Panel 2- Recent Advance in Otitis Media Bioinformatics

Abstract

Objectives.

To update the medical literature on recent large-scale studies employing bioinformatics data analysis tools in otitis media (OM) disease models with a principal focus on developments in the past 5 years.

Data Sources.

Pubmed indexed peer-reviewed articles.

Review Methods.

Comprehensive review of the literature using the following search terms: ‘genomics, inflammasome, microRNA, proteomics, transcriptome, bioinformatics’ with the term ‘otitis media’, and ‘middle ear’. Included articles published in the English language from January 1, 2015- April 1, 2019.

Implications for Practice.

Large scale bioinformatics tools over the past five years lend credence to the paradigm of innate immune response playing a critical role in host defense against bacteria contributing to Otitis Media (OM) progression from acute to chronic. In total, genomic, miRNAomic, and proteomic analyses all point to the need for a tightly regulated innate immune and inflammatory response in the middle ear. Currently, there is an urgent need for developing novel therapeutic strategies to control immunopathology and tissue damage, improve hearing and enhance host defense for both acute and chronic OM based on full understanding of the basic molecular pathogenesis of OM.

Introduction - The usage of Bioinformatics to impact the treatment of human disease

Over the past decade or so, an unprecedented volume of biological data has been produced by large-scale big data efforts such as the human genome project, advances in mass spectrometric based proteomics analyses, next generation genetic sequencing, epigenetics, as well as projects in other organisms, such as the elucidation of health and disease microbiomes. The huge demand for analysis and interpretation of these data is being managed by the evolving science of bioinformatics. Bioinformatics is defined as the application of tools of computation and analysis to the capture and interpretation of biological data and is essential for the understanding, applicability, and management of data in modern biology and medicine. In general, the OM field has lagged relative to other respiratory and infectious disease arenas in terms of large-scale profiling of molecular, genetic, and bacterial processes during disease progression. This is largely due to the fact that OM is a ubiquitous disease process in which discovery and advances span through multiple medical specialties and global geographical regions. The amount of federal funded dollars going to OM basic research is markedly less than other infectious disease entities. Relatively mundane and widely available surgical techniques, such as tympanosotomy tube placement, are considered effective treatments. Yet, OM accounts for the primary reason for antibiotics prescriptions and surgical procedures in the US, representing a burden of more than $2 billion annually to the US healthcare system^1,2. Critically, there are no medications proven effective to reduce mucous accumulation in the ear, leaving only watchful waiting or surgery as treatment options. Beyond concerns related to overutilization of general anesthesia for small children, the over-usage of antibiotics to treat OM with the expanding emergence of resistant pathogens represents a significant societal problem. As such there is a critical gap to identify novel therapeutic approaches for these children. In general, the genetic and molecular underpinnings of OM disease progression from acute to chronic are only recently being elucidated. Knowledge gained from large scale descriptive studies will serve the goal of uncovering much needed novel management strategies. The goal of this review is to update the latest literature in the OM and upper airway field which taps into bioinformatics to further our understanding of OM progression, focusing on the latest knowledge in genomics, miRNAomics, proteomics, microbiomics and transcriptomics.

Methods

A comprehensive review of the biomedical literature was conducted by the 2019 International Symposium for Otitis Media Symposium Research Conference panel 2-Bioinformatics members. Pub-med was used as the search engine. Given instructions provided to the Panel by the symposium organizers, each Panel member conducted an independent search engine query on the basis of his/her area of expertise. Collectively, the following search terms were employed: ‘genomics, inflammasome, microRNA, proteomics, transcriptome, microbiome, bioinformatics’ with the term ‘otitis media’, and ‘middle ear’. Included articles for this state-of-the-art review mostly focused on those published in the English language from January 1, 2015- April 1, 2019.

