Introduction

Pulmonary disease is the most common presentation of infections with nontuberculous mycobacteria (NTM) in otherwise healthy adults. Hence, NTM are considered an emerging public health risk in Australia and around the world (Thomson 2010; Winthrop et al. 2010). Exposure through soil or water is extremely common but only a small subset of the population develops pulmonary NTM disease. Estimates in Queensland, Australia, suggest 22.1 persons per 100,000 in the population will develop the disease (Thomson 2010). Known risk factors include smoking, excessive alcohol intake, past tuberculosis and underlying structural lung disease such as COPD and bronchiectasis (Cassidy et al. 2009; Winthrop et al. 2010). However, these factors do not account for all cases—a notable exception being patients with the “Lady Windermere” phenotype. These individuals are lean and tall, with higher than normal incidence of thoracic abnormalities including scoliosis, pectus excavatum and mitral valve prolapse. The majority of cases are Caucasian (91%) and female (95%), with an average age of 60 years (Kim et al. 2008). Behavioural questionnaires found no links between pulmonary NTM disease and patients’ activities, exposures or habits (Kim et al. 2008). Genetic factors should be investigated further.

The risk of pulmonary NTM disease is increased in individuals carrying certain alleles of genes encoding cytokines including IL-10, TNFα and IL-28B (Affandi et al. 2013). The critical effect of IL10 polymorphisms was replicated in the present cohort (Halstrom et al. 2017). Other studies have linked polymorphisms in natural-resistance-associated macrophage protein 1 (NRAMP1) and cystic fibrosis transmembrane conductance regulator (CFTR) to an increased risk of NTM disease (Jang et al. 2013; Kim et al. 2008; Koh et al. 2005; Tanaka et al. 2007; Ziedalski et al. 2006). Genes of relevance showing no genetic associations with pulmonary NTM disease risk also include VDR and TLR2 (Huang et al. 1998; Park et al. 2008a; Park et al. 2008b; Ryu et al. 2006). An intronic microsatellite repeat polymorphism in TLR2 (Yim et al. 2008) has been linked with pulmonary NTM disease. Mendelian susceptibility to mycobacterial disease (MSMD) is a disease associated with bioactive polymorphisms in genes involved in IFN-γ dependent immunity (IFNGR1, IFNGR2, STAT1, IL12B, IL12RB1, ISG15, IRF8, NEMO and CYBB) (Al-Muhsen and Casanova 2008; Bustamante et al. 2014), but these alleles are not associated with pulmonary NTM disease.

No study yet has investigated possible associations between polymorphisms in P2X ligand gated ion channel 4 (P2X4R), P2X ligand gated ion channel 7 (P2X7R), calcium/calmodulin-dependent protein kinase kinase 2 beta (CAMKK2) and pulmonary NTM disease risk. Their potential for modulation of the inflammatory immune response makes them viable candidates. In particular, P2X7R has been associated with aggressive tuberculosis presentations; P2X4R and CAMKK2 are important in macrophage function and activated killing of phagocytosed bacteria (central to mycobacterial disease pathogenesis), and all three lie in a cluster on chromosome 12 exhibiting some measure of linkage disequilibrium (Goullee et al. 2016).

The P2X7R gene encodes the purinergic receptor P2X7R protein, an ATP-gated ion channel which plays an important role in the innate immune response. When bound by its ligand, extracellular ATP, P2X7R enables the intracellular killing of phagocytosed bacteria (e.g. NTM) in the presence of local inflammatory stress. An important role for the P2X7R has been identified in aggressive forms of tuberculosis (Amaral et al. 2014; Ge and Chen 2016; Wu et al. 2014). A P2X7R knockout mouse model implicated P2X7R in Mycobacterium tuberculosis infection (Santos et al. 2013). These data make P2X7R a candidate for modulation of NTM disease.

P2X4R plays a similar role to P2X7R in immune regulation. The gene encodes the purinergic receptor P2X4R protein, a purinoreceptor activated by ATP allowing the transport of calcium ions across a membrane. P2X4R is generally co-expressed with P2X7R in intracellular lysosomal compartments of macrophages and microglia but can be rapidly trafficked to the surface membrane via the process of endolysosomal secretion (Stokes and Surprenant 2009). P2X4R appears to be involved in the process of phagocytosis, lysosomal destruction and removal of pathogens taken up by macrophages, specifically post-fusion phase exocytosis (Miklavc et al. 2011). Polymorphisms in P2X4R may allow NTM to exploit this system, and so persist in macrophage phagosomes via inhibition of lysosome binding.

