PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of plosonePLoS OneView this ArticleSubmit to PLoSGet E-mail AlertsContact UsPublic Library of Science (PLoS)
 
PLoS One. 2017; 12(8): e0182592.
Published online 2017 August 7. doi:  10.1371/journal.pone.0182592
PMCID: PMC5546581

A network-based method using a random walk with restart algorithm and screening tests to identify novel genes associated with Menière's disease

Lin Li, Formal analysis, Methodology, Writing – original draft,1 YanShu Wang, Formal analysis, Validation,2 Lifeng An, Conceptualization, Funding acquisition, Supervision, Validation, Writing – review & editing,1,* XiangYin Kong, Methodology, Writing – review & editing,3 and Tao Huang, Methodology, Writing – review & editing3
Quan Zou, Editor

Abstract

As a chronic illness derived from hair cells of the inner ear, Menière’s disease (MD) negatively influences the quality of life of individuals and leads to a number of symptoms, such as dizziness, temporary hearing loss, and tinnitus. The complete identification of novel genes related to MD would help elucidate its underlying pathological mechanisms and improve its diagnosis and treatment. In this study, a network-based method was developed to identify novel MD-related genes based on known MD-related genes. A human protein-protein interaction (PPI) network was constructed using the PPI information reported in the STRING database. A classic ranking algorithm, the random walk with restart (RWR) algorithm, was employed to search for novel genes using known genes as seed nodes. To make the identified genes more reliable, a series of screening tests, including a permutation test, an interaction test and an enrichment test, were designed to select essential genes from those obtained by the RWR algorithm. As a result, several inferred genes, such as CD4, NOTCH2 and IL6, were discovered. Finally, a detailed biological analysis was performed on fifteen of the important inferred genes, which indicated their strong associations with MD.

Introduction

Menière’s disease (MD) is a disorder that involves the inner ear with various episodic symptoms, including vertigo, hearing loss, tinnitus and ear fullness, and it is a frustrating condition with a sensation of pressure in the middle ears [1, 2]. In most patients, MD only affects one ear and may induce at least two to three of the symptoms mentioned above, in contrast to other problems in the ear [3].

In the clinic, MD has specific differential diagnosis standards. Two tests have been widely applied in the diagnostic processes, involving two explicit symptoms of MD. A hearing test is one of the most significant diagnostic methods that has been widely applied for preliminary screening, and it tests whether the patients can hear sounds with different pitches and volumes and can tell the difference between similar sounds [46]. A specific test item, named electrocochleography (ECog), is a standard testing method that distinguishes neuropathic and functional hearing disorders and is usually accompanied by an auditory brainstem response test [7, 8]. After the hearing test, to target another specific symptom of MD, vertigo and balance tests are also generally applied for the diagnosis of MD. Balance tests contribute to the functional identification of inner ears. Considering that vertigo is a typical symptom of MD, the function of the inner ears may be disabled in such patients [9]. Generally, during the clinical diagnosis of MD, electronystagmography (ENG) is the most commonly used test and can detect inner ear induced eye movements [10]. In addition to ENG detection, another method, named the rotary chair, has been applied in clinical diagnosis and can reduce the false negative rate of a single test [11]. Other tests, such as the vestibular evoked myogenic potential (VEMP), magnetic resonance imaging (MRI) and computerized tomography (CT), have also been applied to differentially diagnose MD [12].

After the diagnosis of MD, three main treatment methods have been preferred to treat this disease. The first one is effective medication. As mentioned above, the pathogenesis of MD has not been fully revealed, and the usual treatments for such a disease focus on relieving clinical symptoms, such as hearing loss and vertigo [13, 14]. The medicine that may be applied to treat this disease usually provides relief of certain symptoms but does not treat the causes. Therefore, the drug choices rely mainly on diverse symptoms. For example, if nausea and vomiting are main symptoms of MD patients, an antiemetic may be an optimal choice for treatment to avoid vomiting [15]. However, for the patients with severe vertigo symptoms, a diuretic, which contributes to the output of urine, may be the best choice for treatment [16]. In addition to drug therapy, hearing aids have been widely applied as a custom treatment method that contributes to the relief of hearing loss symptoms. With the integration of tutorials and hardware support, hearing aids contribute mainly to balancing the hearing ability in the two ears (healthy and ill), and this also only provides symptom relief [17]. Furthermore, surgery is an option for patients with severe hearing loss and vertigo, targeting the decrease of fluid in the inner ear and relieving specific symptoms [18].

As we have analyzed above, most of the diagnosis and treatment methods for clinical applications concentrate on relieving specific symptoms but do not address the pathogenesis of MD. For the potential pathogenesis of this disease, inflammation induced endolymphatic hydrops have been identified as a secondary pathogenesis of the disease, with the initial triggers remaining not fully understood [19]. Although the initial triggers of the disease have not been validated, various genes and variants have been confirmed to be related to MD, which also suggests a specific family genetic predisposition and implies that genetic factors may play an essential role in the initiation and progression of MD [20, 21]. There are three main groups of genes that have been confirmed to contribute to the initiation and progression of this disease. The first group contains immune-associated genes, particularly innate immune-associated genes that have been widely reported to contribute to the disease. Toll-like receptor coding genes, including TLR3, TLR7, TLR8 and TLR10, have all been directly confirmed to be related to the progression of MD, thus implying a specific role for the immune system during the pathological processes [22]. Apart from immune associated genes, water and ion channel protein coding genes and their regulatory factors have also been widely reported to participate in the pathogenesis. AQP2, AQP4 and AQP5 are three representative water channel protein coding genes that are related to MD and may directly contribute to hearing impairment, deafness, and the severe complications during the pathological processes [23]. In addition, proliferation- and cell survival-associated genes have also been widely reported to contribute to the disease. The NOTCH and NF-KB signaling pathways have been confirmed to be abnormally regulated by specific variants during the initiation and progression of this disease, implying an abnormal proliferation of certain cell subtypes that may also contribute to the pathogenesis [24]. Apart from these three subgroups of genes, functional genes such as COCH and the DFNA family that do not belong to a particular functional category participating in the initiation and progression of MD, thus validating the irreplaceable role of a genetic background for MD [25, 26].

As mentioned above, the genetic background has been confirmed to play a specific role during the initiation and progression of MD and related complications. Currently, the identification and validation of MD associated genes depend mainly on the genetic screening of clinical patients and their families. However, since the pathological mechanisms of such diseases are quite complicated and involve various aspects of the biological processes, it is quite difficult and time-consuming to identify each specific MD by experimental dependent genetic screening [27]. With the development of bioinformatics, some computational methods have been presented to contribute to the identification of similar disease associated genes [28] and other related problems [2931]. Among them, a network-based method is an important type of computational method [3237], such as Guilt-by-association (GBA)-based methods [3840], the random walk with restart (RWR) algorithm [4143], and the shortest path algorithm [4351]. This study also built a network-based method to identify novel MD-related genes. A protein-protein interaction (PPI) network was constructed using the PPI information retrieved from the STRING database [52]. Then, the RWR algorithm was applied to the network to search for possible genes by setting known MD-related genes as seed nodes. Furthermore, a series of screening tests, including a permutation test, an interaction test and an enrichment test, were designed to pick out essential genes from the genes yielded by the RWR algorithm. Several inferred genes were produced and were deemed to be closely related to MD. A biological analysis of fifteen important inferred genes was performed, validating their strong relationships with MD and uncovering the potential molecular processes that these genes may participate in.

Materials and methods

2.1 Materials

MD-related genes were collected from the literature indexed by PubMed (http://www.ncbi.nlm.nih.gov/pubmed/). The keywords “Menière’s disease” and “gene” were used to search the literature in PubMed, resulting in 120 papers (January, 2017). Among these papers, 72 papers reported novel MD associated genes, from which we accessed 84 genes, which are provided in S1 Table. According to these screened out papers, there are three principal methods that have been applied to identify MD-associated genes: (I) sequencing (either high-throughput sequencing or Sanger sequencing) together with pedigree analysis in MD families; (II) in situ immune-histochemical localization of target genes and gene products of clinical samples; and (III) in vitro cloning and expression of target genes in proper models together with functional validation. Because we used the PPI network reported in the STRING database, in which Ensembl IDs were adopted to represent proteins, all 84 genes were mapped to their Ensembl IDs, producing 106 Ensembl IDs, which are also provided in S1 Table. These Ensembl IDs were used to search for novel MD-related genes.

