Molecular heterogeneity of the non-human primate cochlea

Chen, Xin; Che, Yuwei; Qi, Jieyu; Cen, Ming; Gao, Shan; Zhu, Bin; Ao, Yiheng; Ma, Xiangyu; Cheng, Cheng; Guo, Rongrong; Wang, Xinlin; Shen, Tian; Wu, Jiheng; Kong, Fanliang; Si, Wei; Chen, Yongchang; Tan, Tao; Lu, Ling; Chai, Renjie

doi:10.1038/s41467-026-68350-2

Download PDF

Article
Open access
Published: 17 January 2026

Molecular heterogeneity of the non-human primate cochlea

Xin Chen^1,2^na1,
Yuwei Che¹^na1,
Jieyu Qi^3,4^na1,
Ming Cen¹^na1,
Shan Gao¹^na1,
Bin Zhu¹,
Yiheng Ao¹,
Xiangyu Ma ORCID: orcid.org/0000-0002-5657-1991¹,
Cheng Cheng^5,6,
Rongrong Guo^7,8,
Xinlin Wang¹,
Tian Shen⁹,
Jiheng Wu ORCID: orcid.org/0009-0001-6519-7233¹,
Fanliang Kong¹,
Wei Si ORCID: orcid.org/0000-0003-0861-0682^7,8,
Yongchang Chen ORCID: orcid.org/0000-0002-3856-3576^7,8,
Tao Tan ORCID: orcid.org/0000-0001-8650-0388^7,8,
Ling Lu¹ &
…
Renjie Chai ORCID: orcid.org/0000-0002-3885-543X^1,2,3,10,11

Nature Communications volume 17, Article number: 1633 (2026) Cite this article

4936 Accesses
1 Citations
7 Altmetric
Metrics details

Subjects

Abstract

Unraveling the intricate composition and function of the cochlea is paramount to comprehending the mechanisms underlying sound perception and the pathogenesis of auditory disorders. The mammalian cochlea displays a highly organized structure, which contributes to the diversity and complexity of auditory processing. However, the cellular intricacies in non-human primates remain largely unexplored. In the present study, we employed high-throughput transcriptomic sequencing to profile over 36,701 nuclei across virtually all cochlear cell types in both juvenile and adult Macaca fascicularis at single-cell resolution. Our analysis unveiled remarkable heterogeneity both across and within cell types. Despite a largely conserved cellular composition of the cochlea between mouse and macaque species, glial cells exhibited substantial species-specific diversity, while hair cells and spiral ganglion neurons with specialized transcriptional programs were well-mapped onto their murine counterparts, underscoring the similarities that persist despite evolutionary divergence. Furthermore, we constructed a disease map associated with hearing loss, establishing this transcriptomic atlas of the macaque cochlea as an indispensable resource for future investigations in both human and non-human primates.

Single-cell profiling identifies hair cell SLC35F1 deficiency as a signature of primate cochlear aging

Article 20 June 2025

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

Article Open access 23 April 2024

Molecular logic for cellular specializations that initiate the auditory parallel processing pathways

Article Open access 09 January 2025

Introduction

The cochlea, a coiled structure situated in the ventral region of the inner ear, performs unique biophysical functions in vertebrates and serves as the primary sound transducer. Airborne pressure waves induce fluid-based traveling waves at the base of the cochlea, which propagate towards the apex within the scala vestibuli. These waves progressively vibrate the spiral duct in the scala media and subsequently travel back towards the base in the scala tympani¹. During this complex fluctuation process, the sensory epithelium lining the floor of the scala media, known as the cochlear epithelium, perceives mechanical stimuli and transduces physical signals to neurons². The cochlear epithelium comprises two types of mechanosensory hair cells (HCs), namely inner HCs and outer HCs, and at least six types of associated glia-like supporting cells (SCs). The scarcity of HCs and SCs in mammals, with approximately 700 inner HCs and 2000 outer HCs in mice³, and around 15,000 HCs in the human cochlea³, poses significant challenges in investigating the molecular mechanisms underlying the development and physiology of this organ. Furthermore, the vulnerability and sensitivity of cochlear cells, particularly those in the epithelial tissue, coupled with the calcification of the temporal bone in juvenile and adult stages, render cochlear dissection an exceptionally arduous task.

Damage or degeneration of the cochlea can lead to irreversible HC loss in mammals^4,5, with hereditary hearing loss (HL) being the most common disabling sensorineural disorder. The majority of genes known to be associated with deafness are expressed in specific cochlear cell types, such as HCs, and these genes often affect the mechanoelectrical transduction components in the hair bundles that are critical for sound perception, including Cib2 (calcium and integrin-binding family member 2), Tmie (transmembrane inner ear expressed protein), and Tmc1/2 (transmembrane channel-like protein 1/2). Other genes, such as GJB2 (connexin-26), are expressed in several types of SCs⁶. However, a comprehensive disease map of deafness genes in non-human primates remains elusive. Gene therapy strategies, primarily based on the adeno-associated virus delivery system, are increasingly being tested in mouse models that mimic human deafness. Nevertheless, it is imperative to optimize the safety and efficacy of gene therapies in larger animal models, such as non-human primates, for advanced preclinical trials due to species-specific differences in physiology and anatomy between mice and humans.