Genomic associations in otitis media disease susceptibility

Classic studies in twins reveal a substantial heritable component to recurrent acute otitis media (rAOM) and chronic otitis media with effusion (cOME)^3-5. To delineate this genetic component for OM susceptibility, initial widely cited studies by researchers in the United Kingdom identified candidate gene from two mouse models of severe, spontaneous OM ^6,7. The mutations underlying these mouse models mapped to the Evi1 gene and the Fbxo11 gene. Another group in Western Australia evaluated an association between polymorphisms in the human homologues, EVI1 and FBXO11, and susceptibility to rAOM or cOME in a cohort of families treated by the authors (The Western Australian Family Study of OM), and in a unique cohort of children followed longitudinally since birth for 20 years (the Western Australian Pregnancy Cohort Study, or ‘Raine study’)⁸. Using bioinformatics analyses, these studies showed that the association at FBXO11 was independent of associated environmental risk factors for OM (i.e. day care attendance, allergy). As both EVI1 and FBXO11 proteins appear to interact in the transforming growth factorβ (TGFβ) signaling pathway, these data confirm a genetic basis for susceptibility to severe OM and suggest this might be mediated by variants at FBXO11 that result in inflammatory dysfunction by altering regulation of the TGFβ pathway. Two classic studies into genome-wide linkage scans using multi-case families of Caucasian origin identified specific regions of the genome harboring 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 interactions⁹. The second identified two different regions of linkage on 17q12 and 10q22.3.¹⁰. Whilst the region of 19q13.43 has since been refined^11,12, there has been no replication of any of these regions in independent studies and the causal genes underlying them have yet to be identified.

Notably, in the past five years, a Finnish group evaluated 35 candidate genes in 624 cases and 778 controls identified strong association of the TLR4 gene region with risk of childhood OM. The TLR4 gene region identified a novel and previously unrecognized risk haplotype in OM cases, defined by the minor alleles of rs13209060, rs13209057, and rs5030717¹³. The association was strongest in children who had their first OM episode at younger than 6 months which is in accordance with several clinical studies showing that these children will be otitis-prone in over 80 % and in children who had undergone multiple tympanostomy tube insertions. This suggested that the more severe phenotype of OM has the strongest connection with the TLR4 TCG risk haplotype, which further supported a role of TLR4 in OM pathogenesis. The p-value of the association in the subset of patients with multiple tube insertions was of the order of 10⁻⁵, which is one of the most significant associations so far reported in OM candidate gene studies. TLR4 is thought to play a major role in innate immunity against the pathogens causing OM. The major microbial ligand of TLR4 is LPS, a bacterial cell wall component of Gram-negative bacteria. TLR4 also recognizes other microbial antigens such as pneumococcal pneumolysin. TLR4 has a role in the up-regulation of TLR2, which recognizes the peptidoglycan of Gram-positive bacteria. TLR4 is thus important for the innate immune response against the major pathogens causing OM (Haemophilus influenzae, Moraxella catarrhalis and Gram-positive Streptococcus pneumoniae) TLR4 also recognizes endogenous ligands such as heat shock proteins and proteolytically cleaved fibrinogen which may potentially contribute to the pathogenesis of OM. Of note is that TLR4 knock-out mice have decreased pathogen clearance and prolonged inflammation in the middle ear during experimental OM, further supporting a connection between TLR4 and OM risk^14,15. They were unable to replicate this association in a UK cohort or in two US cohorts. This may reflect the heterogeneous nature of OM and differences in different experimental models, and that inter-action of environmental and host factors may modulate which genetic factors contribute to overall predisposition to OM¹³. Similarly, by performing single nucleotide polymorphism (SNP) analyses, a number of groups found that polymorphisms in TLRs along with other regulatory and inflammatory cytokines (IL-10, IL-1α, IL-1β, IL-8) genes are also associated with OM susceptibility^16-18.

Further highlighting the importance of host epithelial and innate immunity in OM responses, recent large-scale genetic susceptibility studies in an indigenous Filipino population and in US children employing novel exome and Sanger sequencing along with bioinformatic data analyses reported a strong association of variants within A2ML1 as significant risk factors for OM. Sixteen novel, damaging A2ML1 variants identified in OM patients were rare or low-frequency in population-matched controls. In the indigenous population, both gingivitis and A2ML1 variants including the known duplication variant and the novel splice variant c.4061+1G>C were independently associated with OM¹⁹. This report suggests a role for A2ML1 in epithelial differentiation within the middle ear as part of OM pathology and the potential application of ROCK inhibition in otitis media. Another recent study from the same multi-institutional group identified variants of the fucose transferase FUT2 gene to also be associated with OM predisposition²⁰. Notably, OM bacterial pathogens were shown to activate FUT2, suggesting that upon bacterial stimulation middle ear glycoproteins, such as mucins, rich in specific glycan fucoses are produced, in turn predictably interacting with neutrophils and other innate immune cells in the middle ear, driving their innate immune defense function.

