Hearing loss, including noise-induced hearing loss, is highly prevalent and severely hinders an individual's quality of life, yet many of the mechanisms that cause hearing loss are unknown. The pannexin (Panx) channel proteins, Panx1 and Panx3, are regionally expressed in many cell types along the auditory pathway, and mice lacking Panx1 in specific cells of the inner ear exhibit hearing loss, suggesting a vital role for Panxs in hearing. We proposed that Panx1 and/or Panx3 null mice would exhibit severe hearing loss and increased susceptibility to noise-induced hearing loss. Using the auditory brainstem response, we surprisingly found that Panx1−/− and Panx3−/− mice did not harbor hearing or cochlear nerve deficits. Furthermore, while Panx1−/− mice displayed no protection against loud noise-induced hearing loss, Panx3−/− mice exhibited enhanced 16- and 24-kHz hearing recovery 7 days after a loud noise exposure (NE; 12 kHz tone, 115 dB sound pressure level, 1 h). Interestingly, Cx26, Cx30, Cx43, and Panx2 were up-regulated in Panx3−/− mice compared with wild-type and/or Panx1−/− mice, and assessment of the auditory tract revealed morphological changes in the middle ear bones of Panx3−/− mice. It is unclear if these changes alone are sufficient to provide protection against loud noise-induced hearing loss. Contrary to what we expected, these data suggest that Panx1 and Panx3 are not essential for baseline hearing in mice tested, but the therapeutic targeting of Panx3 may prove protective against mid-high-frequency hearing loss caused by loud NE.
Hearing loss is one of the most disabling sensory deficits, affecting ∼360 million people worldwide . Currently, cochlear implants and hearing aids are the most commonly used interventions for managing hearing impairment. The lack of drug therapies for hearing loss hinges upon the fact that the molecular aspects and mechanisms that regulate the transduction of sound waves into electrical impulses along the auditory pathway are not fully understood.
Connexin (Cx) channels, which allow the passage of small molecules and ions between cells and their external environments, make up the two gap junction networks of the inner ear . These networks are formed before hearing onset  and are essential for cochlear ionic and metabolic homeostasis and maintenance of the endocochlear potential [2,4–6]. Although Cxs are not expressed in the sensory receptor hair cells of the cochlea [7–10], they are still critical for hearing as Cx26 and Cx30 mutations lead to deafness [2,11,12]. Recently, it was discovered that a unique family of glycoproteins that share similar structure and topology to Cxs, called pannexins (Panxs), are expressed in the cochlea and other resident cell types of the auditory pathway [13,14]. Panxs, of which there are three subtypes (Panx1, -2, and -3), are all proposed to form single membrane channels at the cell surface, but they may also serve other roles that have yet to be characterized [15,16]. These channels connect the extracellular and intracellular environments by allowing the passage of small members of the metabolome across the plasma membrane [17,18]. By far, the most understood and studied member of the Panx family is Panx1. Panx1 plays a role in many cellular processes, such as Ca2+ wave propagation, ATP release, and apoptosis [19–21]. Under normal conditions, Panx1 channels have been reported to be opened via mechanical stimulation, intracellular calcium, extracellular potassium, ATP, membrane depolarization, and caspase3 cleavage [19–23]. In contrast, Panx1 channels are blocked by the negative feedback of ATP and carbon dioxide-mediated acidification, or in the presence of channel blockers such as probenecid and carbenoxolone [23–26]. In the cochlea, Panx1 expression was found in the spiral limbus, the spiral prominence in the cochlear lateral wall, Reissner's membrane, spiral ganglion neurons, and supporting cells of the organ of Corti, including pillar cells, Hensen cells, Boettcher cells, and Claudius cells [13,14,27]. In addition, Panx1 mRNA expression was found in neurons of the auditory cortex . Similar to Cxs, Panx1 was not detected in the hair cells that are responsible for initiating the transduction of electrical impulses through the auditory pathway .
Recently, mice lacking Panx1 in a subset of cells in the cochlea were generated through the crossing of Panx1 floxed mice with paired box 2 (Pax2) Cre mice and were shown to exhibit sensorineural hearing loss . Another study crossed Panx1 floxed mice with a forkhead box G1 (Foxg1, which is required for proper morphogenesis of the cochlea ) Cre mouse line and showed that these mice also exhibited hearing loss . The hearing loss in these mice may be linked to the finding that Panx1 is reported to be a predominant ATP release channel in the cochlea . ATP acts as a signaling molecule in the inner ear, regulates mechanoelectrical transduction processes, and maintains proper cochlear mechanics [32–36]. To further support a possible role for Panx1 in hearing, a human patient was recently identified who harbored a loss-of-function PANX1 gene mutation. This patient exhibited sensorineural hearing loss, required bilateral cochlear implants from a young age, and had neurological deficits among multiple other conditions . Therefore, increasing evidence suggests the importance of Panx1 in hearing when conditionally ablated within the inner ear. However, a conditional ablation does not account for the expression of Panx1 along the entire auditory pathway, including its presence within the auditory cortex . Therefore, it is important to assess the comprehensive and developmental role of Panx1 in the entire auditory pathway from hair cell transduction to the transmission of electrical impulses along the successive relay nuclei in the central auditory system.
