Untargeted Metabolomic Analysis in Endolymphatic Sac Luminal Fluid from Patients with Meniere’s Disease

Abstract

Dysfunction of the endolymphatic sac (ES) is one of the etiologies of Meniere’s disease (MD), the mechanism of which remains unclear. The aim of the present study was to explore the molecular pathological characteristics of ES during the development of MD. Metabolomic profiling of ES luminal fluid from patients with MD and patients with acoustic neuroma (AN) was performed. Diluted ES luminal fluid (ELF) samples were obtained from 10 patients who underwent endolymphatic duct blockage for the treatment of intractable MD and from 6 patients who underwent translabyrinthine surgery for AN. ELF analysis was performed using liquid chromatography-mass spectrometry before the raw data were normalized and subjected to subsequent statistical analysis by MetaboAnalyst. Using thresholds of P ≤ 0.05 and variable important in projection > 1, a total of 111 differential metabolites were screened in the ELF, including 52 metabolites in negative mode and 59 in positive mode. Furthermore, 15 differentially altered metabolites corresponding to 15 compound names were identified using a Student’s t-test, including 7 significant increased metabolites and 8 significant decreased metabolites. Moreover, two differentially altered metabolites, hyaluronic acid (HA) and 4-hydroxynonenal (4-HNE), were validated to be upregulated in the epithelial lining of the ES, as well as in the subepithelial connective-tissue in patients with MD comparing with that in patients with AN. Among these differentially altered metabolites, an upregulated expression of HA detected in the ES lumen of the patients with MD was supposed to be associated with the increased endolymph in ES, while an increased level of 4-HNE found in the ELF of the patients with MD provided direct evidence to support that oxidative damage and inflammatory lesions underlie the mechanism of MD. Furthermore, citrate and ethylenediaminetetraacetic acid were detected to be decreased substantially in the ELF of the patients with MD, suggesting the elevated endolymphatic Ca²⁺ in the ears with chronic endolymphatic hydrops is likely to be associated with the reduction of these two chelators of Ca²⁺ in ES. The results in the present study indicate metabolomic analysis in the ELF of the patients with MD can potentially improve our understanding on the molecular pathophysiological mechanism in the ES during the development of MD.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10162-023-00887-1.

Introduction

Meniere’s disease (MD) is characterized by episodic vertigo, fluctuating sensorineural hearing loss, and aural symptoms. Accumulating evidence has previously shown that dysfunction of the endolymphatic sac (ES) is one of the etiologies of this disease [1]. The ES is a small structure (∼15 mm²) and is an extension of the luminal compartment of the cochlea and vestibule [2]. The luminal epithelium of ES is comprised of the following two different cell types: mitochondria-rich cells, which mainly perform ion transport functions, and ribosomal-rich cells, which mainly perform secretory functions [3]. Therefore, the ES serves an important role in regulating ion homeostasis and endolymphatic fluid volume in the inner ear [4]. In addition, the ES has been reported to be involved in regulating immune function in the inner ear whilst removing cellular debris, such as floating otoconia, from this organ [5, 6]. In cases of pathological changes, dysregulation of the endolymphatic volume in the ES will occur and endolymphatic hydrops (EH) may develop, causing fluctuating hearing loss and recurrent vertigo in patients with MD [7]. Previous studies suggested that a number of etiological factors, including infectious, inflammatory, autoimmune, vascular, genetic, oxidative stress, and endocrine agents, can mediate pathological changes to the ES that may result in the development of EH [8]. The relationship between pathological changes in the ES and the pathophysiology of MD has been previously studied by analyzing histological changes in cadaveric samples of human ES and by animal studies [4–6]. However, functional pathological evidence of MD in primary human ES samples remains insufficient. One of the main reasons of this is that the molecular mechanism underlying the resorption and secretion of endolymph in the ES has yet to be determined [1]. Therefore, identification of biomarkers in the luminal fluid and characterization of the functional activity dynamics in the epithelium of the ES would be of great importance for understanding the etiology of MD. Recently, endolymphatic duct blockage (EDB) was selectively performed for treatment of intractable MD at our University Hospital since January 2019. Based on the classification of the intraoperative pathoanatomic findings of the sac in MD described by Yazawa et al. [9], the patients who were found to have endolymphatic sacs classified into large or intermediate size (hyperplastic or normal ES in morphology) underwent EDB procedure, whereas the patients who were found to have endolymphatic sacs classified into small size and poor in vascularity (hypoplastic or degenerated ES in morphology) underwent endolymphatic sac drainage surgery. Notably, those EDB procedures could provide an opportunity for collecting the samples of ES luminal fluid from the MD patients with hyperplastic or normal ES in morphology for further understanding the pathological changes at the molecular level in the ES of the patients with MD.

