Sulfatase modifying factor 1 (SUMF1) is associated with Chronic Obstructive Pulmonary Disease

Forty controls and 82 COPD patients, defined according to GOLD criteria (forced expiratory volume in 1 second (FEV₁)/forced volume capacity (FVC) <0.7), were included in the Lund cohort (Table 1). All subjects were current smokers or ex-smokers with >15 pack-years, had normal levels of alpha-1 antitrypsin, and had no history of asthma, lung cancer, or any other cardiorespiratory diseases. They did not suffer from any lower respiratory infections within 3 weeks prior to the visit. They were asked to refrain from inhaled bronchodilators for 8 h for short acting beta agonists and short acting muscarinic antagonists and 48 h for long acting beta agonists and long acting muscarinic antagonists before the visit. All study participants performed flow-volume spirometry, body plethysmography (MasterScreen Body, Erich Jaeger GmbH), and single breath helium dilution carbon monoxide diffusion (MasterScreen Diffusion, Erich Jaeger GmbH) after bronchodilation (400 μg salbutamol, Buventol Easyhaler®). Lung function measurements were performed according to manufacturer’s instructions and European Respiratory Society/American Thorax Society recommendations [23‐25]. The reference values used were established by Crapo et al. [26] (spirometry), and from Quanjer et al. [27] (Body plethysmography and carbon monoxide diffusion). All subjects signed written informed consent and the study was approved by the Regional Ethics Review Board in Lund.

Sputum was induced by inhalation of 3% saline for 5 min, and thereafter 4.5% saline for 2x5 min. After each step, patients were asked to try to expectorate sputum. Samples were picked for plugs which were incubated with 4 volumes of cold 0.1% dithiothreitol in phosphate buffered saline. After 30 min incubation in 4 °C, additional 4 volumes of phosphate buffered saline were added, and the sample was filtered (60 μm filters). Cells were pelleted at 1000 × g for 5 minutes (4 °C) and lysed for future RNA analysis [28].

A bronchoscopy was performed in 15 COPD patients. Central lung biopsies were sampled from which fibroblasts were isolated as previously published [29].

For examination of RNA from various body tissues, the Human Total RNA Master Panel II Lot# 1505145A (TakaraBio-Clonetech, Saint-Germain-en-Laye, France) was utilized.

For mRNA analyses, RNA was extracted from whole blood, sputum cells, and lung fibroblasts. cDNA synthesis and quantitative real-time PCR (qPCR) was performed as described previously [29].

Multiple protein coding splice variants have been identified for SUMF1, of which the functional role and tissue specificity remains unknown. We focused on three well-established splice variants in SUMF1 (Splice variants 1 (full length), 2 (lacking exon 3) and 3 (lacking exon 8); For primer sequences and NCBI codes see Additional file 1: Table S1) that were predicted at the time of this study. All mRNA expressions were normalized against expression of the reference genes β-Actin and GAPDH (see Additional file 1: Table S1).

To assess associations between the SNPs and SUMF1 gene expression in lung tissue (i.e., cis-acting expression (RNA) quantitative trait loci (cis-eQTL) analysis), the Lung eQTL consortium was used, including lung tissue samples obtained from patients at three participating sites; University of Groningen (GRN), Laval University (Laval) and University of British Columbia (UBC) [6].

Tissue was obtained from patients that underwent lung resectional surgery. DNA samples were genotyped with Illumina Human1M-Duo BeadChip arrays, and gene expression profiles were obtained using a custom Affymetrix microarray. Gene expression data is available on the Gene Expression Omnibus accession number GSE23546 and platform GPL10379.

Imputed SNP data was available for 1,095 of the 1,111 subjects, covariate data was missing for another 8 subjects. In the current analyses, we included current and ex-smokers >40 years with ≥5 pack-years. COPD was defined as an FEV₁/FVC ratio <0.7. Non-COPD control was defined as an FEV₁/FVC ≥ 0.7. In case lung tissue samples were derived from healthy donors, no data on FEV₁ or FEV₁/FVC ratio were available. For FEV₁ and FEV₁/FVC, pre-bronchodilator values were used when post-bronchodilator values were not available. Subjects with other lung diseases such as asthma, cystic fibrosis or interstitial lung diseases were excluded. The final dataset included 512 subjects. Patients provided written informed consent and the study was approved by the ethics committees of the Institut universitaire de cardiologie et de pneumologie de Québec and the UBC-Providence Health Care Research Institute Ethics Board for Laval and UBC, respectively. The study protocol was consistent with the Research Code of the University Medical Center Groningen and Dutch national ethical and professional guidelines.