Compared with Panx1, Panx3 appears to have a more restricted expression pattern, being found mostly in bone, skin, and cartilage tissues [22,38–41]. Panx3 plays a role in the differentiation of chondrocytes and osteoblasts, and is involved in growth and development of bone [38,39,42]. Panx3 has recently been described in the cochlea, where it was found to be expressed in the cochlear bone and the modiolus of the inner ear . Sound conduction through the middle ear bones and the cochlea can be impaired by improper bone growth, leading to otosclerosis . Patients with otosclerosis often show signs of sensorineural hearing loss due to improper bone growth in the middle ear and/or the cochlear bone, which ultimately affects sound transduction to the hair cells and neurons within the inner ear . In addition, loss of proteins that assist in endochondral ossification in the ear lead to deafness in mice, and abnormalities in middle ear bone formation can leave them in a nonfunctioning state [45,46]. Therefore, because Panx3 is important in bone, it may be essential for proper middle and inner ear bone formation and morphology, but this has yet to be examined.
Since Panxs have recently been implicated in baseline hearing, it is reasonable to predict that they may also be involved in noise-induced hearing loss. Approximately 5% of the population worldwide will suffer from industrial, military, or recreational noise-induced hearing loss, which is most commonly caused by loss of hair cells due to apoptosis [47,48]. Hair cells and cochlear neurons do not regenerate, and as a consequence, excessive exposure to loud noise can result in permanent hearing loss. Currently, there are no completely accepted pharmacological treatments for noise-induced hearing loss, thus emphasizing that the development of molecular targets to prevent or reduce noise-induced hearing loss is essential . Owing to the expression of Panx1 in the inner ear and its role in ATP release and Ca2+ propagation functions, as well as the expression of Panx3 in the cochlear bone, both Panx1 and Panx3 may be potential molecular candidates implicated in noise-induced hearing loss through cochlear homeostasis and/or sound conduction. Thus, we set out to evaluate the impact of Panx1 and Panx3 ablation on acute loud noise damage.
To that end, we have obtained and/or generated Panx1 and Panx3 null mice, to account for the potentially broad distribution of both Panxs along the entire auditory pathway. In the present study, we assessed if these pannexins are crucial in baseline hearing and/or noise-induced hearing loss, and we evaluated the importance of Panx3 in the development of the middle ear bones. In contrast with our expectations, we discovered that Panx1 and Panx3 are not necessary for baseline hearing in these mouse lines. However, mice lacking Panx3, but not Panx1, are partially protected from acute loud noise damage.
Panx1 null mice (Panx1−/−) were obtained from Genentech (San Francisco, CA) and were previously described . Briefly, the targeting construct used loxP sites to flank exon2 and its deletion was achieved by breeding to the ubiquitously active Cre deleter mouse, C57BL/6-Gt(ROSA)26Sortm16(Cre)Arte (TaconicArtemis) and has since been backcrossed several times to C57BL/6 mice, allowing Cre recombinase to have been bred out. Mice were genotyped to confirm Panx1 ablation using the following primers: forward: 5′-GGA GAA GCA GCT TAT CTG G-3′; reverse: 5′-ACT GTG CAA TAC TAC ACG GA-3′. Panx1 expression has been shown to be ablated in several organs of these mice including muscle, lung, liver, kidney, tail, brain, heart, thymus, spleen, and skin tissues, validating this mouse model as a Panx1−/− mouse [49–51]. Panx3 null mice (Panx3−/−) were bred in-house and have been previously described . Briefly, Panx3 floxed mice were crossed with C57BL/6 Cre deleter mice [B6.C-Tg(CMV-cre)1Cgn/J; Jaxmice #006054]. Cre recombinase was bred out following generation of the Panx3−/− mouse line. Mice bred on the same background (C57BL/6) were used as age- and sex-matched wild-type (WT) controls. Mice were genotyped to confirm Panx3 ablation using the following primers: forward: 5′-TGC CCT CCA CAG AAA GCT ACC-3′; reverse: 5′-GGC CAG CCT TGT CCT GCG TAT G-3′. All mice tested were male and 2–3 months of age (P60–P90) unless otherwise stated. Mice were housed in the animal care facility at the University of Western Ontario, and maintained on a 12/12-h light/dark cycle and fed ad libitum. Animal experiments were approved by the Animal Care Committee on animal care of the University of Western Ontario.
