study throughout the years. We also thank associate professor Yong-Ke Cao at the College of Foreign Languages of Nanjing Medical University for professional English-language proofreading of the manuscript.

Additional information

Funding

This study was supported by the grants (No. 81800890 and No. 81970862) from the National Natural Science Foundation of China.