Gene expression analysis was carried out using the induced sputum of 18 asthmatic patients and 10 healthy controls. All the study participants completed a detailed, pre-edited questionnaire. The recent Global Initiative for Asthma (GINA) guidelines (www.ginasthma.org) were used to diagnose asthma by a respiratory medicine specialist. The evaluation of asthma severity was done at the time of acquisition of induced sputum samples from the patients based on patient history, including number of exacerbations per year, lung function test results, medical treatment applied, and response to medication. The asthmatic patients were divided into 4 severity groups; however, Global Initiative for Asthma 1, 2 (mild) and GINA 3, 4 (moderate-severe) were aggregated because of the small number of patients. Of the 18 asthmatic patients, 14 regularly used inhaled corticosteroids (ICS): <500 µg/day beclomethasone dipropionate (BDP) or equivalent (n=5), 500-1,000 µg/day BDP or equivalent (n=7), and >1,000 µg/day BDP or equivalent (n=2); while 4 were considered steroid naive. Controls were healthy volunteers with no previous history of asthma or any airway conditions. All subjects participated in a lung function test (PDD-301/S; Piston Inc., Budapest, Hungary) and were assessed for fractional exhaled nitric oxide (FENO) levels (NIOX MINO; Aerocrine, Solna, Sweden). All the healthy patients had a normal lung function and had no respiratory tract infection 4 weeks prior to the analysis. A skin prick test was also performed for common allergens in order to test the presence of atopy, which is a genetic predisposition to develop allergic rhinitis, atopic dermatitis, and/or asthma. The participants' sex, age, smoking habits, and allergic statuses were compared between cases and controls, and between severity groups, but no statistical significance was found.16 Table 1 shows a summary of this study population.

The genotyping analysis included 1,233 unrelated individuals, out of which 522 were asthmatic children (mean age±standard deviation [SD], 10.20±5.31 years; 218 males and 194 females), and 711 healthy controls (mean age±SD, 14.00±11.22; 429 males and 282 females). The patients were all Hungarian (Caucasian), from which about 5%, based on state population data, were of Gypsy origin. Asthma was diagnosed based on the GINA guidelines as previously described. Atopy was identified by a positive skin prick test and/or positive total or specific serum IgE levels. The total and specific serum IgE levels were tested with the 3gAllergy blood tests in the Immulite 2000 Immunoassay system (Siemens Healthcare Diagnostics, Tarrytown, NY, USA). The resulting serum IgE levels were either normal or high based on age-specific ranges (kU/L).

Different types of asthma were defined in patients, except for 131 subjects (25%), where not enough information was available on asthma phenotypes; hence altogether, 391 asthmatic children were involved in the phenotype analysis. Asthma was divided into allergic and non-allergic asthma subtypes. If asthma was not provoked by allergens but allergy was also present, the allergy types were marked. In allergic patients, depending on the types and quantities of allergens, subgroups of indoor, outdoor, or inhalative allergic phenotypes have been created. If asthma exacerbation was provoked by exercise in the medical history of the patients, asthma was categorized as exercise-induced one. If the onset of asthma or asthma exacerbations have been associated with an infection-related acute respiratory illness, asthma was classified as viral-induced one. Non-atopic patients with viral-induced asthma phenotype composed the non-allergic asthma subgroup.17 Indoor allergens included dust mites, mold, pet dander, and cockroaches, whereas outdoor allergens consisted of different types of pollen. Eosinophil cell counts from blood were measured with the Coulter AXM analyzer, of which the normal relative range was between 1%-6%, and the normal range of absolute eosinophil count was between 50-200/µL. The patients had no change of therapy before the blood samples were taken. Neither did they have exacerbations or respiratory infections for at least 4 weeks prior to the blood test. The detailed characteristics of the asthmatic patients participating in the gene association study are presented in Table 2.

The control children were from the Orthopaedic Department in Budai Children's Hospital or from the Urological Department from Heim Pál Hospital, both in Budapest. The controls had no symptoms of asthma or any airway conditions, nor any need for medication.

Written informed consents were signed by all patients or by their parent/guardian. The study was done according to the Declaration of Helsinki and approved by the Ethics Committee of the Hungarian Medical Research Council.

Induced sputum was used for gene expression assays and Western blot analysis. Sputum samples were taken and prepared as previously described by Ungvári et al.16

RNA was isolated successfully from the induced sputum samples of 18 patients and 10 control subjects with the Qiagen Mini RNeasy Kit according to the manufacturer's instructions (Qiagen, Germantown, MD, USA). The cDNA used in the gene expression analysis was produced with the High Capacity cDNA Reverse Transcription Kit from Thermo Fisher Scientific (Thermo Fisher Scientific, Waltham, MA, USA).

Real-time quantitative polymerase chain reaction (PCR) was performed on LATS1, LATS2, MST1, MST2, SAV1, YAP1, TAZ, and β-actin genes using the 7900HT Fast Real-Time PCR System (Thermo Fisher Scientific) in which β-actin was used as an endogenous control, and all results were normalized to it.

