Regulator of G protein signaling 17 represents a novel target for treating cisplatin induced hearing loss

Regulators of G protein signaling (RGS) accelerate the GTPase activity of G proteins to enable rapid termination of the signals triggered by G protein-coupled receptors (GPCRs). Activation of several GPCRs, including cannabinoid receptor 2 (CB2R) and adenosine A1 receptor (A1AR), protects against noise and drug-induced ototoxicity. One such drug, cisplatin, an anticancer agent used to treat various solid tumors, produces permanent hearing loss in experimental animals and in a high percentage of cancer patients who undergo treatments. In this study we show that cisplatin induces the expression of the RGS17 gene and increases the levels of RGS17 protein which contributes to a significant proportion of the hearing loss. Knockdown of RGS17 suppressed cisplatin-induced hearing loss in male Wistar rats, while overexpression of RGS17 alone produced hearing loss in vivo. Furthermore, RGS17 and CB2R negatively regulate the expression of each other. These data suggest that RGS17 mediates cisplatin ototoxicity by uncoupling cytoprotective GPCRs from their normal G protein interactions, thereby mitigating the otoprotective contributions of endogenous ligands of these receptors. Thus, RGS17 represents a novel mediator of cisplatin ototoxicity and a potential therapeutic target for treating hearing loss.

www.nature.com/scientificreports/ signaling (RGS) promote the inactivation and termination of G protein mediated signaling by facilitating GTP hydrolysis 29 . Transcriptome data analysis of cochleae from control and cisplatin-treated rats performed in our laboratory demonstrate differential expression of RGS proteins. Among these, RGS17, a member of RGS-RZ subfamily, which commonly targets GTP bound Gα i1-3 , Gα o , Gα z and Gα q for hydrolysis 30 was highly induced. RGS17 has been extensively studied in the central nervous system for its role in inhibiting μ-opioid, dopamine and cannabinoid receptors. An interesting function of RGS17 is its role as a potential redox transducer. This is mediated by the interaction of cysteine-rich domain of RGS17 with nitric oxide to release zinc 31 . RGS17 positively influences the protein kinase A-cyclic AMP response element binding protein 32 pathway by inhibiting Gαi/o signaling to promote proliferation, migration and invasion in cancers.
In this study we test the novel hypothesis that RGS17 is a critical regulator of cochlear function and mediator of cisplatin ototoxicity and works by inhibition of the functions of otoprotective GPCRs. Moreover, we examined the possibility that RGS17 antagonizes the protective actions of the CB2R against cisplatin ototoxicity. Accordingly, inhibition of RGS17 could enhance the protection mediated by these GPCRs by extending the duration of the receptor-G protein interaction.

Results
RGS17 is expressed in the rat cochlea. RGS proteins are not widely studied in the auditory system. Among more than 20 RGS proteins, only RGS4 is expressed in SGNs 18 , supporting cells and OC of adult cochlea. Increased level of RGS4 is associated with neuronal defects in PTEN-deficient SGNs 18 . While this manuscript was in the process of being submitted, a published study provided convincing evidence implicating RGS4 in noise-induced hearing loss in rats 33 . Additionally, RGS18 is present in otic vesicles of zebrafish and mouse embryos where it is involved in development of cilia in hair cells of the inner ear 34 . Our study demonstrates the distribution of RGS17 in cochlea using the polyclonal antibody. Validation of this antibody was done in OCderived UB/OC-1 cells 35 transiently transfected with RGS17 overexpressing plasmid, which resulted in increased level of RGS17 protein expression (almost 300% increase) as detected via immunofluorescence and Western blotting (Supplementary Fig. S1A-C). In rodents, RGS17 is predominantly expressed in CNS, specifically in the cerebral cortex, nucleus accumbens and striatum 36,37 but not in normal liver tissue or hepatocytes 38 . Thus, we characterized the specificity of RGS17 antibody using cerebral cortex as a positive control and liver tissue as a negative control. We observed distinct immune-positive fluorescent labelling of RGS17 in the brain section whereas no significant staining was observed in liver sections ( Supplementary Fig. S1D). Furthermore, knockdown and overexpression studies described below show reductions and increases in RGS17 levels, respectively. Our observations confirm the specificity of the antibody we used to detect RGS17 protein is valid and thus we further used it to identify RGS17 in the cochlea.
The mid-modiolar sections of the cochlea from male Wistar rats showed distinct RGS17 immunoreactivity. Positive immunolabeling was observed in OC, SGNs and SV (Fig. 1A). In OC, both sensory and non-sensory cells-outer hair cells (OHCs), inner hair cells (IHCs), inner pillar cells (IPCs), outer pillar cells (OPCs) and Deiters cells (DCs) expressed RGS17 (Fig. 1B). These mid-modiolar sections were also co-labeled with antibodies against Na + /K + /ATPase and RGS17 to investigate the expression of the latter in the lateral wall of the cochlea. Na + /K + /ATPase was used as a marker for the lateral wall as it is expressed in strial cells and fibrocytes of spiral ligament 11,39 . Intense RGS17 immunolabelling was observed in the basal cells of SVA, compared to marginal cells, and the least fluorescent intensity was found in intermediate cells (Fig. 1C). In addition, co-staining of RGS17 with Tuj1, an antibody for class III β-tubulin that labels both type of SGNs, demonstrated the expression of RGS17 in Type I and Type II SGNs (Fig. 1D). Taken together, these data strongly support the presence and localization of RGS17 in the rat cochlea.
