Hearing Is Normal without Connexin30

Abstract

Gjb2 and Gjb6, two contiguous genes respectively encoding the gap junction protein connexin26 (Cx26) and connexin 30 (Cx30) display overlapping expression in the inner ear. Both have been linked to the most frequent monogenic hearing impairment, the recessive isolated deafness DFNB1. Although there is robust evidence for the direct involvement of Cx26 in cochlear functions, the contribution of Cx30 is unclear since deletion of Cx30 strongly downregulates Cx26 both in human and in mouse. Thus, it is imperative that any role of Cx30 in audition be clearly evaluated. Here, we developed a new Cx30 knock-out mouse model (Cx30^Δ/Δ) in which half of Cx26 expression was preserved. Our results show that Cx30 and Cx26 coordinated expression is dependent on the spacing of their surrounding chromosomic region, and that Cx30^Δ/Δ mutants display normal hearing. Thus, in deaf patients with GJB6 deletion as well as in the previous Cx30 knock-out mouse model, defective Cx26 expression is the likely cause of deafness, and in contrast to current opinion, Cx30 is dispensable for cochlear functions.

Introduction

Cx30 and Cx26, which coassemble in gap junction (GJ) channels allowing the direct passage of ions and small molecules between the cytoplasms of cochlear non-sensory cells, are considered crucial for hearing (Forge et al., 2003; Martínez et al., 2009). In humans, >100 mutations in GJB2, nearly all affecting proper translation of Cx26, underlie the recessive non-syndromic hearing impairment DFNB1, the most common cause of prelingual inherited deafness (del Castillo and del Castillo, 2012). Additionally, Cx26 inactivation results in profound deafness in mouse (Cohen-Salmon et al., 2002; Sun et al., 2009; Crispino et al., 2011). In contrast to GJB2, silencing point mutations in GJB6 encoding Cx30 and located just 30 kb upstream from GJB2 in the same DFNB1 locus on chromosome 13, have never been found in deafness cases, although three missense mutations are associated with dominantly inherited hearing loss (Grifa et al., 1999; Nemoto-Hasebe et al., 2009; Wang et al., 2011). Thus far, the only evidence supporting that the absence of Cx30 leads to deafness in human was provided by the discovery of two large deletions truncating GJB6 and segregating in a double heterozygous state with a single GJB2 recessive mutation, hence suggesting the possibility of digenic inheritance for DFNB1 (Lerer et al., 2001; Pallares-Ruiz et al., 2002; del Castillo et al., 2002, 2005). However, skin or buccal biopsies of these patients have revealed that not only Cx30 but also Cx26 expression was dramatically reduced, demonstrating that both genes are coregulated and suggesting the existence of a common cis-regulatory element in the DFNB1 locus (Common et al., 2005; Rodriguez-Paris and Schrijver, 2009). A comparable situation is found in mouse, in which Cx30 inactivation causes profound deafness (Teubner et al., 2003; Cohen-Salmon et al., 2007; Sun et al., 2009), but also strongly reduces the expression of Cx26 (Ortolano et al., 2008; Lynn et al., 2011). Thus, how exactly the lack of Cx30 translates into a deafness phenotype still remains unclear.

Materials and Methods

Animal experimentation.

Experiments on mice were performed according to Institut National de la Santé et de la Recherche Médicale welfare guidelines as well as in compliance with the European Community Council Directive of November 24, 1986 (86/609/EEC).

Mice.

Cx30^fl/fl and Cx30^−/− mice were maintained on a pure C57BL/6 background. Cx30^Δ/Δ and Cx30^+/+ mice used in all experiments were littermates. Mice of either sex were used.

Construction of Cx30 floxed mice.