Ribonucleic acid extraction and reverse transcription–polymerase chain reaction
Following euthanasia by CO2, cochleae were dissected from mice and flash-frozen in liquid nitrogen. Ribonucleic acid (RNA) was extracted from tissues using a Qiagen RNeasy mini kit (catalog #74104) and an RNase-free DNase kit (catalog #79254) or by a combination of Trizol reagent, Qiagen RNeasy mini kit, as well as RNase-free DNase treatment modified from a previous publication . A NanoDrop spectrophotometer was used to measure the absorbance of RNA. Subsequently, a one-step reverse transcription–polymerase chain reaction (RT–PCR) was performed using an RT–PCR Qiagen kit (catalog #210210). The PCR profile was as follows: 50°C for 30 min, 95°C for 15 min, 94°C for 30 s, 64°C for 30 s, and 72°C for 1 min for 30 cycles. Panx1-specific primers (forward: 5′-ACA GGC TGC CTT TGT GGA TTC A-3′ and reverse: 5′-GGG CAG GTA CAG GAG TAT C-3′) and Panx3-specific primers (forward: 5′-TTT CGC CCA GGA GTT CTC ATC-3′ and reverse: 5′-CCT GCC TGA CAC TGA AGT TG-3′) were used. 18S ribosomal ribonucleic acid (rRNA) was used as a control (forward: 5′-GTA ACC CGT TGA ACC CCA TT-3′ and reverse: 5′-CCA TCC AAT CGG TAG TAG CG-3′). The amplified products were run on a 2% agarose gel with ethidium bromide for visualization.
Following euthanasia by CO2, cochleae were dissected from mice and flash-frozen in liquid nitrogen. RNA was extracted as described above and converted into cDNA using a SuperScript VILO cDNA Synthesis kit (Thermo-Fisher catalog #11754050). Two-step quantitative PCRs (qPCRs) were performed using a SYBR Green PCR master mix by Life Technologies (catalog #4309155) in order to quantify Panx1 and Panx3 expression in the cochleae. The Panx1 PCR protocol was as follows: 50°C for 2 min, 95°C for 2 min, 95°C for 3 s, 60°C for 30 s, for 40 cycles followed by a melt curve. The Panx3 PCR protocol was as follows: 50°C for 2 min, 95°C for 2 min, 95°C for 3 s, 65°C for 30 s, for 40 cycles followed by a melt curve. Panx1 and Panx3 transcripts were normalized to the 18S rRNA transcript level. The same primers as mentioned above were used for Panx1 and Panx3. The following primers were also used in the present study: Cx26 (forward: 5′-CCGTCTTCATGTACGTCTTTTACAT-3′ and reverse: 5′-ATACCTAACGAACAAATAGCACAGC-3′), Cx30 (forward: 5′-GGCCGAGTTGTGTTACCTGCT-3′ and reverse: 5′-TCTCTTTCAGGGCATGGTTGG-3′), Cx43 (forward: 5′-ACAACAAGCAAGCCAGCGAG-3′ and reverse: 5′-TCGTCAGGGAAATCAAACGG-3′), and Panx2 (forward: 5′-TGGTACCCATCCTGCTGGT-3′ and reverse: 5′-GGGTGAAGTTGTGCGGAGT-3′). For all three Cxs and Panx2, the PCR profile was as follows: 50°C for 2 min, 95°C for 2 min, 95°C for 5 s, 60°C for 15 s, followed by a melt curve.
Following euthanasia by CO2, cochleae and brain tissues were dissected from mice, flash-frozen in liquid nitrogen, and stored at −80°C. For cochlear tissues, a litter of mice (∼6–10 cochleae) were pooled together for one lysate due to the small amount of tissue. Cochleae were crushed using a mortar and pestle with liquid nitrogen, and protein was extracted using a Triton X-100-based immunoprecipitation buffer as previously described . Lysates (30 μg) were run on 10% polyacrylamide gels using SDS–PAGE and protein was transferred to a nitrocellulose membrane using an iBlot transfer machine (Invitrogen). Membranes were blocked with 3% bovine serum albumin in Tween-PBS for 1 h. Blots were probed overnight with custom-made site-directed antibodies: anti-Panx1 (CT-295) at a 1:2500 dilution and anti-Panx3 (CT-379) at a 1:5000 dilution, as previously described . Blots were then washed and stained with AlexaFluor 680 secondary antibody (1:5000 dilution, Life Technologies) for 2 h and subsequently stained with anti-mouse glyceraldehyde 3-phosphate dehydrogenase (GAPDH; 1:10000 dilution, Chemicon) as a loading control. IRDye800 (1:5000 dilution, Rockland) was used as the secondary antibody for GAPDH. Membranes were visualized with an Odyssey infrared imaging system (LiCor).
Hearing assessment with an auditory brainstem response
Hearing levels were determined using the auditory brainstem response (ABR), which measured the electrical activity in the brainstem evoked by the repeated presentation of a given acoustic stimulus. Mice were anesthetized using a combination of ketamine (100 mg/kg) and xylazine (10 mg/kg) administered via intraperitoneal injections and positioned in a double-walled sound-attenuating chamber. The electrical activity of the ABR was recorded by subdermal electrodes (27 gauge; Rochester Electro-Medical, Lutz, FL) that were positioned at the vertex (active electrode), over the mastoid of the stimulated (right) ear (reference electrode), and on the mid-back (ground electrode). Throughout the ABR recordings, body temperature was maintained at ∼37°C using a homeothermic heating pad (507220F; Harvard Apparatus, Kent, U.K.).