Cisplatin induces RGS17 expression. Our laboratory has previously shown that agonist activation of GPCRs such as A 1 AR and CB2R protect against cisplatin-induced hearing loss 12,14 . Furthermore, we showed that knock down of CB2R in the cochlea aggravated the cisplatin-induced ototoxicity, as evidenced by increased ABR threshold shifts 12 . These data suggest endogenous "cytoprotective" GPCRs like CB2R could protect against cisplatin ototoxicity following activation by endocannabinoids or by administration of exogenous agonists of this receptor. To determine whether cisplatin could inhibit this endogenous defense system, we examined whether this agent could induce the expression of RGS17 in the cochlea. In effect, induction of RGS17 is expected to antagonize GPCR (such as CB2R) -G protein interaction and function by accelerating hydrolysis of bound GTP to GDP. We tested this hypothesis in vitro in UB/OC-1 cells, an organ of Corti-derived clonal cell line, which were treated with cisplatin (20 μM) over different time points, 6 h, 12 h, 24 h, 36 h and 48 h. Cisplatin produced time-dependent increases in RGS17 protein level which peaked by 24 h (149% ± 20) and remained relatively stable at 36 and 48 h post treatment ( Fig. 2A-C). Similar observations were reported using RT q-PCR and immunofluorescence (data not shown). For in vivo studies, we treated male Wistar rats with cisplatin for 24 h, 48 h and 72 h. Cochleae collected from treated animals along with control (PBS-treated) animals were collected for either RNA extraction or fixed for cryo-sectioning. Cisplatin treated animals showed increased RGS17 immunoreactivity as observed in mid-modiolar sections. Immunostaining showed increasing fluorescence intensities from 24 h, 48 h and to 72 h (Fig. 2D). While the intensity of the RGS17 image in the vehicle looks weak, it was captured at a lower intensity (than that of Fig. 1) to better depict the relatively high RGS17 fluorescence in the cisplatin group. We have presented the vehicle/cisplatin comparison images at 72 h in Supplementary Fig. S2, which were captured at comparable increasing intensities. The fluorescence intensities of RGS17 immunolabeling were increased in all three regions (OC, SGs and lateral wall) of the cochlea after cisplatin treatment (Fig. 2E). RGS17 immunoreactivity was significantly higher in OC (4.9 ± 0.5, 11.5 ± 0.3 and 14.9 ± 1.7 intensity units) and the lateral wall, SVA + SL (9.0 ± 1.7, 16.1 ± 2.4 and 21.3 ± 1.6) post 24, 48 and 72 h cisplatin treatment, respectively. RGS17 fluorescence intensities was also significantly increased in SG: 8.9 ± 0.6, 13.0 ± 1.2 and 19.4 ± 3.4 at 24, 48, and 72 h respectively. Similarly, RT-qPCR analysis of RNA obtained from the cochleae showed cisplatin significantly increased RGS17 expression at 24, 48 and 72 h by 2.1 ± 0.4, 4.1 ± 1.1 and 4.6 ± 0.7 fold, respectively (Fig. 2F). These data suggest that the RGS17 gene is susceptible to regulation by cisplatin in the cochlea.

Functional role of RGS17 in hearing.
To elucidate the role of RGS17 in hearing, we overexpressed RGS17 into the cochlea by trans-tympanic injection of an adenoviral vector tagged with m-cherry (m-cherry-Adv-RG17) at a concentration of 4 × 10 8 plaque forming units/cochlea. The contralateral ears were administered the same concentration of adenoviral vector without RGS17 and served as the vehicle-controls. Auditory brain responses (ABRs) were assessed five days after the adenoviral vector treatment (Fig. 3A). Adenoviral infection produced a significant increase (~ threefold) in the expression of RGS17 in the cochlea by 5 days, especially in the SVA and SG ( Supplementary Fig. S3A,B). This produced significant elevations in ABR thresholds of 10.8 ± 3.1 dB (at 16 kHz) and 19.2 ± 3.5 dB (at 32 kHz) when compared to the vehicle control ears (Fig. 3B) but not at 8 kHz frequency.
To investigate the role of RGS17 in cisplatin-induced hearing loss model, we administered siRNA against RGS17 via trans-tympanic route (0.9 μg) into the middle ear of male Wistar rats, two days prior to cisplatin treatment. The control group received a scrambled nucleotide sequence. Cisplatin (11 mg/kg) or vehicle were then administered by intraperitoneal administration over a 30 min period. Pre-ABRs were performed before trans-tympanic administration and post-ABRs were recorded 3 days after cisplatin treatment (Fig. 3C). Efficient knockdown of RGS17 (70%) was observed by siRNA when compared to the scrambled sequence ( Supplementary   Figure 1. Distribution of RGS17 in the rat cochlea. Naïve male Wistar rats were used for the collection of cochleae. Collected cochleae were fixed with 4% PFA, cryo-frozen and processed for mid-modiolar sectioning using cryostats. Mid-modiolar section was immunolabeled with RGS17 (green) and DAPI (blue). (A) Significant immunolabelling of RGS17 observed in the rat cochlea. (B) Expression of RGS17 was observed in OHCs, IHC, supporting cells: DCs, IPC and OPC. (C) Co-immunolabelling with Na + K + ATPase (magenta) shows expression of RGS17 in the lateral wall of the cochlea. Positive immunofluorescence was seen in basal, marginal, intermediate cells and SL. (D) Co-staining with Tuj1 (red) shows RGS17 expression in cochlear neuronal cells, SGNs (Scale 50 μm).  Fig. S3D).
