Background
Otitis media (OM) is a common childhood disease characterised by the presence of infection (acute otitis media; AOM) or fluid (otitis media with effusion; OME) in the middle ear cavity. Most children experience at least one episode of AOM by 1 year of age with up to 40% experiencing recurrent episodes of AOM (rAOM; ≥3 episodes of AOM in 6 months) or chronic episodes of OME (COME; middle ear effusion (MEE) ≥3 months) in childhood [
1]. Recurrent disease can result in perforation of the tympanic membrane and/or conductive hearing loss, leading to deficits in language development and poor educational outcomes. Treatment for recurrent disease may include insertion of tympanostomy tubes. High prevalence rates of OM result in substantial health care related costs and significant childhood morbidity in many countries [
2].
The causal mechanisms that lead to recurrent disease are poorly understood but familial clustering and high heritability estimates point to a genetic component. Using self-reported AOM history from 2,570 Norwegian twin pairs, Kvaerner
et al. [
3] determined heritability estimates of 0.45 in males and 0.74 in females. Two prospective twin and triplet studies have confirmed these estimates. Casselbrant
et al.[
4] reported heritability estimates of 0.73 for time with MEE in the first two years of life. Likewise, Rovers
et al.[
5] reported heritability estimates of 0.49 at 2 years increasing to 0.71 at 4 years with a concomitant decrease in shared environment estimates from 0.41 at 2 years to 0.16 by 4 years of age. Overall these epidemiological studies confirm that susceptibility to OM has a substantial genetic component that increases with age.
To date, there have been few studies undertaken to pinpoint the genes involved. Two genome-wide linkage scans using multi-case families of Caucasian origin have identified specific regions of the genome harbouring putative susceptibility genes. The first identified two regions of linkage on chromosome 10q26.3 and 19q13.43, and a further region on 3p25.3 after conditioning on the linked regions, suggesting epistatic interactions [
6]. The second identified two different regions of linkage on 17q12 and 10q22.3 [
7]. Whilst the region of 19q13.43 has since been refined [
8], there has been no replication of any of these regions in independent studies and the causal genes underlying them have yet to be identified.
To determine whether these genomic regions are important in recurrent/severe OM in a Western Australian population we have carried out linkage analysis using families who contain at least two individuals diagnosed with rAOM or chronic OME (COME) recruited to the Western Australian Family Study of Otitis Media (WAFSOM) [
9]. To identify the putative disease susceptibility locus in the only region that replicated in this population, association using SNPs was undertaken in the Western Australian Pregnancy Cohort (Raine) Study [
10]. Functional and bioinformatic studies were used to further clarify the putative etiological gene.
Discussion
Genome-wide linkage analyses have highlighted five regions containing OM disease susceptibility loci on chromosomes 3p25.3, 10q22.3, 10q26.3, 17q12, and 19q13.43 [
6‐
8]. Here, using a Western Australian cohort of children diagnosed with recurrent OM, we found evidence of replication of linkage at 10q26.3. We also found borderline evidence to support a region of linkage at 10q22.3 but did not find evidence for linkage at 3p25.3, 17q12, or 19q13.43, or for any region after conditioning on 10q26.3. Therefore, we focused attention on mapping the genes under the chromosome 10q26.3 linkage peak.
To identify the etiological gene/variant under the 10q26.3 peak, we performed fine-mapping using directed analysis of 2,270 genotyped SNPs (10,185 after imputation) spanning this region available on 256 cases and 575 controls within the Raine Study cohort. We focused our search in the region 127 Mb to qter, and found statistical support for variants in several genes/regions across this region contributing to the linkage peak. This is not an unusual phenomenon in complex diseases, where previous studies have also highlighted multiple genes contributing to peaks of linkage [
40]. In this case, for the region of linkage at 10q26.3, at least 49 genes lie within the interval, a number of which could be considered as potential functional candidates for OM susceptibility. Statistical support was observed for independent effects of SNPs at
ADAM12,
DOCK1 and the intergenic region between
TCERG1L and
PPP2R2D.
The
ADAM12 (A Disintegrin and Metalloproteinase domain 12) gene is a member of the disintegrin and metalloproteinase (ADAM) family of proteins. This gene has been implicated in the epidermal growth factor receptor (EGFR) signalling pathway [
41], which is upregulated in human middle ear epithelial cells in response to tobacco smoke exposure [
42]. The ADAM12 protein also interacts with the muscle-specific α-actinin-2 protein, with function centred on myoblast/muscle development showing increased expression during muscle generation [
43]. The DOCK1 (Dedicator of cytokinesis 1) protein has roles in phagocytosis of apoptotic cells in concert with ELMO1 during Rac signalling and cellular migration [
44,
45]. The
DOCK1 gene also has a role during embryogenesis in muscle development with knock-out mouse mutants having decreased skeletal and respiratory muscle tissues [
46]. This suggests the apoptotic function of
DOCK1 could have a subtle role in the cell death of inflammatory factors and in apoptosis of mucous cells after an immune response to OM, whilst the ADAM12 protein interacts with pathways that could influence expression of inflammatory mediators within the middle ear.
The strongest statistical support for association was in the
TCERG1L to
PPP2R2D intergenic region, directly below the peak of linkage on chromosome 10q26.3. Very little is known about the function of the transcription elongation regulator like protein (
TCERG1L) gene. In recent GWAS studies, variants at or near
TCERG1L have been associated with fasting insulin, insulin resistance [
47] and attention deficit disorder [
48]. Hypermethylation of the
TCERG1L promoter region leading to gene silencing has also been observed in colon cancer [
49].
