XBP1 mitigates aminoglycoside-induced endoplasmic reticulum stress and neuronal cell death

Abstract

Here we study links between aminoglycoside-induced mistranslation, protein misfolding and neuropathy. We demonstrate that aminoglycosides induce misreading in mammalian cells and assess endoplasmic reticulum (ER) stress and unfolded protein response (UPR) pathways. Genome-wide transcriptome and proteome analyses revealed upregulation of genes related to protein folding and degradation. Quantitative PCR confirmed induction of UPR markers including C/EBP homologous protein, glucose-regulated protein 94, binding immunoglobulin protein and X-box binding protein-1 (XBP1) mRNA splicing, which is crucial for UPR activation. We studied the effect of a compromised UPR on aminoglycoside ototoxicity in haploinsufficient XBP1 (XBP1(+/-)) mice. Intra-tympanic aminoglycoside treatment caused high-frequency hearing loss in XBP1(+/-) mice but not in wild-type littermates. Densities of spiral ganglion cells and synaptic ribbons were decreased in gentamicin-treated XBP1(+/-) mice, while sensory cells were preserved. Co-injection of the chemical chaperone tauroursodeoxycholic acid attenuated hearing loss. These results suggest that aminoglycoside-induced ER stress and cell death in spiral ganglion neurons is mitigated by XBP1, masking aminoglycoside neurotoxicity at the organismal level.

Figure 1

Aminoglycoside-induced mistranslation. (a–b) Misreading and (c–d) readthrough was measured in RRL (a and c) and HEK wild-type cells (b and d). Results are derived from the ratio hFluc/hRluc, given in fold induction. Untreated samples are set as 1 (n=3; ±S.E.M.)

Figure 2

Proteomic analysis of geneticin-treated HEK wild-type cells. (a) Thirty-five upregulated proteins (Bonferroni-corrected P-value <0.05, log₂ FC >0.3) were grouped according to their biological function. (b) Upregulation of the geneticin-induced heat shock proteins, chaperones and foldases (Bonferroni-corrected P-value <0.05, log₂ FC >0.3). (c) Comparison of the significantly regulated proteins (Bonferroni-corrected P-value <0.05) and their corresponding mRNA fold induction. The upregulated proteins of the folding machinery are shown in red

Figure 3

Aminoglycosides induce the UPR. (a–c) qPCR analysis. HEK wild-type cells were treated with geneticin (16 μM) or gentamicin (400 μM) and incubated for the indicated times. Expression of mRNA for CHOP (a), BiP (b) and GRP94 (c) is shown. Means+S.D. of fold induction are presented relative to 0 h (untreated) sample (n=3); *P<0.05; **P<0.01; ***P<0.005; ****P<0.001. (d) XBP1-splicing assay. HEK wild-type or HEK aph(3′) cells were treated with geneticin (16 μM), gentamicin (1250 μM), hygromycin (2 μM), cycloheximide (2 μM), tunicamycin (5 μg/ml) for 24 h or left untreated. NTC, no template control. PCR products of XBP1 were analyzed by gel electrophoresis; unspliced and spliced versions of XBP1 are indicated. Tunicamycin was a positive control to induce ER stress; GAPDH was a loading control. The asterisk indicates the position of a hybrid amplicon. (e) Western blot analysis. HEK wild-type cells were treated with geneticin (16 μM) or gentamicin (400 μM) and incubated for 24 h. Ten micrograms of total protein were loaded and BiP, GRP94 and ATF4 were detected by immunoblotting using specific antibodies. β-Actin was used as a loading control and tunicamycin (2.5 μg/ml) as a positive control. (f and g) Reporter assays. HEK cells were transfected with luciferase reporter plasmids (f) UPRE (reporter for ATF6 activity) or (g) ERSE (reporter for ATF6 and XBP1 activity). Cells were treated with geneticin (16 μM) or gentamicin (800 μM) for 24 h. Cycloheximide (16 μM) was used as a negative control, and tunicamycin (2.5 μg/ml) as a positive control for eliciting UPR. Luciferase activities were determined and the Fluc/Rluc ratios were calculated. Untreated samples are set as 1 and fold inductions are given (n=3–6, ±S.E.M.). **P<0.01, ***P<0.005. (h) Phosphorylated eIF2α was detected by immunofluorescence. HEK wild-type cells were treated with geneticin (16 μM) for 24 h or arsenite (0.5 mM) for 1 h as a positive control. Scale bars: 40 μm. The lower panels show insets in higher magnification. Bar graph indicates quantification of p-eIF2α immunofluorescence (n number of cells; n_Un=540; n_Gen=249; n_Ars=648); ****P<0.001. Ars, arsenite; CHX, cycloheximide; Gen, geneticin; Gm, gentamicin; Hyg, hygromycin; Tm, tunicamycin; Un, untreated