The Role of MicroRNA responses in airway/middle ear inflammation

Beyond genetic association large-scale analyses, the recent discovery of short sequence RNAs as potent regulators of gene transcription has opened a whole new field of molecular research into epigenetic mechanisms of disease susceptibility. MicroRNAs (miRs) are small RNA molecules, approximately 22 nucleotides long that can negatively control their target gene expression post-transcriptionally. There are currently more than 460 human miRNAs known, and the total number is predicted to be much larger. Although there is no question about the biological importance of miRs as key post-transcriptional regulators of gene expression, understanding the function of a single miR or clusters of miRs using bioinformatics tools may be challenging. First, multiple miRs and targets are often involved simultaneously. Second, miR profiles (just as mRNA transcripts) are cell and context specific^21,22. Third, as further detailed below, the presence of non-canonical miRNA-mRNA binding and diverse mechanisms of miRNA action (e.g. mRNA decay) make the traditional bioinformatics approach of identifying “seeding” sites in specific 3’ untranslated regions (3′ UTRs)²³, insufficient to understand or predict overall miR functional effects. The traditional approach of examining the effect of a particular miR, or a set of candidate miRs, begins with the identification of predicted targets. This process usually involves bioinformatics tools for in silico analyses of miRNA–mRNA interactions based on 3′ UTRs-mediated silencing of gene expression at the post-transcriptional level²³. The probability algorithm is usually built based on the perfect complementarity of the miR seed region, the sequence spanning nucleotides 2 to 7 of the 5’ end of the mature miR, with the seed-match at the 3’ UTR, which is a highly evolutionarily conserved region²³. These bioinformatics tools are often linked to simultaneous mRNA profiles to refine possible miR targets (downregulated genes). However, this target-based approach has become problematic as it does not encompass the diverse mechanisms of action of miRs in cell biology. Indeed, miRs affect multiple non-canonical targets indirectly or directly using imperfect base pair matching^24,25. Multiple reports have also established the ability of miRs to target protein-coding sequences (CDS) and not only those located in 3’ UTR regions^25,26. In fact, some seeding sequences for miRs are only functional when present in CDS²⁶. More importantly, it is also clear now that in addition to post-transcriptional silencing, the molecular effects of miRs include mRNA decay²⁴. Specifically, miRs promote mRNA decay by recruiting deadenylases and decapping factors onto the target mRNAs through a mechanism mediated by GW182/TNRC6²⁴. These miRNA-driven complexes deprotect mRNA ends, thus leading to mRNA decay by both 5′ –3′ and 3′ –5′ RNA exonucleases²⁴. Notably, prior studies comparing effects in mRNA levels (by RNA-seq) and translation efficiency (by ribosome profiling) have demonstrated that miR-induced mRNA decay is the dominant functional consequence in mammalian cells^24,27.

Specific to middle ear and respiratory tract, the role of miRs in disease progression is only recently being studied. The presence of extracellular miRs secreted for cell-to-cell reprogramming in distinct areas of the respiratory tract have now been profiled^28,29. This mechanism of action of some miRs is increasingly recognized as a powerful way to amplify and/or sustain autocrine and/or paracrine effects of a single or clusters of miRs^28-30. Given that miR packaging and exporting is a highly complex and regulated process, the determination of the key miRs in a given sample needs to account for the influence of secreted miRs. For instance, that human respiratory and middle ear secretions contain a baseline population of secreted miRs (microRNome) has been recently described^28,29. These miRs likely exert a homeostatic influence on the gene expression of neighboring cells. Notably, acute disturbances (e.g. viral or bacterial infection) can alter this baseline secretory microRNome, and consequently may affect gene expression of homing and recruited immune cells. Thus, it is clear now that in addition to intracellular miRs, it may be required to assess secretory miRs when analyzing samples in the respiratory system. Indeed, extracellular miRs are in direct contact with airway cells (e.g. epithelium and immune cells) and thus may have a crucial regulatory effect in cellular function via autocrine and/or paracrine actions at baseline and in response to different stimuli. Relevant to the middle ear, miR-378 was recently identified as the miRNA most robustly induced in middle ear exosomes upon NTHi bacterial exposure³¹. Notably, miR-378 has been reported to mediate a robust increased expression of MUC5AC, vascular endothelial growth factor (VEGF), interleukin-8 (IL-8), and Ang-1 in lung epithelial cell lines³², suggesting that its presence in the middle ear may similarly be driving a mucogenic response. A separate recent report explored the role of anti-inflammatory miRNAs molecules, namely miR-146, in the middle ear. Findings suggested that miR-146 intracellular levels correlate to increased middle ear mucosal thickness and inflammation³³. Relevant to OM genetic susceptibility studies is the fact that miR-146 exerts an anti-inflammatory role by shutting off the transcription of TLRs³⁴.

