One of the most common disabilities in the US, hearing loss, is reported by The National Institutes of Health to affect approximately 36 million Americans. One of the major contributing factors to this loss in hearing is the loss of the sensory hair cells (HCs) within the cochlea. Also in the mammalian cochlea, six major groups of supporting cell (SC) subtypes reside in close proximity to HCs and may have the potential to regenerate HCs after damage. These subtypes include cells of the greater epithelial ridge, inner phalangeal/border cells, inner and outer pillar cells, Deiters’ cells, Hensen cells, and Claudius cells. During embryonic development, progenitor cells differentiate into HCs or one of the SC subtypes by Notch-mediated lateral inhibition. In the neonatal mouse cochlea, many studies have shown that inhibition of Notch signaling allows SCs to convert into HCs in both normal undamaged cochleae, as well as in drug-damaged cochlear explants. This mechanism is also implicated during spontaneous HC regeneration that occurs in non-mammalian vertebrates. We and others have recently observed that spontaneous HC regeneration can also occur in the neonatal mouse cochlea. However, little is known about the molecular mechanism or the SC subtypes which act as the source of regenerated HCs. In the neonatal mouse cochlea, HCs were killed in vivo at birth using a genetically-modified mouse model to express a toxin in HCs. Subsequently, SCs formed new HCs by either direct transdifferentiation, where no cell division occurred, or by mitotic regeneration. My dissertation investigated the role of Notch signaling in the ability of SC subtypes to regenerate HCs after damage. My central hypothesis is that after HC ablation is induced at birth, Notch signaling is partially eliminated and therefore lateral inhibition is lost in neonatal SCs in a subtype specific manner, which allows some SCs, but not others, to differentiate into and regenerate HCs. Aim 1 focused on changes in the Notch signaling pathway in response to HC damage during the window of spontaneous HC regeneration. Changes in the expression of genes in the Notch pathway were measured using real time qPCR, immunostaining, and in situ hybridization. The Notch effector HeyL was increased in the apical one-third of the cochlea while other Notch players are decreased. The most notable example is the Notch effector Hes5, which is directly responsible for inhibiting HC fate, and was reduced in outer pillar cells and Deiters’ cells, but not in other SC subtypes. From this we conclude that Notch signaling is reduced differentially among SC subtypes. In Aim 2 we investigated whether inhibition of Notch signaling is required for spontaneous HC regeneration to occur by maintaining active Notch signaling in all SCs in the context of HC damage. We hypothesized that maintaining active Notch signaling after HC damage will prevent SC-to-HC conversion thus preventing HC regeneration. We found significantly fewer regenerated HCs while maintaining Notch expression compared to controls with HC damage and no manipulation of Notch signaling. Therefore we conclude loss of Notch mediated lateral inhibition is required for the majority of spontaneous HC regeneration. In Aim 3 we investigated the ability of different SC subtypes to regenerate HCs by fate-mapping SC subtypes during the HC regeneration process. Since fate-mapping creates a permanent label in targeted cells, we can track their potential change in cell fate or reentry in the cell cycle after HC damage. We hypothesized that pillar cells and Deiters’ cells are the source for spontaneously regenerated HC within the neonatal mouse cochlea based on our results from Aim 1. We used three CreER mouse lines to fate-map distinct groups of SC subtypes during the HC damage and regeneration process. More pillar and Deiters’ cells regenerated HCs after damage than other SC populations. We found that outer pillar cells and Deiters’ cells are capable of downregulating the cell cycle inhibitor, p27Kip1, after HC damage. Therefore we investigated the ability of SC subtypes to mitotically regenerate HCs by including a mitotic tracer along with fate-mapping. A larger proportion of mitotically regenerated HCs came from pillar and Deiters’ cells. From these experiments, we conclude that outer pillar and Deiters’ cells are the source for the majority of spontaneously regenerated HCs in vivo. This knowledge will allow targeted investigation into outer pillar cells and Deiters’ cells that maintain regenerative plasticity at postnatal ages. Understanding how these cells change with age will inform efforts to induce HC regeneration in more mature cochleae. Additionally, understanding how Notch signaling regulates this regenerative plasticity will lead to the development of potential targets for the treatment of hearing loss.