First, cohort specific (GRN, Laval and UBC) principal components (PCs) were calculated based on residuals from linear regression models on 2-log transformed gene expression levels (of each probe separately) adjusted for age, gender and smoking status (never/ever/unknown). PCs that explained at least one percent of the total variance were saved and included as covariates in the main analysis, these were 14 PCs for GRN and Laval, and 16 for UBC. Second, in each cohort separately, linear regression analysis was used to test for association between the SNPs and 2-log transformed gene expression levels. SNPs were tested in an additive genetic model and the models were adjusted for disease status, age, gender, smoking status and the cohort specific number of PCs. Finally, SNP effect estimates of the three cohorts were meta-analyzed using fixed effects models with effect estimates weighted by the inverse of the standard errors.

A cis-eQTL was defined as a SNP that was significantly associated with expression levels of a probe (gene) within a 50 Kb distance of that SNP. We focused on SNPs which overlapped between eQTL imputed database and Cyto Chip 12, the array used to genotype the Lifelines cohort.

Associations between SUMF1 SNPs and COPD was performed in a Dutch general-population based cohort, the LifeLines cohort study [30]. Subjects with complete genotype and phenotype data (existing data [30]) were included when having smoked at least 5 pack-years and if over 50 years of age. COPD was defined as having FEV₁/FVC < 0.7 and FEV₁%predicted < 80, based on Quanjer et al.[24] with pre-bronchodilator spirometry following European Respiratory Society/American Thorax Society criteria [24]. Controls were defined as having FEV₁/FVC ≥ 0.7 and FEV₁% predicted > 90.

In the Lifelines cohort, genotyping was performed using IlluminaCytoSNP-12 arrays and SNPs were included that fulfilled the quality control criteria: genotype call-rate ≥95%, minor allele frequency ≥1%, and Hardy-Weinberg equilibrium cut-off p-value ≥10⁻⁴. Samples with call rates below 95% were excluded.

Whole blood was taken from all subjects in the Lund cohort and DNA was extracted. All patients were genotyped for the SUMF1 SNPs identified to be top hits in the eQTL analysis and Lifelines using Agena iPLEX genotyping. Genotyping was performed at the Mutation Analysis Facility at Karolinska University Hospital (Huddinge, Sweden) using iPLEX® Gold chemistry and MassARRAY® mass spectrometry system [31] (Agena Bioscience, San Diego, CA, U.S.A.). Multiplexed assays were designed using MassARRAY® Assay Design v4.0 Software (Agena Bioscience). Protocol for allele-specific base extension was performed according to Agena Bioscience’s recommendation. Analytes were spotted onto a 384-element SpectroCHIP II array (Agena Bioscience) using Nanodispenser RS1000 (Agena Bioscience) and subsequently analyzed by MALDI-TOF on a MassARRAY® Compact mass spectrometer (Agena Bioscience). Genotype calls were manually checked by two persons individually using MassARRAY® TYPER v4.0 Software (Agena Bioscience).

The association between SNPs and COPD in the Dutch cohort was performed using logistic regression models including the SNP in a co-dominant genetic model and adjusted for sex, age, and pack years using SPSS version 22.

Finally, associations between the SNPs (in an additive model) and lung function parameters were tested using linear regression adjusted for COPD, smoking status, and age in the Lund cohort (using SPSS version 22).

Table 1 shows the descriptive statistics of the Lund cohort. An adequate sputum sample, from which RNA could be extracted, could be obtained from 38 subjects (19 controls and 19 COPD patients, Additional file 1: Table S2) in the Lund cohort. Additional file 1: Table S3 shows the descriptive statistics of the 15 COPD patients in the Lund cohort that performed a bronchoscopy, and from which lung fibroblasts were obtained.

We found that SUMF1 mRNA was expressed relatively high in whole lung tissue (Figs. 2a-d), and specifically, Splice variant 3 showed the highest expression in lung tissue compared to all other investigated tissues in the body (Fig. 2d).