Recently, proteomic analysis and microRNA (miRNA) profile characterization of human perilymph samples from patients with MD compared with those in controls have revealed promising results for the identification of molecular biomarkers associated with this disease [8, 10]. Proteomic profiling of the luminal fluid from the human ES has provided direct evidence for the local alteration of protein expression in the ES of patients with MD [11, 12]. Recently, an upregulated expression of interferon-gamma, interleukin-6, and tumor necrosis factor-alpha in the endolymphatic sac of Meniere’s disease was reported, suggesting the local inflammatory response underlies the mechanism of this disease [13]. However, these findings remain insufficient for identifying disease-specific biomarkers for the diagnosis or prognosis of MD, and therefore, its molecular mechanisms of MD remain elusive at present.

Metabolomics is an emerging technique that can be used to discover novel biomarkers for deepening the understanding of disease pathogenesis. Compared with proteomics, which focuses on analyzing protein expression, function, modifications, and interaction, metabolomics can provide insights into changes in the levels of metabolites, which can be used to reflect genomic, transcriptomic and proteomic changes. Therefore, it represents an accurate biochemical phenotyping tool [14–16]. Using high-throughput analytical methods, such as liquid chromatography coupled with high-resolution mass spectrometry, non-targeted metabolomics is able to establish broad metabolic profiles using samples, which can in turn be used to identify novel metabolites and biomarkers. In addition, non-targeted metabolomics can provide valuable information on any changes in a large list of metabolites and alterations in a correspondingly large list of metabolic networks. However, due to the difficulty in obtaining fluid samples from the inner ear, one of the challenges of performing metabolomic studies on inner ear fluid samples is insufficient quantity. Nevertheless, technical analysis on low volumes of fluid samples from the inner ear (perilymph, 0.8–1 µl) has been previously validated in a methodological study [17]. In this particular previous study, metabolomic analysis has been applied to successfully unravel the metabolomic profile in the perilymph underlying sensorineural hearing loss in patients who underwent cochlear implantation [18]. In addition, metabolomic profiles were also reported in animal models of noise-induced hearing loss and cisplatin-induced ototoxicity [19, 20]. Results of the previous perilymph metabolomic analyses aforementioned have been used to explore possible molecular mechanisms in sensorineural hearing loss.

To further the understanding of molecular changes in the endolymph of ES during MD development, we conducted an untargeted metabolomics analysis on diluted ES luminal fluid (ELF) samples harvested from patients with MD and patients with acoustic neuroma (AN). The aim was to identify the differentially altered metabolites in the ELF during MD pathogenesis.

Materials and Methods

Sampling of the ELF

In total, 10 patients diagnosed with unilateral MD (sex, six males and four females; mean age, 49.5 years; age range, 32–66 years; mean duration of the disease, 22.5 months; range, 18–36 months) according to the 2015 Criteria of Classification Committee of the Bárány Society for the diagnosis of MD [21] were enrolled into the patient group, whereas 6 patients with AN (sex, three males and three females,mean age, 54.5 years; age range, 35–65 years) were enrolled into the control group. None of these patients had a history of systemic autoimmune disease, and vestibular migraine has been ruled out using the Bárány diagnostic criteria in all MD patients [21].

At our university hospital between July 2020 and June 2021, samples of ELF were obtained from 10 patients who underwent endolymphatic duct blockage for intractable MD followed by magnetic resonance imaging (MRI)-based visualization of unilateral EH. In addition, ELF samples were collected from 6 patients who underwent translabyrinthine surgery for AN. All patients in the control group with AN had non-serviceable hearing on the ipsilateral side, but had no history of sudden vertigo, who were excluded ipsilateral endolymphatic hydrops by three-dimensional fluid-attenuated inversion recovery (3D-FLAIR) MRI using intratympanic administration of gadolinium contrast agent.

Sampling ELF from the patients with MD who underwent EDB procedure was described previously [13]. Briefly, after blockage of the rugose portion of the ES using the ligating clip applier, the proximal extraosseous ES was opened with an L-shaped incision, following which the lumen of the ES was carefully identified. We infused 2 μl sterile water into the lumen of the ES and then aspirated the 2-μl diluted luminal fluid from the lumen of the ES. Subsequently, the sample was transferred into an Eppendorf tube (Eppendorf, Germany) already containing 100 µl sterile water for storage at − 80 °C until further analysis.

Ethics Statement

All patients provided informed consent for collection of their ES luminal fluid. All study protocols were approved by the Medical Ethics Committee of the Second Xiangya Hospital (approval no. S525), and informed consent was obtained from all participants in accordance with the national legislation and the Declaration of Helsinki.