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第20798号 / 医博第4298号 / 新制||医||1025(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 辻川 明孝, 教授 清水 章, 教授 戸口田 淳也 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

TRPA1 channels are sensors for noxious stimuli in a subset of nociceptive neurons. TRPA1 channels are also expressed in cells of the mammalian inner ear, but their function in this tissue remains unknown given that Trpa1–/– mice exhibit normal hearing, balance and sensory mechanotransduction. Here we show that non-sensory (supporting) cells of the hearing organ in the cochlea detect tissue damage via the activation of TRPA1 channels and subsequently modulate cochlear amplification through active cellshape changes. We found that cochlear supporting cells of wild type but not Trpa1–/– mice generate inward currents and robust long-lasting Ca2+ responses after stimulation with TRPA1 agonists. These Ca2+ responses often propagated between different types of supporting cells and were accompanied by prominent tissue displacements. The most prominent shape changes were observed in pillar cells which here we show possess Ca2+-dependent contractile machinery. Increased oxidative stress following acoustic overstimulation leads to the generation of lipid peroxidation byproducts such as 4-hydroxynonenal (4-HNE) that could directly activate TRPA1. Therefore, we exposed mice to mild noise and found a longer-lasting inhibition of cochlear amplification in wild type than in Trpa1–/– mice. Our results suggest that TRPA1-dependent changes in pillar cell shape can alter the tissue geometry and affect cochlear amplification. We believe this novel mechanism of cochlear regulation may protect or fine-tune the organ of Corti after noise exposure or other cochlear injuries.

A causa mais frequente de surdez de herança autossômica recessiva são as mutações no lócus DFNB1, onde estão os genes GJB2 e GJB6. Dentre os indivíduos com deficiência auditiva associada a esse lócus, 10% a 50% apresentam uma única mutação recessiva no gene GJB2, frequência muito superior à esperada em função da frequência de heterozigotos na população geral. Apesar de alguns desses casos terem sido elucidados após a identificação de grandes deleções no gene GJB6 ou nas suas proximidades, a existência de muitos indivíduos com uma única mutação patogênica no gene GJB2 sugere que a haplo-insuficiência nesse gene possa interagir com outras mutações no mesmo gene, no gene GJB6 vizinho, ou até em outros genes. O objetivo desse estudo foi identificar novos alelos patogênicos, novas proteínas e novos genes que interagem com o lócus DFNB1, do ponto de vista molecular e celular, e que possam ser responsáveis por surdez de herança autossômica recessiva. Desse modo, pretendemos contribuir para o esclarecimento da patogênese da surdez de herança autossômica recessiva. Nesse trabalho, três tipos de estudos foram realizados, com metodologias próprias. Na primeira parte, buscamos identificar novos alelos patogênicos no lócus DFNB1 que poderiam ser responsáveis por surdez quando presentes em heterozigose composta com outros alelos patogênicos nos genes GJB2 e GJB6. Foi realizada a análise do DNA de 16 pacientes surdos portadores de uma única mutação patogênica em um desses dois genes por meio: (i) do sequenciamento das regiões codificadora, promotora e doadora de splicing (intron 1) do gene GJB2, (ii) da triagem de uma deleção de 200 kb localizada a 130 kb da proximidade distal da região 5\' do gene GJB6 e (iii) da pesquisa de variações no número de cópias de um ou mais exons dos genes GJB2, GJB6, GJB3 e WFS1 por MLPA (Multiplex Ligation-dependent Probe Amplification). Detectamos uma segunda mutação provavelmente patogênica em dois dos 16 pacientes heterozigotos: em um deles, a mutação p.L76P (c.C227T) foi identificada na região de código do gene GJB2 e foi por nós descrita pela primeira vez; no segundo caso, uma duplicação (0,4-1,2Kb) que inclui a região de código do gene GJB2 foi detectada, também inédita na literatura. Na segunda parte, tivemos como objetivo obter um modelo experimental para estudos funcionais in vitro da proteína codificada pelo gene GJB2, a conexina 26, em seu local de expressão que são as células de suporte do órgão de Corti. Padronizamos o cultivo in vitro de células progenitoras do órgão de Corti de camundongos e de cobaias e conseguimos obter a diferenciação in vitro das otoesferas dos camundongos em células que expressam marcadores de células ciliadas (Miosina VIIa e Jagged2) e de células de suporte (p27kip e Jagged1). Por fim, na terceira parte, buscamos por proteínas que interagem com a conexina 26 por meio de ensaios de precipitação por afinidade. Para isso, produzimos clones recombinantes de uma proteína de fusão GST-Cx26 e de uma proteína controle (GST), e realizamos sua expressão in vitro em bactérias E.coli B21. Ensaios de precipitação por afinidade entre a proteína de fusão GST-Cx26 ou GST sozinha e proteínas extraídas de