RGS17 is implicated in synaptopathy in cochlea. We next explored whether RGS17 is also implicated in cochlear synaptopathy. Overexpression of RGS17 in the cochlea led to significant reductions in wave 1 amplitudes observed at 80 dB by 58.7% (from 0.63 ± 0.12 to 0.26 ± 0.04 µV) and by 70.0% (from 0.98 ± 0.13 to 0.59 ± 0.08 µV) at 90 dB. No significant changes were observed at 60 and 70 dB intensities. These values were reduced by 63.5% (0.16 ± 0.03 and 0.06 ± 0.02 µV) at 60 dB and 52.3% (0.29 ± 0.05 and 0.14 ± 0.04 µV) at 70 dB by RGS17 overexpression (Fig. 4A). Similar to the RGS17 vector, cisplatin significantly reduced wave 1 amplitudes, while trans-tympanic administration of RGS17 siRNA, which did not affect wave 1 amplitudes by itself, reduced the magnitude of the cisplatin's effects at 80 and 90 dB (Fig. 4B) Peak latency represents the time interval between the onset of acoustic stimuli and the initiation/occurrence of peak of the corresponding wave. Wave I latency signifies the time required for neural impulses to conduct the information from IHC to cochlear nerve. We found that cisplatin significantly delayed the response latencies compared to control animals across a broad range of sound intensities (40-80 dB): 4.56 ± 0.23 ms, 4.68 ± 0.18 ms, 4.17 ± 0.13 ms, 3.83 ± 0.11 ms, 3.45 ± 0.08 ms, respectively, whereas siRGS17 pretreatment significantly shortened www.nature.com/scientificreports/ the peak latencies at 40, 50 and 60 dB by at least 14%: 3.83 ± 0.08 ms, 3.89 ± 0.17 ms, 3.59 ± 0.07 ms respectively (Fig. 4C). Functional characteristics of wave I go in parallel with the integrity of synapse at the junction of IHC and afferent nerves from SGN I neurons 40,41 . Previously, we have shown that cisplatin decrease cochlear ribbon synapses, whereas activation of CB2 receptors prevented the loss of synapses to promote normal hearing 12 . We quantified the number of C-terminal binding protein 2 (CtBP2)-immuno-positive puncta as an indicator for ribbon synapses at IHC and post synaptic glutamate receptor, GluR2, -immuno-positive puncta as a post synaptic marker at the afferent nerve. The functional synapse is reported as a puncta/dot co-stained with both the markers. Cisplatin treatment reduced the number of functional ribbon synapses (as marked by decreased number of bright white puncta shown by the combination of three colors: red (CtBP2), green (GluR2) and blue in region of IHC (Fig. 4D). Rats pretreated with RGS17 siRNA maintained the same number of functional synapses (Fig. 4D,E). Cisplatin significantly reduced the average number of functional ribbon synapses to 10.7 ± 0.3, as compared to 19.7 ± 0.20 for control. However, siRGS17 significantly attenuated this loss of synapses (observed with cisplatin) to 15.7 ± 0.6 ( Fig. 4E). The number of functional synapse were counted manually and results were plotted in the graph. A regression plot analysis of wave I amplitude versus ribbon synapse counts showed that these two variables are positively related across the different treatment groups with correlation coefficient, R2 = 0.94 and slope (m) = 1.01 (Fig. 4F). This strong correlation indicates that the reduction in wave 1 amplitude and functional synapses with cisplatin treatment and the protection afforded by RGS17 knockdown. Our data suggests RGS17 could be a potential therapeutic target for alleviating synaptopathy associated with cisplatin ototoxicity.

Knockdown of RGS17 protects against cisplatin-induced apoptosis in cochlea.
Cisplatin is known to produce apoptosis primarily via ROS generation or by directly intercalating into DNA in cochlea cells 3,4 . Various studies have shown TUNEL positive cells in OHCs, SVA and SGNs 11,42,43 which were more pronounced in basal turn of the cochlea 11,12 . To test if RGS17 is involved in OHC loss, the basal turns from wholemount preparations of cochleae collected from previous treatment groups (described in Fig. 3) were stained for myosin VIIa (magenta) and loss in OHCs was indicated as white dots (Fig. 5A). Fluorescence imaging of stained  www.nature.com/scientificreports/ Overexpression of RGS17 disrupts STAT1/STAT3 balance. STAT1 and STAT3 show differential effects on cisplatin-induced apoptosis. A non-selective inhibitor of STAT1, epigallocatechin gallate (EGCG), protected against cisplatin-induced cell death 11 . Both EGCG and capsaicin inhibit the pro-inflammatory pathway and promote cell survival by increasing the STAT3/STAT1 ratio 11,44 . Since, knockdown of RGS17 ameliorated cisplatin-induced apoptosis in cochlea (see above), we investigated its role in affecting the STAT1/STAT3 balance. We transiently over-expressed RGS17 in UB/OC-1 cells using a plasmid vector. Western blots performed after the transfection showed an increase activation of p-STAT1 but decreased p-STAT3 activation as compared to the empty vector transfected cells (Fig. 6A). The effects of interferon gamma (IFN-γ) and interleukin-6 (IL-6), positive controls for STAT1 and STAT3 activation, respectively, were tested. In UB/OC-1 cultures, RGS17 significantly increased the ratio of p-STAT1/p-STAT3 to 1.6 ± 0.1-fold, compared to empty vector control (Fig. 6B).