Extraction of Metabolites

The collected ELF samples in the MD and control groups were thawed at 4 °C before 100 μl ELF samples were obtained and transferred into a clean tube. In total, 200 μl methanol and 200 μl acetonitrile were added before the mixture was vortexed for 30 s. After sonicating for 10 min, the mixture was incubated at − 20 °C for 1 h to promote protein precipitation and then centrifuged at 13,000 rpm for 15 min at 4 °C. The supernatant was collected and dehydrated to dry in a vacuum concentrator. The dried extracts were then resuspended in 100 μl acetonitrile solution (acetonitrile: H₂O = 1:1), and the mixture was vortexed for 30 s and sonicated for 10 min. After centrifugation at 13,000 rpm for 15 min at 4 °C, the supernatant was collected and used for liquid chromatography (LC)-mass spectrometry (MS).

LC–MS

The supernatant collected as aforementioned was analyzed by LC–MS. LC–MS was performed using an Agilent 1290 LC system (Agilent, USA) equipped with an Acquity ULPC BEH C18 column (1.7 µm, 2.1 × 100 mm; Waters) and a TripleTOF® 6600 plus MS (Sciex, USA). The mobile A phase (A) was 25 mM ammonium acetate and 25 mM ammonium hydroxide dissolved in water, whereas the mobile B phase (B) was 100% acetonitrile. The flow rate was 0.5 ml/min with the column temperature set at 25 °C, and the injection volume was 2 μl. The elution gradient was set as follows: 5% A and 95% B, 0 and 30 s; 35% A and 65% B, 7 min; 60% A and 40% B, 8 and 9 min; 5% A and 95% B, 9.1 and 12 min. The electrospray ionization mass spectra were acquired in both positive and negative ion modes. The ion spray voltages of MS for positive and negative modes were 5000 and 4000 V, respectively. The heated-capillary temperature was maintained at 600 °C. Full scanning was performed with a range of 60–1200 mass-to-charge ratio (m/z). The curtain gas flow, nebulizer, and heater gas were set to 35, 60, and 60 arbitrary units, respectively. The collision energy was set to 30 V.

Data Quality Control

To assess the stability of the detection system, a quality control (QC) sample was prepared by mixing all the test samples at equal volumes during the experiment. During instrumental analysis, one QC sample was inserted for every six test samples. During data analysis, repeatability of the QC samples was tested to inspect the stability of the instrument during the entire analysis process [22].

Immunohistochemical Analysis in Human Endolymphatic Sac

Sac specimens were obtained from the above-mentioned 10 patients with MD who underwent endolymphatic duct blockage for intractable MD, and 6 patients with AN who underwent translabyrinthine surgery for AN. The method for collecting sac specimen was described previously [13]. Subsequently, tissues were fixed with 4% paraformaldehyde solution (Head Biotechnology CO., LTD, Beijing, PRC; Cat No:157–8) for 12 h, and then transferred sequentially to 50, 70, 95, and 100% ethanol baths (1 h for each bath) for dehydration. Then, samples were infiltrated using molten paraffin wax in the oven for 1 h, before being embedded into paraffin wax blocks. Paraffin embedded ES tissues were sectioned at 5-μm thickness for subsequent immunohistochemistry. Slides were deparaffinized and rehydrated in 3% hydrogen peroxide (Abcam, Cambridge, UK; cat. no. AB64218) for 5 min. After antigen retrieval using 1% sodium dodecyl sulfate (Merck & Co., Inc., Rahway, USA; cat. no. L3771) in PBS for 10 min, slides were blocked in 10% normal horse serum (Abcam, Cambridge, UK; cat. no. AB7484) in PBS for 1 h. Primary antibodies (100 µl, 1:400) against hyaluronan binding protein 2 (HABP2) (Proteintech Group, Wuhan, PRC; Cat No: 12863–1-AP; Rabbit Polyclonal), 4-hydroxynonenal (4HNE) Antibody (Abcam, Cambridge, UK; cat. no. AB46545; Mouse Polyclonal), hyaluronan synthase 2 (HAS2) antibody (Yaji Biotechnology, shanghai, PRC; cat. no. 11290R; rabbit monoclonal), and hyaluronidase 1 (Hyal-1) antibody (Yaji Biotechnology, shanghai, PRC; cat. no. 1235R; mouse monoclonal) in 0.1% Triton-PBS (Ziker Biological technology CO., LTD, Shenzhen, PRC; cat. no. ZK-L1578) were applied to the samples overnight at room temperature. Negative control samples were processed simultaneously in an identical manner, with the exception that PBST was used to replace the primary antibody. The sections were incubated with the relevant Invitrogen® fluorescent secondary antibodies (Thermo Fisher Scientific, Inc., Waltham, MA, USA; cat. no. A-11034, 1:400 dilution) for 2 h at room temperature. After washing with 0.1 M PBS, the slides were mounted using antifading medium (with DAPI, Vector Laboratories, Burlingame, CA, USA; cat. no. H-1800). Confocal images were acquired using a laser scanning confocal microscope (Leica TCS SP5, Leica Microsystems GmbH, Mannheim, Germany) equipped with 561 and 633 nm lasers for excitation and a 63 × oil immersion objective (1.4 NA, Leica). Confocal microscopy analysis was used to demonstrate subcellular localization in the membranous, cytoplasmic, and nuclear for the expression of HABP2, 4-HNE, HAS2, and Hyal-1.

