The impact of systemic versus intratympanic dexamethasone administration on the perilymph proteome

Abstract

Glucocorticoids are the first line treatment for sensorineural hearing loss, but little is known about the mechanism of their protective effect or the impact of route of administration. The recent development of hollow microneedles enables safe and reliable sampling of perilymph for proteomic analysis. Using these microneedles, we investigate the effect of intratympanic (IT) vs. intraperitoneal (IP) dexamethasone administration on guinea pig perilymph proteome. Guinea pigs were treated with IT dexamethasone (n=6), IP dexamethasone (n=8), or untreated for control (n=8) six hours prior to aspiration. The round window membrane (RWM) was accessed via a postauricular approach, and hollow microneedles were used to perforate the RWM and aspirate 1 μL of perilymph. Perilymph samples were analyzed by liquid chromatography-mass spectrometry-based label-free quantitative proteomics. Mass spectrometry raw data files have been deposited in an international public repository (MassIVE proteomics repository at https://massive.ucsd.edu/) under dataset # MSV000086887. In the 22 samples of perilymph analyzed, 632 proteins were detected, including the inner ear protein cochlin, a perilymph marker. Of these, 14 proteins were modulated by IP, and 3 proteins were modulated by IT dexamethasone. In both IP and IT dexamethasone groups, VGF nerve growth factor inducible was significantly upregulated compared to control. The remaining adjusted proteins modulate neurons, inflammation, or protein synthesis. Proteome analysis facilitated by the use of hollow microneedles shows that route of dexamethasone administration impacts changes seen in perilymph proteome. Compared to IT administration, the IP route was associated with greater changes in protein expression, including proteins involved in neuroprotection, inflammatory pathway and protein synthesis. Our findings show that microneedles can mediate safe and effective intracochlear sampling and hold promise for inner ear diagnostics.

Graphical Abstract

Heat Map of Mean Abundances for the 15 Proteins Modulated by Dexamethasone Administration, generated using Morpheus (https://software.broadinstitute.org/morpheus).

1. INTRODUCTION

Glucocorticoids, given systemically via oral route or locally by intratympanic (IT) injection, are the first-line treatment choice for sudden sensorineural hearing loss¹, yet little is known about the effects exerted in the inner ear and the mechanism of hearing loss treatment. Immunosuppressive effects via the glucocorticoid receptor have often been credited for their positive effects on hearing loss, but other cellular functions may also play a role². Furthermore, differences in effect between systemic and local administration of glucocorticoids are not fully elucidated. IT administration has been shown to be as effective as systemic administration in treating hearing loss^3,4. Local administration results in greater perilymph concentrations and favors the cochlear base whereas systemic administration favors the apex^5,6. In mice, IT glucocorticoids affect thousands more inner ear genes compared to systemically delivered glucocorticoids on studies of mRNA expression in cochlear tissues⁷.

Prior research on the effects of inner ear glucocorticoid administration has focused on their alteration of gene expression^7,8. Systemically administered glucocorticoids were found to modulate immune and inflammatory pathways, including genes that were associated with cytokine-cytokine receptor interaction and cell adhesion molecules in the immune system⁸. Immune-related genes Ccl12 and Glycam1 were upregulated in noise-exposed cochlea, but downregulated by systemic dexamethasone administration⁹. Anti-inflammatory effects through binding of the glucocorticoid receptor are among possible mechanisms of action⁸; however, effects through binding of the mineralocorticoid receptor as well as effects unrelated to inflammation may also be significant⁷. Glucocorticoids have a strong binding affinity for the mineralocorticoid receptor and may exert effects on ion and water homeostasis within the inner ear^2,10. Another study found that ion homeostasis genes were upregulated after IT administration of glucocorticoids compared to control, while inflammatory cytokines were upregulated in both control and treatment groups, suggesting that effects through the mineralocorticoid receptor maybe predominant¹¹.

At the proteomic level, heat shock protein 70 and myelin protein zero in mice cochleae were found to be modulated by the administration of dexamethasone¹². However, this proteomic study required dissecting the whole cochlea from animals for an adequate tissue sample.