In vivo study indicate increased mRNA levels of STAT1 and reduced levels of STAT3 in the cochlea infected with RGS17 adenoviral vector compared to empty vector controls. Cochleae collected from rats overexpressing RGS17 (refer to Fig. 3B) demonstrated a significantly higher ratio of STAT1/STAT3 (4.6 ± 0.7-fold), compared to control cochleae (Fig. 6C).Thus, overexpression of RGS17 tilted the ratio towards greater dominance of STAT1 over STAT3 signaling in the cultured cells and the cochlea. We propose that RGS17 could therefore alter the cellular environment of the cochlea (via STAT1/STAT3 ratio), which renders it more susceptible to hearing loss. Interestingly, ChIP-seq data indicate that STAT1 binding sites are present in the RGS17 promoter 45 . Furthermore, data from our lab indicate that inhibition of STAT1 reduced the expression of RGS17 (unpublished data).

Role of RGS17 in redox signaling in the cochlea.
To understand the basis of STAT1 activation and induction by RGS17, we focused on the role of ROS. Cisplatin alters the redox balance in cochlea by increasing the production of ROS which causes cochlear cell damage and apoptosis 6,46 . We hypothesized that induction of RGS17 potentially increased oxidative stress, tilting redox balance to cell apoptosis. Transiently transfection of www.nature.com/scientificreports/ UB/OC1 cells with plasmid encoding RGS17 increased ROS generation, assessed via CellROX assays, as compared to empty vector transfected cells (Fig. 7A). Previous studies have linked cisplatin-induced upregulation of iNOS, NOX3 and KIM-1 to cause oxidative damage to the cochlea [47][48][49] . Similarly, we observed increases in these genes in the cochlea following infection with adenoviral vector expressing RGS17. In these studies, overexpression of RGS17 led to increased expression of mRNA level of NOX3, iNOS and KIM1 to 2.5 ± 0.4, 2.4 ± 0.2 and 2.4 ± 0.2 fold, respectively (Fig. 7B). In addition, overexpression of RGS17 in the cochlea significantly increased the levels of inflammatory genes, such as COX2 and TNF-α (see Supplementary Fig. S4A). In additional studies, we administered a scrambled oligonucleotide sequence or siRGS17 into the cochlea by trans-tympanic injections, waited 2 days and then administered intraperitoneal cisplatin (11 mg/kg). Cisplatin increased the levels of NOX3, iNOS and KIM1 by 2.44 ± 0.26, 2.23 ± 0.30 and 2.33 ± 0.29 folds, respectively, while knock down of RGS17 (by siRNA) reduced these changes to 0.93 ± 0.10, 0.97 ± 0.13 and 1.28 ± 0.14 folds, respectively (Fig. 7C). Moreover, knock down of RGS17 blocked the reductions in cytoprotective genes such as Nrf2 and SOD2 by cisplatin and even promoted the expression of these antioxidant genes to 1.77 ± 0.23 and 1.86 ± 0.14 folds, respectively (Fig. 7D). Knockdown of RGS17 in the cochlea also significantly reduced cisplatin-induced increase of inflammatory genes, such as COX2 and TNF-α (see Supplementary Fig. S4B). These data suggest that RGS17 serves as a mediator of cisplatin-induced activation of the inflammatory cascade in the cochlea which contributes to hearing loss.
Reciprocal relation between cannabinoid receptor 2 (CB2) and RGS17. A previous study has characterized the role of RGS17 in the central nervous system. They found that cannabinoid receptor 1 interacts with RGS17-HINT complex and this modulatory interaction protected against NMDAR-induced excitotoxicity 50 . A previous study from our lab has indicated the importance of another subtype of cannabinoid receptor, cannabinoid receptor 2 (CB2), against cisplatin ototoxicity 12 . We therefore investigated potential interaction between RGS17 and CB2 in the cochlea. In this regard, we have previously shown that JWH-015, a CB2 agonist, protected against cisplatin-induced hearing loss 12 . In the present study, we show that cisplatin increased the expression of RGS17 in the cochlea which was abolished by pretreatment of trans-tympanic JWH-015 prior to cisplatin administration (Fig. 8A). qPCR analysis showed a 4.5 ± 1.0-fold increase in RGS17 with cisplatin, while pretreatment with JWH-015 significantly reduced the fold change to 1.1 ± 0.2. JWH-015 did not produce any change when administered alone but reduced the increase in RGS17 immunoreactivity in cochleae from rats exposed to cisplatin (Fig. 8B). This highlights the role of activated CB2R in suppressing RGS17 expression, which normally reduces the duration of GPCR-GTP interaction. In addition, in vitro studies showed that transient over www.nature.com/scientificreports/ expression of RGS17 in UB/OC-1 cells decreased cell viability to 61.2 ± 3.1%, which was partly reversed by treatment with JWH-015 (cell viability increased to 78.8 ± 3.4) as compared to control vector transfected cells (data not shown). Western blot analysis for G proteins, Gαz and Gαi/o, showed decreases in the levels of Gαz and Gαi/o to 80 ± 10.3% and 86 ± 7.1%, respectively, after cisplatin treatment, whereas JWH-015 treatment increased the levels of Gαz and Gαi/o proteins to 117.3 ± 9.3% and 108.1 ± 6.3% respectively (see Supplementary Fig. S5A-D). These data suggest a close association between the regulation of CB2R, G proteins and RGS17 in regulating cisplatin toxicity in the cochlea. We next examined the effect of knockdown of RGS17 on CB2 expression. Trans-tympanic injection of siRGS17 before cisplatin treatment significantly elevated CB2 mRNA level in the cochleae to 2.04 ± 0.20-fold when compared to cisplatin exposed cochleae (0.59 ± 0.10-fold change) (Fig. 8C). Interestingly, knockdown of RGS17 by itself in the cochlea significantly increased CB2 expression to 2.07 ± 0.50-fold. In addition, the mRNA level of GNAZ (Gαz gene) increased to 2.5 ± 0.2-fold compared to cisplatin treatment (0.4 ± 0.09-fold change), following siRGS17 pretreatment. Both knockdown of CB2R 12 and overexpression of RGS17 (current data) has been found to exacerbate hearing loss. In this study, we found knockdown of CB2 with siRNA significantly increased RGS17 mRNA level by 2.4 ± 0.4-fold (Fig. 8D) and infection of cochleae with adenoviral vector over expressing RGS17 led to significant reductions in CB2 expression to 0.2 ± 0.02 fold (Fig. 8E). This shows that the same ototoxic effect observed previously with knockdown of CB2 and overexpression of RGS17 may be attributed to induction of RGS17. These latter findings suggest a reciprocal regulation between RGS17 and CB2, disruption of which might produce hearing loss.