Quantification of Fluorescence Intensity in DAPI-Labeled Epithelial Area

The fluorescence intensity of each type of immunolabeling in ES epithelium of patients with MD and patients with AN was compared. In each immunolabeled section, ten microscopic images were taken in different regions where at least 10 DAPI-labeled epithelial cells were identified. ImageJ software (National Institutes of Health) was used to measure the DAPI-labeled epithelial area in each image. The same color intensity threshold was used to analyze images for each type of immunolabeling. Each type of immunolabeling was performed on three non-consecutive sections from the same specimen in order to determine the mean values and standard deviations of the DAPI-stained epithelial area. The Mann–Whitney U test were used to analyze statistical significance. P < 0.05 was considered to indicate a statistically significant.

Data Analysis

The raw data generated by LC–MS were first converted into files of the.mzXML and.mgf formats using the ProteoWizard tool (version 3.0.6150) and then submitted for data processing. XCMS software (version 1.46.0) was used to perform peak identification, peak filtration, and alignment. The results were subsequently combined with.mgf files and imported into the MetDNA tool (https://metdna.zhulab.cn/) for identifying the metabolites according to their observed physical characteristics, including mass-to-charge ratios (m/z), retention time (RT), and peak intensity. The obtained data matrix was then subjected to statistical analysis to screen the differentially altered metabolites using the MetaboAnalyst tool (https://www.metaboanalyst.ca) [23]. Principal component analysis (PCA) and orthogonal partial least-squares differential analysis (OPLS-DA) were performed using the table of normalized peak total intensity to probe for possible separations in the metabolite profiles between the MD and control samples. Fold-change (FC) and P-values (as assessed using Student’s t-test) were then calculated. The thresholds set to screen out differentially altered metabolites were P ≤ 0.05 and variable important in projection (VIP) > 1 [24].

Results

Patient Characteristics

From July 2020 and June 2021, a total of 28 patients underwent surgical procedure for their intractable unilateral MD, including endolymphatic sac drainage surgery in 17 patients with hypoplastic or degenerated ES and EDB surgery in 11 patients with hyperplastic or normal ES in morphology. A total of 10 diluted ELF samples were collected from 11 patients with MD who underwent endolymphatic duct blockage because one sample was excluded due to possible blood contamination. In addition, 6 control ELF samples were obtained from 6 patients with AN following translabyrinthine surgery. Sex distribution and mean age of patients between the MD and control groups were not found to differ significantly (P < 0.05 for the χ² test and Mann–Whitney U test).

Metabolomic Profiling of the ELF Samples

The total ion chromatograms of the QC samples in the negative and positive modes showed that the spectral peak intensity of the QC samples and retention time were comparable (Fig. 1), suggesting stability of the system and good overall reproducibility of the methodology applied in the present study.

Fig. 1.

Base peak chromatographic analysis of the quality control samples in the negative and positive modes. The spectral peak intensity of the quality control samples and retention time were comparable, implicating stability of the system and good overall reproducibility of the protocol

In total, 5681 peaks in positive ion mode and 5465 peaks in negative ion mode were detected by LC–MS/MS. After de-noising by relative standard deviations (RSD < 0.3), exclusion of metabolites with detection rates < 50%, filling in missing raw data with half the minimum value, and internal standard normalization, 4748 peaks in positive ion mode and 4557 peaks in negative ion mode were retained. Subsequently, the data in the negative and positive modes were submitted to ProteoWizard for further analysis. PCA was first used to assess the overall distribution of each sample and the degree of dispersion between the MD group and control group (Fig. 2A, B). The PCA could not separate the two groups. To improve the resolution and identify the variables responsible for the segregation of these two groups, supervised orthogonal projections to latent structures–discriminant analysis (OPLS-DA) were applied. We obtained models for the positive ion mode (R2Y = 0.771; Q2 = 0.392) and the negative ion mode (R2Y = 0.722; Q2 = 0.357) that were stable, predictive, and that fit the data well. The models were valid, clearly discriminated the two groups, and demonstrated different metabolic profiles between them (Fig. 2C, D). A permutation test validated these models (Fig. 2E, F). Using R2Y = 0.639, and Q2 =  − 0.227 for the positive ion mode and R2Y = 0.636, Q2 =  − 0.165 for the negative ion mode, 200 permutations showed model robustness, low risks and lack of overfitting, and high reliability.

Fig. 2.

