Genetic variants in miR-145 gene are associated with the risk of asthma in Taiwan

Asthma is a chronic airway inflammation disease and the diagnosis and treatment strategies remain difficult. MicroRNAs play important roles in many biological and pathological processes including asthma development. There is no study confirming the contribution of genetic variants in miR-145 to asthma etiology. We hypothesize that single nucleotide polymorphisms (SNPs) in the promoter region of miR-145 may be associated with the risk of asthma in Taiwanese. We used a case–control study to test this hypothesis. In 198 asthma patients and 453 healthy controls, the genotypes of miR-145 rs4705342 and rs4705343 were determined, and the associations of miR-145 genotypes with asthma risk and severity were evaluated. The distribution of miR-145 rs4705342 genotypes between asthma patients and non-asthmatic control groups were significantly different (p = 0.0187). In multivariable logistic regression analysis, compared with the wild-type TT genotype, individuals carrying the variant genotypes had progressively decreased risks of asthma: the odds ratio (OR) for the heterogeneous variant genotype (CT) and homozygous variant genotype (CC) was 0.77 (95% CI 0.55–1.10, p = 0.1788) and 0.41 (95% CI 0.21–0.79, p = 0.0102), respectively (p for trend = 0.0187). In allelic test, the C allele was associated with a 31% reduced risk of asthma (OR = 0.69, 95% CI 0.53–0.90, p = 0.0070). In addition, the rs4705342 variant genotypes were correlated with the symptom severity (p = 3 × 10–5). Furthermore, the variant genotypes correlated with lower miR-145-5p expression level in serum (p = 0.0001). As for rs4705343, there was no differential distribution of genotypes between cases and controls. Our data provide evidence for miR-145 rs4705342 to serve as a novel biomarker for asthma risk prediction.

www.nature.com/scientificreports/ heritability of up to 60-80% [4][5][6] . In 2014, an animal model of human asthma suggested that about two hundred genes may contribute to the etiology of asthma 7 . There have been numerous candidate gene studies investigating the contributions of DNA repair genes, extracellular metabolism genes, cell cycle regulating genes, cytokines and immunological genes to asthma susceptibility, focusing on the gene-environment interactions and endotypes of asthma etiology 6,[8][9][10][11][12][13] . Additional genetic susceptibility loci for asthma remain to be identified. MicroRNAs (miRNAs) are small non-coding RNAs that act as regulators of gene expression as they bind to the 3'-untranslated regions of their target mRNAs and can influence many cellular signaling networks 14 . MiRNA gene SNPs can affect several processes including primary target gene transcription, pri-/pre-miRNA processing, or miRNA-mRNA interactions 15 . MiRNA dysregulation has been associated with various diseases, especially cancer, since miRNA can target genes involved in regulation of cell proliferation and survival, DNA repair and immune response 16 .
MiR-145-5p was previously found to be significantly increased in the plasma of patients with chronic obstructive pulmonary disease and asthma, indicating that plasma miR-145-5p is a specific biomarker of respiratory disease 17 . Ozone, a poisonous form of oxygen, is associated with lots of adverse health effects and significantly increased the expression of a variety of miRNAs, including miR-145 in human bronchial airways 18 . Moreover, in house dust mite (HDM)-induced asthma mice models, HDM increased the expression of miR-145-5p, while miR-145-5p inhibition reduced eosinophilic inflammation, mucus hypersecretion, type 2 helper T cell (Th2) cytokine production, and airway hyperresponsiveness 19 . Furthermore, Liu and his colleagues have found that miR-145-5p was aberrantly overexpressed in airway smooth muscle cells exposed to cytokine stimulation that mimic the etiology of asthma patients 20 . However, although the upregulation of miR-145-5p plays a role in the pathogenesis of asthma, its underlying mechanism remains unclear.
MiR-145 gene is located in the extremely conserved chromosomal region 5q32 21 . In 2013, several SNPs were identified upstream from the transcription start site of miR-145 21 . Among them, rs4705342 and rs4705343 were reported to be functional, with C allele carriers exhibiting relatively higher reporter gene activity by increasing the extent of NF-κB binding [22][23][24] . Although more and more evidence has suggested that miR-145 gene SNPs were associated with cancer risks 25,26 , the influence of miR-145 genotypes on asthma risk has never been reported. In 2021, Tiwari and colleagues reported that the expression of miR-145-5p is associated with the early decline patterns of lung function growth leading to chronic obstructive pulmonary disease (COPD) in children with asthma and additionally increases airway smooth muscle cell proliferation 27 . It adds indirect biological evidence that miR-145 genotypes may play a role in asthma etiology. In this study, we first examine the associations of miR-145 rs4705342 and rs4705343 genotypes with the risk and severity of asthma, then reveal the genotype-phenotype correlation between miR-145 genotypes and serum miR-145-5p expression level.