Discussion
This study provides evidence implicating RGS17 in cisplatin-induced hearing loss. Evidence to support this conclusion include the demonstration that overexpression of RGS17 in the cochlea increases cochlear synaptopathy and hearing loss. Cisplatin induced the expression of RGS17 in the cochlea, which increased oxidative stress, the expression of inflammatory genes and promoted apoptosis of cochlear cells. In contrast, inhibition of RGS17 abrogated cisplatin-mediated oxidative stress and pro-inflammatory genes, reduced cochlear cell apoptosis and preserved synaptic integrity at IHCs. RGS17 and CB2 show reciprocal antagonistic interactions which could influence the otoprotective actions of endocannabinoids and exogenously administered cannabinoids in the cochlea. The transcription factors, STAT1 and STAT3, play a critical role in the reciprocal regulation of RGS17 and CB2 in the cochlea by cisplatin. These data show that RGS17 is an important determinant of cisplatin www.nature.com/scientificreports/ ototoxicity by uncoupling of cochlear GPCRs (such as CB2R) which promotes oxidative stress, inflammation and apoptosis in the cochlea. Induction of RGS17 by cisplatin progressed over a 72 h period and was evident in the different regions of the cochlea known to be affected by cisplatin. Knockdown of RGS17 abrogated cisplatin-induced hearing loss, implicating this protein in hearing loss. Furthermore, overexpression of RGS17 in the cochlea of naïve rats, using a viral vector, produced hearing loss. The pattern of high frequency hearing loss is somewhat less than that observed with cisplatin and noise exposures in our laboratory 51 . RGS17 mimicked cisplatin by increasing ABR thresholds and producing cochlear synaptopathy. We show that these changes stem from the ability of RGS17 to promote oxidative stress and inflammation, presumably through inactivation of otoprotective CB2R, which is coupled to the Gαi and Gz proteins. We previously observed that CB2R in the cochlea is tonically active and that inhibition of this receptor produced hearing loss 12 . Thus, overexpression of RGS17 is expected to antagonize the otoprotective actions of CB2R 12 by rapidly terminating G protein activation mediated by this receptor subtype. Oxidative stress is known to increase lipid peroxidation of membranes, increase inflammation, increase DNA damage and apoptosis, which could contribute to hearing loss. These events could also contribute to synaptopathy, as assessed by reductions in wave I supra-threshold amplitudes 40,41,52 which we observed in this study. Ghosh et al. observed IHC synaptopathy and ABR threshold shifts following cisplatin treatment which were attenuated by activation of CB2R 12 . We similarly observed that inhibition of RGS17 mimicked the effect of CB2 agonist. These data suggest that boosting the active G-protein signaling either by activating GPCRs (by CB2R agonists) or by delaying the inactivation of GPCR signaling (by inhibiting RGS17), could represent an effective strategy Figure 8. Reciprocal regulation between RGS17 and CB2 signaling. (A) Naïve male Wistar rats were treated with CB2 agonist, JWH-015 (0.815 µg/ear) by the trans-tympanic route prior to cisplatin (11 mg/kg, i.p) administration and cochleae were harvested after 3 days. RNA was extracted from the whole cochleae and processed for RT-qPCR. RGS17 expression was significantly elevated by cisplatin, but was significantly attenuated by pretreatment with JWH-015. #p < 0.05 versus control. *p < 0.05 versus cisplatin, N = 4, two way ANOVA. (B) The cochleae collected from same rats were fixed and processed for mid-modiolar sectioning and immunolabelled for RGS17. Cisplatin increased RGS17 immunoreactivity whereas activation of CB2R via JWH-015 reduced this fluorescence. Scale bar 50 μm. (C) siRNA against RGS17, siRGS17 (0.9 µg/ear), was administered into the rat cochleae 48 h prior to cisplatin exposure. RT-qPCR analysis was performed on the cochleae collected from those treated animals to assess CB2 and Gαz expression. siRGS17 increased cochlear CB2 and Gαz expression whereas cisplatin treatment decreased it. Data indicates mean fold change in expression ± SEM, N ≥ 4, p < 0.01, *statistically significant difference from cisplatin treatments, # indicates statistically significant difference from vehicle control, two-way ANOVA. (D) Rats were administered either scrambled RNA (0.9 µg/ear) or siRNA against CB2 (siCB2R, 0.9 µg/ear) trans-tympanically. Cochleae were collected to perform RT-qPCR analysis. siCB2R treated rats showed significantly increased RGS17 mRNA expression. #p < 0.05 versus scrambled RNA. (E) Conversely, overexpression of RGS17 in the rat cochlea via