Genomic DNA was isolated from the whole blood samples of 1,233 individuals using the QIAamp blood DNA midi kit (Qiagen) or the iPrep PureLink gDNA Blood kit on the iPrep Purification Instrument (Invitrogen, Carslbad, CA, USA). KBiosciences Competitive Allele-Specific PCR (KASP) version 4.0 genotyping assays were used (LGC Genomics, Berlin, Germany) to genotype 14 single nucleotide polymorphisms (SNPs) on the YAP1 gene (Table 3) according to the manufacturer's instructions. PCR reactions were carried out using the 7900HT Fast Real-Time PCR System (Thermo Fisher Scientific).

Western blot analysis was carried out with human induced sputum samples. After sample preparation, cells were lysed in RIPA lysis and extraction buffer (Thermo Fisher Scientific) supplied with Halt protease inhibitor (Thermo Fisher Scientific) at 1x final concentration and centrifuged at 14,000×g at 4℃ for 15 minutes. Total protein content was determined by the Pierce BCA protein assay kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Samples were loaded on 4%-20% Tris-glycine precast gels (Thermo Fisher Scientific) and then blotted onto the Immun-Blot PVDF membrane (Bio-Rad Laboratories, Hercules, CA, USA). After blocking, the following antibodies were used. The primary antibodies were anti-FRMD6, anti-YAP1, and anti-GAPDH (Abcam, Cambridge, UK). The secondary antibodies were polyclonal donkey anti-rabbit IgG HRP (Abcam) and polyclonal goat anti-mouse immunoglobulins HRP (Dako, Glostrup, Denmark). The membrane was treated with Pierce ECL plus substrate (Thermo Fisher Scientific) according to the manufacturer's instructions, and bands were visualized on a standard X-ray film (Kodak, Rochester, NY, USA).

For sputum analysis, normalized gene-expression levels were compared by the Mann-Whitney U or Kruskal-Wallis test as appropriate. Contingency tables were analyzed by Fisher's exact test. Correlation studies were performed by the Spearman non-parametric test. Differences were considered significant if P<0.05. Allele frequencies between the groups of case and control subjects were estimated by allele counting and tested for deviation from Hardy-Weinberg equilibrium (HWE) by the software program DeFinetti (https://ihg.gsf.de/cgi-bin/hw/hwa1.pl). For the significant deviation threshold, we used a P value of <0.05.

SNP data were analyzed using SPSS software version 22 (SPSS Inc., Chicago, IL, USA). Logistic regression analyses adjusted for age and gender were used to evaluate the association between YAP1 genotypes and asthma, its intermediate phenotypes, the discretized (normal/high) serum IgE and discretized (normal/high) eosinophil levels (see at Subjects) or different phenotypes. Additionally, multinomial logistic regression adjusted for age and gender was used for the analysis of YAP1 SNPs and GINA statuses. Confidence intervals (CIs) were calculated at the 95% level. Multiple comparisons were corrected for using the Benjamini-Hochberg correction, and a new significance level of P=0.004 with the false discovery rate (FDR) <6.5% was estimated. Haplotype analysis was carried out with the Haploview 4.2 program (Broad Institute of MIT, Cambridge, MA, USA). Odds ratios (ORs) for haplotypes were counted by VassarStats software (http://vassarstats.net/index.html).

Earlier, we have developed an alternative, a systems biological statistical method, named Bayesian network based Bayesian multilevel analysis of relevance (BN-BMLA). Bayesian networks offer a rich language for genetic association studies, because they exhaustively and exactly represent strongly relevant variables and their interactions through the Markov Blanket Set and Markov Blanket Graph features and they are able to evaluate multiple targets. Furthermore, this Bayesian global relevance analysis method provides posterior probabilities, which are direct statements about hypotheses; thus, it can also be used to construct probabilistic data analytic knowledge bases in genetic association studies to support complex querying, off-line meta-analysis, and fusion with background knowledge.18192021

We have previously described the BN-BMLA method in detail,15222324 so that the following only briefly summarizes this approach.

A Bayesian network is a directed acyclic graph (DAG) that aids the discovery of various dependency relations between random variables by representing their joint probability distribution. A node in the network represents a variable, and edges connecting 2 nodes represent direct dependency between those variables. To find the dependence relations of the variables, a DAG that best describes the dataset must be found. In most cases, there are many DAGs with non-negligible posteriors, but certain structural features may be extracted accurately. Such a feature is based on the concept of strong relevance of a single variable or a set of variables. Bayesian learning allows the evaluation of the strength of the data indicating the presence of a certain feature by evaluating its a posteriori probability.

The a posteriori probability can be calculated for strongly relevant variable sets with regard to a target variable. The strongly relevant variables have direct impacts on the target. The a posteriori probability of the strong relevance is between 0 and 1, where 1 means that the target (e.g., phenotypes of asthma) most certainly has a dependency relationship with a predictor (e.g., SNP), on the other hand 0 means that there is no such relationship. Posterior probabilities of strong relevance greater than or equal to 0.5 are regarded as relevant, and above 0.75 as convincing.18192021

In the present study, 29 SNP in the YAP1, FRMD6, and BIRC5 genes (genotyped with the same methods and on the same populations;1516 Supplementary Table S1) and the characteristics of the patients detailed in Table 2 were involved in the BN-BMLA analysis.