To further elucidate the role of these recently profiled middle miR signatures, miR bioinformatics tools must integrate intracellular and extracellular miR population with experimental data to identify the functionally dominant miR-mRNA networks (e.g. TF co-regulatory networks) in a particular cell system. The later can be then used to elucidate miR mechanisms of action in a cell and context specific manner, and thus provide novel and powerful insights into the pivotal role of miRs in biological processes.

Middle ear cell responses to bacterial stimulation- what do broad profiling studies tell us?

As demonstrated by the genetic and miRNA studies described above, many of the genes or molecular mediators involved in OM disease progression appear to be regulated by bacteria. Along these lines, a number of recent studies have been carried out to identify the genetic factors underlying the regulation of innate inflammatory and host defense response in the pathogenesis of OM using in vitro and in vivo approaches. Val et al^31,35 performed quantitative proteomic analyses using human and mouse middle ear epithelial cells (HMEEC and mMEEC) stimulated with nontypeable Haemophilus influenzae (NTHi). They found that the exosomal microRNA (miRNA) content comprised 110 unique miRNAs, with 5 miRNAs that were significantly induced by NTHi (miR-378a-3p + miR-378i, miR-200a-3p, miR-378g, miR30d-5p, and miR-222-3p), all known to target innate immunity genes. Hong et al³⁶ found that calgranulins (S100A8, S100A9 and S100A12) were up-regulated in HMEEC stimulated with Streptococcus pneumoniae. In addition, they were also found to be elevated in the middle ear mucosa of OM, suggesting the important role of calgranulins in the pathogenesis of recurrent and chronic OM. By performing DNA microarray analysis, Hernandez et al ³⁷ found that approximately 8% of transcripts were significantly changed by NTHi in the mouse model of acute OM. Network and pathway enrichment analysis identified sets of genes involved in activation/negative regulation of immune response, changes in epithelial and stromal cell markers, and the recruitment/function of neutrophils and macrophages. Crompton et al³⁸ identified and characterized a novel OM mouse mutant, edison, from N-ethyl-N-nitrosourea (ENU) mutagenesis program at MRC Harwell. A mutation in the Nisch gene caused edison mice to spontaneously develop chronic OM and hearing loss. Further molecular analyses of the mutation revealed that PAK1, RAC1 and downstream signaling in LIMK1 and NF-κB pathways were involved. Harrison et al³⁹ used the chinchilla model of OM and found that NTHi induced the actin-remodeling complex, Arp2/3, which plays a role in NTHi invasion. These broad profiling studies have elucidated novel pathways, molecules, and proteins from both the host and the bacteria; for the first time making temporal connections between pathogens and the metabolic and environmental status of the middle ear early during acute OM progressing to chronic OM. These findings are likely applicable to many other infectious/inflammatory conditions and contribute to an expansive understanding of host responses to common upper airway pathogens.

Explaining the middle ear ‘inflammasome’

Although pathogenesis of otitis media (OM) is multifactorial, the aforementioned studies point to inflammatory dysregulation playing a role in OM response and susceptibility. As such, it is noteworthy to further describe how large-scale profiling studies over the past five years have elucidated a middle ear ‘inflammasome’. To this end, work from Hernandez, et al profiled the transcriptome profile of the middle ear (ME) mucosa isolated from mice during a complete episode of acute bacterial OM event³⁷. This study identified the major cellular pathways including Toll-like receptors (TLRs), NOD-like receptors (NLRs) receptors and inflammasome genes that are activated during OM potentially activating specific downstream regulatory cascades that contribute to the proliferative inflammatory responses and defense observed during OM.

