Large-scale annotated dataset for cochlear hair cell detection and classification.

Our sense of hearing is mediated by cochlear hair cells, of which there are two types organized in one row of inner hair cells and three rows of outer hair cells. Each cochlea contains 5-15 thousand terminally differentiated hair cells, and their survival is essential for hearing as they do not regenerate after insult. It is often desirable in hearing research to quantify the number of hair cells within cochlear samples, in both pathological conditions, and in response to treatment. Machine learning can be used to automate the quantification process but requires a vast and diverse dataset for effective training. In this study, we present a large collection of annotated cochlear hair-cell datasets, labeled with commonly used hair-cell markers and imaged using various fluorescence microscopy techniques. The collection includes samples from mouse, rat, guinea pig, pig, primate, and human cochlear tissue, from normal conditions and following in-vivo and in-vitro ototoxic drug application. The dataset includes over 107,000 hair cells which have been identified and annotated as either inner or outer hair cells. This dataset is the result of a collaborative effort from multiple laboratories and has been carefully curated to represent a variety of imaging techniques. With suggested usage parameters and a well-described annotation procedure, this collection can facilitate the development of generalizable cochlear hair-cell detection models or serve as a starting point for fine-tuning models for other analysis tasks. By providing this dataset, we aim to give other hearing research groups the opportunity to develop their own tools with which to analyze cochlear imaging data more fully, accurately, and with greater ease.

PKHD1L1 is required for stereocilia bundle maintenance, durable hearing function and resilience to noise exposure.

Sensory hair cells of the cochlea are essential for hearing, relying on the mechanosensitive stereocilia bundle at their apical pole for their function. Polycystic Kidney and Hepatic Disease 1-Like 1 (PKHD1L1) is a stereocilia protein required for normal hearing in mice, and for the formation of the transient stereocilia surface coat, expressed during early postnatal development. While the function of the stereocilia coat remains unclear, growing evidence supports PKHD1L1 as a human deafness gene. In this study we carry out in depth characterization of PKHD1L1 expression in mice during development and adulthood, analyze hair-cell bundle morphology and hearing function in aging PKHD1L1-defficient mouse lines, and assess their susceptibility to noise damage. Our findings reveal that PKHD1L1-deficient mice display no disruption to bundle cohesion or tectorial membrane attachment-crown formation during development. However, starting from 6 weeks of age, PKHD1L1-defficient mice display missing stereocilia and disruptions to bundle coherence. Both conditional and constitutive PKHD1L1 knock-out mice develop high-frequency hearing loss progressing to lower frequencies with age. Furthermore, PKHD1L1-deficient mice are susceptible to permanent hearing loss following moderate acoustic overexposure, which induces only temporary hearing threshold shifts in wild-type mice. These results suggest a role for PKHD1L1 in establishing robust sensory hair bundles during development, necessary for maintaining bundle cohesion and function in response to acoustic trauma and aging.

PKHD1L1, a gene involved in the stereocilia coat, causes autosomal recessive nonsyndromic hearing loss.

Identification of genes associated with nonsyndromic hearing loss is a crucial endeavor given the substantial number of individuals who remain without a diagnosis after even the most advanced genetic testing. PKHD1L1 was established as necessary for the formation of the cochlear hair-cell stereociliary coat and causes hearing loss in mice and zebrafish when mutated. We sought to determine if biallelic variants in PKHD1L1 also cause hearing loss in humans. Exome sequencing was performed on DNA of four families segregating autosomal recessive nonsyndromic sensorineural hearing loss. Compound heterozygous p.[(Gly129Ser)];p.[(Gly1314Val)] and p.[(Gly605Arg)];p[(Leu2818TyrfsTer5)], homozygous missense p.(His2479Gln) and nonsense p.(Arg3381Ter) variants were identified in PKHD1L1 that were predicted to be damaging using in silico pathogenicity prediction methods. In vitro functional analysis of two missense variants was performed using purified recombinant PKHD1L1 protein fragments. We then evaluated protein thermodynamic stability with and without the missense variants found in one of the families and performed a minigene splicing assay for another variant. In silico molecular modeling using AlphaFold2 and protein sequence alignment analysis were carried out to further explore potential variant effects on structure. In vitro functional assessment indicated that both engineered PKHD1L1 p.(Gly129Ser) and p.(Gly1314Val) mutant constructs significantly reduced the folding and structural stabilities of the expressed protein fragments, providing further evidence to support pathogenicity of these variants. Minigene assay of the c.1813G>A p.(Gly605Arg) variant, located at the boundary of exon 17, revealed exon skipping leading to an in-frame deletion of 48 amino acids. In silico molecular modeling exposed key structural features that might suggest PKHD1L1 protein destabilization. Multiple lines of evidence collectively associate PKHD1L1 with nonsyndromic mild-moderate to severe sensorineural hearing loss. PKHD1L1 testing in individuals with mild-moderate hearing loss may identify further affected families.