cérebro ou fígado de camundongos foram realizados em diferentes condições. A identificação e a análise das proteínas presentes em bandas de SDS-PAGE, obtidas no ensaio de precipitação com a proteína de fusão GST-Cx26 e ausentes no ensaio com a GST, foram realizadas por espectrometria de massas. Identificamos um total de 49 proteínas candidatas a interagirem com a região C-terminal da Cx26. Realizamos diversas análises in silico e em literatura específica e após exclusão de candidatas por: (i) redundância de representação no ensaio GST-Cx26, (ii) diferença entre a massa molecular esperada e a obtida, (iii) precipitação inespecífica e (iv) localização subcelular incompatível com a conexina 26, selecionamos um total de 22 proteínas candidatas a interagirem com a região C-terminal da conexina 26, para estudos futuros. A confimação da interação entre essas 22 proteínas e a conexina 26 é desejável por meio de estudos de co-localização e imuno-coprecipitação / The most frequent causes of nonsyndromic recessive hearing loss are mutations in locus DFNB1, in the GJB2 and GJB6 genes. Among the individuals with hearing loss with mutations in this locus, 10% to 50% present a single recessive mutation in the GJB2 gene, frequency much higher than expected taking into account the frequency of heterozygotes in the general population. Although some of these cases have been elucidated after the identification of large deletions in GJB6 or its surrounding regions, the existence of many individuals with a single pathogenic mutation in the GJB2 gene suggests that haplo-insufficiency of this gene may interact with other types of mutations in the same gene, in the neighbor gene GJB6, or even in other genes. The aim of this study was to identify new pathogenic alleles, proteins and genes that interact with the locus DFNB1, from the molecular and cellular perspective, and that may be responsible for autosomal recessive deafness. Thus, we aimed to contribute to the understanding of the pathogenesis of autosomal recessive deafness. In this work, three different types of studies were performed, each one with a particular methodology. In the first part, we searched for new pathogenic alleles in the locus DFNB1 that could be responsible for deafness, when present in compound heterozygosis with other pathogenic alleles in GJB2 and GJB6 genes. We performed DNA analysis in samples from 16 deaf patients, carriers of a single pathogenic mutation in one of these two genes by: (i) sequencing the coding, promoter and splice donor (intron 1) regions of the GJB2 gene, (ii) screening for a deletion of 200 kb located 130 kb upstream from GJB6 gene and (iii) investigating copy number variations in of one or more exons of the genes GJB2, GJB6, GJB3 and WFS1 by MLPA (Multiplex Ligation-dependent Probe Amplification). We detected a second mutation, probably pathogenic, in two of the 16 heterozygous patients: in one case, the p.L76P (c.C227T) mutation was identified in the coding region of the GJB2 gene and was firstly described by us; in the second case, a novel duplication (0.4 - 1.2 Mb) that includes the coding region of the GJB2 gene was detected. In the second part, our objective was to obtain an experimental model for in vitro functional studies of the protein encoded by the GJB2 gene, connexin 26, in its site of expression, that is, in the supporting cells of the organ of Corti. We standardized the culturing of guinea pigs and mice progenitor cells of organ of Corti. We were also able to induce differentiation of mice\'s otospheres into cells that express markers of hair (myosin VIIa and Jagged2) and supporting cells (p27kip and Jagged1). Finally, we searched for connexin 26 interacting proteins by pull-down assays. Recombinant clones expressing a fusion protein GST-Cx26 and a control protein (GST) were produced, so that in vitro expression in E. coli B21 could be performed. Pull-down experiments, perfomed with fusion protein GST-Cx26 or GST alone, and with proteins from mice brain or liver extracts were done under several different conditions. The identification and analysis of proteins present in SDS-PAGE bands in experiments performed with the fusion protein GST-Cx26, and absent in the GST assay, were performed by mass spectrometry. We identified a total of 49 candidate proteins for interaction with the C-terminal region of Cx26. In silico analyses performed in several databases and search in the literature allowed exclusion of candidates by: (i) redundancy of representation in the GST-Cx26 experiments; (ii) discrepancy between the expected and the obtained molecular weight; (iii) nonspecific precipitation and (iv) subcellular localization incompatible with connexin 26 localization. Summing up, we selected a total of 22 candidate proteins to interact with the C-terminal region of connexin 26. Confirmation of the interaction between these proteins and connexin 26 is planned to be performed by co-localization studies and by immuno-coprecipitation