2.2 PPI network

Many proteins participating in intracellular and intercellular biological processes are always formed as protein complexes to execute their normal functions, such as the functionally active hemoglobin molecule, which is composed of four subunits, each of which is a protein monomer that has its own tertiary structure [53]. Furthermore, proteins that can form a PPI always share related functions or co-locate in same metabolism pathways [5459]. Therefore, some useful information can be mined from the set of PPIs, which can uncover properties, functions and interaction relationships of proteins. Based on PPI information, several computational methods have been built to predict the properties of proteins, such as protein functions [5658, 60], disease genes [6165], and protein phenotypes [66]. Thus, we also adopted PPI information to infer the novel genes associated with MD in this study.

The STRING database [52] is a well-known public database for both direct (physical) and indirect (functional) PPIs that are derived from (1) genomic contexts, (2) high-throughput experiments, (3) (conserved) co-expression, and (4) previous knowledge. It is easy to see that these PPIs can widely measure the associations between proteins. They were employed in this study to construct the network. All PPIs covering 1,133 organisms in STRING were collected in a file labeled ‘protein.links.v9.1.txt.gz’, from which we extracted 2,425,314 human PPIs involving 20,770 proteins. Each PPI contains two proteins, represented by Ensembl IDs, and one score that indicates the strength of the interaction. For two given proteins pa and pb, the score of the interaction between them was formulated as S(pa, pb), with a larger score value meaning that an interaction between the proteins was more likely to occur.

The constructed network, denoted as G, defined 20,770 proteins as nodes, and each edge in G represented a human PPI; i.e., two nodes were connected if and only if their corresponding proteins were composed of a PPI reported in STRING. In addition, to indicate different roles of edges in G, each edge was assigned a weight that was defined as the score of its corresponding PPI.

2.3 RWR algorithm

The RWR algorithm [41] is a type of ranking algorithm. It has been deemed a useful tool to expand novel objects from known ones. This algorithm always simulates a random walker starting from a seed node or a set of seed nodes, representing known objects, and it calculates the probability of each node being a novel object. For the identification of novel genes associated with MD, the known genes mentioned in Section 2.1 were deemed to be seed nodes, on which the RWR algorithm would be used to infer novel ones. The RWR algorithm repeatedly updated a probability vector Pi that contained 20,770 components, each of which indicated the probability of a node in G being a novel MD gene. In the initialization of the RWR algorithm, P0 was constructed by setting the components of the corresponding seed nodes to 1/106 and the others to zero. The subscript of Pi represents the number of loops that had been run; i.e., Pi representing the probabilities after the i-th round of the loop had been run. Pi can be updated by the following formula:

Pi+1 = (1 − c)ATPicP0
(1)

where A was the column-normalized adjacency matrix of G, and c was the restart probability (it was set to 0.8 in this study to indicate the importance of known MD genes). The loop stopped when || Pi+1Pi || < 1E-06 [41], indicating the probability vector was stable. The probability vector Pi+1 was output as the outcome of the RWR algorithm.

Based on the outcome of the RWR algorithm, each node received a probability of being a novel gene associated with MD. A higher probability meant that the corresponding gene was more likely to relate to MD. For wide detection, we set the threshold of 1E-05 to the probability; i.e., genes with output probabilities larger than 1E-05 were selected as possible genes. For convenience, we called them RWR genes in the following context.

2.4 Screening tests

After the RWR algorithm was executed on the PPI network mentioned in Section 2.3, some RWR genes with probabilities higher than 1E-05 were found. However, there may be false positives among them, so this section presents a series of screening tests to control for this possibility, thereby obtaining the most related genes.

Permutation test

It is clear that the utility of the RWR algorithm is strongly based on the PPI network. The topological structure of the PPI network may cause the selection of some false positives. Obviously, these types of RWR genes are not closely related to MD. To exclude these genes, a permutation test [64, 67, 68] was utilized. First, 1,000 Ensembl ID sets (namely, S1, S2,…, S1000) were constructed, and each of the sets consisted of 106 randomly selected Ensembl IDs from the network. Second, for each set, the RWR algorithm was applied on the PPI network with Ensembl IDs in the set as seed nodes, thus providing a probability for each RWR gene. Finally, a measurement called the p-value was calculated for each RWR gene based on the probability yielded by the RWR algorithm on 106 MD associated genes and 1,000 probabilities yielded by the RWR algorithm on 1,000 randomly produced sets. It can be computed by

p − value(g) = Θ/1000
(2)

where Θ is the number of randomly produced Ensembl ID sets on which the probability of the RWR gene g is higher than that of the 106 MD associated genes. It is clear that RWR genes with high p-values are not special for MD because they can be produced by several randomly produced sets. According to the widely accepted significance level in statistical analysis, 0.05 was used as the threshold of the p-value; i.e., RWR genes with p-values greater than or equal to 0.05 were screened out. The remaining RWR genes were called candidate genes, which would be further checked by the tests mentioned below.

Interaction and enrichment test

The purpose of this study was to identify novel genes associated with MD. Among the candidate genes, some had a strong association with MD, while others had a weak association. To mine the most related candidate genes, two tests, namely, the interaction test and the enrichment test, were built in this section to directly or indirectly measure the association between the candidate genes and MD.

The first test was built based on the PPI information mentioned in Section 2.2. It has been widely accepted that two proteins that can interact with each other are more likely to share related functions. Thus, candidate genes that can interact with at least one MD associated gene are more likely to be novel MD associated genes. For each candidate gene g, a measurement, namely, the maximum interaction score (MIS), was computed by

MIS(g) = max{S(gg′):g′ is a MD-related gene}
(3)

where S(g, g′) represents the interaction score of g and g′. Clearly, candidate genes with high MISs can interact with an MD-related gene with a high probability, implying they may be novel MD-related genes. Because 900 is set to be the cutoff of highest confidence in STRING, it was also set to be the threshold of MIS; i.e., candidate genes with MISs greater than or equal to 900 were selected.

Gene ontology (GO) [69] can clearly describe a given gene and its product based on three aspects: molecular function, biological process, and cellular component. On the other hand, the Kyoto Encyclopedia of Genes and Genomes (KEGG) [70] provides many biological pathways that include several genes. The second test was built based on the GO terms and KEGG pathways of candidate genes and MD-related genes. It is clear that the MD associated genes must be related to some common GO terms and KEGG pathways. Additionally, some GO terms and KEGG pathways have no relationship with these MD associated genes. If a candidate gene exhibits a similar relationship with GO terms and KEGG pathways to those of MD-related genes, it is more likely to be a novel MD-related gene. According to the enrichment theory of GO terms and KEGG pathways [30, 31, 71, 72], the relationship between a gene g and GO terms or KEGG pathways can be encoded as a numeric vector, denoted by FV(g). The proximity of two genes g and g′ on GO terms and KEGG pathways can be measured by the direction cosine of vectors FV(g) and FV(g′), which can be formulated as

MFS(g)=FV(g)FV(g)FV(g)FV(g)
(4)

According to the arguments mentioned above, for each candidate gene, we should measure its relationship to all MD-related genes in this regard and take the maximum to imply its association with MD. Thus, another measurement, namely, the maximum enrichment score (MES), was calculated for each candidate gene g, which was defined by:

MES(g) = max{Γ(gg′):g′ is a MD-related gene}
(5)

Obviously, a larger MES means that several overlapping GO terms and KEGG pathways are shared by the candidate gene g and an MD-related gene. A threshold of 0.8 was set for the MES in this study; i.e., candidate genes with MESs larger than 0.8 were selected.

Of the candidate genes filtered by the permutation test, those with MISs greater than or equal to 900 and MESs larger than 0.8 were finally selected. They were deemed to be of special interest for MD. For convenience, they are called inferred genes.

Results

To clearly illustrate all procedures of the network-based method for the identification of genes associated with MD, a flowchart is shown in Fig 1. This section shows the detailed results yielded by the different procedures of this method.

Fig 1
The flowchart of the network-based method to identify novel MD-related genes.

The RWR algorithm was applied on the PPI network constructed in Section 2.2 by setting the 106 Ensembl IDs of MD-related genes as seed nodes. As a result, each node in the network was assigned a probability of it being a novel MD-related gene. 1E-05 was set as the threshold of probability, resulting in 4,514 RWR genes. These genes, together with the probabilities yielded by the RWR algorithm, are listed in S2 Table.

As mentioned above, a large number of RWR genes were found by the RWR algorithm. However, some false positives were selected due to the structure of the network. In fact, they have little relationship with MD. Thus, the permutation test mentioned in Section 2.4 was applied to evaluate each RWR gene. A measurement called the p-value was calculated for each RWR gene, which is listed in S2 Table. Genes with p-values less than 0.05 were selected, producing 1,069 candidate genes. Compared with the RWR genes, the candidate genes are more likely to relate to MD. The 1,069 candidate genes are provided in S3 Table.