PKHD1L1, A Gene Involved in the Stereocilia Coat, Causes Autosomal Recessive Nonsyndromic Hearing Loss.

Identification of genes associated with nonsyndromic hearing loss is a crucial endeavor given the substantial number of individuals who remain without a diagnosis after even the most advanced genetic testing. PKHD1L1 was established as necessary for the formation of the cochlear hair-cell stereociliary coat and causes hearing loss in mice and zebrafish when mutated. We sought to determine if biallelic variants in PKHD1L1 also cause hearing loss in humans. Exome sequencing was performed on DNA of four families segregating autosomal recessive nonsyndromic sensorineural hearing loss. Compound heterozygous p.[(Gly129Ser)];p.[(Gly1314Val)] and p.[(Gly605Arg)];p[(Leu2818TyrfsTer5)], homozygous missense p.(His2479Gln) and nonsense p.(Arg3381Ter) variants were identified in PKHD1L1 that were predicted to be damaging using in silico pathogenicity prediction methods. In vitro functional analysis of two missense variants was performed using purified recombinant PKHD1L1 protein fragments. We then evaluated protein thermodynamic stability with and without the missense variants found in one of the families and performed a minigene splicing assay for another variant. In silico molecular modelling using AlphaFold2 and protein sequence alignment analysis were carried out to further explore potential variant effects on structure. In vitro functional assessment indicated that both engineered PKHD1L1 p.(Gly129Ser) and p.(Gly1314Val) mutant constructs significantly reduced the folding and structural stabilities of the expressed protein fragments, providing further evidence to support pathogenicity of these variants. Minigene assay of the c.1813G>A p.(Gly605Arg) variant, located at the boundary of exon 17, revealed exon skipping leading to an in-frame deletion of 48 amino acids. In silico molecular modelling exposed key structural features that might suggest PKHD1L1 protein destabilization. Multiple lines of evidence collectively associate PKHD1L1 with nonsyndromic mild-moderate to severe sensorineural hearing loss. PKHD1L1 testing in individuals with mild-moderate hearing loss may identify further affected families.

Large-scale annotated dataset for cochlear hair cell detection and classification.

Our sense of hearing is mediated by cochlear hair cells, localized within the sensory epithelium called the organ of Corti. There are two types of hair cells in the cochlea, which are organized in one row of inner hair cells and three rows of outer hair cells. Each cochlea contains a few thousands of hair cells, and their survival is essential for our perception of sound because they are terminally differentiated and do not regenerate after insult. It is often desirable in hearing research to quantify the number of hair cells within cochlear samples, in both pathological conditions, and in response to treatment. However, the sheer number of cells along the cochlea makes manual quantification impractical. Machine learning can be used to overcome this challenge by automating the quantification process but requires a vast and diverse dataset for effective training. In this study, we present a large collection of annotated cochlear hair-cell datasets, labeled with commonly used hair-cell markers and imaged using various fluorescence microscopy techniques. The collection includes samples from mouse, human, pig and guinea pig cochlear tissue, from normal conditions and following in-vivo and in-vitro ototoxic drug application. The dataset includes over 90'000 hair cells, all of which have been manually identified and annotated as one of two cell types: inner hair cells and outer hair cells. This dataset is the result of a collaborative effort from multiple laboratories and has been carefully curated to represent a variety of imaging techniques. With suggested usage parameters and a well-described annotation procedure, this collection can facilitate the development of generalizable cochlear hair cell detection models or serve as a starting point for fine-tuning models for other analysis tasks. By providing this dataset, we aim to supply other groups within the hearing research community with the opportunity to develop their own tools with which to analyze cochlear imaging data more fully, accurately, and with greater ease.