During OM, the innate immune pattern recognition receptors (PRR) are triggered by pathogen- and danger- associated molecular patters (PAMPs and DAMPs). There are five general classes of PRRs: TLRs, NLRs, absent in melanoma 2 (AIM2)-like receptors, retinoic acid-inducible gene 1 (RIG1)-like receptors, and C-type Lectin receptors. Several other molecules can also serve as PRRs, including RNA POL-III, RAGE and P2XR7⁴⁰. Most TLRs are transmembrane and surveil the extracellular environment, but TLRs 7/8 and 9 are endosomal. TLRs signal thought MyD88 and/or TRIF cascades. As mentioned above, Hafrén et al¹³ and Toivonen et al⁴¹ found that polymorphisms in Toll-like receptors (specifically TLR4 and TLR2) gene loci influences the genetic predisposition to childhood OM. On the other hand, the NLRs family are intracellular receptors. Members of the NLR family (NLRP1, NLRP3, NLRC4, NLRP6, NLRP7, and NLRP12) and the DNA-sensing AIM2 and RIG1 receptors assemble into multiprotein complexes – i.e., the ‘inflammasome’. These cytosolic oligomeric complexes consist of the receptor/sensor molecule, the adapter ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain), and procaspase-1. That subsequently promote caspase-1 activation, a protease controlling the maturation of two inflammatory cytokines (IL-1β and IL-18), as well as pyroptosis, an unusually rapid form of cell death. The inflammasomes are a key component of host-defense and major contributors to a myriad of inflammatory diseases⁴². Using Asc^−/− mice in a bacterial induced OM animal model, Kurabi et al⁴³ showed that the inflammasome plays a critical role in the bacterial clearance, resolution of inflammation and mucosal defense. Lacking the Asc gene resulted in near absence of IL-1β maturation in the middle ear and a reduction in leukocyte recruitment and infiltration to the cavity. In another study, lipopolysaccharide (LPS) injection into the ME significantly elevated the expression levels of IL-1β, NLRP3, ASC and caspase-1⁴⁴. Kariya et al⁴⁵ examined the gene expression of different inflammasome related genes using middle ear tissues from cholesteatoma and chronic otitis patients and found levels of NLRP3, ASC, and caspase-1 were significantly upregulated. It is conceivable that in these patients the increase in levels of NLRP3 expression influences the inflammasome activation threshold. Taken together, these studies indicate that modulation of inflammasome-mediated inflammation in the middle ear represents a novel therapeutic strategy for OM.

Proteomics in otitis media progression

Another approach for the large-scale unbiased profiling of cellular and disease processes using bioinformatics tools is proteomics. Mass spectrometry allows for the extensive identification of proteins and metabolites in samples qualitatively (label free) or quantitatively (with labeled amino acids). The label free analysis of patient middle ear fluids (MEF) has brought abundant information on chronic otitis media (COM) research. In 2010 the use of LC-MS/MS permitted to identify MUC5B mucin protein as the predominant mucin in mucoid MEF⁴⁶. A more extensive analysis of mucoid samples showed the presence of neutrophil extracellular traps (NETs) as an abundant innate immune response found in MEF⁴⁷. Neutrophil proteins such as neutrophil elastase, myeloperoxidase and some other proteases were found to be abundant and account for almost 20% of the identified peptide counts in MEF. Notably, histone proteins were among the most abundant proteins of these samples, that were linked to the presence of NETs. NETs were then characterized by immunofluorescence staining, showing decompacted DNA structures comprising an enlarged neutrophil nucleus and extracellular DNA in shape of long filaments bound to antibacterial proteins (neutrophil elastase and myeloperoxidase). Interestingly, NETs were found to colocalize with MUC5B mucin.

