This study was approved by the Animal Research Committee of Seoul National University Hospital, and all animal care was supervised by the Institutional Animal Care and Use Committee Institute (IACUC 18–0168-S1A0). The detailed methods on experimental animals and drug/vehicle preparation are described elsewhere (Lee J. H et al., 2021; Li et al., 2021). In brief, Albino Hartley guinea pigs (Cavia porcellus) weighing 300–350 g were used (22 ears). Perilymph was collected twice, 2 μL each with a time interval of approximately 30 s. When the first sample was drained from the round window, it took approximately 10 s for the perilymph to refill the ST enabling the second sampling. The ears from guinea pigs were categorized into three groups, according to the treatment strategy. The first two groups were treated with the same concentration (12 mg/ml) of IT dexamethasone phosphate; however, the vehicles for delivering dexamethasone differed between the two groups. Dexamethasone was mixed inside the high and low molecular weight hyaluronic neuronic acid (HA) in the DVV group (n = 10), whereas in the saline group (n = 7), the vehicle was physiologic saline. The control group (n = 5) did not receive any treatment. Therefore, 22 samples were harvested from 22 ears of 15 animals (Table 1). For each animal, the left and right ears were allocated in such a way that individual animal-associated variability was minimized. The specific composition and preparation process of the drug/vehicle are reported elsewhere (Kang et al., 2020; Li et al., 2021). In brief, dexamethasone disodium phosphate (water-soluble salt form) was prepared in 10 mM phosphate buffer at a concentration of 1.2% (w/v). Low-molecular-weight HA (10–100 kDa, MNH Bio Co. Ltd. Hwaseong-si, Korea) was completely dissolved in the solution at a concentration of 1.0% (w/v) to coat the drug. Dexamethasone mixed in low molecular-weight HA was prepared. High-molecular-weight HA (5,000–10,000 kDa, MNH Bio Co. Ltd.) was dissolved in the dexamethasone-low-molecular weight HA solution by vigorous mixing until a concentration of 2.0% (w/v) was reached. Hyaluronic acid was filtered with a 0.2 μm cellulose acetate membrane to prevent bacterial and mold contamination. As for the dexamethasone sodium phosphate, the total aerobic microbial count was less than 103 cfu/g, total combined mold and yeast count was less than 102 cfu/g, and bacterial endotoxins were less than 0.25 EU/mg. Loading the drug in the vehicle was performed in a sterilized container and environment.

The detailed methods on IT drug delivery are described elsewhere (Lee J. H et al., 2021). Animals were anesthetized with 2 mL/kg of xylazine (Rompun, Bayer-Korea, Seoul, Korea) and 1.2 mL/kg of zoletil (Zoletil 50, Virbac, Seoul, Korea) using intramuscular administration. IT dexamethasone administration was performed under a surgical microscope (OPMI Pico, Carl-Zeiss, Oberkochen, Germany). An Angiocath Plus 24-gauge needle (BD, Sandy, UT, United States) was connected to a 1-ml syringe (KovaxSyringe 1 ml, Korea Vaccine Co., Seoul, Korea) with a miniextension tube (Mini-Volume Line, Insung Medical, Seoul, Korea). An air vent was first made in the anterior superior quadrant of the tympanic membrane (TM). The needle was inserted carefully and the drug/vehicle was injected at a low speed. The injection was stopped when the drug/vehicle completely filled the middle ear (bulla) or the drug/vehicle leaked out through the air vent. The volume of the drug/vehicle injected into the middle ear cavity was similar 40–60 μL between the DVV group and saline group. After injecting the drug/vehicle into one ear, the other ear was also immediately injected with a different drug/vehicle; the time interval between the first and second injections was <5 min. To prevent position effects, the animals were placed in a straight prone position, without leaning toward the left or right side, until the experiment was concluded. The drug/vehicle that was injected, and the order in which the drug/vehicle was injected, were randomized.