Methods
Recruiting asthmatic cases and non-asthmatic healthy controls in Taiwan. One hundred and ninety-eight asthmatic cases were recruited at China Medical University Hospital in central Taiwan. Simultaneously, 453 non-asthmatic individuals matched by gender and age were enrolled as controls 10 . The study was approved and supervised by Research Ethics Committee of China Medical University Hospital (CMUH106-REC1-004), and performed in accordance with the Declaration of Helsinki. The symptom severity for asthma was verified by at least two experienced pulmonary physicians according to the Global Initiative for Asthma guidelines 2 . Specifically, the patients are separated into 4 groups based on the level of treatment required to control the symptoms and exacerbations: treated with as-needed inhaled corticosteroid (ICS)-formoterol alone (group 1, mildest), with low-intensity maintenance controller treatment of ICS-formoterol, leukotriene receptor antagonists or chromones (group 2), with low dose ICS-long acting beta2 agonist (LABA) (group 3), and with high dose ICS-LABA (group 4, severest) 2 .
Genotyping. Peripheral blood was collected from all subjects, and their genomic DNA was extracted 28 .
Genotyping methods were described previously 24 . Briefly, the rs4705342 genotype was determined by a TaqMan Assay on an ABI 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA), and the rs4705343 genotype was identified via the polymerase chain reaction-restriction fragment length polymorphism methodology.
Measuring serum miR-145-5p expression. Total RNA was extracted from 45 serum samples using Trizol Reagent (Invitrogen, Carlsbad, CA, USA). The expression levels of miR-145-5p were measured by real-time quantitative reverse transcription-PCR on FTC-3000 real-time quantitative PCR instrument (Funglyn Biotech Inc., Canada) [29][30][31] . The levels of miR-145-5p were normalized to the levels of GAPDH expression and compared with each other. The TT wild-types of miR-145 rs4705342 and rs4705343 were set as 1.0 as reference. Each sample was measured three times. Statistical analysis. The frequencies of rs4705342 and rs4705343 of the control group were estimated by good-of-fit Chi-square test, examining for fitness of Hardy-Weinberg equilibrium. The Student's t-test was used to examine the differential distributions of ages between the case and control groups. The Pearson's Chi-square test was used to examine the differential distribution of various genotypes and the interaction between genotypes with symptom severity. Multivariable logistic regression analysis was used to estimate the adjusted odds ratios (aORs) and 95% confidence intervals (CIs) adjusting for age, gender and smoking behavior. Any p-value less than 0.05 is considered as statistically significant.