Rescue of hearing by adenine base editing in a humanized mouse model of Usher syndrome type 1F.

Usher syndrome type 1F (USH1F), characterized by congenital lack of hearing and balance and progressive loss of vision, is caused by mutations in the PCDH15 gene. In the Ashkenazi population, a recessive truncation mutation accounts for a large proportion of USH1F cases. The truncation is caused by a single C→T mutation, which converts an arginine codon to a stop (R245X). To test the potential for base editors to revert this mutation, we developed a humanized Pcdh15R245X mouse model for USH1F. Mice homozygous for the R245X mutation were deaf and exhibited profound balance deficits, while heterozygous mice were unaffected. Here we show that an adenine base editor (ABE) is capable of reversing the R245X mutation to restore the PCDH15 sequence and function. We packaged a split-intein ABE into dual adeno-associated virus (AAV) vectors and delivered them into cochleas of neonatal USH1F mice. Hearing was not restored in a Pcdh15 constitutive null mouse despite base editing, perhaps because of early disorganization of cochlear hair cells. However, injection of vectors encoding the split ABE into a late-deletion conditional Pcdh15 knockout rescued hearing. This study demonstrates the ability of an ABE to correct the PCDH15 R245X mutation in the cochlea and restore hearing.

SKOOTS: Skeleton oriented object segmentation for mitochondria.

The segmentation of individual instances of mitochondria from imaging datasets is informative, yet time-consuming to do by hand, sparking interest in developing automated algorithms using deep neural networks. Existing solutions for various segmentation tasks are largely optimized for one of two types of biomedical imaging: high resolution three-dimensional (whole neuron segmentation in volumetric electron microscopy datasets) or two-dimensional low resolution (whole cell segmentation of light microscopy images). The former requires consistently predictable boundaries to segment large structures, while the latter is boundary invariant but struggles with segmentation of large 3D objects without downscaling. Mitochondria in whole cell 3D EM datasets often occupy the challenging middle ground: large with ambiguous borders, limiting accuracy with existing tools. To rectify this, we have developed skeleton oriented object segmentation (SKOOTS); a new segmentation approach which efficiently handles large, densely packed mitochondria. We show that SKOOTS can accurately, and efficiently, segment 3D mitochondria in previously difficult situations. Furthermore, we will release a new, manually annotated, 3D mitochondria segmentation dataset. Finally, we show this approach can be extended to segment objects in 3D light microscopy datasets. These results bridge the gap between existing segmentation approaches and increases the accessibility for three-dimensional biomedical image analysis.

Reprogramming by drug-like molecules leads to regeneration of cochlear hair cell-like cells in adult mice.

Strategies to overcome irreversible cochlear hair cell (HC) damage and loss in mammals are of vital importance to hearing recovery in patients with permanent hearing loss. In mature mammalian cochlea, co-activation of Myc and Notch1 reprograms supporting cells (SC) and promotes HC regeneration. Understanding of the underlying mechanisms may aid the development of a clinically relevant approach to achieve HC regeneration in the nontransgenic mature cochlea. By single-cell RNAseq, we show that MYC/NICD "rejuvenates" the adult mouse cochlea by activating multiple pathways including Wnt and cyclase activator of cyclic AMP (cAMP), whose blockade suppresses HC-like cell regeneration despite Myc/Notch activation. We screened and identified a combination (the cocktail) of drug-like molecules composing of small molecules and small interfering RNAs to activate the pathways of Myc, Notch1, Wnt and cAMP. We show that the cocktail effectively replaces Myc and Notch1 transgenes and reprograms fully mature wild-type (WT) SCs for HC-like cells regeneration in vitro. Finally, we demonstrate the cocktail is capable of reprogramming adult cochlea for HC-like cells regeneration in WT mice with HC loss in vivo. Our study identifies a strategy by a clinically relevant approach to reprogram mature inner ear for HC-like cells regeneration, laying the foundation for hearing restoration by HC regeneration.

Mini-PCDH15 gene therapy rescues hearing in a mouse model of Usher syndrome type 1F.