In the induced sputum of 18 asthmatic patients and 10 control subjects, we measured the gene expression level of 7 members of the Hippo/YAP1 pathway. The expression of all genes was detected in both cases and controls. The mean gene expression level of YAP1 was slightly lower in the asthmatic patients than in the control subjects (P=0.032, Fig. 2A). No other differences were detected in this respect. We investigated whether within the asthma group there were differences in gene expression between the subgroups of patients defined by GINA status, but no significant differences were found.

During the correlation studies, we found a significant and positive correlation between YAP1 mRNA levels and sputum bronchial epithelial cells (r=0.575, P=0.003; Fig. 2B). There was a significant and negative correlation between TAZ mRNA and sputum neutrophils (r=−0.509, P=0.009), and STK4 showed a significant and positive correlation with sputum eosinophils (r=0.425, P=0.034). There was no significant correlation of YAP1, TAZ, LATS1, LATS2, SAV1, STK3, or STK4 gene expression with other cellular components, asthma severity, age, gender, airway inflammation, or ICS dose.

We examined whether any of the SNPs in the YAP1 gene influence the susceptibility of asthma or different phenotypes. The statistically significant genotyping results are summarized in Table 4. There was no significant association of the SNPs with asthma susceptibility, allergic status, inhalative, outdoor, indoor allergies, allergic and non-allergic asthma, comorbidities of rhinitis and conjunctivitis or serum IgE and eosinophil levels. However, SNP rs2846836 was significantly associated with exercise-induced asthma (OR=2.1 [1.3-3.4], P=0.004, power=0.83; Table 4 and Supplementary Fig. S1A). Additionally, the distribution of genotypes of SNP rs11225138 showed a significant difference between GINA 1-2 and GINA 3-4 statuses in a dominant model (OR=2.8 [1.4-5.6], P=0.003, power=0.83; Table 4 and Supplementary Fig. S1B).

In order to find more evidence for the associations, we also conducted haplotype analyses. We found a significant difference between patients of GINA 2 and GINA 3 when we compared the frequencies of a haplotype formed by the rare alleles of SNPs rs1426398 and rs11225138, where the frequency of TC haplotype was more prevalent in GINA 3 than in GINA 2 (28% vs 8%, P=10⁻⁷). Furthermore, the CA haplotype from SNPs rs11225138 and rs1426394, also showed a significant difference when patients with GINA 3 were compared to GINA 2 asthmatics (26% vs. 7%, P=10⁻⁷). When we included more than 2 SNPs in the analysis, additional associations were found. Corresponding results are shown in Supplementary Table S2 and Supplementary Fig. S2.

Western blots were carried out on the proteins of the Hippo/YAP1 pathway whose genetic variations showed significant associations with asthma or phenotypes. Earlier, we found a strong significant association between a genetic variation in the FRMD6 gene and asthma (P<0.001, Supplementary Table S1),15 and in the present study genetic variations and haplotypes in the YAP1 gene was associated with different phenotypes of asthma.

The signal for the FRMD6 protein could be detected in all sputum samples from both asthmatic patients and control subjects. Interestingly, although the YAP1 protein was not detected in the sputum samples of the healthy controls, it was seen in the sputum samples of the mild asthmatics (GINA 1, 2) and was absent in the sputum of severe (GINA 3, 4) asthmatics (Supplementary Fig. S3).

Based on the genotyping results of 29 SNPs in YAP1, FRMD6, and BIRC5 genes, the laboratory data and characteristics of the asthmatic patients detailed in Table 2, the posterior probabilities of relevance between the variables with respect to target variables were calculated by BN-BMLA.

Table 5 shows the most relevant variables with high posterior probabilities according to the BN-BMLA analysis. As expected, e.g., IgE levels or inhalative allergy was highly relevant to allergic asthma or eosinophil levels and allergic conjunctivitis to allergic rhinitis.

In the case of genetic variations, no direct SNP-SNP or gene-gene interactions were found. The most relevant association was between rs9671722 in the FRMD6 gene and exercise-induced asthma with a posteriori probability of strong relevance of 0.99. Fig. 3 shows the most likely subgraph of the dependence structure of the variables. This structure suggests a direct relevance of rs9671722 to exercise-induced asthma, while another SNP (rs3751464) of the FRMD6 gene was found to be directly relevant to allergic rhinitis and transitively associated through allergic rhinitis with exercise-induced asthma.

The relationship was confirmed by logistic regression which showed that patients with GG genotypes of the rs9671722 SNP had the highest odds of having both allergic rhinitis and exercise-induced asthma (OR=18.0 [5.9–54.9], P=3.7E-7; Supplementary Table S3). The interaction term in the logistic regression model was significant (P=0.010).

In Supplementary Fig. S4, another graph is presented which shows a dendrogram of the subsets of strongly relevant variables with respect to allergic rhinitis.