HIF-VEGF pathways are critical for chronic otitis media in Junbo and Jeff mouse mutants

Abstract

Otitis media with effusion (OME) is the commonest cause of hearing loss in children, yet the underlying genetic pathways and mechanisms involved are incompletely understood. Ventilation of the middle ear with tympanostomy tubes is the commonest surgical procedure in children and the best treatment for chronic OME, but the mechanism by which they work remains uncertain. As hypoxia is a common feature of inflamed microenvironments, moderation of hypoxia may be a significant contributory mechanism. We have investigated the occurrence of hypoxia and hypoxia-inducible factor (HIF) mediated responses in Junbo and Jeff mouse mutant models, which develop spontaneous chronic otitis media. We found that Jeff and Junbo mice labeled in vivo with pimonidazole showed cellular hypoxia in inflammatory cells in the bulla lumen, and in Junbo the middle ear mucosa was also hypoxic. The bulla fluid inflammatory cell numbers were greater and the upregulation of inflammatory gene networks were more pronounced in Junbo than Jeff. Hif-1α gene expression was elevated in bulla fluid inflammatory cells, and there was upregulation of its target genes including Vegfa in Junbo and Jeff. We therefore investigated the effects in Junbo of small-molecule inhibitors of VEGFR signaling (PTK787, SU-11248, and BAY 43-9006) and destabilizing HIF by inhibiting its chaperone HSP90 with 17-DMAG. We found that both classes of inhibitor significantly reduced hearing loss and the occurrence of bulla fluid and that VEGFR inhibitors moderated angiogenesis and lymphangiogenesis in the inflamed middle ear mucosa. The effectiveness of HSP90 and VEGFR signaling inhibitors in suppressing OM in the Junbo model implicates HIF-mediated VEGF as playing a pivotal role in OM pathogenesis. Our analysis of the Junbo and Jeff mutants highlights the role of hypoxia and HIF-mediated pathways, and we conclude that targeting molecules in HIF-VEGF signaling pathways has therapeutic potential in the treatment of chronic OM.

Figure 1. The inflamed middle ear of the Jbo/+ and Jf/+ mice is hypoxic.

(A) Jbo/+ mouse labeled with pimonidazole (PIMO), arrows indicate hypoxia in [ep] epithelium, [f] connective tissue fibrocyte, [mΦ] foamy macrophage; [tm] temporomandibular bone, [m] thickened inflamed mucosa, [ex] exudate. Note * the cleft is an artifact produced by tissue processing. (B) The normal thin mucosa [m] is not labeled in Junbo wild type (+/+) mice. (C) An unlabeled Jbo/+ mouse is a negative control for anti-PIMO antibody. (D) Hypoxia in foamy mΦ. (E) The middle ear in Jbo/+ mice is chronically hypoxic. The labeling index scores one point each for PIMO-positive staining in inflammatory cells in the bulla; mucosal epithelium; and mucosal connective tissues. Histogram bars are mean ± SEM. 4 wk group size n = 8, 7–8 wk n = 10, 13–15 wk n = 5, 31–37 wk n = 7. (F) The Eustachian tube epithelium [et] is hypoxic in a +/+ mouse but the adjacent nasopharynx epithelium [np] is normoxic. (G) Bulla fluid cytology from a Jbo/+ mouse shows F4/80 foamy mΦ and polymorphonuclear cells (PMN). (H) FACS analysis of PIMO-labeled Jbo/+ bulla fluids stained with Ly6G and Ly6C (PMN marker), for PIMO (hypoxia), and Annexin V (apoptosis marker). The PMN population was gated on the Ly6G and Ly6C signal. Population (1) normoxic viable PMN, (2) hypoxic viable PMN, (3) hypoxic apoptotic PMN. (I) Jf/+ mouse PIMO labeling was restricted to inflammatory cells in the bulla fluids and there was no detectable mucosal labeling, (J) Jeff +/+ mouse does not show PIMO labeling. Scale bars: A,B,C,I,J = 50 µm; D,G =  20 µm; F = 100 µm.

Figure 2. Vegfa titers are elevated in bulla fluids of 8-week-old Jbo/+ and Jf/+ mice compared to serum.

Protein titers of Vegfa in serum and bulla fluids of 8 wk Junbo (+/+ and Jbo/+) and Jeff (+/+ and Jf/+) mice. The gray zone represents the lowest assay standard (7 pg/ml). Each box represents the median with 25 and 75% inter-quartile ranges, with whiskers representing the data range (minimum and maximum). A Kruskall-Wallis test was performed followed by Dunn's multiple comparison tests for post hoc testing.

Figure 3. Il-1β and Tnf-α titers in bulla fluids are elevated in Jbo/+ but not Jf/+ mice.

Protein titers of (A) Tnf-α and (B) Il-1β in serum and bulla fluids of 8 wk Junbo (+/+ and Jbo/+) and Jeff (+/+ and Jf/+) mice. Figures in the gray zone represent the numbers of samples with values beneath the lowest assay standard (23 pg/ml for Tnf-α and 12 pg/ml for Il-1β. Each box and whisker symbol represents the minimum, 25% quartile, median, 75% quartile and maximum for reportable measurements while single values are represented by dots. Since the level of Tnf-α and Il-1β in many Jf/+ and Jbo/+ sera and Jf/+ bulla fluid samples was beneath detection limits, a statistical analysis was not performed.

Figure 4. Treatment of Jbo/+ mice with VEGF receptor inhibitors and the HSP90 inhibitor 17-DMAG moderates hearing loss.

(A) Change in Auditory Brain Stem response (ΔABR) in decibels (dB) in 15 d treatment with BAY 43-9006 (+/+, sham Jbo/+, drug Jbo/+ n =  5, 11, 11 respectively); (B) ΔABR in 28 d treatment with SU-11248 (+/+, sham Jbo/+, drug Jbo/+ n = 10, 15, 15 respectively); (C) ΔABR in 21 d treatment with 50 mg/kg PTK787 (+/+, sham Jbo/+, drug Jbo/+ n = 9, 13, 15 respectively); (D) ΔABR in 28 d treatment with 75 mg/kg PTK787 (+/+, sham Jbo/+, drug Jbo/+ n = 40, 60, 40 respectively); (E) ΔABR in 28 d treatment with 17-DMAG (+/+, sham Jbo/+, drug Jbo/+ n = 9, 15, 15 respectively). In each experiment, the response to drug treatment was compared to the sham control. Histogram bars are mean ± SEM. Statistics were conducted using 1-tailed Mann Whitney U tests.

Figure 5. Treatment of Jbo/+ mice with PTK787 reduces angiogenesis and lymphangiogenesis in the inflamed middle ear mucosa.

Junbo (+/+ and Jbo/+) mice were treated with either 50 mg/kg or 75 mg/kg of PTK787 for 4 wk, the sham control Jbo/+ groups received vehicle alone. The middle ear mucosa in treated Jbo/+ mice had (A) fewer blood vessels in the 75 mg/kg treatment group and (B) fewer lymphatic vessels in the 50 mg/kg or 75 mg/kg treatment groups. (C) The mucosa thickness and (D) mucosal lymphatic vessel diameter did not differ significantly from sham treated controls. The +/+ group sizes were n = 8; Jbo/+ drug treatment groups n = 15; Jbo/+ sham group n = 28. Histogram bars represent mean ± SEM. Data in panels A, B and C were analyzed by one-way ANOVAs and Bonferroni's multiple comparison tests for post hoc testing. Lymphatic vessel number (panel B) was not normally distributed and a Kruskall-Wallis test was performed followed by Dunn's multiple comparison tests for post hoc testing.