Our laboratory has developed novel microneedle technology for inner ear diagnosis and therapy based on the mechanical properties of the round window membrane (RWM)^13–22. An additive manufacturing technique, two-photon polymerization lithography (2PP), was used to direct-write the microneedles via 3D printing. Solid microneedles fabricated using this technique introduced microperforations in vivo that healed within one week without functional consequences^13,23. Hollow microneedles (Figure 1) were then successfully used for atraumatic aspiration of perilymph across the RWM at a quality and volume adequate for proteomic analysis without causing lasting anatomic and functional dysfunction²⁴. These hollow microneedles could be used to sample perilymph of untreated guinea pigs and glucocorticoid-treated guinea pigs for comparison using liquid chromatography/mass spectrometry. In guinea pigs, perilymph aspirates may contain cerebrospinal fluid, but changes in protein expression should primarily reflect changes in the perilymph proteome. The purpose of this study is to determine the effects of IT and intraperitoneal (IP) dexamethasone on the perilymph proteome by utilizing 3D printed hollow microneedles to aspirate perilymph from dexamethasone-treated and untreated guinea pigs for proteomic analyses.

Figure 1. SEM image of hollow surgical microneedle designed for perilymph sampling at 134X magnification.

The hollow microneedle is mounted on a 30-gauge syringe needle. Outer diameter: 100 μm, Inner diameter: 35 μm. Lumen indicated by arrow.

2. MATERIALS AND METHODS

With the exception of dexamethasone treatment, the majority of these methods have been previously published²⁴, and are summarized here. Experiments were completed over the course of four months, alternating between control, IP dexamethasone (IP Dex)-treated, and IT dexamethasone (IT Dex)-treated guinea pigs to avoid freeze time bias of results. Animals were treated and sampled independently, and perilymph samples were stored at −80°C and processed for proteomics analysis upon collection of all samples.

2.1. Microneedle Design and Fabrication

SolidWorks software (Dassault Systems SolidWorks Corporation, Concord, NH, USA) was used for computer-aided design of microneedles. Stereolithography files were generated and parsed using the Describe software (Nanoscribe GmbH, Karlsruhe, Germany), with a slicing distance of 1 μm and laser intensity of 80%. Microneedles were fabricated using 2PP by Photonic Professional GT system (Nanoscribe GmbH) using photoresist IP-S (Nanoscribe GmbH). The inner diameter is set to 35 μm and the outer diameter is set to 100 μm. The lumen extends throughout the length of the needle shaft in the central axis of the microneedle and widens at the base of the needle to decrease fluidic resistance. At the tip of the needle, the lumen curves from the center of the shaft and opens at the side of the microneedle. Details of microneedle design have been previously reported²⁴.

2.2. Dexamethasone Treatment of Animals

All animal procedures described in this study were reviewed and approved by the Columbia University Institutional Animal Care and Use Committee. Twenty-two juvenile guinea pigs of either sex weighing between 150 and 300 grams were obtained from a commercial vendor (Charles River Laboratories, Inc., Wilmington, MA).

Eight animals were assigned to the control group and were not given any treatment prior to aspiration. Eight animals were assigned to the IP Dex group, and six animals were assigned to the IT Dex group. The audiometric and healing data of four guinea pigs (two control, two IP Dex) were not available for analysis because of equipment malfunction unrelated to aspiration; thus, two additional guinea pigs were assigned to the control and IP Dex groups. Dexamethasone sodium phosphate (10 mg/mL, MWI Veterinary Supply, Boise, ID) was administered to IP Dex guinea pigs 6 hours prior to the time of aspiration, by IP injection at a dose of 10 mg/kg¹². It was demonstrated that both IT and IP dexamethasone administration resulted in dexamethasone labelling of the cochlea at 6 hours²⁵, and the time point of 6 hours has previously been chosen for dexamethasone administration experiements¹². For IT Dex guinea pigs, all IT injections were applied to the right ear 6 hours prior to the time of aspiration. Guinea pigs were anesthetized with isoflurane gas (3.0% for induction and 1.0–3.0% for maintenance). A 27-gauge needle was used to first create a ventilation hole in the anterior-superior quadrant of the tympanic membrane. It was then used to inject 50 μL of the 10 mg/ml dexamethasone sodium phosphate solution (equating to 0.5 mg) through the posterior-superior quadrant of the tympanic membrane into the middle ear space. The guinea pig remained under isoflurane anesthesia in a left lateral decubitus position for 30 min. The guinea pig was then allowed to recover in its cage until the aspiration procedure.