Multivariate analyses of metabolic profiles of ELFs in the MD group and control group. PCA analysis of all samples in the positive A and negative B modes, revealing that all samples were tightly clustered within the PCA space. The red and green circles represent the 95% confidence intervals of the scores calculated from each patient. OPLS-DA results showing that the samples in the control and Meniere’s disease groups significantly differentiated using positive C and negative D ion model datasets. Permutation test (n = 200) of OPLS-DA model in positive E and negative F ion models. PCA, principal component analysis; OPLS-DA, orthogonal partial least squares differential analysis. Green circles in A, B, C, and D represent the samples from patients with Meniere’s disease [A]. Red circles in A, B, C, and D represent the samples from patients with acoustic neuroma [N]

Identification of the Differentially Altered Metabolites in ELF

Using P ≤ 0.05 and variable importance in the projection (VIP) > 1 as thresholds, a total of 111 differentially altered metabolites were found in the ELF samples between the patients with MD and those in the patients with AN (or control group), including 52 metabolites in the negative mode (Table S1) and 59 metabolites in the positive mode (Table S2). The volcano plots show the upregulated and downregulated metabolites in the negative mode (Fig. 3A) and in the positive mode (Fig. 3B). Furthermore, following alignment and annotation, 15 differentially altered metabolites corresponding to 15 compound names were identified using Student’s t-test (Table 1). Compared with that in the ELF of the control group, the levels of Hyaluronic acid, 4-hydroxynonenal, 2,3-diaminopropanoate, (5-L-glutamyl)-L-amino acid, D-ribulose 1,5-bisphosphate, 3-hydroxy-5-phosphonooxypentane-2,4-dione, and L-capreomycidine were significantly increased in the ELF of patients in the MD group, whereas the levels of citrate, ethylenediaminetetraacetic acid (EDTA), inosine 5′-tetraphosphate, D-octopine, N-acetyl-D-glucosamine (GlcNAc), D-glucuronic acid (GlcUA), L-arginine, and 1-hydroxy-2-methyl-2-butenyl 4-diphosphate were significantly reduced in the ELF of patients in the MD group.

Fig. 3.

Screening of differentially altered metabolites based on the thresholds of P ≤ 0.05 and variable important in projection > 1. Volcano plots of differentially altered metabolites in the negative A and positive modes B. Blue dots represent reduced metabolites, whereas red dots represent increased metabolites in the ELF samples of the patients with MD compared with that in the patients with AN. MD, Meniere’s disease; AN, acoustic neuroma; ELF: endolymphatic sac luminal fluid

Table 1.

Fifteen differentially altered metabolites corresponding to 15 compound names

Name	ID	Compound name	Rt (s)	m/z	VIP	FC	log2FC	P-value	− log10p
M141T388_NEG	C06393	2,3-Diaminopropanoate	388.49	141.02	1.65	2.11	1.0772	0.022	1.6576
M193T533_NEG	C00158	Citrate	532.89	193.02	1.33	0.09	− 3.4739	0.031	1.5086
M777T254_NEG	C00518	Hyaluronic acid	2546.73	777.23	1.38	3.36	1.7485	0.007	2.1549
M243T411_NEG	C11811	1-Hydroxy-2-methyl-2-butenyl 4-diphosphate	411.38	242.98	1.33	0.47	− 1.0893	0.047	1.3279
M248T326_NEG	C20960	3-Hydroxy-5-phosphonooxypentane-2,4-dione	326.02	247.98	1.59	2.1	1.0704	0.026	1.5850
M292T419_NEG	C00284	EDTA	418.92	292.08	1.66	0.18	− 2.4739	0.028	1.5528
M570T421_NEG	C03614	Inosine 5′-tetraphosphate	420.503	569.92	1.7	0.47	− 1.0893	0.02	1.6990
M175T455_POS	C00062	L-Arginine	455.47	175.12	1.34	0.26	− 1.9434	0.046	1.3372
M155T165_POS	C21642	4-Hydroxynonenal	165.26	155.11	1.56	3.56	1.8319	0.0001	4.0000
M236T378_POS	C18472	L-Capreomycidine	377.6	236.11	1.53	2.04	1.0286	0.029	1.5376
M245T255_POS	C00140	N-Acetyl-D-glucosamine	254.75	245.08	1.61	0.49	− 1.0291	0.025	1.6021
M247T341_POS	C04137	D-Octopine	341.11	247.14	1.32	0.33	− 1.5995	0.048	1.3188
M378T349_POS	C01182	D-Ribulose 1,5-bisphosphate	348.84	377.94	1.48	2.47	1.3045	0.033	1.4815
M193T424_POS	C16245	D-Glucuronic acid	424.93	193.03	1.38	0.43	− 1.2176	0.001	3.0000
M201T390_POS	C03740	(5-L-Glutamyl)-L-amino acid	389.8	201.09	1.34	2.56	1.3561	0.045	1.3468

Immunohistochemical Validation of the Expression of 4-HNE and HA in the ES of the Patients with MD and AN

Among 15 differentially altered metabolites in the ELF of MD, two metabolites were selected as candidates for further validation: hyaluronic acid and 4-hydroxynonenal. These metabolites were selected because HA was known to be crucial for fluid and ion regulation within the ES due to their capacity to bind water [25], and 4-HNE was one of the most common markers for evaluating oxidative stress [26], which was supposed to be involved in the development of EH [27].

