Hearing loss and hair cell death in mice given the cholesterol-chelating agent hydroxypropyl-β-cyclodextrin

Abstract

Cyclodextrins are sugar compounds that are increasingly finding medicinal uses due to their ability to complex with hydrophobic molecules. One cyclodextrin in particular, 2-hydroxypropyl-β-cyclodextrin (HPβCD), is used as a carrier to solubilize lipophilic drugs and is itself being considered as a therapeutic agent for treatment of Niemann-Pick Type C disease, due to its ability to mobilize cholesterol. Results from toxicological studies suggest that HPβCD is generally safe, but a recent study has found that it causes hearing loss in cats. Whether the hearing loss occurred via death of cochlear hair cells, rendering it permanent, was unexplored. In the present study, we examined peripheral auditory function and cochlear histology in mice after subcutaneous injection of HPβCD to test for hearing loss and correlate any observed auditory deficits with histological findings. On average, auditory brainstem response thresholds were elevated at 4, 16, and 32 kHz in mice one week after treatment with 8,000 mg/kg. In severely affected mice all outer hair cells were missing in the basal half of the cochlea. In many cases, surviving hair cells in the cochlear apex exhibited abnormal punctate distribution of the motor protein prestin, suggesting long term changes to membrane composition and integrity. Mice given a lower dose of 4,000 mg/kg exhibited hearing loss only after repeated doses, but these threshold shifts were temporary. Therefore, cyclodextrin-induced hearing loss was complex, involving cell death and other more subtle influences on cochlear physiology.

Figure 1. ABR thresholds are elevated one week following treatment with HPβCD.

(A) Mean thresholds are shown for 4, 16, and 32 kHz. Control, N = 13; 4,000 mg/kg, N = 12; 8,000 mg/kg, N = 15. Error bars represent one standard error of the mean. * P<0.05. (B) Scattergram of thresholds shown by subject for control and 8,000 mg/kg treatment groups. Closed and open symbols in the drug treated group represent animals assigned to “unaffected” and “affected” categories, respectively.

Figure 2. DPOAE responses are decreased one week after treatment with 8,000 mg/kg HPβCD.

Treated animals were broken into two groups according to whether their ABR thresholds were affected by HPβCD. DPOAE amplitude is plotted against the F1 stimulus level for each of the three test frequency pairs with geometric means of 8, 16, and 24 kHz. Error bars represent one standard error of the mean.

Figure 3. Cytocochleograms reveal substantial OHC loss one to two weeks after treatment with HPβCD.

Scatter plots reflect averaged data from 4 to 10 animals per condition. Data from the 8,000 mg/kg treatment group were obtained from animals exhibiting elevated ABR thresholds (affected). Legend applies to both plots. Mapping of characteristic frequency to cochlear location is based on the formula d = 82.5*log(f)–56.5 , where d is the percent distance from the cochlear apex and f is characteristic frequency. The total length of the cochlea used for these calculations was 5.7 mm. The original mapping was derived from single-unit auditory nerve recordings with characteristic frequencies between 7.2 and 61.8 kHz, which correspond to locations between 10% and 90% from the apex. Accordingly, the frequency axis reflects only data from this range.

Figure 4. HPβCD-induced pathology appears restricted to the sensory epithelium.

A mid-modiolar plastic section of the basal cochlear turn is shown by light microscopy for a control mouse (A) and an affected 8,000 mg/kg treated mouse (B). Insets show higher magnification view of the organ of Corti. Bracketed areas indicate the OHC region and arrowheads point to surviving IHCs. Images reflect similar results from three separate preparations in each treatment group.

Figure 5. Cholesterol levels are unchanged one week after treatment with 8,000 mg/kg HPβCD.

(A) Total cholesterol in liver and cochlea of control and HPβCD treated animals (N = 3 for each). Error bars represent one standard deviation from the mean. (B) Representative images of filipin stained organ of Corti. Mid-apical turn of control and HPβCD treated animals is shown.

Figure 6. Weekly dosing with 4,000 mg/kg HPβCD results in a temporary elevation of low-frequency ABR thresholds.

Thresholds for 4, 16, and 32 kHz stimuli were recorded weekly over a 16-week time period. Baseline responses were measured at Week 0. Injections began 5 days before the ABR measured on Week 1 and continued in this manner up to Week 12 (gray box). ABR measurements continued for an additional 4 weeks after cessation of the HPβCD injections. Error bars reflect one standard error of the mean (N = 6). Statistical significance was examined with one-way ANOVA (P values shown on plots). A Dunnett post-hoc test was used for pairwise comparisons between the baseline response on Week 0 and later time points (asterisk indicates P<0.05).