Acoustic stimuli were generated by a Tucker-Davis Technologies (TDT, Alachua, FL) RZ6 processing module at a 100 kHz sampling rate and delivered by a magnetic speaker (MF1; TDT) positioned 10 cm from the animal's right ear. The left ear was occluded with a custom foam earplug. Acoustic stimuli for the click ABR and noise exposure (NE) procedure (described below) were calibrated with custom MATLAB software (The MathWorks, Natick, MA) using a ¼-inch microphone (2530; Larson Davis, Depew, NY) and a preamplifier (2221; Larson Davis), and tonal acoustic stimuli were calibrated using BioSig software by TDT (Alachua, FL) using a ¼-inch microphone (2530; Larson Davis, Depew, NY) and a preamplifier (2221; Larson Davis).
The auditory evoked activity of the ABR was collected using a low-impedance headstage (RA4L1; TDT), preamplified, and digitized (RA16SD Medusa preamp; TDT), and sent to a RZ6 processing module via a fiber-optic cable. The signal was filtered (300–3000 Hz) and averaged using the BioSig software (TDT). Acoustic stimuli consisted of a click (0.1 ms) and four tones (4, 8, 16, and 24 kHz; 5 ms duration and 1 ms rise/fall time). The click stimulus served as an indicator of general hearing, as it encompasses a range of frequencies that stimulate the cochlea. Mouse hearing frequency ranges from ∼1 to 100 kHz ; thus, the different tonal stimuli tested provided a measure of frequency-specific hearing sensitivity, as each tonal frequency stimulates a restricted portion of the cochlea. The click and tonal stimuli were each presented 1000 times (11 times/s) at decreasing intensities from the 90 to 50 dB sound pressure level (SPL) in 10 dB SPL steps. From 50 to ∼15 dB SPL, the successive steps were at 5 dB, with each sound level presented twice in order to best determine the ABR threshold offline. Ultimately, for each of the stimuli (click and 4–24 kHz tones), an experimenter who was blinded to the animal cohort and treatment condition determined the ABR threshold using the criteria of just noticeable deflection of the averaged electrical activity within the 10-ms time window .
For a subset of mice, the initial ABR was followed immediately by a loud NE while the mice were maintained under anesthesia. While positioned in the double-walled sound-attenuating chamber, mice were bilaterally exposed to a calibrated 12 kHz tone at 115 dB SPL for 1 h. The tone was generated with TDT software and hardware (RPvdsEx; RZ6 module) and delivered by a super tweeter (T90A; Fostex, Tokyo, Japan), which was placed 10 cm in front of the mouse. A homeothermic heating pad was used to maintain body temperature at ∼37°C, and supplemental doses of ketamine and xylazine were administered as needed to maintain anesthesia. Immediately following the NE, ABR thresholds were reassessed. Mice were then given antipamezole (1 mg/kg) to reverse anesthesia effects and allowed to recover in their home cage. Seven days after NE, mice were once again anesthetized and a final ABR recording was performed to determine the extent of permanent hearing damage incurred by the loud NE. Upon completion of the final ABR recordings, mice were then killed by cervical dislocation while under deep anesthesia, and tissue samples were collected for further processing.
Auditory ossicle dissection and imaging
In a subset of 2- to 3-month-old WT and Panx3−/− mice, the three auditory ossicles (malleus, incus, and stapes) contained within the middle ear were carefully microdissected. Ossicles were fixed in 4% PFA overnight and then imaged using a Zeiss light microscope. An experimenter blinded to the treatment groups used ImageJ software to make measurements of the ossicles, including the length of the malleus, distance between the tips of the incus, and length of the stapes window.
Two-tailed independent unpaired Student's t-tests were used for comparisons for ABR wave I analysis between WT and Panx1−/− as well as WT and Panx3−/− mice, and the measurements of middle ear bone morphology between WT and Panx3−/− mice. One-way analysis of variance (ANOVA) with a Tukey's post hoc test was used to compare Panx1, Panx2, Panx3, Cx26, Cx30, and Cx43 expression by qPCR in WT, Panx1−/−, and Panx3−/− cochleae. Two-way ANOVA was used to compare WT, Panx1−/−, and Panx3−/− groups for ABR thresholds with a Sidak's post hoc test. Two-way repeated-measure ANOVAs were used for NE analysis (where each individual frequency was analyzed separately), followed by a Sidak's post hoc test. All statistical analyses were performed using GraphPad Prism 6.0, *P < 0.05, N ≥ 3 for all experiments.
Panx1 and Panx3 are expressed in the cochlea and are ablated in knockout mice
To confirm that both Panx1 and Panx3 knockout mice had Panxs ablated from the cochlea, we used RT–PCR and quantitative real-time polymerase chain reaction (qRT-PCR) to determine the mRNA levels for Panx1 and Panx3 in 2- to 3-month-old WT, Panx1−/−, and Panx3−/− cochleae. We found that both Panx1 and Panx3 mRNA transcripts were expressed in the cochlea of WT mice and were ablated in their respective knockout cochleae as revealed when normalized to 18S rRNA (Figure 1A–D). Interestingly, there was no evidence of Panx1 or Panx3 compensation in the cochleae of their reciprocal null mice (Figure 1C,D). Western blots for Panx1 at postnatal days 1 and 8 (P1 and P8) revealed the loss of Panx1 in mutant mice, where age- and sex-matched brains were also used as controls (Figure 1E). Panx1 protein expression revealed the Gly0, Gly1, and Gly2 glycosylated species  in the cochleae of WT mice (Figure 1E). There was a nonspecific band observed that appeared above Gly2 species, which was also evident in the Panx1−/− mice. Western blots further confirmed that Panx3 was not elevated in the cochleae of mice that were devoid of Panx1 (Figure 1F). Interestingly, Panx3 was highly expressed in young mice (P8), before decreasing at P16 and P60–P90. These studies revealed that both Panx1 and Panx3 are expressed in the cochleae of WT mice and are ablated in the cochleae of their respective knockout mice. Moreover, Panx3 does not appear to compensate for the loss of Panx1 and vice versa.