administration of adenoviral vector significantly decreased CB2R expression. Data represents fold change ± SEM, N ≥ 5t-test p < 0.05, #indicates statistically significant difference from Adv vector control. www.nature.com/scientificreports/ for preserving hearing. Furthermore, these findings highlight the importance of active GPCR signaling process in maintaining normal hearing. Based on its canonical pathway, RGS17 binds to and promotes the GTPase activity of Gαz, Gαi 1-3, Gαo and Gαq 53 . As such, RGS17 can profoundly increase the rate of GTP hydrolysis of these G proteins. For example, RGS17 blocked dopamine (D2) receptor mediated inhibition of cyclic AMP accumulation presumably by acting as a GAP for Gαi/o protein 30 . RGS17 also attenuates μ opioid receptor signaling via Gαz in periaqueductal grey matter 54 and is able to regulate thyrotropin-releasing hormone receptor by Gα q/11 . We speculate that a similar action at the level of CB2/Gα could account for the observation that overexpression of RGS17 by viral vectors or by cisplatin could promote hearing loss. A non-canonical pathway for RGS17 described in the brain shows its interaction with HINT1 complex which couples mu opioid receptors to protein kinase C-γ and facilitates activation of the ERK-MAP kinase pathway through NO-dependent release Zn 2+ ions 55 . Such a mechanism is linked to opioid tolerance 55 and could similarly contribute to desensitization of CB2R to endocannabinoids in the cochlea.
We show that overexpression of RGS17 in the cochlea is associated with increased oxidative stress and inflammatory proteins, NOX3, KIM1 and iNOS, an action which is characteristic of inhibition of CB2R in the cochlea 12 . This likely represents uncoupling of CB2R from its cytoprotective effectors which is blunted by knockdown of RGS17. We show that one of the downstream targets of RGS17 is STAT1 which contributes to cochlear inflammation 56 . Studies have established STAT1 as a common mediator for inflammation, apoptosis and hearing loss 11,56 . In addition, the cellular STAT1/STAT3 balance is crucial to maintain apoptosis-survival homeostasis in the cochlea 11 . Similar to cisplatin, we show that overexpression of RGS17 increased Ser 727 STAT1 phosphorylation and decreased Tyr 701 STAT3 phosphorylation, elevating the STAT1/STAT3 ratio which promotes cell apoptosis. Interestingly, the promoter of RGS17 gene has a STAT1 binding site 45 , suggesting that this mechanism could confer induction of RGS17. In preliminary studies, we show that the STAT1 inhibitor, EGCG, significantly decreased RGS17 expression (unpublished data). Thus, there likely exists a pathway from ROS to STAT1 and the induction of RGS17, which ultimately generates more ROS and RGS17.
Glutamate excitotoxicity is suggested to be a key mediator in noise-induced cochlear synaptopathy 57,58 . Glutamate agonists produced swollen cochlear nerve terminals and neuronal damage similar to noise exposure, whereas glutamate antagonist protected cochlear auditory neurons against excitotoxicity 57,59,60 . Thus, the cochlear synaptopathy observed in cisplatin-treated rats could also be due to the excessive glutamate release or NMDAR hyperactivity at IHC. In addition, activation of presynaptic CB2R in the CNS inhibits glutamate release at subthalamo-nigral synapses via the Gαi subunit 61 . It is possible that presynaptic inhibition of glutamate release at IHCs by CB2R activation prevents cisplatin-induced synapse loss, as observed by Ghosh et al. 12 . Chronic activation of NMDAR upregulates TNF-α, IL-1β, iNOS and arachidonic acid levels 62,63 which can, in turn, block reuptake of glutamate at synapses, thereby compounding glutamate excitotoxicity at synapses 64 . Thus, a pathway exists between inflammation and excitotoxicity which can promote neuronal damage and synapse loss 64 . In this study, we observed increased inflammatory markers in the cochlea following overexpression of RGS17, whereas knockdown of RGS17 reduced the inflammation and also promoted synaptic preservation (comparable to activation of CB2R). How RGS17 regulates the inflammatory process and the identity of immune cells involved in synaptic remodeling was not addressed in this study. However, since reductions in tonic CB2R activity using an antagonist produce inflammation, oxidative and synaptic loss 12 , one can reasonably assume that a similar sequence of events would accompany overexpression of RGS17. A likely scenario is that RGS17 uncouples CB2R/G protein and reduces its function and thereby acts indirectly like a CB2R antagonist. Taken together, we believe that these results demonstrate a constitutive role of CB2R/RGS17/G protein signaling in preserving IHC synapses.