The detailed methods on experimental animals (Lee J. H et al., 2021) and drug/vehicle preparation are described elsewhere (Li et al., 2021). Perilymph sampling was performed under anesthesia. After retroauricular incision, skin and fibromuscular layers were removed for verifying the landmark structure under an operating microscope. Round cutting burr was used to open the bulla and expose the round window. Without damaging the round window membrane, the bone dust and tissue fluid were cleared with paper gauze from the bulla and round window niche. The collection of perilymph was performed 3 days after IT injection. This time delay was determined based on our previous publication (Li et al., 2021), considering a significant hearing difference that was found between 1 and 4 days. A 10 μL LongReach Pipette Tip (Multimax, USA) with distal end core diameter ranging from 480.65 to 640.87 μm was introduced slowly into the round window. The round window was not ruptured until the final sampling moment. After sampling the perilymph very slowly with a micropipette (Eppendorf Research plus, Eppendorf, Germany) which was set at the volume of 2 μL, the volume of perilymph was double checked with a separate gauge. Purity of perilymph was screened (from grade 1 to grade 4), by comparing the colour of the sample to a standardized colour map.

The guinea pig perilymph sample preparation of DVV group (n = 10), capsulated dexamethasone within DVV (n = 10), saline mixed dexamethasone treated (n = 7), and non-treated group (n = 2) was performed as described previously with some modifications (Lee J. H et al., 2021). The perilymph sample was extracted from one ear or both ears of the guinea pig. Among them, the perilymph with sufficient amounts were extracted twice, while those with insufficient amounts were extracted only once. The protein digestion process was optimized to 2–5 μL of perilymph samples. Digestion buffer (8 M UREA, 5 mM tris [2-carboxyethyl] phosphine, and 20 mM chloroacetamide in 50 mM ammonium bicarbonate) was added to the perilymph sample for a brief period. The mixture was boiled for 30 min at 60°C to denature and alkylate proteins. Trypsin digestion was performed at 37°C overnight using a trypsin/LysC mixture at a 100:1 protein-to-protease ratio. All resulting peptides were acidified with 10% trifluoroacetic acid and desalted using homemade C18-StageTips as previously described (Lee J. H et al., 2021). Desalted samples were completely dried with a speed-vac and stored at −80°C.

The data-dependent acquisition methods were conducted with an Ultimate 3000 UHPLC system (Dionex, Sunnyvale, CA, United States) coupled to a Q-Exactive HF-X mass spectrometer (Thermo Scientific, Hamburg, Germany) as previously cited modifications (Han et al., 2014; Lee et al., 2018). The digested peptides were re-suspended in 0.1% formic acid after C-18 desalting procedure (Han et al., 2014). Peptide samples were separated on a two-column system with a trap column and an analytical column (75 µm * 50 cm) with 120-min gradients from 2% to 32% acetonitrile at 300 nl/min. The column temperature was maintained at 60°C using a column heater. The column eluent was delivered to the Q-Exactive HF-x via nanoelectrospray. For label-free quantification, a survey scan (350–1,450 m/z) was acquired with a resolution of 70,000 at m/z 200. A top-20 method was used to select the precursor ion with an isolation window of 1.2 m/z. The MS/MS spectrum was acquired at a high collision dissociated-normalized collision energy of 30 V with a resolution of 17,500 at m/z 200. The maximum ion injection times for the full and MS/MS scans were 20 and 100 m, respectively.