2.3. Aspiration procedure

All guinea pigs underwent the aspiration procedure detailed below. For the IP Dex and IT Dex animals, aspiration was performed 6 hours after dexamethasone treatment. All procedures were performed on the right ear. Guinea pigs were anesthetized with isoflurane gas (3.0% for induction and 1.0–3.0% for maintenance). Lidocaine was injected subcutaneously for local anesthesia. Buprenorphine sustained release (0.1 mg/kg) and meloxicam (0.5 mg/kg) were both administered subcutaneously prior to surgery for post-operative analgesia. To reduce head motion from breath cycles, head fixation was achieved using a modular 3D-printed head holder with two pointed screws placed anterior to the external auditory meatus and posterior to the orbit without piercing the skin²⁶.

A scalpel was used to create a 1 cm postauricular incision. Blunt dissection was used to expose the bulla. A 1 mm drill tip fixed onto a Stryker S2 πDrive drill (Stryker, Kalamazoo, MI) was used to create a small opening into the middle ear space. Additional bone was removed using forceps to enlarge the opening to 2–3 mm in diameter for optimal visualization of the RWM. A hollow microneedle, mounted onto the tip of a 2-inch long, 30-gauge, blunt small hub removable needle (Hamilton Company, Reno, NV), was secured onto a 10 μL Gastight Hamilton syringe (Model 1701 RN, Hamilton Company, Reno, NV). The syringe was mounted onto a UMP3 UltraMicroPump (World Precise Instruments, Sarasota, FL), and the pump was fixed to a micromanipulator. Using the micromanipulator, the mounted hollow microneedle was slowly advanced towards the RWM and the perforation was confirmed by visualization.

Using the UltraMicroPump, 1 μL of perilymph was aspirated across the RWM over 45 seconds. The sample was ejected into a 0.5 mL LoBind Microcentrifuge tube (Eppendorf, Hamburg, Germany) containing 2 μL of 1% protease cocktail inhibitor solution (P8340, Sigma Aldrich, St. Louis, Missouri) in liquid chromatography/mass spectrometry grade water (Optima, ThermoFisher Scientific, Fair Lawn, NJ) and stored at −80°C. 72 hours after aspiration, animals were sacrificed with pentobarbital overdose.

2.4. Audiometric Testing

Audiometric testing was performed under anesthesia using methods previously described^23,27,28. Compound action potential (CAP) was used to evaluate the animal’s hearing before RWM perforation and also immediately prior to euthanasia at 72 h after completion of the procedure. Distortion product otoacoustic emissions (DPOAE) were used to evaluate hearing before and after RWM perforation and immediately prior to animal sacrifice at 72 h after the procedure.

CAP measures the activity of the auditory nerve by recording the synchronous firing of the sum of each individual unit action potential near the RWM after sound stimulation. Tone pips were played into the ear and the neural responses were measured by a silver electrode placed at the base of the cochlea. CAP responses were measured for 18 frequencies ranging from 0.5 to 40 kHz. Stimulus intensity was steadily increased in 5 dB increments to determine a hearing threshold. The lowest stimulus level that induced a recognizable response curve was used to determine the threshold. To minimize bias, experimenters were blinded to previous threshold measurements. CAP threshold shifts were considered significantly different than zero at p<0.05 using two-tailed paired t-tests.

DPOAE are responses produced by the cochlea upon simultaneous stimulation by two pure tone frequencies. They are used to measure the health of outer hair cells and to evaluate potential levels of hearing loss. To perform DPOAE measurements, an ear tube containing a speaker and a low-noise Sokolich ultrasonic probe microphone was placed into the ear canal. The speaker played sound stimuli at sound pressure levels of 70 dB SPL with a fixed frequency ratio of f2/f1 = 1.2 at 1 kHz increments between 1 kHz and 32 kHz. The microphone detected resulting distortion products from the ear. A DPOAE at 2f1 – f2 that is 3 dB above the noise floor level was identified as a positive response.