Figure 4

ER stress in cochlear tissues. (a) Tunicamycin but not gentamicin causes ER stress in hair cells. Tunicamycin (0.07 μg/ml) induced the specific ER stress-associated pro-apoptotic factor, CHOP (green), in the nuclei of hair cells in organ of Corti explants by 24 h. In contrast, CHOP was not observed in any part of the organ of Corti throughout the entire time course of gentamicin treatment (3.5 μM) until hair cell death. Segments shown are from the basal turn. Green: CHOP (GADD 153 antibody), red: myosin 7a antibody, blue: Hoechst 33342 staining for nuclei. The focal plane is at the nuclear level of outer hair cells leaving some regions stained against myo7a out of focus. The figure represents three different explants at each time point. Scale bar (Gm): 10 μm. (b) Gentamicin induces ER stress in SGCs. Tunicamycin (0.07 μg/ml) treatment for 24 h induced CHOP in the nuclei of SGCs (arrows). With gentamicin treatment (3.5 μM), CHOP appeared in the nuclei of SGCs by 48 h (arrows). Green: CHOP (GADD 153 antibody), red: neuronal class III β-tubulin staining for SGCs, blue: Hoechst 33342 staining for nuclei. The figure represents three different explants at each time point. Scale bar, 10 μm

Figure 5

Gentamicin reduces the number of SGCs and IHC synapses in the basal turn of XBP1^+/− mice cochleae. Gentamicin (0.56 M) was locally injected into the middle ear through the bulla as described in the Materials and Methods section ‘Drug administration in vivo.' (a and b) Gentamicin reduces SGCs in XBP1^+/− but not in wild-type (XBP1^+/+) littermates. (a) The number of SGCs was counted from high-magnification images of Rosenthal's canal of saline- or gentamicin-injected wild-type and XBP1^+/− mice. Red: neuronal class III β-tubulin staining for neural cells, blue: Hoechst 33342 staining for nuclei. The figure represents five different animals at each condition. Scale bar: 50 μm. (b) Quantitative evaluations revealed that SGC density in the basal turn of XBP1^+/− mice but not in wild-type mice was significantly decreased by gentamicin. Filled bars, controls; open bars, gentamicin treatment. n=5 in each group; **P<0.01. Middle and apical turns were not affected. (c and d) Gentamicin reduces synaptic ribbons in XBP1^+/− but not in wild-type mice. (c) Hair cells were stained with anti-myo7 antibodies (red) and synaptic ribbons with antibodies to CtBP2 (green). The number of synaptic ribbons per IHC in the basal turn was quantified from 3-D images created by using Imaris software. Staining of some nuclei is consistent with a partial nuclear localization of CtBP2, which has been confirmed for IHCs. The figure represents three different animals at each condition. Scale bar: 20 μm. (d) Quantitative evaluations demonstrated that synaptic ribbon density of XBP1^+/− mice but not of wild-type littermates was diminished by local injection of gentamicin. Filled bars, controls; open bars, gentamicin treatment. n=3 in each group; **P<0.01

Figure 6

Auditory threshold shifts are induced by gentamicin and protected by TUDCA. (a) Gentamicin (0.56 M) was locally injected into the middle ear through the bulla as described in ‘Materials and Methods' section. Three weeks after treatment, large threshold shifts had developed at 32 kHz in XBP1^+/− mice (square symbols) but not in wild-type littermates (circles). Data are presented as mean+S.D. for XBP1^+/− mice and mean–S.D. for wild-types. n=6 in each group; **P<0.01. (b) TUDCA attenuates gentamicin ototoxicity in XBP1^+/− mice. Animals in all three groups received the local injection of gentamicin and, as indicated, TUDCA co-treatment (500 mg/kg sc.) at 6 days, 3 days and 3 h before gentamicin injection. Data are presented as means+S.D. of threshold shifts at 32 kHz, determined 3 weeks after treatment. n=6 in each group; **P<0.01