All MS raw files were processed by using interface of MaxQuant (version 1.5.3.1) (Tyanova et al., 2016a). MS/MS spectra were searched from the UniProt protein sequence data set (December 2014, 88,657 entries) using the Andromeda search engine (Cox et al., 2011). Primary searches were done using a 6-ppm precursor ion tolerance. The MS/MS ion tolerance of 20 ppm was used. Carbamido-methylation of cysteine was specified as the control modification, and N-acetylation of protein and oxidation of methionine were considered as variable modifications. Enzyme specificity was set to full tryptic digestion. Peptides with a minimum length of six amino acids and up to two missed cleavages were included. The acceptable false discovery rate (FDR) was set to 1% at the peptide, protein, and modification levels. To maximize quantification events across samples, we enabled the ‘Match between Runs’ option of the MaxQuant platform. For label-free quantification, the Intensity Based Absolute quantification (iBAQ) algorithm (Schwanhäusser et al., 2013) was used as part of the MaxQuant platform. Briefly, the iBAQ values calculated by MaxQuant are the raw intensities divided by the number of theoretical peptides. Thus, the iBAQ values are proportional to the molar quantities of the proteins.

Perseus software (version 1.6.43.0) was used for all statistical analyses of MS data (Tyanova et al., 2016b). In case of label-free quantification, iBAQ intensities were log2-transformed. After filtering out proteins with at least 70% valid values in each group, missing values were imputed assuming a normal distribution of 0.5 width and 1.8 downshifts. The data were then normalized via width adjustment normalization. The multiple comparison test were performed with one-way ANOVA and t-test was also applied for individual group comparison. The differentially expressed proteins (DEPs) from both analysis filtered with p-value lower than 0.05. In the case of the individual t-test, the fold-change (FC) values were calculated to refine the DEPs with a union of log   2 F C lower than −0.584963 and greater than 0.584963.

Most biological process, molecular function, and pathway analyses were performed using DAVID (Huang et al., 2009) and STRING version 11 (Szklarczyk et al., 2019). The identified biological processes were clustered using the EnrichmentMap (Merico et al., 2010) and AutoAnnotate (Kucera et al., 2016) application in Cytoscape tool (version 3.8.2) (Shannon et al., 2003). Proteins detected by both quantification methods and common interacted proteins using two different topological methods were analyzed using Venny 2.1.0 (bioinfogp.cnb.csic.es/tools/venny). Protein–protein interactions (PPIs) and visualization of protein expression with p-values were mapped using Cytoscape (version 3.8.2). For interactome analysis, PPI analysis were combined with STRING version 11.5 (Szklarczyk et al., 2019). The up-stream signaling or canonical pathway analysis was considered within Ingenuity and KEGG pathway analysis (Kanehisa and Goto, 2000).

Our study was designed to discover the molecular changes and clinical application of DVV carriers that embedded dexamethasone through the biological process of and related signaling pathway analysis of DEPs. Consequently, we performed global proteome profiling of guinea pig perilymphs that were treated with DVV capsulated with dexamethasone, dexamethasone mixed in saline, and a non-treated group (Figure 1). The LC-MS/MS analysis identified 973, 1,047, and 1,038 proteins from the control, dexamethasone mixed with saline, and dexamethasone carried in the DVV groups, respectively (Figure 2A). Among those, 593 quantifiable proteins that were detected more than 70% among all of the samples in each group were filtered for further processing (Figure 2B). The number of quantifiable proteins and the dynamic range of identified proteins from each group are also provided in Figures 2C–E. The protein and peptide FDRs of the identified proteins were below 1% (Q-value <0.01). Before statistical comparison, a principle component analysis (PCA) plot was presented after width adjustment normalization to explore the discrimination power by identified protein abundance. The result shows that the DVV-treated group was distinctively clustered, but the other groups were not specifically clustered, although there was a slight overlap between them (Figure 2F). To compare the protein expressions of three different groups, we performed one-way ANOVA multiple-sample comparison test and determined 46 DEPs that were statistically significant (p-value <0.05). Hierarchical clustering represented three specific clusters over each group. Based on protein expression patterns, each cluster was distinctly separated within two clusters that were DVV, specifically upregulated pattern and downregulated patterns, with no specific difference between the saline and control groups (Figure 3A).