For the 1,069 candidate genes, an interaction test and an enrichment test were used to further evaluate each candidate gene and select the most important candidate genes among them. Two measurements: MIS and MES (cf. Eqs 3 and 5), were yielded by these two tests. Each of the 1,069 candidate genes received these two measurements, with the results provided in S3 Table. Values of 900 and 0.8 were set to be the thresholds of MIS and MES, respectively, yielding the 43 inferred genes listed in S4 Table. These 43 genes are deemed to be tightly associated with MD. To partly elaborate this fact, a bipartite subgraph of the PPI network, using inferred and known MD-related genes as nodes, is shown in Fig 2, which indicates that each inferred gene is closely related to at least one known gene, indicating its close relationship with MD. The interactions used in Fig 2 are available in S5 Table.

Fig 2
The relationship between the inferred genes and known MD-related genes in the PPI network.

Discussion

Relying on the network-based method, we identified 43 functional genes that may participate in MD associated biological processes. As described in Section 1, three specific biological processes have been widely reported to participate in MD: (1) immune associated biological processes; (2) cell surface channel associated processes, and (3) proliferation and cell survival associated biological processes, involving different groups of functional genes. From the 43 inferred genes, we chose fifteen for our analysis, which are listed in Table 1. These genes have been shown to contribute to two of the aforementioned three biological processes according to recent publications.

Table 1
The details of fifteen important inferred genes.

4.1 Immune associated genes

Among the fifteen genes listed in Table 1, eleven genes were shown to contribute to immune associated biological processes. The specific gene CD4 (ENSP00000011653) is a potential pathological gene for MD. As a membrane glycoprotein of T lymphocytes, CD4 mainly contributes to MHC class-II antigen/T-cell receptor interactions, regulating the activation of T cells [73]. Based on recent publications, the interactions between MHC II molecules of antigen presenting cells (APCs) and specific T cell receptors have been confirmed to contribute to MD associated immune reactions [74]. During the initiation and progression of MD, it has been reported that the proportion of CD4+ T cells have been increased, indicating the specific function of such interactions for MD [74]. Further, CD4 molecules have been reported to participate in allergy associated biological processes [75]. Because allergy is definitely associated with MD, it is quite reasonable to conclude that CD4 may be a potential MD-related gene [76, 77].

Apart from CD4, there are three genes encoding functional components of the interleukin family, together with one specific gene encoding a respective receptor. IL6 (ENSP00000258743), as the abbreviation of interleukin 6, is one member of the interleukin family that may contribute to the initiation and progression of MD. As a specific functional cytokine, it mainly participates in the regulation of immune responses, hematopoiesis, platelet production and bone resorption [7881]. Although few publications have revealed the relationships between IL6 and MD, a recent publication confirmed that during the pathological processes of MD, the serum level of IL6 together with IL1 is directly related to the specific complications of MD; e.g., vertigo, implying the underlying relationships between IL6 and MD [82]. Apart from IL6, IL1 may also contribute to MD in similar ways. IL-1R1 (ENSP00000233946), encoding a function component of the IL1 receptor complex, has also been predicted to be an inferred gene that may participate in MD. Binding to interleukin-1, which just has been confirmed to contribute to the pathogenesis of MD, such genes mainly participate in the regulation of IL-1 associated activation of NF-kappa-B, MAPK and other functional signaling pathways [8385]. There are still no direct interactions that can be revealed between IL-1R1 and MD. However, a recent publication confirmed that genes and proteins that contribute to interleukin-1 signaling pathways may be related to a specific clinical symptom of MD, i.e., sudden sensorineural hearing loss [86]. Another component of the interleukin family that has also been screened to be an inferred gene is IL10 (ENSP00000412237). IL10 is a cytokine synthesis inhibitory factor and a co-stimulator for the proliferation and differentiation of T and B mast cells [87, 88]. For the relationships between IL10 and MD, although no direct relationships between such genes and the disease have been reported, a specific report on autoimmune hearing loss validated that the abnormal secretion of IL-10 may contribute to specific hearing loss symptoms of experimental autoimmune hearing loss [89]. Because hearing loss is a typical symptom of MD, which is also widely considered an autoimmune disorder, it is quite reasonable that IL10 may also participate in the pathological processes of MD [89, 90]. Another gene IL18 (ENSP00000280357) has been widely reported to contribute to the Th-1 mediated cellular immunity and may stimulate interferon gamma production in Th-1 cells [91]. According to recent publications, it has been confirmed that specific Th-1 mediated immunological responses may be associated with sensorineural hearing loss and MD, implying that as a key regulator of Th1 cells, IL18, may also be a specific MD-related gene [92]. Therefore, these four interleukins indicate that interleukins may definitely be a specific group of functional regulators during the initiation and progression of MD.

Three chemokines encoding genes have also been predicted as MD-related genes by our network-based method. CCL2 (ENSP00000225831), as a functional chemokine that attracts monocyte and basophils, has been widely reported to participate in monocyte proliferation associated disease, such as psoriasis and rheumatoid arthritis [93, 94]. Although the gene CCL2 has no direct relationship with MD, a publication confirmed that the distribution and proliferation of monocytes are regulated by CCL2. Therefore, it may cause pathological processes in the human endolymphatic sac, thereby inducing vertigo, tinnitus and hearing loss [95]. Considering that the human endolymphatic sac associated pathological processes has been confirmed to be associated with MD, it is quite reasonable to suggest that CCL2 may also play an irreplaceable role in MD [96, 97]. For another chemokine-associated gene, CXCL9 (ENSP00000354901), which has been widely reported to participate in the regulation of cell growth, movement or activation status was also identified as an inferred gene for MD [98100]. Similar to CCL2, CXCL9 has been reported to participate in monocyte proliferation associated disease, including psoriasis [93]. CXCL9, as a specific monocyte-associated gene, may also participate in MD in a similar way to CCL2, as we analyzed above. For CXCL10 (ENSP00000305651), similar to CCL2 and CXCL9, it gene has also been widely reported to contribute to the chemotactic regulation of monocytes and T lymphocytes, thus indicating its potential function during the pathological processes of MD [101, 102]. Apart from such potential regulatory mechanisms, recent publications have also confirmed that CXCL10 may directly contribute to immune mediated apoptosis in the ear, inducing human presbycusis, which has also been considered a severe complication of MD, thus implying the potential role of CXCL10 during the pathological processes of MD [103].

Furthermore, we also identified three functional components of the innate immune response. TLR2 (ENSP00000260010) is a member of the Toll-like receptor family, which mainly contributes to pathogen recognition and innate immune activation [104, 105]. Recruiting MYD88 (ENSP00000379625), another inferred gene, TLR2 is mainly involved in the innate immunity against Gram positive bacteria [106, 107]. According to recent publications, the initiation and progression of MD have been widely confirmed to be associated with the innate immune system and bacterial infection [22, 76, 108]. Although no direct relationship between TLR2 and MD has been reported, considering the specific relationship between gram-positive bacteria and MD, it is quite reasonable to regard TLR2, which mediates innate immunity against gram-positive bacteria, as a potential MD-related gene [109, 110]. Encoding the downstream recruited functional component of TLR2, another inferred gene MYD88 may also be a candidate gene of MD [111]. Another innate immune associated gene, as the homologue of TLR2, TLR9 (ENSP00000353874) has also been screened out as an inferred gene. Different from TLR2, which always interacts with gram-positive bacteria, TLR9 is a nucleotide-sensing TLR that identifies unmethylated cytidine-phosphate-guanosine (CpG) dinucleotides [112, 113]. Recent publications have confirmed that polymorphisms in toll-like receptor (including TLR9) is related to MD, which indicates that TLR9 may be an MD-associated gene [22]. In addition, TLR9 has been confirmed to contribute to the recognition of auto-antigens and induce auto-immune inner ear disease with hearing loss, a specific complication of MD, thus validating its special role in MD [114].