CAMKK2 encodes the CaMKK2 protein, a calcium ion activated protein kinase involved in many different pathways, including lymphocyte activation. CaMKK2 may act downstream of P2X4R in the process of exocytosis in macrophages that have taken up NTM, with P2X4R allowing calcium ions to traverse cell membranes activating Calmodulin-dependant pathways. This function may be affected by the total or partial loss-of-function mutations leading to an increased susceptibility to NTM. CAMKK2 is selectively expressed at higher than base levels in monocytes, macrophages and B-lymphocytes in response to lipopolysaccharide (LPS) stimulation (Racioppi et al. 2012). In knockout mice, CAMKK2-null macrophages displayed deficiencies in motility, phagocytosis of bacteria and synthesis of cytokines in response to the TLR4 agonist LPS (Racioppi et al. 2012). Loss of CaMKK2 function uncouples the TLR4 cascade, thereby failing to properly activate the inflammatory response. Poor macrophage function in CAMKK2 knockout-mice establishes the potential for mutations in this gene to affect NTM disease.

Here, single nucleotide polymorphisms (SNPs) and haplotypes spanning P2X7R, P2X4R and CAMKK2 are investigated for associations with the incidence of pulmonary NTM disease in Caucasians. Associations were corrected for the disproportionate incidence of pulmonary NTM disease in females.

Materials and methods

Patients and controls

DNA was obtained from patients attending Greenslopes Private Hospital and Prince Charles Hospital with pulmonary NTM disease between 2005 and 2014. All participants provided a written informed consent to participate in this work, which was approved by the Greenslopes Research and Ethics Committee (Protocol 12/12) in accordance with the National Statement on Ethical Conduct in Human Research. Control samples were supplied by the Institute for Respiratory Health, the University of Western Australia, in accordance with the Royal Perth Hospital Human Research Ethics Committee. This cohort is a representative of the Caucasian Australian population from which patients were also derived.

Genotyping

DNA samples were quantified with Qubit Fluorometric Quantitation technology (Thermo Fisher Scientific, Waltham, MA, USA) and adjusted to 50 ng/uL. Genotyping was performed using TaqMan OpenArray Genotyping Plates (Life Technologies, Grand Island, NY, USA) (Goullee et al. 2016). DNA samples were diluted at 1:1 in TaqMan OpenArray Genotyping Master Mix for 50 cycles of PCR amplification. The output was viewed using OpenArray™ SNP Genotyping Analysis software, and genotypes were allocated manually. Several SNPs were excluded from further analysis, as they did not meet Hardy-Weinberg Equilibrium (HWE) or had a high percentage of no call readings (rs11065456, rs208294, rs208307, rs12299020, rs504677, rs10160951, rs2230912, rs2668252, rs11608486, rs1169719, rs11065502, rs11065503, rs11837114, rs1718120 and rs2686367).

Haplotype construction

Haplotypes were identified using the fastPHASE algorithm (Scheet and Stephens 2006). As per convention, the most frequent allele in the population is denoted 0, with the “minor” allele denoted 1. This system is used to portray haplotypes in this study. During haplotype construction, multiple possible haplotypes were generated with fastPHASE for each individual based on their genotypes. The most probable combination of haplotypes for each was selected. From these, 188 different haplotypes were identified from the combined patient and control samples (n = 353). Haplotypes with a combined frequency of 1% or less across patient and control populations were excluded from further analysis.

Statistical analyses

Statistical analyses were performed in Stata 12 (StataCorp, Collage Station, TX, USA). Univariate analyses evaluating associations between pulmonary NTM lung disease and gender, age, SNP and haplotypes were performed using two-tailed Fisher’s exact or Chi2 tests (χ 2) as appropriate. Multivariate analyses were performed using logistic regression modelling where all included factors associated weakly with NTM in univariate analysis (p < 0.20 cut-off chosen for this exploratory study), followed by a stepwise removal procedure to obtain the model of best fit.

Results

The demographic profile of patients with pulmonary NTM disease

Samples were available from 124 patients with pulmonary NTM disease (attending clinics at the Prince Charles and Greenslopes Private Hospitals, Queensland, Australia) and 229 healthy controls (resident in Western Australia). All participants declared Caucasian ethnicity. Seventy-three percent of the patients and 55% of the healthy controls were female. As this is a significant difference (χ 2, p = 0.001), gender was included in logistic regression models for SNP and haplotypes of interest.

NTM patients were older than healthy controls [median (range) 67 (25–89) vs. 44 (21–75) years respectively]. Whilst some control donors may develop pulmonary NTM lung disease later in life, the impact on the data is considered minor as pulmonary NTM disease is rare. Patients displayed infections from Mycobacterium intracellulare (n = 64), Mycobacterium avium (n = 13), Mycobacterium abscessus (n = 10), Mycobacterium kansasii (n = 3), Mycobacterium triplex (n = 2), Mycobacterium xenopi (n = 1), Mycobacterium simiae (n = 1), Mycobacterium interjectum (n = 1), Mycobacterium shimoidei (n = 1), Mycobacterium terrae (n = 1), Mycobacterium gordonae (n = 1) or M. lentiflavum (n = 1). Some individuals were infected with more than one mycobacterial species (n = 12).