Usher syndrome type 1 F (USH1F), caused by mutations in the protocadherin-15 gene (PCDH15), is characterized by congenital deafness, lack of balance, and progressive blindness. In hair cells, the receptor cells of the inner ear, PCDH15 is a component of tip links, fine filaments which pull open mechanosensory transduction channels. A simple gene addition therapy for USH1F is challenging because the PCDH15 coding sequence is too large for adeno-associated virus (AAV) vectors. We use rational, structure-based design to engineer mini-PCDH15s in which 3-5 of the 11 extracellular cadherin repeats are deleted, but which still bind a partner protein. Some mini-PCDH15s can fit in an AAV. An AAV encoding one of these, injected into the inner ears of mouse models of USH1F, produces a mini-PCDH15 which properly forms tip links, prevents the degeneration of hair cell bundles, and rescues hearing. Mini-PCDH15s may be a useful therapy for the deafness of USH1F.

The hair cell analysis toolbox is a precise and fully automated pipeline for whole cochlea hair cell quantification.

Our sense of hearing is mediated by sensory hair cells, precisely arranged and highly specialized cells subdivided into outer hair cells (OHCs) and inner hair cells (IHCs). Light microscopy tools allow for imaging of auditory hair cells along the full length of the cochlea, often yielding more data than feasible to manually analyze. Currently, there are no widely applicable tools for fast, unsupervised, unbiased, and comprehensive image analysis of auditory hair cells that work well either with imaging datasets containing an entire cochlea or smaller sampled regions. Here, we present a highly accurate machine learning-based hair cell analysis toolbox (HCAT) for the comprehensive analysis of whole cochleae (or smaller regions of interest) across light microscopy imaging modalities and species. The HCAT is a software that automates common image analysis tasks such as counting hair cells, classifying them by subtype (IHCs versus OHCs), determining their best frequency based on their location along the cochlea, and generating cochleograms. These automated tools remove a considerable barrier in cochlear image analysis, allowing for faster, unbiased, and more comprehensive data analysis practices. Furthermore, HCAT can serve as a template for deep learning-based detection tasks in other types of biological tissue: With some training data, HCAT's core codebase can be trained to develop a custom deep learning detection model for any object on an image.

Electron Microscopy Techniques for Investigating Structure and Composition of Hair-Cell Stereociliary Bundles.

Hair cells-the sensory cells of the vertebrate inner ear-bear at their apical surfaces a bundle of actin-filled protrusions called stereocilia, which mediate the cells' mechanosensitivity. Hereditary deafness is often associated with morphological disorganization of stereocilia bundles, with the absence or mislocalization within stereocilia of specific proteins. Thus, stereocilia bundles are closely examined to understand most animal models of hereditary hearing loss. Because stereocilia have a diameter less than a wavelength of light, light microscopy is not adequate to reveal subtle changes in morphology or protein localization. Instead, electron microscopy (EM) has proven essential for understanding stereocilia bundle development, maintenance, normal function, and dysfunction in disease. Here we review a set of EM imaging techniques commonly used to study stereocilia, including optimal sample preparation and best imaging practices. These include conventional and immunogold transmission electron microscopy (TEM) and scanning electron microscopy (SEM), as well as focused-ion-beam scanning electron microscopy (FIB-SEM), which enables 3-D serial reconstruction of resin-embedded biological structures at a resolution of a few nanometers. Parameters for optimal sample preparation, fixation, immunogold labeling, metal coating and imaging are discussed. Special attention is given to protein localization in stereocilia using immunogold labeling. Finally, we describe the advantages and limitations of these EM techniques and their suitability for different types of studies.

Mechanical overstimulation causes acute injury and synapse loss followed by fast recovery in lateral-line neuromasts of larval zebrafish.

Excess noise damages sensory hair cells, resulting in loss of synaptic connections with auditory nerves and, in some cases, hair-cell death. The cellular mechanisms underlying mechanically induced hair-cell damage and subsequent repair are not completely understood. Hair cells in neuromasts of larval zebrafish are structurally and functionally comparable to mammalian hair cells but undergo robust regeneration following ototoxic damage. We therefore developed a model for mechanically induced hair-cell damage in this highly tractable system. Free swimming larvae exposed to strong water wave stimulus for 2 hr displayed mechanical injury to neuromasts, including afferent neurite retraction, damaged hair bundles, and reduced mechanotransduction. Synapse loss was observed in apparently intact exposed neuromasts, and this loss was exacerbated by inhibiting glutamate uptake. Mechanical damage also elicited an inflammatory response and macrophage recruitment. Remarkably, neuromast hair-cell morphology and mechanotransduction recovered within hours following exposure, suggesting severely damaged neuromasts undergo repair. Our results indicate functional changes and synapse loss in mechanically damaged lateral-line neuromasts that share key features of damage observed in noise-exposed mammalian ear. Yet, unlike the mammalian ear, mechanical damage to neuromasts is rapidly reversible.