4.2 Proliferation and cell survival associated genes

In Table 1, four specific genes were confirmed to contribute to the proliferation and cell survival of certain cell subtypes, inducing their special contributions to the pathological processes of MD. NOTCH2 (ENSP00000256646), a member of the Notch family, mediates cell-cell interactions and contributes to the cell fate decisions of certain cell subtypes [115, 116]. No direct contributions of NOTCH2 have been made to the initiation and progression of MD. However, based on recent publications, it is quite interesting that the development of auditory hair cells may be quite significant for congenital MD, related to hearing loss and vertigo [117, 118]. Therefore, as the core regulator of auditory hair cells, NOTCH2 and its related biological processes may definitely participate in the pathological processes of MD [119]. Two specific glutathione peroxidase encoded genes, GPX4 (ENSP00000346103) and GPX5 (ENSP00000392398), have also been predicted to be inferred MD-related genes. These two genes encode two functional glutathiones contribute to the catabolic pathway of activated oxygen species, free radical detoxification [120, 121]. Although they seem to be homologues, they can participate in quite different biological processes. GPX4 plays a functional role during the regulation of primary T cell responses against viruses [122, 123]. Considering the underlying relationship between MD and T cell mediated anti-virus immune response, which has been widely reported, it is quite reasonable to regard GPX4 as a potential MD-related gene [124126]. For GPX5, such genes protect cells and enzymes from oxidative damage, especially in the sperm membrane lipids [127, 128]. Considering that oxidative stress and damage have been widely reported to contribute to the pathogenesis of MD, as a regulator and protector against oxidative damage, GPX4 may contribute to MD [129131]. Another inferred gene, PTGS2 (ENSP00000356438), has also been predicted to be an inferred MD-related gene. Also known as cyclooxygenase, this gene mainly contributes to the biosynthesis of prostaglandin as both dioxygenase and peroxidase [132]. As a hormone regulator, PTGS2 contributes to the synthesis of prostaglandin, regulating its specific biological functions [132, 133]. According to recent publications, prostaglandin, which is regulated and synthesized by PTGS2, has a direct relationship with fluctuating hearing loss, a typical symptom of MD, showing the underlying interactions between PTGS2 and MD [134].

Based on the above analysis of fifteen inferred genes, they directly or indirectly participate in the biological processes associated with MD, implying high probabilities of them being novel MD-related genes. For the rest of the inferred genes, we did not discuss our analysis in this report and only provided this in S4 Table. Interested investigators can perform further validations.

Conclusions

In this study, a network-based method was built to predict putative genes related to MD. Forty-three inferred genes were obtained that could play important roles in the pathogenesis of MD. These newly obtained genes, together with the already-known genes, may not only broaden the scope of known MD genes in human but also clarify the potential pathogenic mechanisms of MD. Furthermore, they also shed light on the diagnosis and therapy of this disease.

Supporting information

S1 Table

Genes associated with Menière’s disease and their Ensembl IDs, sources.

(DOCX)

S2 Table

The 4,514 RWR genes with probabilities higher than 1E-05.

(DOCX)

S3 Table

The 1,069 candidate genes with permutation FDRs less than 0.05.

(DOCX)

S4 Table

The 43 inferred genes with probabilities higher than 1E-05, permutation FDRs less than 0.05, MISs greater than or equal to 900 and MESs larger than 0.8.

(DOCX)

S5 Table

The interactions used for drawing Fig 2.

(DOCX)

Acknowledgments

This study was supported by the National Natural Science Foundation of China (81100702), Health and Family Planning Foundation of Jilin Province (20152046), and Science and technology Development plan Foundation of Jilin Province (20160101070JC), Science and technology Development plan Foundation of Jilin Province (20150204084SF).

Funding Statement

This work was supported by the National Natural Science Foundation of China (81100702), Health and Family Planning Foundation of Jilin Province (20152046), and Science and technology Development plan Foundation of Jilin Province (20160101070JC), Science and technology Development plan Foundation of Jilin Province (20150204084SF). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability

Data Availability

All relevant data are within the paper and its Supporting Information files.