We subcloned a 10,950-bp fragment of 129/Sv mouse genomic DNA, containing Cx30 exon 3, the single Cx30 coding exon, by using gap-repair homologous recombination (Lee et al., 2001) on BAC 166gt19 from the CITB-CJ7B mouse BAC library (Research Genetics). To prepare the targeting vector, we introduced a loxP site upstream and a floxed PGK-hyg-resistance cassette downstream of Cx30 exon 3, by means of bacterial recombinogenic engineering techniques (Liu et al., 2003). Embryonic stem (ES) cells derived from a 129/Sv embryos were electroporated with the purified, linearized targeting vector and plated on hygromycin selective medium. Two independently obtained ES cell clones, in which one Cx30 wild-type allele had been replaced with the Cx30 floxed allele, were used to generate chimeric mice by injecting C57BL/6 host blastocysts. All chimaeras tested were fertile and transmitted the transgene to their offspring. Cx30 floxed mice (Cx30^fl/fl) from both lines were born at the expected mendelian frequency and showed no differences in growth, development or fertility compared with their Cx30^+/+ littermates.

Genotyping.

Genotyping of the Cx30^fl allele was performed by PCR analysis using a primer binding in the Cx30 transcribed sequence Gjb6R 5′-TTCCCTATGCTGGTAGAGTGCTTGT-3′ and a primer binding upstream of the first loxP site Gjb6F 5′-GCAGTAACTTATTGAAACCCTTCACCT-3′. Genotyping of the Cx30^Δ allele was performed using the Gjb6F and a primer binding downstream of the third loxP site Gjb6ΔR 5′-CCCACCATCAAGGTTGAACT-3′.

Quantitative PCR analyses of Cx26 and Cx30 transcription.

Postnatal day 30 (P30) mice were anesthetized by lethal injection of pentobarbital and decapitated. Total RNA was extracted from whole dissected inner ear using the RNeasy kit (Qiagen). Reverse transcription was performed using 1 μg of RNA. Quantitative PCR (qPCR) was conducted on cDNAs using Cx26 and Cx30 primers previously designed (Ortolano et al., 2008) (working concentration 300 nm) and Hprt primers: Hprt.f 5′-GTTGGATACAGGCCAGACTTTGTTG-3′ and Hprt.r 5′-GATTCAACTTGCGCTCATCTTAGGC-3′ (working concentration 300 nm), using SYBR Green PCR master kit (Applied Biosystems). PCR cycling conditions were 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. Samples were analyzed in triplicate on a LC480 Roche LightCycler. The relative abundance of amplified Cx30 or Cx26 cDNA was calculated as 2^−ΔCt, where ΔCt (change in cycle threshold) equals Ct of Cx30 or Cx26 in mutants minus wild-type mice.

Western blot analysis of Cx26 and Cx30 expression.

P30 mice were anesthetized by lethal injection of pentobarbital and decapitated. Inner ears were dissected, reduced to powder at −80°C, and immediately dissolved in PBS with 2% SDS, and 1× EDTA-free Complete Protease Inhibitor (Roche). Lysates were sonicated 3 times at 10 Hz (Vibra-Cell VCX130) and centrifuged 20 min at 10,000 × g at 4°C. Supernatants were boiled in 5× Laemmli loading buffer. Protein content was measured using the Pierce 660 nm protein assay reagent (Thermo Scientific). Equal amounts of proteins were separated by denaturing electrophoresis in NuPAGE gel (Invitrogen), electrotransferred to nitrocellulose membranes, first analyzed using either the primary rabbit anti-Cx30 antibody (Invitrogen, 1:500) or the rabbit anti-Cx26 antibody (Invitrogen, 1:500), and HRP-conjugated secondary antibodies. HRP activity was visualized by enhanced chemiluminescence using Western Lightning Plus enhanced ECL system (PerkinElmer). Blots were reprobed with mouse monoclonal anti-GAPDH-peroxidase (Sigma, 1:10,000) to check the protein load. Chemiluminescence imaging was performed on a LAS4000 (Fujifilm). Semiquantitative densitometric analysis was performed with ImageJ software.

Assessment of hearing impairment.

Auditory brainstem response (ABR) reflects the electrical response of the cochlear ganglion neurons and the nuclei of the central auditory pathway to sound stimulation. Their threshold assesses the cochlear sensitivity. The distortion product otoacoustic emissions (DPOAEs) are a noninvasive measure of outer hair cell amplification activity. ABRs and DPOAEs were recorded and analyzed as described previously (Verpy et al., 2008).