Norrie disease protein is essential for cochlear hair cell maturation.

Mutations in the gene for Norrie disease protein (Ndp) cause syndromic deafness and blindness. We show here that cochlear function in an Ndp knockout mouse deteriorated with age: At P3-P4, hair cells (HCs) showed progressive loss of Pou4f3 and Gfi1, key transcription factors for HC maturation, and Myo7a, a specialized myosin required for normal function of HC stereocilia. Loss of expression of these genes correlated to increasing HC loss and profound hearing loss by 2 mo. We show that overexpression of the Ndp gene in neonatal supporting cells or, remarkably, up-regulation of canonical Wnt signaling in HCs rescued HCs and cochlear function. We conclude that Ndp secreted from supporting cells orchestrates a transcriptional network for the maintenance and survival of HCs and that increasing the level of ß-catenin, the intracellular effector of Wnt signaling, is sufficient to replace the functional requirement for Ndp in the cochlea.

Serial scanning electron microscopy of anti-PKHD1L1 immuno-gold labeled mouse hair cell stereocilia bundles.

Serial electron microscopy techniques have proven to be a powerful tool in biology. Unfortunately, the data sets they generate lack robust and accurate automated segmentation algorithms. In this data descriptor publication, we introduce a serial focused ion beam scanning electron microscopy (FIB-SEM) dataset consisting of six outer hair cell (OHC) stereocilia bundles, and the supranuclear part of the hair cell bodies. Also presented are the manual segmentations of stereocilia bundles and the gold bead labeling of PKHD1L1, a coat protein of hair cell stereocilia important for hearing in mice. This depository includes all original data and several intermediate steps of the manual analysis, as well as the MATLAB algorithm used to generate a three-dimensional distribution map of gold labels. They serve as a reference dataset, and they enable reproduction of our analysis, evaluation and improvement of current methods of protein localization, and training of algorithms for accurate automated segmentation.

Renewed proliferation in adult mouse cochlea and regeneration of hair cells.

The adult mammalian inner ear lacks the capacity to divide or regenerate. Damage to inner ear generally leads to permanent hearing loss in humans. Here, we present that reprogramming of the adult inner ear induces renewed proliferation and regeneration of inner ear cell types. Co-activation of cell cycle activator Myc and inner ear progenitor gene Notch1 induces robust proliferation of diverse adult cochlear sensory epithelial cell types. Transient MYC and NOTCH activities enable adult supporting cells to respond to transcription factor Atoh1 and efficiently transdifferentiate into hair cell-like cells. Furthermore, we uncover that mTOR pathway participates in MYC/NOTCH-mediated proliferation and regeneration. These regenerated hair cell-like cells take up the styryl dye FM1-43 and are likely to form connections with adult spiral ganglion neurons, supporting that Myc and Notch1 co-activation is sufficient to reprogram fully mature supporting cells to proliferate and regenerate hair cell-like cells in adult mammalian auditory organs.

PKHD1L1 is a coat protein of hair-cell stereocilia and is required for normal hearing.

The bundle of stereocilia on inner ear hair cells responds to subnanometer deflections produced by sound or head movement. Stereocilia are interconnected by a variety of links and also carry an electron-dense surface coat. The coat may contribute to stereocilia adhesion or protect from stereocilia fusion, but its molecular identity remains unknown. From a database of hair-cell-enriched translated proteins, we identify Polycystic Kidney and Hepatic Disease 1-Like 1 (PKHD1L1), a large, mostly extracellular protein of 4249 amino acids with a single transmembrane domain. Using serial immunogold scanning electron microscopy, we show that PKHD1L1 is expressed at the tips of stereocilia, especially in the high-frequency regions of the cochlea. PKHD1L1-deficient mice lack the surface coat at the upper but not lower regions of stereocilia, and they develop progressive hearing loss. We conclude that PKHD1L1 is a component of the surface coat and is required for normal hearing in mice.