References

1. Luzzi R, Belcaro G, Hu S, Dugall M, Hosoi M, Cacchio M, et al. Improvement in symptoms and cochlear flow with Pycnogenol (R) in patients with Meniere's disease and tinnitus. Minerva Med. 2014;105(3):245–54. [PubMed]
2. Mick P, Amoodi H, Arnoldner C, Shipp D, Friesen L, Lin V, et al. Cochlear Implantation in Patients With Advanced Meniere's Disease. Otol Neurotol. 2014;35(7):1172–8. [PubMed]
3. Doobe G, Ernst A, Ramalingam R, Mittmann P, Todt I. Simultaneous Labyrinthectomy and Cochlear Implantation for Patients with Single-Sided Meniere's Disease and Profound Sensorineural Hearing Loss. Biomed Research International. 2015;2015:457318 doi: 10.1155/2015/457318 [PMC free article] [PubMed]
4. McNeill C, Freeman SRM, McMahon C. Short-term hearing fluctuation in Meniere's disease. International Journal Of Audiology. 2009;48(8):594–600. doi: 10.1080/14992020802716778 [PubMed]
5. Silverstein H, Wazen J, Van Ess MJ, Daugherty J, Alameda YA. Intratympanic gentamicin treatment of patients with Meniere's disease with normal hearing. Otolaryngology-Head And Neck Surgery. 2010;142(4):570–5. doi: 10.1016/j.otohns.2009.12.009 [PubMed]
6. Perez N, Rama-Lopez J. Vestibular function at the end of intratympanic gentamicin treatment of patients with Meniere's disease. J Vestibul Res-Equil. 2005;15(1):49–58. [PubMed]
7. Hornibrook J, Bird P, Flook E, O'Beirne GA. Electrocochleography for the Diagnosis of Meniere's Disease: The Wrong Stimulus. Otol Neurotol. 2016;37(10):1677–8. doi: 10.1097/MAO.0000000000001206 . [PubMed]
8. Lamounier P, de Souza TS, Gobbo DA, Bahmad F Jr. Evaluation of vestibular evoked myogenic potentials (VEMP) and electrocochleography for the diagnosis of Meniere's disease. Brazilian journal of otorhinolaryngology. 2016. doi: 10.1016/j.bjorl.2016.04.021 . [PubMed]
9. Tassinari M, Mandrioli D, Gaggioli N, Di Sarsina PR. Meniere's Disease Treatment: A Patient-Centered Systematic Review. Audiol Neuro-Otol. 2015;20(3):153–65. doi: 10.1159/000375393 [PubMed]
10. Le CH, Truong AQ, Diaz RC. Novel techniques for the diagnosis of Meniere's disease. Curr Opin Otolaryngo. 2013;21(5):492–6. doi: 10.1097/MOO.0b013e328364869b [PubMed]
11. Trine MB, Lynn SG, Facer GW, Kasperbauer JL. Intratympanic gentamicin treatment: preliminary results in two patients with Meniere's disease. Journal of the American Academy of Audiology. 1995;6(3):264–70. . [PubMed]
12. Seo YJ, Kim J, Kim SH. The change of hippocampal volume and its relevance with inner ear function in Meniere's disease patients. Auris, nasus, larynx. 2016;43(6):620–5. doi: 10.1016/j.anl.2016.01.006 . [PubMed]
13. Hoa M, Friedman RA, Fisher LM, Derebery MJ. Prognostic implications of and audiometric evidence for hearing fluctuation in Meniere's disease. The Laryngoscope. 2015;125 Suppl 12:S1–12. doi: 10.1002/lary.25579 . [PubMed]
14. Tyrrell JS, Whinney DJ, Ukoumunne OC, Fleming LE, Osborne NJ. Prevalence, associated factors, and comorbid conditions for Meniere's disease. Ear and hearing. 2014;35(4):e162–9. doi: 10.1097/AUD.0000000000000041 . [PubMed]
15. Atrache Al Attrache N, Krstulovic C, Perez Guillen V, Morera Perez C, Perez Garrigues H. Response Over Time of Vertigo Spells to Intratympanic Dexamethasone Treatment in Meniere's Disease Patients. The journal of international advanced otology. 2016;12(1):92–7. doi: 10.5152/iao.2016.2177 . [PubMed]
16. Basura GJ, Lin GC, Telian SA. Comparison of second-echelon treatments for Meniere's disease. JAMA otolaryngology—head & neck surgery. 2014;140(8):754–61. doi: 10.1001/jamaoto.2014.1249 . [PubMed]
17. McNeill C, Freeman SR, McMahon C. Short-term hearing fluctuation in Meniere's disease. Int J Audiol. 2009;48(8):594–600. . [PubMed]
18. Yokota Y, Kitahara T, Sakagami M, Ito T, Kimura T, Okayasu T, et al. Surgical results and psychological status in patients with intractable Meniere's disease. Auris, nasus, larynx. 2016;43(3):287–91. doi: 10.1016/j.anl.2015.10.007 . [PubMed]
19. Gurkov R, Pyyko I, Zou J, Kentala E. What is Meniere's disease? A contemporary re-evaluation of endolymphatic hydrops. Journal of neurology. 2016;263 Suppl 1:S71–81. doi: 10.1007/s00415-015-7930-1 ; [PMC free article] [PubMed]
20. Lee JM, Kim MJ, Jung J, Kim HJ, Seo YJ, Kim SH. Genetic aspects and clinical characteristics of familial Meniere's disease in a South Korean population. The Laryngoscope. 2015;125(9):2175–80. doi: 10.1002/lary.25207 . [PubMed]
21. Teranishi M, Uchida Y, Nishio N, Kato K, Otake H, Yoshida T, et al. Polymorphisms in genes involved in oxidative stress response in patients with sudden sensorineural hearing loss and Meniere's disease in a Japanese population. DNA Cell Biol. 2012;31(10):1555–62. doi: 10.1089/dna.2012.1631 ; [PMC free article] [PubMed]
22. Requena T, Gazquez I, Moreno A, Batuecas A, Aran I, Soto-Varela A, et al. Allelic variants in TLR10 gene may influence bilateral affectation and clinical course of Meniere's disease. Immunogenetics. 2013;65(5):345–55. doi: 10.1007/s00251-013-0683-z [PubMed]
23. Eckhard A, Gleiser C, Arnold H, Rask-Andersen H, Kumagami H, Muller M, et al. Water channel proteins in the inner ear and their link to hearing impairment and deafness. Molecular Aspects Of Medicine. 2012;33(5–6):612–37. doi: 10.1016/j.mam.2012.06.004 [PubMed]
24. Cabrera S, Sanchez E, Requena T, Martinez-Bueno M, Benitez J, Perez N, et al. Intronic Variants in the NFKB1 Gene May Influence Hearing Forecast in Patients with Unilateral Sensorineural Hearing Loss in Meniere's Disease. Plos One. 2014;9(11):e112171 doi: 10.1371/journal.pone.0112171 [PMC free article] [PubMed]
25. Calzada AP, Lopez IA, Parrazal LB, Ishiyama A, Ishiyama G. Cochlin expression in vestibular endorgans obtained from patients with Meniere's disease. Cell Tissue Res. 2012;350(2):373–84. doi: 10.1007/s00441-012-1481-x [PMC free article] [PubMed]
26. Frykholm C, Larsen HC, Dahl N, Klar J, Rask-Andersen H, Friberg U. Familial Meniere's disease in five generations. Otol Neurotol. 2006;27(5):681–6. doi: 10.1097/01.mao.0000226315.27811.c8 [PubMed]
27. Gazquez I, Moreno A, Requena T, Ohmen J, Santos-Perez S, Aran I, et al. Functional variants of MIF, INFG and TFNA genes are not associated with disease susceptibility or hearing loss progression in patients with Meniere's disease. Eur Arch Otorhinolaryngol. 2013;270(4):1521–9. doi: 10.1007/s00405-012-2268-0 . [PubMed]
28. Friedland DR, Tarima S, Erbe C, Miles A. Development of a Statistical Model for the Prediction of Common Vestibular Diagnoses. Jama Otolaryngol. 2016;142(4):351–6. doi: 10.1001/jamaoto.2015.3663 [PMC free article] [PubMed]
29. Zou Q, Wan S, Ju Y, Tang J, Zeng X. Pretata: predicting TATA binding proteins with novel features and dimensionality reduction strategy. BMC Syst Biol. 2016;10(Suppl 4):114 doi: 10.1186/s12918-016-0353-5 ; [PMC free article] [PubMed]
30. Chen L, Zhang YH, Zheng M, Huang T, Cai YD. Identification of compound-protein interactions through the analysis of gene ontology, KEGG enrichment for proteins and molecular fragments of compounds. Molecular genetics and genomics: MGG. 2016;291(6):2065–79. Epub 2016/08/18. doi: 10.1007/s00438-016-1240-x . [PubMed]
31. Chen L, Zhang Y-H, Lu G, Huang T, Cai Y-D. Analysis of cancer-related lncRNAs using gene ontology and KEGG pathways. Artificial Intelligence in Medicine. 2017;76:27–36. doi: 10.1016/j.artmed.2017.02.001 [PubMed]
32. Barabasi AL, Gulbahce N, Loscalzo J. Network medicine: a network-based approach to human disease. Nat Rev Genet. 2011;12(1):56–68. doi: 10.1038/nrg2918 ; [PMC free article] [PubMed]
33. Liu Y, Zeng X, He Z, Zou Q. Inferring microRNA-disease associations by random walk on a heterogeneous network with multiple data sources. IEEE/ACM Transactions on Computational Biology and Bioinformatics. 2016. doi: 10.1109/TCBB.2016.2550432 [PubMed]
34. Zou Q, Li J, Song L, Zeng X, Wang G. Similarity computation strategies in the microRNA-disease network: a survey. Briefings in Functional Genomics. 2016;15(1):55–64. doi: 10.1093/bfgp/elv024 [PubMed]
35. Zeng X, Liao Y, Liu Y, Zou Q. Prediction and validation of disease genes using HeteSim Scores. IEEE/ACM Trans Comput Biol Bioinform. 2016. doi: 10.1109/TCBB.2016.2520947 . [PubMed]