Results

We developed a novel knock-out mouse model (Cx30^Δ/Δ) in which, in contrast to the previous model (referred as Cx30^−/− in the literature) (Teubner et al., 2003), Cx30 was removed without perturbing the surrounding sequences. Indeed, in Cx30^−/− mice, the Cx30 coding exon was both removed and replaced by a lacZ reporter gene and a neo resistance cassette, and we suspected that such strong manipulation may impact on the expression of the nearby Cx26 encoding gene and on the phenotypic outcome. In our model, the Cx30-coding exon was replaced with a Cx30 floxed allele by homologous recombination and the Cx30^fl/fl mouse strain was crossed with Pgk-Cre mice which ubiquitously express the Cre recombinase (Lallemand et al., 1998), allowing for the global deletion of the Cx30 transcript sequence (see Materials and Methods; Fig. 1A,B). To test for Cx30 inactivation, we measured the level of Cx30 expression in the inner ear by qPCR (see Materials and Methods; Fig. 1C). As expected, no Cx30 transcript was detected in Cx30^Δ/Δ mice, whereas Cx30^fl/fl displayed reduced expression of Cx30 mRNA to 64 ± 3% of Cx30^+/+ mice used as controls, suggesting an effect of the 3′ hygromycin gene insertion on Cx30 expression. Consistent with the mRNA results, Western blot analysis of protein levels in the inner ear of Cx30^fl/fl mice showed a reduced level (64 ± 13%) of Cx30, which was undetectable in Cx30^Δ/Δ mice (Fig. 1D).

Figure 1.

Conditional Cx30 inactivation model. A, Homologous recombination resulting in the Cx30^fl allele where Cx30 transcript sequence (coding sequence, dark gray box; untranslated 5′ and 3′ sequences, black boxes) is flanked by LoxP sites (arrowheads) followed by the hygromycin resistance gene (hyg) and a LoxP site. Crossing Cx30^fl/fl mice with Pgk-Cre mice generated Cx30^Δ/Δ mice. Arrows indicate genotyping primers. B, PCR genotyping: GjB6 R and F produce a 350 bp band in Cx30^+/+ mice, and a 488 bp band in Cx30^fl/fl mice. GjB6F and GjB6ΔR produce a 521 bp band in Cx30^Δ/Δ mice. C, D, Cx30 expression was quantified by qPCR (C) and immunoblot (D) in whole inner ear. Mean ± SEM; 3 independent experiments; 3 mice of each genotype per experiment; t test. *p < 0.05, ***p < 0.0001.

To investigate the issue of Cx30 and Cx26 coregulation, we quantified the mRNA level of Cx26 in the inner ear of Cx30^Δ/Δ, Cx30^fl/fl and Cx30^−/− mice relative to that of Cx30^+/+ controls (see Materials and Methods; Fig. 2). In Cx30^−/− mice, we found a rather drastic reduction to 27 ± 6%. In Cx30^fl/fl mice, Cx26 mRNA was decreased to 54 ± 6%, a level comparable to that of Cx30 mRNA (Fig. 2A). Of note, in Cx30^Δ/Δ mice, Cx26 transcription was 62 ± 5% of that measured in Cx30^+/+ (Fig. 2B), a level comparable to Cx30^+/− mice (Rodriguez et al., 2012). These results were confirmed by Western blot analysis (Fig. 2B). Compared with Cx30^+/+ mice, Cx26 protein was significantly reduced to 10 ± 3% in Cx30^−/− mice, to 35 ± 13% in Cx30^fl/fl mice, and to 52 ± 4% in Cx30^Δ/Δ mice. Thus, the level of residual Cx26 protein is 5 times higher in Cx30^Δ/Δ than in Cx30^−/− mice.

Figure 2.

Cx26 expression in the inner ear of Cx30^+/+, Cx30^fl/fl, Cx30^Δ/Δ, and Cx30^−/− mice. Cx26 expression quantified by qPCR (A) and quantitative immunoblot (B). Mean ± SEM; 3 independent experiments; 3 mice of each genotype per experiment; t test. *p < 0.05, **p < 0.001, ***p < 0.0001.