Results
Demographics of asthmatic and non-asthmatic groups. The 198 asthmatic cases and 453 noasthmatic controls were frequency-matched on age and gender. There was no significant difference in smoking behavior between the cases and controls (p = 0.7161). For pulmonary functions, both the average ratio of forced expiratory volume in the first second (FEV1) to forced vital capacity (FVC) (FEV1/FVC, %) and the percentage of predicted FEV1 (FEV1%), were lower among the asthmatic cases than among the control subjects (both p < 0.0001). There were 60 (30.3%), 65 (32.8%), 34 (17.2%) and 39 (19.7%) patients belonging to the symptom severity group 1 (mild), 2, 3 and 4 (severe), respectively ( Table 1).
Association of miR-145 genotypes with the risk of asthma. First, the genotypic frequencies of rs4705342 in the control group fit well with the Hardy-Weinberg equilibrium (p = 0.4451, Table 2). Second, the genotypic frequencies of rs4705342 were differentially distributed among the asthmatic cases and the nonasthmatic healthy controls (p for trend = 0.0187). In multivariable logistic regression analysis adjusting for age, gender and smoking behavior, compared with the wild-type TT genotype, individuals carrying the variant genotypes had progressively decreased risks of asthma: the ORs for the heterogeneous variant genotype (CT)  Table 2). The rs4705343 genotypes were not significantly associated with the risk of asthma in any models (Table 3).
Allelic frequency distribution analysis. The allelic frequency analysis showed that individuals with the C allele at rs4705342 were at a significantly lower risk of asthma than those with the T allele (OR = 0.69, 95% CI 0.53-0.90, p = 0.0070) ( Table 4). The rs4705343 alleles were not significantly associated with asthma risk (Table 4).
MiR-145 rs4705342 genotypes were associated with symptom severity. We are interested in whether the rs4705342 genotypes are associated with symptom severity. To answer this question, the asthmatic cases were stratified according to their rs4705342 genotypes and symptom severity. The asthma patients with CT and CC genotypes were pulled together. The results showed that variant genotype (CT or CC) carriers were at a lower risk to suffer from severe symptom than those wild-type (TT) ones (p = 3 × 10 -5 ) ( Table 5). There were no significant associations between rs4705343 genotypes and symptom severity (Table 5).  The results showed that the rs4705342 variant genotypes (CC and CT) were associated with progressively reduced serum miR-145-5p levels than the wild-type genotype (TT): the serum miR-145-5p levels of CT (0.7742) and CC (0.5429) genotype carriers were 23% and 46% lower than that of the wild-type TT genotype carriers (p < 0.0001 for both comparison) (Fig. 1A). The comparison of homozygous variant genotype (CC) with heterogeneous variant genotype (CT) carriers was also significantly different (0.5429 versus 0.7742, p = 0.0074) (Fig. 1A). There was no significant difference of the expression levels of miR-145-5p for the rs4705343 TT, CT or CC genotype carriers (all p > 0.05) (Fig. 1B).