Two interesting observations from our study is that the expression of RGS17 is under control of cisplatin and that RGS17 negatively influences CB2R expression. These data suggest a reciprocal interaction between RGS17 and CB2R such that both proteins could likely serve as therapeutic drug targets for treating cisplatin-induced and possibly other forms of hearing loss. In this regard, our data show that siRGS17 inhibited cisplatin-induced hearing loss, synapse loss and apoptosis of cochlear cells while trans-tympanic administration of CB2R agonist (JWH-015) protected against cisplatin-induced hearing loss 12 . In contrast, overexpression of RGS17 in the cochlea or blockade of CB2R produced hearing loss. An unanticipated finding is that trans-tympanic administration of siRNA against RGS17, by itself, stimulates the expression of CB2R, Gzα, Nrf2 and SOD2. These finding are intriguing in that they suggest that these genes are all under tonic regulation by RGS17. While the mechanism(s) underlying induction of these genes is unclear, increases in these proteins could enhance the endogenous otoprotective actions of endocannabinoids in the cochlea.
In summary, our study identified a new gene, RGS17, which is implicated in cisplatin-induced hearing loss (proposed mechanism see Fig. 9). We showed that RGS17 is induced by cisplatin which could render GPCRs (such as CB2R) less responsive to endogenous and exogenous agonists. Furthermore, knockdown or inhibition of this protein protected against cisplatin-induced hearing loss. Importantly, RGS17 antagonizes the otoprotective action of CB2R and suppresses its expression. Thus, RGS17 could represent a novel target for treating cisplatininduced hearing loss and potentially other toxicities associated with this drug.

Materials and methods
Drugs and antibodies. Cisplatin  Animals. This study was performed in compliance with the ARRIVE guidelines. Adult male Wistar rats (weighing ~ 200 g) were purchased from Envigo, Indianapolis (USA) and housed at the animal facility at the Southern Illinois University School of Medicine, IL. Rats were housed in a temperature-controlled room with normal light/dark cycles (12 h) with easy access to food and water ad libitum. Auditory brainstem recordings (ABRs) were performed in rats anesthetized using a 3:1 mixture of ketamine (90 mg/kg) and xylazine (17 mg/ kg). These studies were generally performed between 9 AM and 12 Noon and animals were returned to the vivarium after they had fully recovered from anesthesia. Pre-ABRs were performed prior to any trans-tympanic treatments on anesthetized rats. Rats were administered cisplatin ( Trans-tympanic administration of siRNA and adenoviral vector. Trans-tympanic injections were performed in the anterior inferior region of rat's tympanic membrane using a 28-30G needle with the aid of a Zeiss operating microscope. Following injections, rats were kept undisturbed with the injected ear facing upward Figure 9. Proposed mechanism underlying the involvement of RGS17 in cisplatin ototoxicity. In the cochlea, cisplatin-induced increase in RGS17 negatively modulates otoprotective GPCR/Gα signaling by facilitating the inactive Gα-GDP state. This, we believe relieves a tonic suppression of ROS and reactive nitrogen species (RNS) generation via NOX3 and iNOS, respectively, increases cellular STAT1/STAT3 ratios and increases proinflammatory and apoptotic pathways. We believe these changes promote hearing loss. Inhibition of RGS17 (by celastrol, knockdown of RGS17) or activation of CB2 relieves the oxidative stress, reduced the STAT1/STAT3 ratios and inhibits cochlear inflammation and apoptosis. www.nature.com/scientificreports/ for at least twenty min. Injections included 0.9 μg of siRNA in 50 μl of nuclease free water per cochlea or 4 × 10 8 p.f.u/cochlea of adenoviral vector. This protocol has been previously described by our lab 65 . Auditory brain response (ABR). ABRs were determined as described previously 66 . Male Wistar rats were anesthetized (as above) and electrode placement locations include the vertex of the skull (active), hind flank muscle (ground) and pinna (two reference electrodes under each pinna). Sound stimului were tone bursts at 8, 16 and 32 kHz with a 5 ms plateau and a 1 ms rise/fall time presented at a rate of 5/s. The auditory stimulus intensity was initiated at 10 dB sound pressure level (SPL) and reaching to a maximum intensity at 90 dB SPL with 10 dB increments. The auditory threshold was determined as a minimum sound intensity evoking visually detectable waves II and III, with a minimum amplitude of 0.5 μV. Threshold shift are calculated based as the difference between pre-and post-treatment thresholds obtained at each frequency.
Detection of ROS. ROS at cellular level was measured in UB/OC-1 cells using CellROX assay (Invitrogen), as described previously 11 . Cells were plated in 12 well plate with coverslips and pre-treated with vehicle or drugs or transfected with different agents. CellROX reagent was added at a final concentration of 5 μM and coverslips were incubated for 30 min at 33 °C. Coverslips were washed with PBS and fixed using 4% paraformaldehyde for 15 min and mounted on glass slides using VECTASHIELD mounting medium with DAPI (Vector Laboratories). ROS fluorescence was captured with a Leica Confocal microscope at 488 nm.