The protein expression was distinctively clustered within two different patterns, as described in the colorimetric scheme presented with the expression patterns. The first cluster included the relatively upregulated proteins within DVV-treated groups compared with the saline mixed dexamethasone and control groups. The other clusters were specifically downregulated proteins within the DVV group. To determine the biological processes and physiological traits after DVV treatment, we proceeded to a gene ontology analysis of DEPs categorized in those two different clusters. The results showed that the upregulated proteins after DVV treatment were related to the response to cytokine stimulus, cytoskeleton organization, and the negative regulation of actin filament depolymerization. On the other hand, the downregulated proteins after DVV treatment were categorized in the protein activation cascade, wound healing, and regulation of the acute inflammatory response (Figure 3B). Moreover, PPI analysis was performed to examine the interaction between the overall identified DEPs. The STRING database (version 11.5) (Szklarczyk et al., 2019) was applied to plot interaction with a medium confidence of interaction score (<0.4), and functional analysis results (or visualization) were refined within the Cytoscape tool (Figure 3C) (Shannon et al., 2003). Within the interaction, some upregulated proteins after DVV treatment were clustered, and functional analysis resulted in cytoskeletal organization. Among these, many proteins acted to inhibit actin filament formation and release actin biogenesis. For instance, lamin A/C’s (LMNA) deficient cells increased myosin bipolar filament accumulation (Wiggan et al., 2020) and gelsolin (GSN) lowered actin filament formation (Feldt et al., 2018). On the other hand, the downregulated proteins after DVV treatments were classified into terms of acute inflammatory response and response to external stimulus. Interestingly, the downregulated proteins are associated with a reduction in the frequency of the immediate defensive reaction to infection or injury caused by DVV treatment. Specifically, the kininogen 1 (KNG1) knockout model decreases cytokine production and inflammatory monocytes that depress the kallikrein-kininogen system (Stadnicki et al., 1998; Wang et al., 2018). C4b binding protein alpha (C4BPA) chain is principally employed in critical inflammatory and coagulation process (Iqbal et al., 2022). Therefore, our results suggest that DVV treatment induces a decrease in the acute inflammatory response within the adaptive immune process and inhibits actin cytoskeletal organization.

Although, one-way ANOVA test presented significant expression patterns within the DVV group, there were much similar expression patterns between control and dexamethasone mixed in saline group. Therefore, we also performed a paired t-test between the three different groups for a detailed comparison. The DEPs with upregulated (p-value <0.05 and fold-change ≥1.5) and downregulated (p-value <0.05 and fold-change ≤ −1.5) expressions were selected after each comparison. The MS analysis identified 16 (up = 6, down = 10), 54 (up = 32, down = 22), and 70 (up = 36, down = 34) DEPs from a comparison between the control and saline groups, the saline and DVV-treated groups, and the control and DVV-treated groups, respectively. All identified DEPs from each comparison group were respectively visualized using volcano plots (Figures 4A–C), and all of the DEPs are listed in Supplementary Tables S1–3. These statistical results revealed that there was not much difference between the control and saline groups, while protein expression comparisons between the control or saline groups and the DVV-treated group showed more remarkable changes. These results proved that dexamethasone carried by DVV treatment exhibits better therapeutic effects in pre-clinical level and suggests the possibility that DVV treatment could widely affect pathogenically than saline-mixed dexamethasone treatment, which is the traditional way in present.