2.5. Proteomics Analysis

All perilymph samples (1 μL) were stored at −80°C and then processed as described previously²⁴. Briefly, samples were purified with a methanol chloroform protein precipitation as previously described²⁹. Proteins were resuspended in 8M urea, 3 mM DTT, 100 mM ammonium bicarbonate in liquid chromatography/mass spectrometry grade water, reduced with dithiothreitol and alkylated with iodoacetamide. For proteolytic digestion, samples were diluted five-fold in 100 mM ammonium bicarbonate and then digested using sequencing grade trypsin (Promega V511) at a protease:protein ratio of 1:50 at 37°C for 16 h as described previously³⁰. Samples were then desalted with Nest Group C18 Macrospin columns (Southborough, MA). Resulting peptides were analyzed by liquid chromatography/mass spectrometry as described previously³⁰. Peptide concentration was evaluated by NanoDrop spectrophotometry (ThermoFisher Scientific) at 205 nm and LC/MS inject loading amounts were adjusted (normalized) based on amino acid concentration. Specifically, separations were performed with an Ultimate 3000 RSLCnano liquid chromatograph with a 75 μm ID × 50 cm Acclaim PepMap C18 column (Thermo Scientific P/N 164942) coupled to a Q Exactive HF mass spectrometer (Thermo Scientific) in positive ion mode using data-dependent acquisition. Acquisition settings included top 15 precursors, resolution 120,000 for MS scan, 15 000 for MS/MS scan, dynamic exclusion for 20 s and maximum injection time of 30 ms for MS and 100 ms for MS/MS. For fragmentation, NCE was set at 28.0. The Nanospray Flex Ion Source was operated at 2.2 kV with heated capillary set at 250 °C and the S-Lens RF level at 55.0 %.

Raw data files were searched with MaxQuant Version 1.6.10.43 with Andromeda search engine with FDR 0.01 for PSM and Protein. Fixed modification was carboxyamidomethylation of cysteine and variable modifications were Oxidation (M) and Acetyl (Protein N-term). The database was UniProt release 2019_10, published November 13, 2019 with 25,731 sequences 14,311,265 residues of reviewed and unreviewed Cavia porcellus sequences and isoforms. Cavia porcellus sequences were from reference proteome # up000005447 and included porcine trypsin, human keratins and lab contaminants. Search parameters included MS tolerance of 20 ppm MS/MS tolerance of 20 ppm. Statistical post processing was with Perseus Version 1.6.10.50. Label-free quantitation (LFQ) in MaxQuant was used. Normalization of data across runs is performed by the LFQ algorithm.^31,32 Statistical analysis of LFQ data was performed in Perseus³³ with ANOVA and Tukey’s honestly significant difference post –hoc test.

All mass spectrometry raw data files generated in this work have been deposited in an international public repository (MassIVE proteomics repository at https://massive.ucsd.edu/) under dataset # MSV000086887.

3. RESULTS

3.1. Characterizing the guinea pig perilymph proteome in treated and untreated guinea pigs

Perilymph from 22 guinea pigs were included for proteomic analysis. Identifications were returned for 632 proteins with a 1% false discovery rate for both peptide sequence matches and protein identifications. Of the 632 proteins, 414 proteins represented by a single peptide or with insufficient data (<3 data points per treatment group) were deleted from the final analysis. An additional 43 proteins were contaminants or added proteins. 175 proteins represented by two or more peptides were included in the analysis (Table S1).

Data were analyzed with the PANTHER ontology tool (http://www.pantherdb.org/) using gene names (Table S1) searched against a Mus musculus background³⁴. Figure 2 shows the number of proteins per functional category and the distribution of these functional categories, accounting for protein abundances. Out of the 175 proteins included in the analysis, 146 proteins were mapped onto a functional category in PANTHER. Among these 146 proteins, the most common functional categories are protease inhibitors, protein-binding activity modulators, extracellular matrix proteins, and peroxidases.

Figure 2. Composition of guinea pig perilymph proteome, by functional categories.

The 175 gene names were searched against the mouse gene list in PANTHER (http://www/pantherdb.org) to determine the distribution of proteins across functional classes. 146 proteins mapped to a functional class. A: The protein class fold-enrichment is displayed. B: The protein class −log10 FDR enrichment is displayed.

At the individual level, preproalbumin, globin A1, hemoglobin subunit alpha, transthyretin, alpha-2-HS-glycoprotein, and Serine proteinase inhibitor A3K were the most abundant proteins in all groups (Table S1). The well-known inner-ear protein cochlin (A0A2C9F1F1_CAVPO) was identified based on 10 unique peptides, with 26% sequence coverage, was also a high abundance protein.