Most MEF from COM are mucoid and very viscous, and a minor number of samples are not viscous, called serous effusion. A study compared these two types of COM MEF with label free proteomics techniques in order to generate hypothesis on different pathogenesis resulting in different fluid characteristics⁴⁸. Mucoid and serous MEF were found to have different compositions and biological signatures. Ingenuity pathway analysis was used to find pathways activated or repressed in these sets of samples. Mucoid MEF showed a neutrophilic signature associated with MUC5B presence and extracellular DNA, confirming the implication of NETs in COM. On the other hand, serous MEF contained a much lower amount of mucins and neutrophil markers, but a higher amount of early innate immunity markers (complement and immunoglobulin proteins) and serum proteins. It was suggested that this difference could be due to the middle ear epithelium propensity to remodel in patients exhibiting mucoid MEF. Proteomics techniques have also been used with in vitro models to study both the acute effect of bacterial infection on the middle ear epithelium and to characterize the effect of chronic bacterial infection on differentiated middle ear epithelial cells. The SILAC method (stable isotope labeling of amino acid in cell culture) was used as a reliable quantitative proteomic method to evaluate the effect of bacterial exposure of middle ear epithelial cells. For this method, cells were cultured in the presence of 13C6, 15N2-labeled lysine and 13C6-labeled arginine that can be specifically detected by mass spectrometry as they have a higher mass than native lysine and arginine. SuperSILAC was used to compare several sets of samples, consisting in mixing unlabeled samples of interest with a common reference of labeled proteins from the same cell type (spike-in). The first study involved mouse middle ear epithelial cells (mMEEC) differentiated at air-liquid interface (ALI), exposed to Non-typeable Haemophilus influenzae (NTHi) lysates chronically during the course of 3 weeks³⁵. Cell lysates were recovered to analyze the change in protein expression overtime in response to the bacteria lysates. Cultures at ALI that formed a thin and junctive epithelium was found to thicken and get loser between 1 and 3 weeks of chronic exposure to NTHi lysates, and the quantitative proteomics analysis showed an induction of proliferation markers as well as proteins involved in epithelial remodeling such as keratins and plakins. The second study used human middle ear epithelial cells (HMEEC) in submerged conditions with the SuperSILAC method, exposed to NTHi lysates between 24hrs and 48hrs, therefore mimicking an acute exposure to bacteria³¹. Interestingly, the most upregulated proteins detected by mass spectrometry were extracellular matrix proteins and RNA processing proteins heterogeneous ribonucleoproteins (hnRNPs) that normally reside in the cell nucleus. The presence of hnRNPs in HMEEC secretions was proved to be due to the secretion of exosomes in response to NTHi lysates exposure, and especially hnRNPA2B1 known to process miRNAs in exosomes.

Mass spectrometry has also been useful to characterize a new primary mouse middle ear epithelial model created from Mulay et al⁴⁹. This study used fresh mouse middle ear cells cultured at ALI that showed phenotypical characteristics of middle ear epithelium in vivo. A proteomic analysis was performed to determine apical secretions specific to this in vitro model, and showed the presence of innate defense proteins such as serotransferrin, neutrophil gelatinase-associated lipocalin, cathepsin D and pulmonary surfactant-associated protein D also present in MEF from COM patients⁴⁷. In addition, as described above, an in vivo model of middle ear infection to NTHi was performed with chinchilla, infecting their middle ear with live NTHi to then collect the middle ear mucosa for proteomic and metabolomic analysis with high-resolution quantitative two-dimensional liquid chromatography-tandem mass spectrometry³⁹. The middle ear epithelium that thickened and showed formation of a bacterial layer on its apical side also demonstrated modification of protein expression. Among the 105 proteins regulated by NTHi infection, 25 proteins were also found in middle ear fluid proteome⁴⁷ including extracellular matrix and innate immunity proteins. Interestingly, neutrophil associated proteins were detected, indicating neutrophil infiltration and relevant with the observation of abundant NETs in middle ear fluids⁴⁷. Interestingly, 14 of the 105 proteins identified had functions in actin morphogenesis, which was shown to participate in mucosal defense against bacterial invasion. This study showed that the actin remodeling in the middle ear epithelium involved the protein Arp2/3 and was required to limit NTHi invasion in epithelial cells.