To examine the physiological consequences of these reduced connexin levels, we next examined hearing in Cx30^Δ/Δ mutants by measuring ABR thresholds as well as DPOAEs (see Materials and Methods). To our surprise, Cx30^Δ/Δ mutants showed ABR thresholds and DPOAEs indistinguishable from Cx30^+/+ mice or Cx30^fl/fl mice (Fig. 3). Thus, micromechanics and transduction, the two basic functions underpinning normal hearing (Ashmore and Gale, 2000) are maintained without Cx30, provided that the Cx30 coding exon is deleted without introducing other perturbation of the surrounding sequences. Accordingly, the simple insertion of Hyg gene into the 3′ end of Cx30 in Cx30^fl/fl mice decreased the expression of both Cx30 and Cx26 (Figs. 1, 2), although their residual level was sufficient to generate normal ABR and DPOAEs (Fig. 3).

Figure 3.

Assessment of hearing in Cx30^Δ/Δ mice. A, ABR thresholds (mean ± SEM) recorded at 20–100 dB SPL across a frequency range of 10–32 kHz. No significant difference in threshold was found between Cx30^+/+ and Cx30^Δ/Δ or Cx30^fl/fl mice. In contrast, ABR waveforms showed up only above 80 dB in the Cx30^−/− mice. B, DPOAEs at 10 kHz for equilevel stimuli (mean ± SEM), increasing stepwise from 30 to 75 dB SPL. No significant difference in threshold or level was found between Cx30^+/+ and Cx30^Δ/Δ or Cx30^fl/fl mice. In contrast, Cx30^−/− mice had no DPOAE up to 75 dB SPL. A, B, Kruskal–Wallis nonparametric test, p > 0.1.

Discussion

Together, our results demonstrate that hearing loss phenotype exhibited by Cx30^−/− mice depends on the cumulative effect of deletion of Cx30 and 3′ insertion of the lacZ and neo genes, which are associated with dramatically reduced Cx26 levels (Figs. 1, 2). They also suggest that further investigation of two recent knock-in mouse models for connexin mutations linked to human deafness (Schütz et al., 2010, 2011) is needed, as both models present with lacZ insertion downstream of the mutated exon. Our data also support the notion of a cis-acting element coregulating Cx30 and Cx26, whose efficiency depends on the spacing of the surrounding sequences. Similarly, large deletions in the DFNB1 locus in humans may perturb the functionality of such element, leading to the downregulation of both GJB6 and GJB2 and to profound deafness (del Castillo and del Castillo, 2012). This hypothesis is further supported by the recent discovery of deletions causing DFNB1 that exclude both GJB2 and GJB6 (Wilch et al., 2006, 2010).

In conclusion, our results show the following. (1) They demonstrate that Cx30 is dispensable for hearing. In this light, what role does Cx30 play in the cochlea? First, Cx30 itself modulates the expression of Cx26. Indeed, reexpression of Cx30 in organotypic Cx30^−/− cochlear cultures partially restores Cx26 expression (Ortolano et al., 2008). Second, Cx26 and Cx30 coassemble in gap junctions in the cochlea, and their coexpression results in channels with dye-transfer properties that differ from those of homomeric Cx26 channels (Forge et al., 2003). Therefore, deafness linked to the three known Cx30 dominant missense mutations (Grifa et al., 1999; Nemoto-Hasebe et al., 2009; Wang et al., 2011) might reflect functional alterations in heteromeric channels comprising normal Cx26 and mutated Cx30. (2) Our results indicate that defective Cx26 expression is the likely cause of hearing loss previously reported in Cx30^−/− mice, which may explain why overexpression of Cx26 rescues hearing in these mice (Ahmad et al., 2007). Similarly, the hearing loss of patients harboring one of the known large deletions encompassing GJB6 would be linked to abnormal expression of GJB2. In this case, the current digenic inheritance hypothesis including cooperative functions between Cx30 and Cx26 would no longer be tenable. Of note, Cx30^Δ/Δ and Cx30^+/− mice (Rodriguez et al., 2012) display normal hearing with Cx26 expression reduced by half. Thus, it remains to be determined at which level of Cx26 expression hearing is lost. (3) Our results explain why no GJB6 silencing mutations have been found in human deafness cases so far.