From the pairwise comparison between each group by t-test, we discovered that protein expression levels were relatively not much different between saline-treated and control groups in comparison with the difference between DVV-treated and control groups. Therefore, there are specific physiological or biological traits that could reflect such differences. To discover biological functions among different groups after DVV treatment, we performed network analysis using PPIs of DEPs from a comparison between the DVV-treated group with saline-treated and control groups. These were also embedded with protein expressions and presented as log2FC values and interaction scores. The gene ontology (GO) analysis of those were also performed to explore and compare the biological process over each group (Supplementary Tables S5, S6). The PPI plot of the comparison between the DVV group and the saline-treated group is described in Figure 5A. Consistent with the results obtained from the one-way ANOVA analysis, biological processes, such as actin filament organization or response to cytokine, are enriched in the sub-network. Furthermore, responses to oxygen-containing compounds were specifically observed in the PPI network. Interestingly, most of the proteins included in this term were downregulated after DVV treatment. The absence of vimentin (VIM), the only upregulated protein in response to the oxygen-containing compound process, induced reactive oxygen species (ROS) production and intensified bacterial killing (Mor-Vaknin et al., 2013). With the reported functions and other downregulated proteins, it is possible to propose that DVV treatment decreases the ROS production-related process or activation.

Distinct physiological processes were identified from the DEPs compared to the DVV-treated and saline-treated groups. In the PPI network, the actin cytoskeletal organization or response to oxidative stress was also clustered similarly to the comparison between the DVV-treated and control groups. Notably, proteins related to the regulation of wound healing were clustered (Figure 5B). This biological process is known to include the rate or extent of events that recover integrity to a damaged tissue. Despite almost all proteins being downregulated, fibromodulin (FMOD) was the only upregulated protein in the regulation of the wound healing process reported to repair cutaneous wounds from adult FMOD knock-out mice with the development of scar formation, block wound closure, and reduced angiogenesis. A previous study suggested that this could be caused by exogenous FMOD administration (Zheng et al., 2017). Fibulin-1 (FBLN1) affects several disease pathogenesises and related tissue remodeling. The genetic inhibition of FBLN1 decreases the inflammatory cells by reducing pro-inflammatory cytokines within chronic obstructive pulmonary disease (Liu et al., 2016) and its expression increases infiltrating macrophages by elevating the stromal and immune scores in high FBLN1 tissues (Gong et al., 2020).

Moreover, to investigate the molecular interaction, reaction, or related physiological pathways of identified DEPs after IT dexamethasone treatment with a DVV carrier, a KEGG pathway analysis was applied (Kanehisa and Goto, 2000). Specifically, the antigen processing and presentation pathway (map04612) and proteasome complex (map03050), which is also a part of the antigen processing pathway, were identified (Figure 6A). The cathepsin B (CTSB) protein—the downregulated protein in our study, which is categorized in the antigen processing pathway—is well known for controlling mediators of Th1 immune reaction in antigen-presenting cells (Gonzalez-Leal et al., 2014). During antigen processing, the proteasome complex transforms several subunits and forms an immunoproteasome. These immunosubunit formations adapt to serve the immune system as well (Kloetzel, 2001). The 26 S proteasome regulatory subunit 6B (PSMC4) located in regulatory particles, proteasome subunit alpha type 3 (PSMA3), and proteasome subunit beta types 6 (PSMB6) and 7 (PSMB7) included in core particles were upregulated after IT dexamethasone treatments. The proteasomes play an important role in the formation of antigenic peptides for presentation on the major histocompatibility (MHC) class Ⅰ and the activation of the immune process as well (McCarthy and Weinberg, 2015). It has also been reported that proteasome mutation, which results in loss of activation, causes autoinflammation managed by the interferon gene signature (Goetzke et al., 2021). Therefore, the identified DEPs announced increased proteasome activity after IT dexamethasone treatment and supported antigen presentation. On the other hand, the downregulated proteins in the analysis (complement C1s [C1S], C4bp alpha-chain [C4BPA], and complement C8 alpha chain [C8A]) are categorized in complement and coagulation cascade (map04610; Figure 6B). The identified DEPs are especially consisted within the classical complement pathway, which usually acts as an antigen-antibody complex formation. Specifically, the downregulation of C4BPA inhibits C3b construction by suppressing C3 convertase, and downregulation of C8a would also lower complement activation and interfere with the development of a membrane attack complex (MAC) (Oikonomopoulou et al., 2012).