Microbiome studies in Otitis Media

Over the past 5 years several reports have attempted to characterize the global bacterial profile in the middle ear both in health and disease. In 2017, employing 16S rRNA gene sequencing, Mirami et. al. compared the microbiome of 67 normal middle ear washes vs. 88 COME fluid samples. This study in fact is one the few to report on the microbiome of healthy middle ear cavities. They found that ears with chronic otitis media without active inflammation (dry ear) had similar middle ear microbiomes as the normal middle ears group. Subjects with chronic otitis media with active inflammation (wet ear) had a lower prevalence of Proteobacteria and a higher prevalence of Firmicutes than the normal middle ears. The report concluded that the middle ear microbiome indeed contribute to OM pathogenesis and inflammation, but that longitudinal studies to assess the transition of the bacterial community of the middle ear before and after COM treatment, are warranted⁵⁰. Along those same lines, using similar 16s rRNA gene sequencing techniques, Lappan et. al. characterized the nasopharyngeal microbiome of healthy children in comparison to children with rAOM. They found that some bacterial genera, namely Corynebacterium and Dolosigranulum, are characteristic of a healthy nasopharynx, and that indeed these bacteria may be protective⁵¹. Finally, Krueger et al., also employed 16s rRNA sequencing to characterize the microbiome of 55 middle ear fluid samples. They found that the microbiome of middle ear fluid from children with COME differs according to specific clinical features, such as mucin content, age and presence of hearing loss. Notably, Haemophilus species were found to be more abundant in patients with conductive hearing loss, or in those with more abundant mucin glyocoproteins⁵².

As seen, most of the middle ear microbiome studies to date have employed 16s rRNA sequencing followed by metagen dataset analysis as a methodology. New bioinformatics methods papers have been recently published discussing how low biomass issues can often be a problem in 16s rRNA sequencing/metagen dataset analysis. For OM studies specifically, low bacterial biomass is often a feature of nasopharyngeal and middle ear fluid specimens, and this may in fact confer bacterial detection more difficult to ascertain with contaminants posing a risk to data interpretation^53,54. Other identified bioinformatics problems with microbiome studies have been recently described. These include statistical and computational challenges in the analysis of metagenomics data, such as issues with normalization and quantification of relative taxa, bacterial genes, and metabolic abundances; and multivariate analysis of high-dimensional compositional data⁵⁵. Taken together these issues indicate findings from middle ear microbiome studies need to be interpreted with caution.

Putting it all Together – Current Challenges and Future Directions”

Taken together, large scale bioinformatics tools over the past five years lend credence to the paradigm of innate immune response playing a critical role in host defense against bacteria causing otitis media (OM). Studies of particular relevance in the past 5 years which employed bioinformatics strategies are summarized in Table 1. In total, genomic, miRNAomic, and proteomic analyses all point to the need for a tightly regulated innate immune and inflammatory response in the middle ear. Currently, there is an urgent need for developing novel therapeutic strategies to control immunopathology and tissue damage, improve hearing and enhance host defense for both acute and chronic OM based on full understanding of the molecular pathogenesis of OM. Further bioinformatics challenges include increasing the knowledge of middle ear metabolomics and glycomics and determining how these mediators regulate immunity in the middle ear. A better understanding of the microbiota in the upper airways and the differences in different populations and over time would be needed to design effective strategies to counteract antibiotic resistance. Increasing understanding of the functional implications of some of the novel genetics variants recently uncovered will be important. Profiling middle ear epithelial maturation and response to inflammation employing big data, state-of-the-art methods such as single cell RNA sequencing will also build on the fundamental knowledge of host response specific to the upper airway and middle ear. Finally, developing means of controlling inflammatory and immune dysregulation through molecular targeting, employing recently developed transtympanic delivery systems could radically alter the treatment of this condition by diminishing the need for systemic drug dosing and mitigating unwanted side effects of molecular therapeutics.

Table 1.