3.2. Modulated proteins of each treatment group

Overall, 15 proteins were modulated by either IP Dex or IT Dex compared to control. Compared to control, 14 proteins were modulated in the IP Dex group; 8 were upregulated, and 6 were downregulated (Table 1). Only three proteins were modulated in the IT Dex group; one was upregulated and two were downregulated (Table 2). VGF nerve growth factor inducible (VGF) was significantly upregulated compared to control in both IP (abundance ratio 3.20, p=0.0016) and IT (abundance ratio 2.23, p=0.007) Dex groups. Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein gamma was significantly downregulated in both treatment groups (IP: abundance ratio 0.42, p=0.0081; IT: abundance ratio 0.41, p=0.03).

Table 1:

Proteins modulated with systemically administered dexamethasone

Uniprot ID	Protein	Gene name	Abundance Ratio (IP Dex/Control)	P-Value (IP Dex vs. Control)
H0VZA9_CAVPO	VGF nerve growth factor inducible	VGF	3.20	1.6E-03
H0V8E0_CAVPO	Amyloid beta precursor like protein 1	APLP1	2.44	6.9E-03
A0A286XP66_CAVPO	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein gamma	YWHAG	0.42	8.1E-03
H0W9I8_CAVPO	Elongation factor 1-alpha		0.38	0.01
H0W3T8_CAVPO	Fructose-bisphosphate aldolase	ALDOC	0.26	0.02
H0UW89_CAVPO	SPARC like 1	SPARCL1	3.82	0.03
H0VLM3_CAVPO	Heat shock protein family A (Hsp80) member 5	Hspa5	0.67	0.03
H0V7P8_CAVPO	Annexin	ANXA2	0.59	0.04
H0UY41_CAVPO	Complement C5	C5	3.56	0.04
H0V7V7_CAVPO	Dickkopf WNT signaling pathway inhibitor 3	DKK3	1.97	0.04
H0VN20_CAVPO	Enolase 3	ENO3	0.09	0.04
H0V7Q6_CAVPO	Neuronal cell adhesion molecule	NRCAM	3.34	0.04
H0V077_CAVPO	SMB domain-containing protein	Vtn	1.57	0.05
H0UV55_CAVPO	Neural EGFL like 2	NELL2	1.78	0.05

Table 2:

Proteins modulated with intratympanically administered dexamethasone

Uniprot ID	Protein	Gene name	Abundance Ratio (IT Dex/Control)	P-Value (IT Dex vs. Control)
H0VZA9_CAVPO	VGF nerve growth factor inducible	VGF	2.23	7.0E-03
A0A286XP66_CAVPO	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein gamma	YWHAG	0.41	0.03
H0W263_CAVPO	Insulin like growth factor binding protein 2	IGFBP2	0.60	0.03

3.3. Audiometric testing and RWM healing

Audiometric testing and RWM healing data have been previously reported for untreated guinea pigs²⁴. Aspiration resulted in no permanent damage to hearing at 72 h and all perforations healed completely within 72 h. Of the 18 guinea pigs (n=6 per treatment group) that underwent all audiometric measurements successfully, there was no permanent damage to hearing at 72 h after perforation as assessed by CAP and DPOAE (Figure 3), and all perforations healed completely within 72 h. Results were similar between control, IP Dex, and IT Dex groups.

Figure 3. Compound Action Potential (CAP) and Distortion product otoacoustic emissions (DPOAE).

A: Mean CAP threshold shifts at 72 h compared to baseline, for all tested frequencies. The shaded area is bounded by the upper and lower limits of the 95% confidence interval. There was no significant shift in CAP measured at 72 h after perforation, for all 18 frequencies (n=18). B and C: DPOAE at the frequency 2f1 – f2 in response to a 70 dB stimulus averaged over all survival experiments at 0 h after perforation (A, n=18) and 72 h after perforation (B, n=18). Solid gray lines show the average noise for the experiments; Solid blue lines show the mean measured baseline DPOAE signal; and dotted red lines show mean DPOAE signal at 0 h (B) or 72 h (C) post-perforation. Shaded areas are bounded by the upper and lower limits of the 95% confidence interval for the average noise, and mean DPOAE signals. All DPOAE signals remained out of noise.