Immunofluorescent staining showed 4-HNE was strongly expressed in the cytoplasm and nucleus of the ES epithelium and the subepithelial connective-tissue in the patients with MD, as shown in Fig. 4; while HA was found to be strongly expressed in the cytomembrane and cytoplasm of the ES epithelium forming ring-like structure around the well visualized cell nucleus and stripe-like structure in the subepithelial matrix in the patients with MD, as shown in Fig. 4. In contrast, 4-HNE staining was weak or absent and HA signal was moderate in the ES epithelium and the subepithelial connective-tissue in the patients with AN, as shown in Fig. 4. Coexpression of HA synthase 2 (HAS2) and hyaluronidase 1 (HYAL1) in the ES tissue showed HAS2 was strongly expressed in the cytoplasm and nucleus of the ES epithelium in the patients with MD and weakly expressed in the patients with AN, as shown in Fig. 5, whereas HYAL1 was moderately stained in the cytoplasm and nucleus of the ES epithelium in both the patients with MD and the patients with AN, as shown in Fig. 5. Quantification analysis of fluorescence intensity revealed that the mean values and standard deviations of the fluorescence intensity of 4-HNE, HA, HAS2, and HYAL1 in the DAPI-stained epithelial area were 1.22 ± 0.23, 2.47 ± 0.70, 1.61 ± 0.46, and 1.21 ± 0.33 in the patients with MD, and 0.59 ± 0.12, 0.93 ± 0.32, 0.83 ± 0.25, and 1.27 ± 0.29 in the patients with AN, respectively. The fluorescence intensity of 4-HNE, HA, and HAS2 in the patients with MD was significantly higher than that in the patients with AN (P < 0.05), whereas there was no statistically significant difference in the fluorescence intensity of HYAL1 between the patients with MD and the AN control, as shown in Fig. 6.

Fig. 4.

An example of the co-expression of 4-HNE and HA in ES tissue sections of the patients with MD and AN. Immunofluorescence staining showed a strong signal for 4-HNE (green) A and an intensive labeling for HABP (red) B in the ES epithelium of the patients with MD. Co-expression of 4-HNE and HABP along with the marker of nuclei, DAPI (blue) C showed 4-HNE was localized in the cytoplasm and nucleus of the ES epithelium D, while HABP2 was localized in the cytomembrane and cytoplasm of the ES epithelium forming ring-like structure (white arrow) around the well visualized cell nucleus and stripe-like structure (white arrow head) in the subepithelial matrix in the patients with MD D. In contrast, a weak or absent staining for 4-HNE (green) E and a moderate signal for HABP2 (red) F were detected in the ES epithelium and the subepithelial connective-tissue in the patients with AN, whereas merged image of immunofluorescent labeling for 4-HNE and HABP2 showed red-green color weak in ES epithelia cells H in the patients with AN. HA, hyaluronic acid; 4-HNE, 4-hydroxynonenal; HABP2, hyaluronan binding protein 2; MD, Meniere’s disease; AN, acoustic neuroma; ES, endolymphatic sac. Scale bars: A–H 12.5 µm

Fig. 5.

An example of the co-expression of HAS2 and HYAL1 in ES tissue sections of the patients with MD and AN. A moderate signal for HYAL1 was detected in the cytoplasm and nucleus of the ES epithelium in both the patients with MD A and the patients with AN E, whereas immunofluorescent staining for HAS2 showed an intensive labeling in the patients with MD B and a weak staining in the patients with AN F in the cytoplasm and nucleus of the ES epithelium. Merged image of immunofluorescent labeling for HAS2 and HYAL1 showed color in pink in ES epithelium of the patients with MD D and color in green in ES epithelium of the patients with AN H. MD, Meniere’s disease; AN, acoustic neuroma; ES, endolymphatic sac; HAS2, hyaluronan synthase 2; HYAL1, hyaluronidase 1. Scale bars: A–H 20 µm

Fig. 6.

The values of fluorescence intensity of 4-HNE, HA, HAS2, and HYAL1 in the DAPI-stained ES epithelial area in the MD group compared with those in the AN group. MD, Meniere’s disease; AN, acoustic neuroma; ES, endolymphatic sac; HA, hyaluronic acid; 4-HNE, 4-hydroxynonenal; HAS2, hyaluronan synthase 2; HYAL1, hyaluronidase 1

Discussion

In the present study, we applied non-targeted MS-based metabolomic profiling to examine local changes in metabolite levels in diluted ELF samples collected from patients with MD and those with AN. To the best of our knowledge, the present study was the first to explore MD-specific changes in metabolites in the ES. OPLS-DA was used to evaluate the robustness of the experimental data, demonstrating clear separation between the two groups. Subsequently, we verified the differentially altered metabolites using Student’s t-test, which identified 7 significant increased metabolites and 8 significant decreased metabolites. Using immunohistochemical analysis in the sac specimens obtained from the above-mentioned 10 patients with MD and 6 patients with AN, two differentially altered metabolites, HA and 4-HNE, have been validated to be upregulated in the epithelial lining of the ES, as well as in the subepithelial connective-tissue in patients with MD comparing with that in patients with AN. These metabolites found to be significantly altered in the present study may further the understanding of the molecular mechanism underlying MD.