Bioinformatics Studies of Relevance 2015-2019

Journal Information	Major findings of the Research	Bioinformatics Strategy
Larson ED, Magno JPM, Steritz MJ et al. A2ML1 and otitis media: novel variants, differential expression and relevant pathways. Hum Mutat 2019	Sixteen novel, damaging A2ML1 variants identified in OM patients were rare or low-frequency in population-matched controls. In the indigenous population, both gingivitis and A2ML1 variants including the known duplication variant and the novel splice variant c.4061+1G>C were independently associated with OM. This report suggests a role for A2ML1 in epithelial differentiation within the middle ear as part of OM pathology and the potential application of ROCK inhibition in otitis media.	Exome and Sanger sequencing of human DNA samples
Santos-Cortez RLP, Chiong CM, Frank DN et al. FUT2 Variants Confer Susceptibility to Familial Otitis Media. Am J Hum Genet 2018; 103:679-690.	Notably, OM bacterial pathogens were shown to activate FUT2, suggesting that upon bacterial stimulation middle ear glycoproteins, such as mucins, rich in specific glycan fucoses are produced, in turn predictably interacting with neutrophils and other innate immune cells in the middle ear, driving their innate immune defense function.	Exome and Sanger sequencing, linkage analysis, and Fisher exact and transmission disequilibrium tests (TDT) of human DNA samples
Val S, Jeong S, Poley M, Krueger A, Nino G, Brown K, Preciado D. Purification and Characterization of MicroRNAs within Middle Ear Fluid Exosomes: Implication in Otitis Media Pathophysiology. Pediatr Res 2017; 81: 911-918.	Val et al performed Mass Spectrometry (MS) and miRNA identification analyses using middle ear effusions (exosomes) from chronic OM patients, and detected 445 unique proteins including mucin MUC5B, heat shock proteins, neutrophil mediators such as neutrophil elastase, and miRNAs (miR-223-3p, miR-451a, miR-16-2p, miR-320e, miR-25-3p). Interestingly, Ingenuity Pathways Analysis (IPA) predicted glycolysis, hypoxia signaling, epithelial junctions remodeling, and actin cytoskeleton signaling activation, indicating the significance of these signaling pathways in the chronicity of OM.	Nanostring, Ingenuity Pathway Analysis, mass spectrometry, uniprot database
Crompton M, Purnell T, Tyrer HEet al. A mutation in Nischarin causes otitis media via LIMK1 and NF-kappaB pathways. PLoS Genet 2017;	Crompton et al identified and characterized a novel OM mouse mutant, edison, from N-ethyl-N-nitrosourea (ENU) mutagenesis program at MRC Harwell. A mutation in the Nisch gene caused edison mice to spontaneously develop chronic OM and hearing loss. Further molecular analyses of the mutation revealed that PAK1, RAC1 and downstream signaling in LIMK1 and NF-κB pathways were involved.	N-ethyl-N-nitrosourea mutagenesis with Next Gen Sequencing
Harrison A, Dubois LG, St John-Williams Let al. Comprehensive Proteomic and Metabolomic Signatures of Nontypeable Haemophilus influenzae-Induced Acute Otitis Media Reveal Bacterial Aerobic Respiration in an Immunosuppressed Environment. Mol Cell Proteomics 2016; 15:1117-1138.	Harrison et al used the chinchilla model of OM and found that NTHi induced the actin-remodeling complex, Arp2/3, which plays a role in NTHi invasion.	high-resolution quantitative two-dimensional liquid chromatography-tandem mass spectrometry
Toivonen L, Vuononvirta J, Mertsola J, et l. Polymorphisms of Mannose-binding Lectin and Toll-like Receptors 2, 3, 4, 7 and 8 and the Risk of Respiratory Infections and Acute Otitis Media in Children. Pediatr Infect Dis J 2017; 36:e114-e122.	The authors found that polymorphisms in Mannose-binding Lecting and Toll-like receptors (TLR2, 3, 4, 7 and 8) gene locus influences the genetic predisposition to childhood OM.	Pyrosequencing for genetic polymorphisms in Mannose-binding lectin structural gene
Rachael Lappan, Kara Imbrogno, Chisha Sikazwe, Denise Anderson, Danny Mok, Harvey Coates, Shyan Vijayasekaran, Paul Bumbak, Christopher C. Blyth, Sarra E. Jamieson, Christopher S. Peacock. A microbiome case-control study of recurrent acute otitis media identified potentially protective becteria genera BMC Microbiol. 2018; 18: 13	They characterized the nasopharyngeal microbiome of healthy children in comparison to children with rAOM. They found that some bacterial genera, namely Corynebacterium and Dolosigranulum, are characteristic of a healthy nasopharynx, and that indeed these bacteria may be protective	16s rRNA sequencing
Minami SB, Mutai H, Suzuki T, Horii A, Oishi N, Wasano K, Katsura M, Tanaka F, Takiguchi T, Fujii M, Kaga K. Microbiomes of the normal middle ear and ears with chronic otitis media. Laryngoscope. 2017 Oct;127(10):E371-E377. doi: 10.1002/lary.26579. Epub 2017 Apr 11	ears with chronic otitis media without active inflammation (dry ear) had similar middle ear microbiomes as the normal middle ears group. Subjects with chronic otitis media with active inflammation (wet ear) had a lower prevalence of Proteobacteria and a higher prevalence of Firmicutes than the normal middle ears. The report concluded that the middle ear microbiome may contribute to OM pathogeneisis and inflammation.	16s rRNA sequencing