Panx1 and Panx3 are expressed in the cochlea and are ablated in knockout mice.
Panx1−/− and Panx3−/− mice do not exhibit reduced hearing sensitivity or cochlear nerve deficits
To assess baseline hearing sensitivity in mice lacking Panx1 or Panx3, we measured ABR thresholds in response to various acoustic stimuli (click and 4–24 kHz tones). ABR trace recordings from a representative WT, Panx1−/−, and Panx3−/− mouse had similar waveform characteristics in response to a click stimulus at decreasing sound levels (90–30 dB SPL; Figure 2A). To further validate the finding that mutant mice had normal hearing, the ABR waveform traces for a 90 dB SPL click stimulus were averaged for each group (WT: n = 17, Panx1−/−: n = 17, and Panx3−/−: n = 18) to create a composite ABR waveform for each group of mice (Figure 2B). There were no observable deficits in ABR waveforms in either Panx1−/− or Panx3−/− mice (Figure 2B). Furthermore, there were no significant differences in ABR thresholds between WT and Panx1−/− or Panx3−/− mice for any of the stimuli tested (click and 4–24 kHz tones; Figure 2C,D). Additionally, we examined the amplitudes and latencies of the first wave of the ABR (wave I), which is representative of the activity at the level of the cochlear nerve, in response to a click stimulus at 90 dB SPL as well as 20 dB above the animal's ABR threshold, to determine if there were any deficits in neuronal synchrony or speed, respectively, at this point in the auditory pathway. We found that there were no differences in cochlear nerve amplitude or latency between WT and Panx1−/− mice, indicating no neuronal deficits within this region (Figure 2E,G). To our surprise, Panx3−/− mice had significantly increased wave I amplitudes compared with WTs in response to the 90 dB SPL click stimulus; however, there were no differences observed in wave I latencies (Figure 2F,H). Taken together, these results suggest that gene ablation of Panx1 or Panx3 does not lead to sensorineural hearing loss nor any deficits at the level of the cochlear nerve.
Panx1−/− and Panx3−/− mice do not exhibit hearing or vestibulocochlear nerve deficits.
WT and Panx1−/− mice are equally susceptible to noise-induced hearing loss
To assess whether Panx1 channels play a role in noise-induced hearing loss, we exposed WT and Panx1−/− mice to a loud noise (12 kHz tone at 115 dB SPL for 1 h) and let them recover for 7 days following the exposure. ABR thresholds were recorded for each stimulus (click and 4–24 kHz tones) at three time points: (1) baseline, (2) immediately after NE (post-NE), and (3) 7 days after NE, in an attempt to discern whether ablation of Panx1 channels would exacerbate or protect against acoustic trauma. At baseline, there were no differences observed in click, 4, 8, 16, or 24 kHz ABR thresholds between WT and Panx1−/− mice (Figure 3A–D; 8 kHz data not shown). Immediately after NE, ABR thresholds were significantly elevated (P < 0.05 for all stimuli) in both WT and Panx1−/− mice compared with baseline (Figure 3A–D), with the most severe elevation in ABR threshold being at the higher frequencies, 16 and 24 kHz tones (Figure 3C,D). There were no differences between WT and Panx1−/− ABR thresholds immediately after NE at any of the stimuli tested (Figure 3A–D), suggesting that a similar auditory insult occurred in both groups. Seven days after NE, ABR thresholds recovered to near baseline levels at the low-frequency stimuli (Figure 3A,B), with no difference between WT and Panx1−/− ABR thresholds. As expected after exposure to a loud 12 kHz tone for 1 h, ABR thresholds were still elevated 7 days later at the 16 kHz (WT: 86 ± 2 dB SPL; Panx1−/−: 86.5 ± 3 dB SPL) and the 24 kHz (WT: 86 ± 1.6 dB SPL; Panx1−/−: 86.5 ± 2 dB SPL) stimuli, but there were no differences found between WT and Panx1−/− mice. Collectively, these results suggest that ablation of Panx1 does not have an impact on noise-induced hearing loss.
WT and Panx1−/− mice show similar susceptibility to noise-induced hearing loss.