Although numerous causative factors have been considered in the development of hydrops and in the pathogenesis of related cochleovestibular dysfunction [1], the etiology of MD remains elusive. However, increasing evidences have suggested that oxidative stress is involved in the development of EH and apoptotic cell death, in turn contributing to sensorineural hearing loss during the latter stages of MD [27]. An upregulated expression of inducible nitric oxide synthase (iNOS), a marker of oxidative stress, and high level of nitric oxide (NO), the product of arginine catalyzed by iNOS [28], and accumulation in the hydropic vestibule were observed in both animal model [29] and MD patients [30, 31], which was supposed to be associated with the development of EH. A decreased level of L-arginine and an increased level of L-capreomycidin, one of downstream products of L-arginine [32], in the ELF of patients with MD found in the present study were consistent with the upregulated expression of iNOS in the hydropic vestibule reported in previous studies [30, 31]. Moreover, previous studies have shown a chronic inflammatory process underlying MD in two inflammatory subtypes of patients according to the proinflammatory cytokines and methylation profile status [33–35], and an increased immunologic activity in the microenvironment of the ES of the patients with MD [13], suggesting that an increase in oxidative stress-associated inflammation that can strongly induce the expression of iNOS may play a critical role in pathogenesis of MD. Moreover, an increased level of protein carbonyls and 4-HNE, the most common markers for evaluating oxidative stress, have been found in lymphocytes from MD patients with respect to control group, suggesting patients affected by MD are under condition of systemic oxidative stress [27]. In the present study, we found that the level of 4-HNE in the ELF of the patients with MD was significantly increased, and the expression of 4-HNE in the epithelium and the subepithelial connective-tissue of the ES in patients with MD was upregulated compared with that in patients with AN, further supporting oxidative stress product may be involved in the pathology of MD. As 4-HNE is a major aldehydic end-product measured as an indicator of oxidative stress-induced lipid peroxidation, studies have corroborated its ability to disrupt signal transduction and protein activity, as well as induce inflammation and trigger fibrogenesis in conditions of oxidative stress [36, 37]. Therefore, we speculated that an increased level of 4-HNE in the ELF of the patients with MD might mediate the damage to the ES leading to the dysfunction of ES that have been directly linked to the pathogenesis of EH [38].

HA is a non-sulfated glycosaminoglycan in which GlcNAc and GlcUA are linked together by alternating β-1,3 and β-1,4 linkages [39]. As HA synthesis is controlled by HA synthases (HASs) and HA degradation is controlled by hyaluronidases (HYALs), those high HA levels could be associated with high expression of HASs and low expression of HYALs [39]. Previous studies have shown that HA could be synthesized in the ES epithelial cells [40] and subsequently secreted out into the ES lumen following treatment with hyperosmolar solutions, loop diuretics or after obstruction of the endolymphatic duct [41, 42]. Due to its water-binding capability, HA may be involved in inner ear fluid volume and pressure regulation [43]. In the present study, HA was found to be increased about threefold in the ELF of the patients with MD compared with those in patients with AN, and validated to be expressed increasingly in the epithelium and subepithelial matrix of the ES from the patients with MD (Fig. 4), suggesting the synthesis of HA was enhanced in MD, which was consistent with the upregulated expression of HAS2 in the ES epithelium of the patients with MD (Fig. 5). While GlcUA and GlcNAc, two substrates for synthesis of HA, decreased in the ELF of the patients with MD could be explained by the excess synthesis of HA in the ES epithelium, which consumed large quantities of these nucleotide sugar precursors. However, an increased level of HA in the ES lumen could attract more ions and water by its anionic charges and hydrophilic properties and further the influx of water to the endolymphatic space for the development of EH. Recently, we reported the reversal of EH in some patients following endolymphatic duct blockage for treatment of intractable MD [44, 45]. A plausible explanation is that this surgical procedure may prevent the diseased ES from communicating excessive endolymph volumes to the cochlea. Such clinical evidence strongly suggests that increased endolymph is associated with ES in at least some patients with MD. However, this overproduction of endolymph in ES is possibly associated with the upregulation of HA in the ES lumen.