4. DISCUSSION

In this study, using novel hollow microneedle technology, we show that route of dexamethasone administration greatly impacts changes seen in perilymph proteome. The hollow surgical microneedles facilitated the aspiration of 1 μL of perilymph from the scala tympani in a series of survival experiments on guinea pigs; all RWMs healed within 72 hours and aspiration did not cause lasting physiological dysfunction, further confirming previously published results in untreated guinea pigs²⁴. Among these 175 proteins represented by two or more peptides and detected in at least 3 samples per treatment group, 14 proteins were modulated by systemic dexamethasone and 3 proteins were modulated by IT dexamethasone. In both treatment groups, VGF was found to be upregulated and tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein gamma (also known as 14-3-3 gamma) was found to be downregulated.

The VGF polypeptide is the precursor of several biologically active peptides with expression restricted to a subset of neurons in the central and peripheral nervous system and specific populations of endocrine cells³⁵. VGF and its peptides are found in large dense core vesicles and released from neuronal and endocrine cells through the secretory pathway. VGF peptides include NERP-1 and NERP-2, NAPP129, TPGH, TLQP-21, TLQP-62, HHPD-41, AQEE-11, AQEE-30, and LQEQ-10, and have been shown to regulate neuronal activity such as synaptic plasticity, neurogenesis and neurite growth³⁶. VGF also exhibits neuroendocrine effects through two of its peptides, NERP-1 and NERP-2, which modulate antidiuretic hormone release and may be important for fluid balance³⁷. VGF expression levels are very low under normal physiological conditions but are rapidly upregulated in various situations, such as nerve injury³⁶.

VGF is induced by neurotrophic factors, including brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT3)^38,39. BDNF and NT3 have both been shown to exert otoprotective effects through preservation of spiral ganglion neuron and regeneration of cochlear synapses^40–45. It has also been suggested that VGF may have roles in neuroprotection. In an in vitro model of amyotrophic lateral sclerosis, VGF was shown to be a mediator of the protective effects of the free radical scavenger SUN N8075 on ER stress-induced cell death⁴⁶. In another study, VGF was shown to be induced by optic nerve crush⁴⁷. The VGF peptide, AQEE-30, suppressed the loss of retinal ganglion cells in mice with optic nerve crush and promoted outgrowth of neurites of rat- and human induced pluripotent stem cells-derived retinal ganglion cells in vitro. In our study, we found that dexamethasone, a glucocorticoid commonly used to treat sudden sensorineural hearing loss, when systemically or intratympanically administered, induced protein expression of VGF. VGF may be a mechanism for the protective effects of dexamethasone, but further studies involving animal models for hearing loss, including VGF knock-outs and over-producers, are needed.

Interestingly, 14-3-3 gamma was downregulated after both IP and IT dexamethasone administration. Similarly in mice, 14-3-3 has been shown to increase in the meninges and CSF of mice with eosinophilic meningitis after infection, and was downregulated upon dexamethasone administration⁴⁸. It has also been found that 14-3-3 gamma upregulates glucocorticoid receptor in the liver⁴⁹. Our findings show that treatment with dexamethasone, a glucocorticoid, 14-3-3 levels in the perilymph are downregulated. Taken together, these results suggest a negative feedback loop between the 14-3-3 gamma and glucocorticoids whereby 14-3-3 upregulates the glucocorticoid receptor, and is in turn downregulated when glucocorticoid levels are high. However, it is unknown if 14-3-3 gamma also upregulates the glucocorticoid receptor in the inner ear. Further studies of 14-3-3 gamma in the inner ear are needed to elucidate its possible role as a mediator of the protective effect of glucocorticoids on hearing.

The remaining proteins modulated by either systemic dexamethasone or intratympanic dexamethasone can be broadly divided into three general categories. Six of these proteins, all upregulated following IP Dex, have roles in neurogenesis, neural development, or neuroprotection: APLP1, NRCAM, NELL2, DKK3, VTN, AND SPARCL1. Amyloid beta precursor like protein 1 (APLP1), part of the amyloid precursor protein family essential to central nervous system development and exclusively expressed in neurons⁵⁰, has roles in proper synapse formation and maintenance⁵¹. Neuronal cell adhesion molecule (NRCAM) is expressed by sensory cells in the cochlea and is necessary for proper cochlear innervation during development⁵². Neural EGFL-like 2 (NELL2) has roles in neural cell growth and differentiation⁵³. Dickkopf WNT signaling pathway inhibitor 3 (DKK3) is anti-inflammatory and neuroprotective in intracranial hemorrhage⁵⁴. Vitronectin (VTN) supports cell attachment and promotes neurite extension and has been shown to be upregulated in optic nerve injury⁵⁵. SPARCL1 regulates cell migration, proliferation, and differentiation, and is upregulated in brain injury⁵⁶.