Gene Transfer with AAV9-PHP.B Rescues Hearing in a Mouse Model of Usher Syndrome 3A and Transduces Hair Cells in a Non-human Primate.

Hereditary hearing loss often results from mutation of genes expressed by cochlear hair cells. Gene addition using AAV vectors has shown some efficacy in mouse models, but clinical application requires two additional advances. First, new AAV capsids must mediate efficient transgene expression in both inner and outer hair cells of the cochlea. Second, to have the best chance of clinical translation, these new vectors must also transduce hair cells in non-human primates. Here, we show that an AAV9 capsid variant, PHP.B, produces efficient transgene expression of a GFP reporter in both inner and outer hair cells of neonatal mice. We show also that AAV9-PHP.B mediates almost complete transduction of inner and outer HCs in a non-human primate. In a mouse model of Usher syndrome type 3A deafness (gene CLRN1), we use AAV9-PHP.B encoding Clrn1 to partially rescue hearing. Thus, we have identified a vector with promise for clinical treatment of hereditary hearing disorders, and we demonstrate, for the first time, viral transduction of the inner ear of a primate with an AAV vector.

Mechanotransduction current is essential for stability of the transducing stereocilia in mammalian auditory hair cells.

Mechanotransducer channels at the tips of sensory stereocilia of inner ear hair cells are gated by the tension of 'tip links' interconnecting stereocilia. To ensure maximal sensitivity, tip links are tensioned at rest, resulting in a continuous influx of Ca2+ into the cell. Here, we show that this constitutive Ca2+ influx, usually considered as potentially deleterious for hair cells, is in fact essential for stereocilia stability. In the auditory hair cells of young postnatal mice and rats, a reduction in mechanotransducer current, via pharmacological channel blockers or disruption of tip links, leads to stereocilia shape changes and shortening. These effects occur only in stereocilia that harbor mechanotransducer channels, recover upon blocker washout or tip link regeneration and can be replicated by manipulations of extracellular Ca2+ or intracellular Ca2+ buffering. Thus, our data provide the first experimental evidence for the dynamic control of stereocilia morphology by the mechanotransduction current.

Gene therapy restores auditory and vestibular function in a mouse model of Usher syndrome type 1c.

Because there are currently no biological treatments for hearing loss, we sought to advance gene therapy approaches to treat genetic deafness. We focused on Usher syndrome, a devastating genetic disorder that causes blindness, balance disorders and profound deafness, and studied a knock-in mouse model, Ush1c c.216G>A, for Usher syndrome type IC (USH1C). As restoration of complex auditory and balance function is likely to require gene delivery systems that target auditory and vestibular sensory cells with high efficiency, we delivered wild-type Ush1c into the inner ear of Ush1c c.216G>A mice using a synthetic adeno-associated viral vector, Anc80L65, shown to transduce 80-90% of sensory hair cells. We demonstrate recovery of gene and protein expression, restoration of sensory cell function, rescue of complex auditory function and recovery of hearing and balance behavior to near wild-type levels. The data represent unprecedented recovery of inner ear function and suggest that biological therapies to treat deafness may be suitable for translation to humans with genetic inner ear disorders.

Rescue of Hearing by Gene Delivery to Inner-Ear Hair Cells Using Exosome-Associated AAV.

Adeno-associated virus (AAV) is a safe and effective vector for gene therapy for retinal disorders. Gene therapy for hearing disorders is not as advanced, in part because gene delivery to sensory hair cells of the inner ear is inefficient. Although AAV transduces the inner hair cells of the mouse cochlea, outer hair cells remain refractory to transduction. Here, we demonstrate that a vector, exosome-associated AAV (exo-AAV), is a potent carrier of transgenes to all inner ear hair cells. Exo-AAV1-GFP is more efficient than conventional AAV1-GFP, both in mouse cochlear explants in vitro and with direct cochlear injection in vivo. Exo-AAV shows no toxicity in vivo, as assayed by tests of auditory and vestibular function. Finally, exo-AAV1 gene therapy partially rescues hearing in a mouse model of hereditary deafness (lipoma HMGIC fusion partner-like 5/tetraspan membrane protein of hair cell stereocilia [Lhfpl5/Tmhs-/-]). Exo-AAV is a powerful gene delivery system for hair cell research and may be useful for gene therapy for deafness.