Panx3−/− mice exhibit enhanced 16 and 24 kHz hearing recovery after loud NE
Panx3−/− mice were exposed to a loud 12 kHz tone for 1 h to assess whether Panx3 plays an important role in noise-induced hearing loss. At baseline, there were no differences observed in ABR thresholds of click, 4, 8, 16, or 24 kHz tones between WT and Panx3−/− mice (Figure 4A–D; 8 kHz data not shown). Immediately after NE, ABR thresholds were significantly elevated (P < 0.05) compared with baseline levels at all stimuli, with the most severe elevation occurring at the higher frequencies, 16 and 24 kHz tones (Figure 4C,D). WT and Panx3−/− mice showed a similar elevation of ABR threshold for all of the stimuli tested (Figure 4A–D). Seven days after NE, ABR thresholds recovered to baseline levels at the lower frequencies, with no differences found between WT and Panx3−/− mice (Figure 4A,B). Interestingly, 7 days after NE, Panx3−/− mice had significantly less hearing loss at both the higher frequency stimuli (16 and 24 kHz) tested, as indicated by lower ABR thresholds than those observed in the WT mice (Figure 4C,D). The most profound enhancement of post-NE recovery was found at the 16 kHz stimulus (WT: 86 ± 2 dB, Panx3−/−: 66 ± 5 dB), where the Panx3−/− mice showed an ∼20 dB improvement in hearing recovery compared with the WT mice (Figure 4C). Collectively, these data suggest that ablation of Panx3 has a protective function against exposure to a single bout of loud noise.
Panx3−/− mice have enhanced recovery 7 days after auditory insult.
Panx3−/− mice have morphological changes in the middle ear bones
We examined the morphological characteristics of the middle ear bones (malleus, incus, and stapes) to determine if ablation of Panx3 would alter their formation during development. Representative images of each middle ear bone from both WT and Panx3−/− mice revealed that there were no gross changes in anatomical structures (Figure 5A). However, analysis of the total length of the malleus (i.e. distance from the head of the malleus to the end of the malleus) was significantly decreased in Panx3−/− mice (WT: 1.6 ± 0.014 mm, Panx3−/−: 1.54 ± 0.014 mm; Figure 5A,B). In addition, the distance between the processes of the incus was significantly increased in Panx3−/− mice (WT: 0.59 ± 0.009 mm, Panx3−/−: 0.62 ± 0.006 mm; Figure 5A–C). The length of the stapes window was significantly decreased in Panx3−/− mice (WT: 0.36 ± 0.01 mm, Panx3−/−: 0.33 ± 0.008 mm; Figure 5A–C). Thus, the ablation of Panx3 leads to small morphological changes in middle ear bones.
Panx3−/− mice have morphological alterations in middle ear bones.
Panx3−/− mice have enhanced Cx26, Cx30, Cx43, and Panx2 mRNA transcript levels
To assess for possible compensatory mechanisms of Cxs in Panx1 or Panx3 null mice, we examined Cx26, Cx30, and Cx43 mRNA expression levels in WT, Panx1−/−, and Panx3−/− mice at 2–3 months of age. Interestingly, all three Cxs were up-regulated in Panx3−/− mice compared with WT or Panx1−/− mice, while their expression in Panx1−/− remained similar to WT levels (Figure 6A–C). In addition, we assessed Panx2 mRNA expression in all three mice to determine if Panx2 may be up-regulated in either Panx1 or Panx3 null mice. Panx2 mRNA expression was up-regulated in Panx3−/− mice compared with Panx1−/− mice; however, no significant differences were observed in comparison with WT mice (Figure 6D). Panx1−/− mice had similar Panx2 expression to WT mice (Figure 6D).
Assessment of Cx and Panx2 mRNA transcript levels in pannexin-ablated mice.
Since their discovery in 2000, Panx large-pore channels have been implicated in a large number of human diseases [15,18,56]. Most recently, Panx1 was found to be linked to hearing loss in two tissue-specific mouse models where Panx1 was selectively ablated from resident cells of the cochlea [29,31]. In the current study, we addressed the possibility that Panx1 null mice, where Panx1 has been ablated throughout the entire animal, would exhibit even more severe hearing loss. Furthermore, we are the first group to assess if Panx3 is essential for normal hearing, and whether either Panx1 or Panx3 exacerbates or provides protection against noise-induced hearing loss. Surprisingly, the Panx1−/− mice tested here had normal hearing, and exhibited no protection from, nor increased sensitivity to, loud NE. Panx3−/− mice also exhibited normal hearing, but were partially protected from noise-induced hearing loss, possibly through morphological changes that were observed in the middle ear bone structures and/or compensatory up-regulation of other large-pore channels found in the cochlea. Collectively, the present findings challenge the position that Panx1 ablation will lead to hearing impairment, but support the concept that Panx3 ablation partially protects against loud noise-induced hearing loss.