Associations between SNP genotypes and pulmonary NTM disease

Of the SNPs tested, four in P2X7R (rs208288, rs1718125, rs1186055 and rs2857585) were weakly associated with pulmonary NTM disease (p ≤ 0.20) in univariate χ 2 analysis. These SNP and gender met our criteria for inclusion in logistic regression modelling (Tables 1 and 3). The optimal model (model p = 0.0004; R 2 = 0.027) did not retain any SNP, only gender. No SNP from P2X4R or CAMKK2 were associated with NTM in univariate analyses (p > 0.20).

Table 1 No SNP associated with NTM in univariate analyses
Full size table

Associations between haplotypes and pulmonary NTM disease

From the 30 SNPs analysed, 188 unique haplotypes were generated by fastPHASE. Haplotypes present in less than 1% of the cohort were excluded, leaving 27 haplotypes of interest. One haplotype (SH95) was significantly associated with pulmonary NTM disease (χ 2, p = 0.019) on univariate analysis (Table 2). Seven haplotypes (SH5, SH11, SH17, SH70, SH83, SH95 and SH97) displayed associations above our exploratory cut-off (p ≤ 0.20) on univariate analysis and were included in logistic regression models with gender (Table 2). After a stepwise removal process, the final model (model p < 0.0001; R 2 = 0.05) included haplotypes SH95, SH70 and gender (Table 3). The haplotype most strongly associated with NTM disease in multivariate analysis was SH95. Associations between haplotypes and NTM disease were similar when gender was excluded as a variable (model p = 0.0013; R 2 = 0.04) (Table 3).

Table 2 One haplotype (SH95) associated with NTM in univariate analyses
Full size table
Table 3 Logistic regression models define gender and haplotypes as predictors of NTM disease
Full size table

The results of patient radiological findings (nodular bronchiectasis, cavitary or mixed) (χ 2, p = 0.631) and mycobacterial isolates from the lung (fast or slow growing species) (χ 2, p = 0.590) were tested for an elevated co-incidence with SH95. No associations were observed.

Discussion

Exposure to potentially pathogenic mycobacteria from the environment is ubiquitous, so the mechanism behind the rare and selective distribution of infection in the population presents an apparent paradox. Genetic predisposition seems the most likely influence. Here, polymorphisms within three genes involved in the inflammatory immune response were investigated. As pulmonary NTM disease generally develops during the later years of life (average patient age 60 years (Kim et al. 2008)) in individuals without histories of bacterial infection, it is unlikely that factors increasing susceptibility reflect a primary immunodeficiency. Hence, loss-of-function mutations in any of the genes of interest would be poor candidate polymorphisms. The SNP investigated in this study are likely to have more subtle effects. Here, the clearest confounding factor was gender, with 59–95% of the reported cases being in females (Huang et al. 1998; Kim et al. 2008; Winthrop et al. 2010). Gender was included in multivariate analyses here in order to confirm an independent effect from the genes under study.

Of the seven haplotypes, meeting the p ≤ 0.20 cut-off in univariate analyses, SH95 was most clearly associated with pulmonary NTM disease (Tables 2 and 3). None of the three SNP displaying minor alleles (rs12301635, rs7975295 and rs7314454) in this haplotype (present in P2X7R and CAMKK2) were independently associated with the disease (i.e. none met the p < 0.20 exploratory cut-off).

SH70 and SH83 associated weakly with risk and protection (respectively) in univariate analysis and shared several SNP alleles. The three alleles (rs10849849, rs1718125 and rs1186055) distinguishing these haplotypes lie within P2X7R and do not explain the association observed at the haplotype level. We also compared SH97 (potentially protective) with the three haplotypes most closely linked with risk (SH95, SH70 and SH17). No minor alleles of SH97 were shared with either SH95 or SH17, but single minor allele was present in both SH97 and SH70, rs1653609 (P2X7R). As SH97 and SH70 are associated with protection and risk respectively, the impact of a minor allele at rs1653609 remains unclear.

These discrepancies between haplotypes and their contained SNP allele associations may be explained by linkage disequilibrium between the alleles identified and other as yet unidentified alleles. These may be within P2X7R or CAMKK2 regions of the genome or in strong linkage disequilibrium with them. Alternatively, these alleles may require a second polymorphic allele in a pathway-related gene before a contextual loss-of-function phenotype is observed. Further work is required to characterise the relationships between these haplotypes and pulmonary NTM disease.

Whilst haplotypes trending to an association with NTM disease carried minor alleles of P2X7R and CAMKK2, there was no evidence of a similar relationship for P2X4R alleles. Moreover no SNP alleles investigated in P2X4R met the exploratory cut-off (p < 0.20) in univariate χ 2 analysis.

This exploratory study implicates P2X7R and perhaps CAMKK2 (and their protein products P2X7R and CaMKK2) in pathways affecting pulmonary NTM disease, whilst P2X4R was not associated with disease protection or risk. Confirmatory studies in additional cohorts are required to confirm the results obtained. Future work would investigate P2X7R and CAMKK2, genes in linkage disequilibrium with them, and genes encoding products in the same pathways in more depth and in larger cohorts.