36. Zou Q, Li J, Wang C, Zeng X. Approaches for Recognizing Disease Genes Based on Network. BioMed Research International. 2014;2014:10 doi: 10.1155/2014/416323 [PMC free article] [PubMed]
37. Ju Y, Zhang S, Ding N, Zeng X, Zhang X. Complex Network Clustering by a Multi-objective Evolutionary Algorithm Based on Decomposition and Membrane Structure. Sci Rep. 2016;6:33870 doi: 10.1038/srep33870 ; [PMC free article] [PubMed]
38. Oti M, Snel B, Huynen MA, Brunner HG. Predicting disease genes using protein-protein interactions. Journal of medical genetics. 2006;43(8):691–8. doi: 10.1136/jmg.2006.041376 [PMC free article] [PubMed]
39. Krauthammer M, Kaufmann CA, Gilliam TC, Rzhetsky A. Molecular triangulation: Bridging linkage and molecular-network information for identifying candidate genes in Alzheimer's disease. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(42):15148–53. doi: 10.1073/pnas.0404315101 [PubMed]
40. Franke L, van Bakel H, Fokkens L, de Jong ED, Egmont-Petersen M, Wijmenga C. Reconstruction of a functional human gene network, with an application for prioritizing positional candidate genes. American journal of human genetics. 2006;78(6):1011–25. doi: 10.1086/504300 [PubMed]
41. Kohler S, Bauer S, Horn D, Robinson PN. Walking the interactome for prioritization of candidate disease genes. The Amerian Journal of Human Genetics. 2008;82(4):949–58. doi: 10.1016/j.ajhg.2008.02.013 [PubMed]
42. Jiang R, Gan MX, He P. Constructing a gene semantic similarity network for the inference of disease genes. BMC Systems Biology. 2011;5(Suppl 2):S2 doi: 10.1186/1752-0509-5-S2-S2 [PMC free article] [PubMed]
43. Chen L, Yang J, Xing Z, Yuan F, Shu Y, Zhang Y, et al. An integrated method for the identification of novel genes related to oral cancer. PLoS ONE. 2017;12(4):e0175185 doi: 10.1371/journal.pone.0175185 [PMC free article] [PubMed]
44. Chen L, Hao Xing Z, Huang T, Shu Y, Huang G, Li H-P. Application of the Shortest Path Algorithm for the Discovery of Breast Cancer-Related Genes. Current Bioinformatics. 2016;11(1):51–8.
45. Gui T, Dong X, Li R, Li Y, Wang Z. Identification of Hepatocellular Carcinoma—Related Genes with a Machine Learning and Network Analysis. Journal of Computational Biology. 2015;22(1):63–71. doi: 10.1089/cmb.2014.0122 [PubMed]
46. Chen L, Yang J, Huang T, Kong XY, Lu L, Cai YD. Mining for novel tumor suppressor genes using a shortest path approach. Journal of Biomolecular Structure and Dynamics. 2016;34(3):664–75. doi: 10.1080/07391102.2015.1042915 [PubMed]
47. Zhang J, Yang J, Huang T, Shu Y, Chen L. Identification of novel proliferative diabetic retinopathy related genes on protein—protein interaction network. Neurocomputing. 2016;217:63–72. doi: 10.1016/j.neucom.2015.09.136
48. Zhu LC, Chen XJ, Kong XY, Cai YD. Investigation of the roles of trace elements during hepatitis C virus infection using protein-protein interactions and a shortest path algorithm. Bba-Gen Subjects. 2016;1860(11):2756–68. doi: 10.1016/j.bbagen.2016.05.018 [PubMed]
49. Cai Y-D, Zhang Q, Zhang Y-H, Chen L, Huang T. Identification of genes associated with breast cancer metastasis to bone on a protein-protein interaction network with a shortest path algorithm. J Proteome Res. 2017;16(2):1027–38. doi: 10.1021/acs.jproteome.6b00950 [PubMed]
50. Zhu L, Zhang YH, Su F, Chen L, Huang T, Cai YD. A Shortest-Path-Based Method for the Analysis and Prediction of Fruit-Related Genes in Arabidopsis thaliana. PLoS One. 2016;11(7):e0159519 doi: 10.1371/journal.pone.0159519 ; [PMC free article] [PubMed]
51. Chen L, Huang T, Zhang Y-H, Jiang Y, Zheng M, Cai Y-D. Identification of novel candidate drivers connecting different dysfunctional levels for lung adenocarcinoma using protein-protein interactions and a shortest path approach. Scientific Reports. 2016;6:29849 doi: 10.1038/srep29849 [PMC free article] [PubMed]
52. Franceschini A, Szklarczyk D, Frankild S, Kuhn M, Simonovic M, Roth A, et al. STRING v9.1: protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Research. 2013;41(Database issue):D808–15. Epub 2012/12/04. doi: 10.1093/nar/gks1094 ; [PMC free article] [PubMed]
53. Hsia CC. Respiratory function of hemoglobin. N Engl J Med. 1998;338(4):239–47. Epub 1998/01/22. doi: 10.1056/NEJM199801223380407 . [PubMed]
54. Ng KL, Ciou JS, Huang CH. Prediction of protein functions based on function-function correlation relations. Comput Biol Med. 2010;40(3):300–5. doi: 10.1016/j.compbiomed.2010.01.001 [PubMed]
55. Gao YF, Chen L, Cai YD, Feng KY, Huang T, Jiang Y. Predicting Metabolic Pathways of Small Molecules and Enzymes Based on Interaction Information of Chemicals and Proteins. PLoS ONE. 2012;7(9):e45944 doi: 10.1371/journal.pone.0045944 [PMC free article] [PubMed]
56. Hu L, Huang T, Liu XJ, Cai YD. Predicting protein phenotypes based on protein-protein interaction network. PLoS One. 2011;6(3):e17668 Epub 2011/03/23. doi: 10.1371/journal.pone.0017668 ; [PMC free article] [PubMed]
57. Gao P, Wang QP, Chen L, Huang T. Prediction of Human Genes Regulatory Functions Based on Proteinprotein Interaction Network. Protein and Peptide Letters. 2012;19(9):910–6. [PubMed]
58. Hu L, Huang T, Shi X, Lu WC, Cai YD, Chou KC. Predicting functions of proteins in mouse based on weighted protein-protein interaction network and protein hybrid properties. PLoS One. 2011;6(1):e14556 Epub 2011/02/02. doi: 10.1371/journal.pone.0014556 ; [PMC free article] [PubMed]
59. Zhang Q, Zhang PW, Cai YD. The Use of Protein-Protein Interactions for the Analysis of the Associations between PM2.5 and Some Diseases. Biomed Research International. 2016;2016:4895476 doi: 10.1155/2016/4895476 [PMC free article] [PubMed]
60. Huang G, Chu C, Huang T, Kong X, Zhang Y, Zhang N, et al. Exploring Mouse Protein Function via Multiple Approaches. PLoS One. 2016;11(11):e0166580 doi: 10.1371/journal.pone.0166580 ; [PMC free article] [PubMed]
61. Gui T, Dong X, Li R, Li Y, Wang Z. Identification of hepatocellular carcinoma-related genes with a machine learning and network analysis. Journal of computational biology: a journal of computational molecular cell biology. 2015;22(1):63–71. Epub 2014/09/24. doi: 10.1089/cmb.2014.0122 . [PubMed]
62. Li Z, An L, Li H, Wang S, Zhou Y, Yuan F, et al. Identifying novel genes and chemicals related to nasopharyngeal cancer in a heterogeneous network. Sci Rep. 2016;6:25515 Epub 2016/05/07. doi: 10.1038/srep25515 ; [PMC free article] [PubMed]
63. Chen L, Xing Z, Huang T, Shu Y, Huang G, Li H-P. Application of the shortest path algorithm for the discovery of breast cancer related genes. Current Bioinformatics. 2016;11(1):51–8.
64. Wang S, Huang G, Hu Q, Zou Q. A network-based method for the identification of putative genes related to infertility. Biochim Biophys Acta. 2016;1860(11 Pt B):2716–24. Epub 2016/04/23. doi: 10.1016/j.bbagen.2016.04.010 . [PubMed]
65. Chen L, Wang B, Wang S, Yang J, Hu J, Xie Z, et al. OPMSP: A computational method integrating protein interaction and sequence information for the identification of novel putative oncogenes. Protein Pept Lett. 2016;23(12):1081–94. . [PubMed]
66. Chen L, Zhang YH, Huang T, Cai YD. Identifying novel protein phenotype annotations by hybridizing protein-protein interactions and protein sequence similarities. Molecular Genetics and Genomics. 2016;291(2):913–34. doi: 10.1007/s00438-015-1157-9 [PubMed]
67. Yuan F, Zhang YH, Wan S, Wang S, Kong XY. Mining for Candidate Genes Related to Pancreatic Cancer Using Protein-Protein Interactions and a Shortest Path Approach. Biomed Res Int. 2015;2015:623121 Epub 2015/11/28. doi: 10.1155/2015/623121 ; [PMC free article] [PubMed]
68. Wang B, Yuan F, Kong X, Hu LD, Cai YD. Identifying Novel Candidate Genes Related to Apoptosis from a Protein-Protein Interaction Network. Comput Math Methods Med. 2015;2015:715639 Epub 2015/11/07. doi: 10.1155/2015/715639 ; [PMC free article] [PubMed]
69. Gene Ontology Consortium: going forward. Nucleic Acids Res. 2015;43(Database issue):D1049–56. Epub 2014/11/28. doi: 10.1093/nar/gku1179 ; [PMC free article] [PubMed]
70. Kanehisa M, Goto S. KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Research. 2000;28(1):27–30. [PMC free article] [PubMed]