Recently, it has been suggested that the most extensively studied Panx, Panx1, is the predominant ATP release channel in the cochlea , although Cx hemichannels have also been implicated in this process . ATP acts as a signaling molecule in the cochlea and plays a role in the mechanoelectrical transduction processes throughout the auditory pathway  and in propagating Ca2+ waves in response to hair cell damage . ATP also causes discrete bursts of action potentials in primary auditory neurons, an essential component in the development of central auditory pathways [32,59,60]. Panxs have also been implicated in Ca2+ wave propagation, and Ca2+ is essential for inner hair cell development through its role in creating spontaneous Ca2+-driven action potentials in early stages of development [61,62]. Within the cochlea, Panx1 is expressed in the spiral limbus, supporting cells of the organ of Corti, the spiral prominence in the cochlear lateral wall, Reisner's membrane, and the spiral ganglion neurons [13,14]. Moreover, Panx1 mRNA transcripts were detected in neurons of the auditory cortex [13,14,28]. Mounting evidence supports a role for Panx1 in hearing as a patient with a loss-of-function PANX1 germline mutation exhibited multisystem defects, including severe hearing loss . In addition mutant mice lacking Panx1 in Pax2- and Foxg1-expressing cells of the cochlear duct exhibited sensorineural hearing loss [29,31]. Pax2 is a transcription factor expressed in the endolymphatic duct and in hair cells of the cochlea [63–65]. Foxg1 is part of a forkhead gene family and is required for morphogenesis of the cochlea, and is expressed in most cell types of the inner ear . To account for the potentially broad distribution of Panx1 within the peripheral and central auditory systems, we predicted that Panx1−/− mice would have even more severe hearing loss than those lacking Panx1 in Pax2- or Foxg1-expressing cells only, but to our surprise both Panx1−/− and Panx3−/− mice tested did not exhibit sensorineural hearing loss. Moreover, the functional integrity of the auditory nerve was intact with no detrimental changes in neural transmission, which was observed by amplitude and latency analysis of ABR waveforms . Consistent with Panx1 not playing a major role in the auditory pathway, Panx1 null mice were equally susceptible to noise-induced hearing damage compared with controls, further validating the conclusion that Panx1 does not contribute to impairments in hearing sensitivity of the mice tested. Thus, the ablation of Panx1 does not lead to sensorineural hearing loss, cochlear nerve deficits, or altered susceptibility to noise-induced hearing loss.
Our findings were surprising and in contrast with previous studies that showed conditional ablation of Panx1 in the cochlea led to sensorineural hearing loss [29,31]. On first pass, one might predict that a Panx1−/− mouse would have a more severe hearing loss phenotype than when Panx1 is only ablated from a subset of resident cochlea cells; however, systemic gene-ablated mice have more opportunity for compensatory mechanisms to rescue a potential disease phenotype. For example, previous studies have shown that when Panx1 was ablated in mice, Panx3 was up-regulated in dorsal skin and blood vessels, findings that support a compensatory mechanism [50,67]. In the present study, we confirmed Panx1 ablation in the cochlea of Panx1−/− mice by assessing both mRNA transcripts and protein expression levels. We also showed that cochleae of Panx1−/− mice did not exhibit up-regulation of Panx2 mRNA expression or Panx3 at either the mRNA transcript or protein levels, minimizing a potential compensatory role for the residual Panxs in the Panx1−/− mice. That said, we cannot exclude the possibility that baseline levels of Panx2 and/or Panx3 channels are hyperactive. Alternatively, it is worth noting that because Cx hemichannels have been proposed to have similar functions to Panx channels , it is possible that Cxs in the cochlea are acting to compensate for the loss of Panx1. However, no increase in the levels of Cx26, Cx30, or Cx43 was observed in Panx1−/− mice. Thus, the reasons for the phenotypic differences in hearing competence identified in the previously published conditional Panx1 ablation mouse models versus the present global ablation model remain elusive, but could possibly be linked to mouse breeding strategies or aging. It is interesting that a recent paper suggested that many genetically modified mouse models harbor passenger mutations when bred back to different backgrounds, potentially influencing phenotypic outcomes . It is unknown if this issue may be a complicating factor in Panx1 mouse ablation studies. In the human context, we recently discovered the first patient worldwide harboring a PANX1 gene mutation linked to severe hearing loss . This suggests a possible link between PANX1 mutants and hearing loss that may be mechanistically distinct from cases where the PANX1 gene is completely ablated. Future studies will need to resolve what findings in Panx1−/− mice best translate to the human condition.