To examine if SUMF1 expression was systemic or lung specific, sputum cells and whole blood from COPD patients and controls from the Lund cohort were examined for differences in SUMF1 expression. In sputum cells (Figs. 2e-h), all three splice variants examined were detectable and showed significantly lower levels in COPD patients than controls in Splice variant 2 (p = 0.018) and Splice variant 3 (p = 0.0086). While in contrast, in whole blood there was no significant difference in total SUMF1 expression between controls and COPD patients (p = 0.39, Additional file 2: Figure S1), and the three splice variants were unable to be detected in the majority of individuals.

We next performed an expression quantitative trait loci (eQTL) analysis in lung tissue in order to determine whether the differential gene expression of SUMF1 were associated with genetic polymorphisms. In the three large cohorts (Groningen, Laval, and UBC; n = 512) examined, twelve of the SNPs, that overlapped with the array used to genotype the Lifelines cohort, showed significant expression differences (Table 2).

A linkage disequilibrium (LD) analysis (HaploView 4.2) show the associations between the twelve SUMF1 SNPs identified (Fig. 3a).

The top hit SNP from the eQTL analysis, rs11915920 (Fig. 4a) provided a strong eQTLs (Table 2). For further data presentation in this study, the most significant SNP associated with gene expression, i.e., rs11915920, is used for further data presentation in this study.

The SNP rs793391 (Fig. 4b), the most significant SNP from the Lifelines cohort (see below), was also a significant eQTL (Table 2), but to a much smaller extent.

In the Lund cohort, the SUMF1 mRNA levels, of total SUMF1 and the different splice variants, were examined in sputum cells from controls and COPD patients as well as in lung fibroblasts from COPD patients in relation to the SUMF1 genotypes of SNPs rs11915920 and rs793391.

Similar trends in SUMF1 mRNA expression were seen in both sputum cells and fibroblasts with SNP rs11915920 (Fig. 5). Significant differences were observed among the rs11915920 genotypes, with a higher expression level in subjects homozygous for the reference allele (C), in all splice variants in sputum cells (Fig. 5b-d; Splice variant 1: p = 0.017, Splice variant 2: p = 0.038, Splice variant 3: p = 0.015). In lung fibroblasts, the expression of Splice variant 3 was significantly different between the genotypes (Fig. 5h, p = 0.014). These in vitro findings validate the eQTL analysis where there were also higher levels of mRNA expression observed in subjects with the reference allele (C) of rs11915920 (Fig. 4a). The top candidate from our SNP analyses of the Lifelines cohort (see below), rs793391, did not show any association with SUMF1 expression in sputum cells or lung fibroblasts (Additional file 2: Figure S2). rs793391 was a much weaker candidate than rs11915920 in the eQTL analysis and the in vitro analysis corroborates these results.

We also investigated the association between the SUMF1 SNPs associated with eQTLs and COPD in the Dutch cohort, LifeLines (n = 1483, for descriptive statistics see (Additional file 1: Table S4)). Convincingly, the reference allele (A) of SNP rs793391 was associated with a higher risk for COPD in the LifeLines cohort (Table 3). In addition, the SNP rs308739 was also associated with COPD. (For allele frequencies, see Table 4). The most significant SNP from the association between SUMF1 and COPD, rs793391, was chosen to be followed up pathophysiologically in this study.

When examining advanced lung physiology in subjects from the Lund Cohort, including controls and COPD patients, we found that among the rs793391 genotypes there was an overall difference in FEV₁/FVC (p = 0.031), FEV₁%predicted (p = 0.035), diffusion capacity (D _LCO = lung diffusion capacity for carbon monoxide)%predicted (p = 0.027) and alveolar volume (VA)%predicted (p = 0.040). Specifically, subjects homozygous for the reference allele of rs793391 had lower FEV₁/FVC and FEV₁%predicted compared to heterozygous subjects (Fig. 6a and b, respectively). A similar pattern was seen in D _LCO%predicted (Fig. 6c) and VA%predicted among the different SUMF1 rs793391 genotypes, but not in D _LCO/VA%predicted. Interestingly, even after correction for COPD, smoking status and age, the association between rs793391 and D _LCO%predicted remained significant, while the association between rs793391 and FEV₁/FVC, FEV₁%predicted and VA%predicted did not (Additional file 1: Table S5)

.

Neither residual volume, total lung capacity, nor air trapping index (residual volume/total lung capacity) showed any difference among the different genotypes of rs793391 (data not shown).

The SNP rs11915920, highly significant in the eQTL analyses, did not have any significant association with measured lung function parameters (Additional file 2: Figure S3), neither had the SNP rs308739.