Discussion
Asthma, a chronic and allergic respiratory illness, is caused by the combination of internal and external factors, while miRNAs may serve as critical internal factors 20,32 . MiR-145-5p, together with miR-138, miR-214, miR-371 and miR-544, can modulate the balance of T helper cells in asthma etiology 32 . MiR-145-5p has been suggested as a suppressor of tumorigenesis, although its expression levels were not conclusive in certain types of cancer. Most importantly, miR-145-5p has been found to be significantly up-regulated in asthma 32 . In that study, among 30 asthmatic patients and 25 healthy subjects, serum miR-145-5p level was found to be significantly higher in asthma patients than in healthy subjects 32 . In addition, its over-expression can suppress the expression of Runx3 and regulate the balance between Th1 and Th2 32,33 . To our surprise, there was no report on either the association of miR-145 genotypes with asthma risk or asthma severity.
In the current study, we revealed that the genotypic proportions for TT, CT, and CC of rs4705342 were 44.4, 43.3, and 12.3% in the control Taiwanese population (Table 2). For the first time, the variant genotypes (CT and CC) and the C variant allele of rs4705342 were found to be significantly associated with reduced asthma risks (Tables 2 and 4). For the first time, rs4705342 were associated with asthma severity (Table 5). Moreover, the genotype-phenotype correlation analysis revealed that the rs4705342 C allele is correlated with a lower expression of miR-145-5p in serum from the healthy controls (Fig. 1A). Our studies demonstrated that the miR-145 rs4705342 genotypes can serve as not only a novel predictor for asthma risk, but also a biomarker of asthma severe symptoms. Although the detailed mechanisms of how miR-145 genotypes contribute to the severe symptoms remain elusive, this association may assist in predicting the prognosis of asthma patients for more precise therapy.
The detailed signaling network between miR-145-5p and asthma etiology remains unclear, but there are some clues. In 2015, Liu and colleagues reported that miR-145-5p up-regulation in airway smooth muscle cells can inhibit KLF4, and affect downstream expressions of p21, MMP-2 and MMP-9 20 . In 2017, miR-145-5p and other 4 miRNAs were confirmed to modulate the Th1/Th2 balance in asthma via regulating the expression level of Runx3 in a combinational manner 32 . In 2019, Xiong and colleagues found that miR-145-5p was up-regulated in airway epithelial cells of asthmatic mice and an miR-145-5p antagonist can significantly improve the asthmatic symptoms 34 . MiR-145-5p can promote the HDM-induced release of cytokines and epithelial barrier dysfunction via KIF3A 34 . MiR-145-5p-induced signaling pathways are complex and warrant more investigations. We provided evidence that the variant genotypes and alleles were associated with lower expression of miR-145-5p (Fig. 1A), consistent with a previous report that the presence of the C allele of rs4705342 would attenuate miR-145-5p transcription ability 22 . Since miR-145-5p is up-regulated in asthma patient 32 and is a biomarker of reduced lung function, COPD, asthma, and other respiratory diseases 17 , it follows that lower expression of miR-145-5p would reduce the risk of asthma. In this regard, our observation that the variant genotypes of rs4705342 are correlated with lower serum levels of miR-145-5p is consistent with the observed protective effect of the variant genotypes on asthma risk.
Rs4705342 is located in the intron region of a long non-coding RNA (CARMN) and the promoter region of miR-143/miR-145 cluster. Previous studies and our own data have shown it correlated with miR-145-5p www.nature.com/scientificreports/ expression. GTEx did not find any expression quantitative trait locus (eQTL). Phenoscanner and REALGAR database did not find any significant correlations either, suggesting that this SNP mainly affects miR-145-5p expression, not other transcripts. Bios Consortium database found a significant methylation quantitative trait loci (meQTL) (cg03543120, p = 1.35 × 10 -8 ), and GoDMC query found five highly significant meQTLa for rs4705342 (all p < 10 -77 ) (cg00226225, cg04317047, cg03370704, cg03543120, and cg09660867). All these CpG sites are cis-meQTL, suggesting rs4705342 may induce differential DNA methylation on neighboring CpG sites thereby regulate miR-145-5p expression. It would be important to validate our observation in other populations. We have downloaded the summery statistics of genome-wide association study (GWAS) data of FinnGen Study and GABRIEL and queried the association of miR-145 rs4705342 with asthma risk in these populations of European descent. The FinnGen dataset has 224,737 genotyped and phenotyped participants (including over 20,000 asthma patients) 42 . The genotype dataset included 16,383,262 SNPs (genotyped and inputted). Rs4705342 was included in the dataset and the C allele frequency in asthma cases is 18.94%, and in controls 19.06% (OR = 0.97, 95% CI 0.94-1.01, p = 0.11). This association was in the same direction as ours, although the effect size was much smaller. The GABRIEL study is a large-scale, consortium-based GWAS of asthma 43 . The dataset contains 582,892 SNPs in 10,365 cases and 16,110 controls from 36 studies. Rs4705342 was not genotyped and not included in the dataset. However, we found a tag SNP (rs3733845) of rs4705342 that was in the dataset. These two SNPs and another SNP (rs17723799) are in high linkage disequilibrium (LD) and form a small haplotype block (Fig. 2). MiR-145 rs3733845 was not associated with asthma risk in Taiwan population (OR = 1.02, 95% CI 0.96-1.08, p = 0.54). It should be pointed out that the C allele frequency of rs4705342 is quite different across different ethnic groups (Table 6). East Asians have by far the highest frequency (more than double other ethnic groups). Different ethnic groups often have both unique and common genetic susceptibility loci. It is not surprising that rs4705342 is strongly associated with asthma risk in Taiwanese but not in other ethnic groups. Nevertheless, further validation in independent populations are warranted to confirm the association of miR-145 rs4705342 with the risk of asthma in East Asians and Taiwanese.

Conclusion
In conclusion, this study provides evidence for the first time that the genotypes at miR-145 rs4705342 may serve as a predictor of asthma risk and symptom severity. There is an obvious genotype-phenotype correlation between rs4705342 genotypes and the serum levels of miR-145-5p. MiR-145-5p is a promising target of asthma, and may facilitate the prediction of asthma occurrence and severity (Supplementary information S1).

Figure 2.
Haplotype block structure of miR-145 genomic region. Three SNPs surrounding rs4705342 are in high linkage disequilibrium (LD) and forms a small block. Rs3733845 was genotyped in GABRIEL study and serves as a tag SNP for rs4705342.