TCERG1L expression is documented [
50] in a variety of tissues, including the brain, lung and eye. In this study, we have also demonstrated that
TCERG1L is expressed in adenoids but not in macrophage or epithelial cell lines, either with or without otopathogen infection. Whilst expression in these cells may have been downregulated during immortalization, analogous to downregulation of this gene in cancer cells [
49], the data suggest that any role
TCERG1L may play in OM susceptibility is unlikely to occur through the innate inflammatory response to otopathogens. On balance,
TCERG1L does not appear to be a strong candidate for OM susceptibility.
In contrast,
PPP2R2D is a particularly interesting candidate gene for OM.
PPP2R2D is a member of the B family of regulatory subunits of the protein phosphatase 2A (PP2A) and is widely expressed at the protein level in the brain, heart, placenta, skeletal muscle, testis and thymus [
51]. This protein has a role as a modulator of the TGF-β/Activin/Nodal pathway [
52], where knockdown of the protein was shown to increase nuclear accumulation and phosphorylation of Smad2. This involvement with Smad2 is of specific interest, as this gene and others in the TGFb pathway have previously been highlighted as candidate susceptibility genes for OM within the WAFSOM cohort [
9]. Furthermore, a GWAS carried out in the Raine Study highlighted at least five other members of the TGFb pathway (
BMP5,
GALNT13, NELL1, TGFB3 and
BPIFA1) as candidates for OM susceptibility [
11].
The strongest signal for association in our study lay within the intergenic region between the
TCERG1L and
PPP2R2D genes. Many association signals in complex diseases have been found to lie within intergenic regions [
40], leading to a search for potential regulatory functions within those regions. Whilst the top SNP (rs7922424) in the intergenic region does not itself lie within a conserved non-coding region, it is in strong LD with rs7087384 and rs7914323, which do lie in highly conserved non-coding regions. In addition, the minor disease associated alleles at rs7922424 and rs7914323 alter putative binding sites for the upstream stimulatory factor 1 (USF1). The USF1 transcription factor is a member of the helix-loop-helix leucine zipper family and is ubiquitously expressed in a variety of cells [
53]. The role of this protein is widespread, ranging from roles in embryonic development [
54] to promoters for a number of activity-induced genes within neuronal nuclei [
55] and as transcription factors that regulate cell-type dependent cellular proliferation [
56]. However, our data favours rs7914323 as the most likely regulatory polymorphism in this region. Not only does this SNP lie within a conserved, non-coding region that could harbor regulatory elements, but our analysis shows that the presence of the minor allele potentially eliminates a TFBS for the cAMP response element binding (CREB) transcription factor and its binding protein (CREBP or CBP). This is interesting in the context of OM as the function of CBP has been linked to the TGFb pathway via the recruitment of EVI1 [
57], which is mutated in a mouse model of OM [
58], although the
EVI1 gene has not been associated with human susceptibility to OM to date [
9,
59].
Overall, analysis of conserved, non-coding regions and putative TFBS sites indicate a number of regulatory elements that lie within this intergenic region that can potentially be influenced by polymorphisms associated with OM susceptibility. It is not possible to determine from these data which gene these regulatory elements may influence. However, the TFBS consensus sequences disrupted by these polymorphisms all lie on the forward strand, upstream of the PPP2R2D promoter. The TCERG1L gene on the other hand is encoded on the reverse strand. Taken together these observations strengthen the evidence for PPP2R2D as the likely gene contributing to linkage at 10q26.3.
Acknowledgements
We would like to thank all the families who have participated in the Western Australian Family Study of Otitis Media. We would also like to thank the Raine Study participants for their contribution to the Raine Study as well as the Raine Study Team for cohort co-ordination and data collection. Thanks also go to Ms Nicole Warrington for her work with the QC analysis of the Raine Study GWAS data. The work was supported by iVEC through the use of advanced computing resources located at iVEC@Murdoch.
This work was supported by funds to SEJ from a Brightspark Foundation Fellowship, a Raine Medical Research Foundation Priming Grant and from core funds to JMB from The Stan Perron Foundation, the University of Western Australia (UWA) and the Western Australian State Government. MSR is supported by an Australian Post Graduate Scholarship. The Raine Study has been supported by the NH&MRC over the last 20 years with funding for Core Management provided by UWA, The Raine Medical Research Foundation at UWA, the UWA Faculty of Medicine, Dentistry and Health Sciences, the Telethon Institute for Child Health Research, the Women and Infants Research Foundation and Curtin University. Raine Study Illumina 660 W-Quad Beadchip Data was supported by the NH&MRC (ID 572613).
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
Conceived and designed the experiments: SEJ JMB MSR. Clinical care and characterization for WAFSOM: SV HLC. WAFSOM recruitment and preparation: ESHS SEJ MSR. Raine Study management: CEP. Adenoid/Tonsil collection and preparation: RT SV HLC. WAFSOM genotyping, cell culture, RNA preparation and RT-PCR expression: MSR. Design and maintenance of in-house genetic database for WAFSOM and bioinformatics CNS pipeline: RWF. Analyzed the data: MSR SEJ JMB. Supervised the work: SEJ JMB CEP. Wrote the paper: MSR SEJ JMB. Reviewed the manuscript. All authors read and approved the final manuscript.