Panx3, the second Panx examined in our study, is involved in proper bone formation and development. Panx3 is expressed in prehypertrophic chondrocytes and is predicted to accelerate their terminal differentiation through ATP release . Recently, it was shown that Panx3−/− mice were protected against surgically induced osteoarthritis and had alterations in their long bones [52,70]. Panx3 is found mostly in bone and cartilage tissues, as well as in the cochlear bone and modiolus of the inner ear [14,18]. In the present study, we found that Panx3 expression in WT cochleae was developmentally regulated, peaking at P8 and there were morphological alterations to the middle ear bones. Interestingly, this finding correlates with previous findings that Panx3 regulates osteoblast differentiation , which our study suggests may be essential in the cochlear bone and auditory ossicles. The middle ear bones are essential for conveying sound from the external environment into the inner ear, as malformations of the middle ear bones can lead to hearing loss . We found that Panx3−/− mice, despite having morphological alterations in the three ossicles, did not exhibit impaired hearing sensitivity. In fact, compared with WTs, Panx3−/− mice showed enhanced neural synchrony at the level of the cochlear nerve. At present, however, it is not possible to reach conclusions about the physiological mechanism underlying this finding. Since the developing cochlea is formed through signaling between the developing bone and sensorineural structures , and the auditory transduction pathway becomes embedded in bone, diseases changing bone remodeling can affect hearing. For example, patients with the human genetic condition cleidocranial dysplasia, resulting from RUNX2 haploinsufficiency, develop bone deformities leading to hearing loss, which can be sensorineural, conductive, or a combination of the two . Interestingly, the Panx3 promoter can be transactivated by RUNX2 . This suggests that Panx3 expression could be regulated by RUNX2 and may explain some of the symptoms associated with cleidocranial dysplasia. Similarly, some patients with cochlear otosclerosis exhibit sensorineural hearing loss without morphological damage to the sensory tissues [44,73]. Thus, the enhancement of cochlear nerve synchrony that was observed in Panx3−/− mice could be due to cochlear and/or middle ear bone shape changes. Alternatively, the enhanced neural synchrony observed in the cochlear nerve of Panx3−/− mice may be due to ablation of Panx3 in cell types that directly contribute to sensorineural processing. While Panx2 levels were increased in Panx3-ablated mice, this increase was not statistically significant compared with WT mice, suggesting that the compensating role of Panx2 in assisting in neural synchrony remains to be firmly established. Finally, it is possible that the up-regulation of cochlear Cxs in Panx3−/− mice could be contributing to the increased synchrony observed in their cochlear nerve, as demonstrated by increased wave I amplitudes in response to the 90 dB click stimulus.
To our surprise, when we challenged the auditory system of Panx3−/− mice with a loud NE, they showed decreased susceptibility to 16 and 24 kHz hearing loss 7 days after the acoustic insult. Similar to the enhanced neural synchrony in the auditory nerve of these mice, the mechanism by which Panx3 ablation allowed the auditory system to be more resilient to noise-induced hearing loss remains unknown and difficult to dissect. We suggest that various mechanisms, including morphological changes in the middle ear bones and/or compensatory up-regulation of Cxs, may underlie the increased resistance to acoustic trauma observed in Panx3−/− mice in the present study. For example, since Panx3 is expressed in bone, the enhanced protection observed in Panx3−/− mice may also be due to differences in conveying mechanical sound through the middle ear bones during NE. Furthermore, the morphological alterations observed in middle ear bones of Panx3−/− mice could perhaps contribute to the enhanced 16 and 24 kHz hearing protection through differences in sound conduction correlating to better tolerance of loud noise.
It is important to acknowledge that the hearing protection observed in the Panx3−/− mice may have occurred by mechanisms associated with Panx3 that extend beyond its expression in bone, such as Ca2+ or ATP release. It is well established that loud NE causes an increase in metabolic activity that alters the cellular redox state of the cochlea, leading to reactive oxygen species that cause excessive cochlear damage [74,75]. This damage can ultimately lead to hair cell death, which is associated with intracellular accumulation of potassium, as well as auditory nerve degeneration [76,77]. Interestingly, dexamethasone administration in the cochlea was shown to have a protective function by preserving hair cell survival and reducing ototoxic injury through inhibition of Ca2+ entry in auditory cells [75,78–80]. Since Panx3, among other functions, has been reported to release Ca2+, it is reasonable to propose that ablation of Panx3 may have provided protection against noise-induced hearing loss in the present study, by reducing ototoxic injury through inhibition of Ca2+ release within the cochlea.
In considering the potential role that Cx compensation may play in Panx3−/−mice, it is worth noting that previous studies have shown that reduced Cx26 expression leads to increased susceptibility to noise-induced hearing loss, and following a loud NE, Cx26 expression decreases [81,82]. In the current study, we found that Cx26, Cx30, and Cx43 were all up-regulated in Panx3−/− mice at baseline, which could potentially act in a protective manner against an acute loud NE. Alternatively, because the expression pattern of Panx3 in the auditory system is not fully characterized, it is possible that the hearing protection observed in the Panx3−/− mice occurred via mechanisms associated with Panx3 that extend beyond its expression in bone. Ultimately, future studies will be needed to uncover the mechanism(s) by which ablation of Panx3 provides hearing protection from excessive NE, and thus, whether Panx3 could serve as a novel molecular target in the clinical protection against traumatic hearing loss.
auditory brainstem response
analysis of variance
forkhead box G1
glyceraldehyde 3-phosphate dehydrogenase
paired box 2
quantitative real-time polymerase chain reaction
ribosomal ribonucleic acid
reverse transcription–polymerase chain reaction
sound pressure level
J.M.A. conducted all experiments, co-designed the project, and drafted the manuscript. J.J.K. co-designed the project and experiments, and assisted in data collection and training of J.M.A.; K.B. bred and maintained all mice used in the present study. A.L.S. assisted in the setup of the mouse ABR apparatus. D.W.L. co-designed the experiments, analyzed data, and edited the manuscript. B.L.A. co-designed the experiments, analyzed data, and edited the manuscript.
The present study was funded by an Action on Hearing Loss Flexi Grant awarded to J.J.K. and D.W.L. Additional funding was provided by the Canadian Institutes of Health Research to D.W.L. [130530 and 148584] and B.L.A. .
The authors thank Dr Silvia Penuela for consulting on this project.
The Authors declare that there are no competing interests associated with the manuscript.