71. Yang J, Chen L, Kong X, Huang T, Cai YD. Analysis of tumor suppressor genes based on gene ontology and the KEGG pathway. PLoS One. 2014;9(9):e107202 Epub 2014/09/11. doi: 10.1371/journal.pone.0107202 ; [PMC free article] [PubMed]
72. Zhang J, Xing Z, Ma M, Wang N, Cai YD, Chen L, et al. Gene ontology and KEGG enrichment analyses of genes related to age-related macular degeneration. Biomed Res Int. 2014;2014:450386 Epub 2014/08/29. doi: 10.1155/2014/450386 ; [PMC free article] [PubMed]
73. Zhao YX, Scott NA, Quah HS, Krishnamurthy B, Bond F, Loudovaris T, et al. Mouse pancreatic beta cells express MHC class II and stimulate CD4(+) T cells to proliferate. European Journal Of Immunology. 2015;45(9):2494–503. doi: 10.1002/eji.201445378 [PubMed]
74. Mamikoglu B, Wiet RJ, Hain T, Check IJ. Increased CD4+ T cells during acute attack of Meniere's disease. Acta oto-laryngologica. 2002;122(8):857–60. . [PubMed]
75. Derebery MJ. Allergic and Immunologic Features of Meniere's Disease. Otolaryngologic clinics of North America. 2011;44(3):655–66. doi: 10.1016/j.otc.2011.03.004 [PubMed]
76. Derebery MJ, Berliner KI. Allergy and Its Relation to Meniere's Disease. Otolaryngologic clinics of North America. 2010;43(5):1047–58. doi: 10.1016/j.otc.2010.05.004 [PubMed]
77. Keles E, Godekmerdan A, Kalidag T, Kaygusuz I, Yalcin I, Alpay HC, et al. Meniere's disease and allergy: allergens and cytokines. Journal Of Laryngology And Otology. 2004;118(9):688–93. doi: 10.1258/0022215042244822 [PubMed]
78. Wong CP, Rinaldi NA, Ho E. Zinc deficiency enhanced inflammatory response by increasing immune cell activation and inducing IL6 promoter demethylation. Molecular Nutrition & Food Research. 2015;59(5):991–9. doi: 10.1002/mnfr.201400761 [PMC free article] [PubMed]
79. Waalen J, von Lohneysen K, Lee P, Xu XL, Friedman JS. Erythropoietin, GDF15, IL6, hepcidin and testosterone levels in a large cohort of elderly individuals with anaemia of known and unknown cause. Eur J Haematol. 2011;87(2):107–16. doi: 10.1111/j.1600-0609.2011.01631.x [PubMed]
80. Mattia G, Milazzo L, Vulcano F, Pascuccio M, Macioce G, Hassan HJ, et al. Long-term platelet production assessed in NOD/SCID mice injected with cord blood CD34(+) cells, thrombopoietin-amplified in clinical grade serum-free culture. Experimental hematology. 2008;36(2):244–52. doi: 10.1016/j.exphem.2007.09.006 [PubMed]
81. Lin FY, Hsiao FP, Huang CY, Shih CM, Tsao NW, Tsai CS, et al. Porphyromonas gingivalis GroEL Induces Osteoclastogenesis of Periodontal Ligament Cells and Enhances Alveolar Bone Resorption in Rats. Plos One. 2014;9(7):e102450 doi: 10.1371/journal.pone.0102450 [PMC free article] [PubMed]
82. Dinarello CA, Simon A, van der Meer JW. Treating inflammation by blocking interleukin-1 in a broad spectrum of diseases. Nat Rev Drug Discov. 2012;11(8):633–52. doi: 10.1038/nrd3800 ; [PMC free article] [PubMed]
83. Song C, He L, Zhang J, Ma H, Yuan X, Hu G, et al. Fluorofenidone attenuates pulmonary inflammation and fibrosis via inhibiting the activation of NALP3 inflammasome and IL-1beta/IL-1R1/MyD88/NF-kappaB pathway. J Cell Mol Med. 2016;20(11):2064–77. doi: 10.1111/jcmm.12898 ; [PMC free article] [PubMed]
84. Kadariya Y, Menges CW, Talarchek J, Cai KQ, Klein-Szanto AJ, Pietrofesa RA, et al. Inflammation-Related IL1beta/IL1R Signaling Promotes the Development of Asbestos-Induced Malignant Mesothelioma. Cancer Prev Res (Phila). 2016;9(5):406–14. doi: 10.1158/1940-6207.CAPR-15-0347 ; [PMC free article] [PubMed]
85. Mahmoudi M, Amirzargar AA, Jamshidi AR, Farhadi E, Noori S, Avraee M, et al. Association of IL1R polymorphism with HLA-B27 positive in Iranian patients with ankylosing spondylitis. Eur Cytokine Netw. 2011;22(4):175–80. doi: 10.1684/ecn.2011.0293 . [PubMed]
86. Furuta T, Teranishi M, Uchida Y, Nishio N, Kato K, Otake H, et al. Association of interleukin-1 gene polymorphisms with sudden sensorineural hearing loss and Meniere's disease. International journal of immunogenetics. 2011;38(3):249–54. doi: 10.1111/j.1744-313X.2011.01004.x . [PubMed]
87. Mosmann TR. Regulation of immune responses by T cells with different cytokine secretion phenotypes: role of a new cytokine, cytokine synthesis inhibitory factor (IL10). International archives of allergy and applied immunology. 1991;94(1–4):110–5. . [PubMed]
88. Sakthivel P, Gereke M, Breithaupt A, Fuchs D, Gigliotti L, Gruber AD, et al. Attenuation of immune-mediated influenza pneumonia by targeting the inducible co-stimulator (ICOS) molecule on T cells. PLoS One. 2014;9(7):e100970 doi: 10.1371/journal.pone.0100970 ; [PMC free article] [PubMed]
89. Zhou B, Kermany MH, Cai Q, Cai C, Zhou Y, Nair U, et al. Experimental autoimmune hearing loss is exacerbated in IL-10-deficient mice and reversed by IL-10 gene transfer. Gene Therapy. 2012;19(2):228–35. doi: 10.1038/gt.2011.88 [PubMed]
90. Barna BP, Hughes GB. Autoimmunity And Otologic Disease—Clinical And Experimental Aspects. Clinics In Laboratory Medicine. 1988;8(2):385–98. [PubMed]
91. Jegaskanda S, Ahn SH, Skinner N, Thompson AJ, Ngyuen T, Holmes J, et al. Downregulation of Interleukin-18-Mediated Cell Signaling and Interferon Gamma Expression by the Hepatitis B Virus e Antigen. J Virol. 2014;88(18):10412–20. doi: 10.1128/JVI.00111-14 [PMC free article] [PubMed]
92. Fuse T, Hayashi T, Oota N, Fukase S, Asano S, Kato T, et al. Immunological responses in acute low-tone sensorineural hearing loss and Meniere's disease. Acta oto-laryngologica. 2003;123(1):26–31. . [PubMed]
93. Duarte GV, Boeira V, Correia T, Porto-Silva L, Cardoso T, Macedo MN, et al. Osteopontin, CCL5 and CXCL9 are independently associated with psoriasis, regardless of the presence of obesity. Cytokine. 2015;74(2):287–92. doi: 10.1016/j.cyto.2015.04.015 . [PubMed]
94. Zhang L, Yu M, Deng J, Lv X, Liu J, Xiao Y, et al. Chemokine Signaling Pathway Involved in CCL2 Expression in Patients with Rheumatoid Arthritis. Yonsei Med J. 2015;56(4):1134–42. doi: 10.3349/ymj.2015.56.4.1134 ; [PMC free article] [PubMed]
95. Rask-Andersen H, Danckwardt-Lilliestrom N, Linthicum FH Jr., House WF. Ultrastructural evidence of a merocrine secretion in the human endolymphatic sac. The Annals of otology, rhinology, and laryngology. 1991;100(2):148–56. doi: 10.1177/000348949110000211 . [PubMed]
96. Moller MN, Kirkeby S, Vikesa J, Nielsen FC, Caye-Thomasen P. Expression of histamine receptors in the human endolymphatic sac: the molecular rationale for betahistine use in Menieres disease. European Archives Of Oto-Rhino-Laryngology. 2016;273(7):1705–10. doi: 10.1007/s00405-015-3731-5 [PubMed]
97. Moller MN, Kirkeby S, Vikesa J, Nielsen FC, Caye-Thomasen P. Gene expression demonstrates an immunological capacity of the human endolymphatic sac. The Laryngoscope. 2015;125(8):E269–E75. doi: 10.1002/lary.25242 [PubMed]
98. Lu HL, Liu HY, Wang JX, Shen JQ, Weng SY, Han L, et al. The chemokine CXCL9 exacerbates chemotherapy-induced acute intestinal damage through inhibition of mucosal restitution. Journal Of Cancer Research And Clinical Oncology. 2015;141(6):983–92. doi: 10.1007/s00432-014-1869-y [PubMed]
99. Shen JQ, Gao J, Chen CY, Lu HL, Hu GY, Shen J, et al. Antifibrotic role of chemokine CXCL9 in experimental chronic pancreatitis induced by trinitrobenzene sulfonic acid in rats. Cytokine. 2013;64(1):382–94. doi: 10.1016/j.cyto.2013.05.012 [PubMed]
100. Sahin H, Borkham-Kamphorst E, Kuppe C, Zaldivar MM, Grouls C, Al-samman M, et al. Chemokine Cxcl9 attenuates liver fibrosis-associated angiogenesis in mice. Hepatology. 2012;55(5):1610–9. Epub 2012/01/13. doi: 10.1002/hep.25545 . [PubMed]
101. Petrovic-Djergovic D, Popovic M, Chittiprol S, Cortado H, Ransom RF, Partida-Sanchez S. CXCL10 induces the recruitment of monocyte-derived macrophages into kidney, which aggravate puromycin aminonucleoside nephrosis. Clinical And Experimental Immunology. 2015;180(2):305–15. doi: 10.1111/cei.12579 [PubMed]
102. Padovan E, Spagnoli GC, Ferrantini M, Heberer M. IFN-alpha2a induces IP-10/CXCL10 and MIG/CXCL9 production in monocyte-derived dendritic cells and enhances their capacity to attract and stimulate CD8+ effector T cells. Journal of leukocyte biology. 2002;71(4):669–76. . [PubMed]