Heat shock protein, a protein involved in the inflammatory process, was downregulated in guinea pigs treated with systemic dexamethasone. Previously, consistent with our finding, heat shock protein family A (Hsp80) member 5 (Hspa5), was shown to be downregulated by dexamethasone in the mouse cochlea of a mouse model of hearing loss¹². Annexin A2 (ANXA2) was also downregulated in guinea pigs treated by IP Dex in the present study. ANXA2 is pro-inflammatory protein that activates inflammatory cytokines⁵⁷, in contrast to its counterpart Annexin A1, which has anti-inflammatory roles⁵⁸. In addition, Complement C5, the fifth component of complement with an important role in inflammation and cell killing processes, was upregulated by systemic dexamethasone in this study.

Finally, proteins involved in protein synthesis (Elongation factor 1-alpha), glucose metabolism (fructose-bisphosphate aldolase and enolase 3), and adipogenesis (Insulin like growth factor binding protein 2)⁵⁹, were all downregulated by dexamethasone in this study.

In comparing the effects of systemically delivered and IT administered dexamethasone, 14 proteins were modulated by IP Dex, while only three proteins were modulated by IT Dex. This suggests that IT Dex has narrower effects on the perilymph while IP Dex imparts broader results due to its systemic delivery route. There are few studies for comparison in this largely unexplored area; additionally, prior studies used different methodology for assessing changes (e.g. inner ear tissues rather than perilymph), thus making direct comparison difficult. For example, Trune et al., showed that intratympanic dexamethasone administration changed the expression of many more inner ear genes than systemic dexamethasone injection based on mRNA expression in tissues⁷; protein expression was not studied. Similar to our study, they also used a 6-hour time point following dexamethasone administration; however, while they used whole cochlear tissue for analysis, we utilized perilymph for proteomic analysis. The authors pointed out that theirs was the first to examine the full inner ear genome impact of steroids, and that 75% of the genes affected, while statistically significant, were upregulated less than 2-fold or downregulated by less than 50%. The authors suggest that the actual change in expression was often minor and perhaps even unimportant in the overall function of the inner ear. No gene list was included in that paper so a detailed comparison to this work is not possible.

IT delivery of dexamethasone is limited by the inconsistent diffusion of drug across the RWM and inadvertent clearance of medication via the eustachian tube^60,61. Given these limitations, it is unclear whether the results of the current study suggest robustness of systemic delivery over local delivery, or the vagaries of IT delivery. The rate of diffusion of medication delivered by IT injection may be improved by introducing perforations into the RWM using solid microneedles^18,62. Alternatively, direct intracochlear injection of medication may be considered to reduce the variability associated with RWM diffusion and eustachian tube leakage.

Furthermore, changes to protein expression may be time-dependent; thus, proteomic analyses of perilymph conducted over time may yield valuable information and overcome the limitation of single time point study. In principle, our sampling method permits multiple perilymph aspiration from the same animal over time as perforations heal and cause no lasting damage to hearing. Future studies should focus on direct delivery across RWM and repeated sampling of perilymph at various time points to determine the effect of steroids over time.

5. CONCLUSION

In this study, dexamethasone was found to upregulate proteins involved in neural development, neuronal regulation, and neuroprotection; modulate proteins involved in inflammatory processes; and downregulate proteins involved in protein synthesis, glucose metabolism, and adipogenesis. VGF was upregulated in the perilymph of guinea pigs treated with IT and systemic dexamethasone and is possibly a mechanism of protective effect of glucocorticoids on hearing loss. Hollow microneedles facilitated the aspiration of perilymph across the RWM at microliter volumes without causing permanent anatomic or physiologic dysfunction and enabled proteomic analysis. Future studies may investigate the effects of direct intracochlear injection of dexamethasone repeated over multiple time points.

Supplementary Material

Supplementary Material Table

Table S1. Proteins detected in guinea pig perilymph.