Furthermore, there were two notable differentially altered metabolites between MD group and control group: Citrate and EDTA. Citrate was found to be decreased about 0.09 fold, and EDTA was found to be reduced about 0.18 fold in the ELF of the patients with MD compared with those in patients with AN. As we known, EDTA is a chelator of divalent metal ions such as copper, iron, calcium, and zinc. As its chelating capability of Ca²⁺, EDTA has been used to prevent metal ions from participating in the generation of reactive oxygen species, lipid peroxidation, and detrimental calcium-dependent cascades [46, 47]. Citrate is well known as a key metabolic intermediate in the mitochondrial tricarboxylic acid (TCA) cycle of living organisms for maintaining energy homeostasis. However, ionic chelation potential of citrate, similar to EDTA, has been considered as the major mechanism underlying citrate’s biological function [48]. In the central nervous system (CNS), high concentrations of citrate are detected in cerebrospinal fluid, which are believed to play a pivotal role in regulating neuronal excitability through its chelating capability of Ca²⁺ and Mg²⁺ [49]. In renal system, citrate chelation of Ca²⁺ is helpful for preventing its precipitation and reducing urinary supersaturation of calcium salts, and plays the protective effect on renal stone formation [50]. However, in the inner ear, an exceptionally low calcium (Ca²⁺) concentration in the endolymph is crucial for proper auditory and vestibular function [51], which is believed to be maintained through modulating the calcium-sensing receptor that was exclusively localized in the ES epithelium [52]. Loss of the proposed Ca²⁺-homeostatic function of the ES may be of significance in various pathological conditions, such as otoconial disorders and MD [53, 54]. Although the functional importance of citrate and EDTA in the ES endolymph still remains elusive, its chelating capability of Ca²⁺, greatly altering the availability of Ca²⁺ concentration, can be expected to play an important role for maintaining extracellular Ca²⁺ homeostasis in the ES endolymph. A remarkably decreased levels of citrate and EDTA in the luminal fluid of ES of the patients with MD found in the present study could be suggested to be associated with the elevated endolymphatic Ca²⁺ in the ears with chronic endolymphatic hydrops, which is consistently observed in animal model [51, 55]. As this nonphysiological increase in [Ca²⁺] endolymph was presumed to be the critical etiopathogenic factor in MD, a significantly reduced levels of citrate and EDTA in the luminal fluid of ES in patients with MD might be an important biomarker underlying the mechanism of this disease.

However, limitations exist in the present study. Theoretically, the use of the ELF from healthy individuals with normal hearing as control would have improved our study. However, sampling ELF from patients with normal function of the inner ear is almost impossible due to ethical reasons. Although an MRI sign of mild EH (a saccular dilation on the side of the tumor) was reported in some patients with AN using non-enhanced T2-weighted images at 3 T [56], the exclusion of ipsilateral EH in all patients with AN using 3D-FLAIR MRI in the present study can minimize the impact of the ELF from patients with AN used as a control for a comparative analysis of the differentially altered metabolites. In addition, as the samples of ELFs were obtained from the patients with MD who were selected to undergo endolymphatic duct blockage based on the intraoperative pathoanatomic findings of the sac, the results in the present study reflected the differentially altered metabolites of ES in part of the patients with MD with hyperplastic or normal ES in morphology rather than the patients with MD with hypoplastic or degenerated ES in morphology. Moreover, the sample size in the present study is limited because only a small group of patients with MD were indicated for endolymphatic duct blockage after the failure of conservative treatment. Furthermore, as the pathological changes at the molecular level may not be the same in the patients with different time of the disease, additional studies between the patients with different time of onset of EH development are necessary to validate our findings in larger cohorts.

In summary, this first metabolomics study conducted on the ELF obtained from 10 patients with MD and 6 patients with AN identified 7 metabolites that their expression increased and 8 metabolites that their expression decreased in the patients with MD. Among these differentially altered metabolites, an upregulated expression of HA detected in the ES lumen of the patients with MD was likely to be associated with the increased endolymph in ES, while an increased level of 4-HNE found in the ELF of the patients with MD provided direct evidence to support that oxidative damage and inflammatory lesions underlie the mechanism of MD. In addition, citrate and EDTA were detected to be decreased substantially in the ELF of the patients with MD, indicating the disturbed endolymphatic Ca²⁺ in MD is likely to be associated with the reduction of these two chelators of Ca²⁺ in ES. The results in the present study indicate that these variations in the metabolic signatures in the ES of patients with MD may facilitate the understanding of molecular mechanisms involved in the etiology of MD and possible therapeutic targets for treatment of MD. However, the role of these metabolites of interest in the development of MD warrants further exploration.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contribution

Li Huang connected data and prepared the first draft. Qin Wang, Huang Chao, and Zhou were responsible for the design, material preparation, data collection, and analysis. Zhiwen Zhang and Anquan Peng designed the study conception and revised the manuscript. All authors read the manuscript and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (Grant No. 81570928).

Data Availability

The authors confirm that the data supporting the findings of this study are available within the article [and/or its supplementary materials].

Declarations

Conflict of Interest

The authors declare no competing interests.