103. Dong Y, Li M, Liu P, Song H, Zhao Y, Shi J. Genes involved in immunity and apoptosis are associated with human presbycusis based on microarray analysis. Acta oto-laryngologica. 2014;134(6):601–8. doi: 10.3109/00016489.2014.880795 . [PubMed]
104. Henrick BM, Yao XD, Rosenthal KL, Team IS. HIV-1 structural proteins serve as PAMPs for TLR2 heterodimers significantly increasing infection and innate immune activation. Front Immunol. 2015;6:426 [PMC free article] [PubMed]
105. Benakanakere M, Abdolhosseini M, Hosur K, Finoti LS, Kinane DF. TLR2 Promoter Hypermethylation Creates Innate Immune Dysbiosis. Journal Of Dental Research. 2015;94(1):183–91. doi: 10.1177/0022034514557545 [PMC free article] [PubMed]
106. Moreira LO, El Kasmi KC, Smith AM, Finkelstein D, Fillon S, Kim YG, et al. The TLR2-MyD88-NOD2-RIPK2 signalling axis regulates a balanced pro-inflammatory and IL-10-mediated anti-inflammatory cytokine response to Gram-positive cell walls. Cellular microbiology. 2008;10(10):2067–77. doi: 10.1111/j.1462-5822.2008.01189.x [PMC free article] [PubMed]
107. Zhang JP, Yang Y, Levy O, Chen C. Human Neonatal Peripheral Blood Leukocytes Demonstrate Pathogen-Specific Coordinate Expression of TLR2, TLR4/MD2, and MyD88 During Bacterial Infection In Vivo. Pediatric research. 2010;68(6):479–83. doi: 10.1203/PDR.0b013e3181f90810 [PMC free article] [PubMed]
108. Moller MN, Kirkeby S, Caye-Thomasen P. Innate immune defense in the inner ear—mucines are expressed by the human endolymphatic sac. J Anat. 2017;230(2):297–302. doi: 10.1111/joa.12559 [PubMed]
109. Jablonka-Strom A, Pospiech L, Zatonski M, Bochnia M. Dynamics of pure tone audiometry and DPOAE changes induced by glycerol in Meniere's disease. European Archives Of Oto-Rhino-Laryngology. 2013;270(5):1751–6. doi: 10.1007/s00405-012-2246-6 [PMC free article] [PubMed]
110. Thomsen J, Vesterhauge S. Critical-Evaluation Of the Glycerol Test In Menieres-Disease. J Otolaryngol. 1979;8(2):145–50. [PubMed]
111. Talreja D, Singh PK, Kumar A. In Vivo Role of TLR2 and MyD88 Signaling in Eliciting Innate Immune Responses in Staphylococcal Endophthalmitis. Invest Ophth Vis Sci. 2015;56(3):1719–32. doi: 10.1167/iovs.14-16087 [PMC free article] [PubMed]
112. Pan XC, Li B, Kuang M, Liu X, Cen YY, Qin RX, et al. Synthetic Human TLR9-LRR11 Peptide Attenuates TLR9 Signaling by Binding to and thus Decreasing Internalization of CpG Oligodeoxynucleotides. International journal of molecular sciences. 2016;17(2):242 doi: 10.3390/ijms17020242 [PMC free article] [PubMed]
113. Ma C, Spies NP, Gong T, Jones CX, Chu WM. Involvement of DNA-PKcs in the type I IFN response to CpG-ODNs in conventional dendritic cells in TLR9-dependent or -independent manners. PLoS One. 2015;10(3):e0121371 doi: 10.1371/journal.pone.0121371 ; [PMC free article] [PubMed]
114. Pathak S, Hatam LJ, Bonagura V, Vambutas A. Innate Immune Recognition of Molds and Homology to the Inner Ear Protein, Cochlin, in Patients with Autoimmune Inner Ear Disease. J Clin Immunol. 2013;33(7):1204–15. doi: 10.1007/s10875-013-9926-x [PMC free article] [PubMed]
115. Nelson BR, Hodge RD, Bedogni F, Hevner RF. Dynamic Interactions between Intermediate Neurogenic Progenitors and Radial Glia in Embryonic Mouse Neocortex: Potential Role in Dll1-Notch Signaling. J Neurosci. 2013;33(21):9122–39. doi: 10.1523/JNEUROSCI.0791-13.2013 [PMC free article] [PubMed]
116. Rouault H, Hakim V. Different Cell Fates from Cell-Cell Interactions: Core Architectures of Two-Cell Bistable Networks. Biophys J. 2012;102(3):417–26. doi: 10.1016/j.bpj.2011.11.4022 [PubMed]
117. Ciuman RR. Inner ear symptoms and disease: Pathophysiological understanding and therapeutic options. Med Sci Monitor. 2013;19:1195–210. [PMC free article] [PubMed]
118. Agrup C, Gleeson M, Rudge P. The inner ear and the neurologist. J Neurol Neurosur Ps. 2007;78(2):114–22. doi: 10.1136/jnnp.2006.092064 [PMC free article] [PubMed]
119. Kiernan AE, Cordes R, Kopan R, Gossler A, Gridley T. The Notch ligands DLL1 and JAG2 act synergistically to regulate hair cell development in the mammalian inner ear. Development. 2005;132(19):4353–62. doi: 10.1242/dev.02002 [PubMed]
120. Liddell JR, Hoepken HH, Crack PJ, Robinson SR, Dringen R. Glutathione peroxidase 1 and glutathione are required to protect mouse astrocytes from iron-mediated hydrogen peroxide toxicity. Journal of neuroscience research. 2006;84(3):578–86. doi: 10.1002/jnr.20957 [PubMed]
121. Giuca MR, Giuggioli E, Metelli MR, Pasini M, Iezzi G, D'Ercole S, et al. Effects Of Cigarette Smoke on Salivary Superoxide Dismutase And Glutathione Peroxidase Activity. J Biol Reg Homeos Ag. 2010;24(3):359–66. [PubMed]
122. Matsushita M, Freigang S, Schneider C, Conrad M, Bornkamm GW, Kopf M. T cell lipid peroxidation induces ferroptosis and prevents immunity to infection. Journal Of Experimental Medicine. 2015;212(4):555–68. doi: 10.1084/jem.20140857 [PMC free article] [PubMed]
123. Gutsche M, Sobotta MC, Wabnitz GH, Ballikaya S, Meyer AJ, Samstag Y, et al. Proximity-based Protein Thiol Oxidation by H2O2-scavenging Peroxidases. Journal Of Biological Chemistry. 2009;284(46):31532–40. doi: 10.1074/jbc.M109.059246 [PMC free article] [PubMed]
124. van Esch BF, van Benthem PPG, van der Zaag-Loonen HJ, Bruintjes TD. Two Common Second Causes of Dizziness in Patients With Meniere's Disease. Otol Neurotol. 2016;37(10):1620–4. doi: 10.1097/MAO.0000000000001215 [PubMed]
125. Taura A, Funabiki K, Ohgita H, Ogino E, Torii H, Matsunaga M, et al. One-third of vertiginous episodes during the follow-up period are caused by benign paroxysmal positional vertigo in patients with Meniere's disease. Acta oto-laryngologica. 2014;134(11):1140–5. doi: 10.3109/00016489.2014.936624 [PubMed]
126. Yamane H, Sunami K, Iguchi H, Sakamoto H, Imoto T, Rask-Andersen H. Assessment of Meniere's disease from a radiological aspect—saccular otoconia as a cause of Meniere's disease? Acta oto-laryngologica. 2012;132(10):1054–60. doi: 10.3109/00016489.2012.680980 [PMC free article] [PubMed]
127. Taylor A, Robson A, Houghton BC, Jepson CA, Ford WCL, Frayne J. Epididymal specific, selenium-independent GPX5 protects cells from oxidative stress-induced lipid peroxidation and DNA mutation. Human Reproduction. 2013;28(9):2332–42. doi: 10.1093/humrep/det237 [PubMed]
128. Gharagozloo P, Gutierrez-Adan A, Champroux A, Noblanc A, Kocer A, Calle A, et al. A novel antioxidant formulation designed to treat male infertility associated with oxidative stress: promising preclinical evidence from animal models. Human Reproduction. 2016;31(2):252–62. doi: 10.1093/humrep/dev302 [PubMed]
129. Calabrese V, Cornelius C, Maiolino L, Luca M, Chiaramonte R, Toscano MA, et al. Oxidative Stress, Redox Homeostasis and Cellular Stress Response in Meniere's Disease: Role of Vitagenes. Neurochemical Research. 2010;35(12):2208–17. doi: 10.1007/s11064-010-0304-2 [PubMed]
130. Labbe D, Teranishi M, Hess A, Bloch W, Michel O. Activation of caspase-3 Is associated with oxidative stress in the hydropic guinea pig cochlea. Hearing Res. 2005;202(1–2):21–7. doi: 10.1016/j.heares.2004.10.002 [PubMed]
131. Asher BF, Guilford FT. Oxidative Stress and Low Glutathione in Common Ear, Nose, and Throat Conditions: A Systematic Review. Altern Ther Health M. 2016;22(5):44–50. [PubMed]
132. Santulli P, Borghese B, Noel JC, Fayt I, Anaf V, de Ziegler D, et al. Hormonal Therapy Deregulates Prostaglandin-Endoperoxidase Synthase 2 (PTGS2) Expression in Endometriotic Tissues. Journal Of Clinical Endocrinology & Metabolism. 2014;99(3):881–90. doi: 10.1210/jc.2013-2950 [PubMed]
133. Stamatakis K, Jimenez-Martinez M, Jimenez-Segovia A, Chico-Calero I, Conde E, Galan-Martinez J, et al. Prostaglandins induce early growth response 1 transcription factor mediated microsomal prostaglandin E-2 synthase up-regulation for colorectal cancer progression. Oncotarget. 2015;6(37):39941–59. doi: 10.18632/oncotarget.5402 [PMC free article] [PubMed]
134. Michel O, Matthias R. Effects of prostaglandin E2 on the fluctuating hearing loss in Meniere's disease. Auris, nasus, larynx. 1992;19(1):7–16. . [PubMed]

Articles from PLoS ONE are provided here courtesy of Public Library of Science