Apoptosis detection by flow cytometric analysis of cells. Apoptosis in cell cultures was detected
using an FITC Annexin V Apoptosis detection kit (BD Pharmingen, BD Biosciences) as described previously 11 . Briefly, following treatments, UB/OC1 cells were washed with ice cold 1× PBS, trypsinized and then resuspended in 1× binding buffer provided in the kit. Cells were transferred into a 5 ml falcon tubes to which were added 5 µl of both FITC Annexin V and propidium iodide (PI) and gently vortexed. The cells were then incubated for 15 min at room temperature in the dark. Binding buffer was mixed into each tube prior to flow cytometry (BD FACSCalibur Western blot. Western blotting in cell cultures was performed as described previously 11 . UB/OC1 cells were lysed using RIPA buffer (50 mM Tris, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% NP-40, pH 8.0) containing protease inhibitor (1:100, Sigma-Aldrich) and phosphatase inhibitors (Cocktail A and B, 1:100, Sigma-Aldrich), centrifuged at 4 °C 14,000 rpm for 10 min and supernatant was collected. Protein concentrations were determined using Bradford reagent 67 (MilliporeSigma). 60-80 μg of the collected whole cell lysate was then mixed with 5× Laemmli buffer (0.5 mM Tris-HCl pH6.8, glycerol, SDS, 0.25% bromophenol blue and beta-merceptoethanol) and heated at 90 °C for 5 min, loaded onto a 10-14% SDS PAGE gel. Proteins were then transferred to nitrocellulose membranes which were incubated with blocking buffer (10 mM PBS, 10 mM EDTA, 20% TritonX-100 and 5% bovine serum albumin) for 1 h and followed by incubation with primary antibody overnight and at 4 °C. The membranes were washed with 1× TBST (20 mM Tris pH 7.5, 150 mM NaCl, 0.1% Tween) and exposed to secondary antibodies for 1 h at room temperature and then imaged using LICOR Odyssey Image X. Proteins bands were analyzed using the Odyssey software normalized to actin (or total STAT1 or total STAT3 where relevant).

RNA isolation.
This procedure was performed essentially as described previously 11 . Snap frozen cochleae were crushed in 1 ml of TRIzol reagent (Sigma-Aldrich) using a homogenizer and incubated in the reagent at room temperature for 15 min. RNA isolation from cells, TRIzol was added to each well of the cell cultured dish and the solution transferred to a microfuge tube and vigorously vortexed for 30 s. This was followed by a chloroform extraction step. Sample tubes were then centrifuged at 12,000×g for 15 min at 4 °C. The aqueous phase (clear solution) containing RNA was collected carefully and transferred to DNase RNase free clear microfuge tube and 100% ice-cold isopropanol (500 µl) was added. Tubes were incubated overnight at − 80 °C for precipitation of RNA, centrifuged at 12,000×g for 20 min at 4 °C and formed RNA pellet was washed once with 100% molecular grade ice-cold ethanol and twice with 70% ethanol followed by 15 min centrifugation each times. The ethanol was removed, the RNA pellet resuspended in warm nuclease-free water and its concentration determined by a NanoDrop ND-1000 spectrophotometer. The pure RNA samples (260/230 wavelengths ranging from 1.8 to 2) were used for PCR.
Real time Q-PCR. These assays were performed as described previously 11 . Total RNA (500 ng) was used to prepare gene specific cDNA by using iScript Select cDNA Synthesis Kit (Bio-Rad). cDNA was synthesized using thermal cycler following reaction conditions at 42 °C for 60 min and 85 °C for 5 min. This cDNA reaction mix was used for q RT-PCR which was performed by using StepOnePlus Real-Time PCR System (Thermo www.nature.com/scientificreports/ Fisher Scientific). The cycling conditions used were 95 °C for 3 min followed by 45 cycles at 95 °C for 15 s, 58-64 °C for 30 s (depending on primer) and 95 °C for 3 s. Upon completion of amplification protocol, the melting curves were analyzed by cooling the reaction to 60 °C and slowly reheating to 95 °C. The mRNA expression were normalized to the levels of housekeeping gene, GAPDH. Negative controls were set up for both the target and housekeeping gene and no template cDNA was added to this mixture. The relative change in mRNA levels of a specific gene between the vehicle control and treated group was quantified by using the formula: 2 (Ct Target gene1-Ct GAPDH)-(Ct Target) gene)2-GAPDH2) 68 . The nucleotide sequences of rodent primer sets were based on homologous sequences of rat and mouse cDNA sequence (see Supplementary Table S1).
Cochlear whole-mount preparation for OHC and ribbon synapse counts. The studies were performed as described previously 12 . Rat cochleae was decalcified using 120 mM ethylenediaminetetraacetic acid (EDTA) for about 3 weeks in a rotating device. The organ of Corti was carefully microdissected from each decalcified cochlea and divided into the apical, middle and basal turns. Sections were immunolabeled with antibodies against myosin VIIa, CtBP2, and GluR2 for staining of hair cells, presynaptic ribbon and post-synaptic glutamate receptors, respectively. The number of OHCs per every 100 µm of the sections was counted manually and reported as an average number of OHC/100 µm. Zeiss LSM800 (× 63 magnification) was used to capture 10 IHCs per image. Three random sections from one cochlear whole mount sample were chosen to evaluate presence of functional synapse and aired synaptic ribbons (CtBP2 + GluR2 immunolabeling overlapping each other) were counted manually from each image.
TUNEL staining of cochlear sections. Apoptosis will be detected, as previously described 11  Statistics. Data are presented as mean ± standard error of mean (SEM). Statistical significance of differences among groups were tested using t-test, one-way or two-way analysis of variance (ANOVA) depending upon experiments, followed by TUKEY's post hoc analysis, using GraphPad Prism version 6.07 for Windows. p value < 0.05 was considered significant. www.nature.com/scientificreports/