Over the past two decades, developmental biologists have made significant strides in uncovering the cellular and genetic processes that regulate the formation of this remarkable structure, including the major axes of cochlear development^7,8, HC fate determination and trans-differentiation^9,10, and the identification and function of the mechanoelectrical transduction complex^11,12. In rodents, the sensory epithelium and spiral ganglion neurons (SGNs) remain immature prior to and at birth, undergoing remodeling and maturation around postnatal days (P) 21^13,14, concurrent with the regression of the greater epithelial ridge, detachment of the tectorial membrane, and the formation of the endocochlear potential^15,16. Recent advances in microfluidic-based single-cell/single-nucleus transcriptomic approaches, which enable the capture of large numbers of cells, have facilitated the development of cell atlases of the mouse cochlea from embryonic to adult stages^17,18,19. However, a comprehensive cell transcriptomic atlas of the cochlea in non-human primates has yet to be undertaken²⁰. Given their evolutionary proximity to humans, constructing a non-human primate cell atlas as a reference for the cochlea is a critical step in advancing our understanding of human auditory physiology, disease, and aging.

In the present study, we utilized juvenile and adult Macaca fascicularis cochleae to generate a high-quality single-nucleus transcriptomic atlas. This map revealed remarkable cellular diversity in the spiral ganglion region, particularly among glial cells. Furthermore, we identified conserved and divergent features between the mouse and macaque cochlea. We anticipate that this atlas will be invaluable for understanding evolutionary innovations in the mammalian cochlea and for developing organoid models and gene therapy of human diseases associated with deafness.

Results

A cell type atlas of the macaque cochlea

To determine the cellular composition of the macaque cochlea, we performed single-nucleus RNA sequencing (snRNA-seq; Droplet-based 10× Genomics) on 36,701 nuclei after quality control, doublet removal, and filtering (Fig. 1a–d and Supplementary Fig. 1a–c; for details see “Method”). We profiled the entire cochlea, with nuclei sampled from the cochlear epithelium, spiral ganglion, stria vascularis, spiral ligament, and other tissues, including macrophages and pericytes, in a 1-, 5-, and 11-year-old macaques across approximately juvenile and adult stages (Supplementary Figs. 1–3b, c).

**Fig. 1: A single-cell transcriptomic atlas of the macaque cochlea.**

To fine-tune the characterization of macaque cochlear cell types, we first utilized the published single-nucleus and single-cell transcriptome of murine cochleae as a reference for assessing well-established marker genes used for manual cell-type annotation (Supplementary Fig. 4). These gene markers used in the referenced dataset have been thoroughly validated in the murine cochlea across juvenile and adult periods¹⁹. We found that our in-house single-nucleus sequencing data of the murine cochlea at P25 exhibited approximately similar cellular diversity comparable to the referenced dataset from Jean et al.¹⁹ (Supplementary Figs. 5 and 6) and observed accurate cell-type mapping onto our macaque cochleae (Fig. 1d, e and Supplementary Figs. 1d, e, g, h and 2a, b). However, some differences exist as circulating cells including erythrocytes, neutrophils, B cells, T cells, NK cells, monocytes, and mast cells are undetectable in our in-house dataset compared to Jean et al’s dataset¹⁹ (Fig. 1d and Supplementary Figs. 1, 2 and 5). We reasoned that it is likely due to cochlear dissection process for single-nucleus suspension we used, which was different from Jean et al.¹⁹.

By using a support vector machine classifier trained on the macaque cells, we find strong correspondence between the transcriptomic signatures of the sensory organ cell types in macaques and mice underscoring a highly conserved molecular identity between macaque and murine species (Supplementary Fig. 3d), as well as high quality and accuracy within these cell types (Supplementary Figs. 1f, i, 2c, and 5c, g). Moreover, our atlas also displays five main cell-type ensembles (Supplementary Fig. 3c) as noted previously¹⁹. The cell ensembles of the neurosensory epithelium, for instance, consist of SCs (defined by OTOG expression) and HCs (defined by MYO7A expression; Supplementary Fig. 3c). For further validation of these sequencing data, we applied immunofluorescence assays to confirm the localization of selectively the most differentially expressed genes. CALB1 encoding calbindin, a hair cell marker⁹, is localized in both inner and outer HCs. Otoferlin (OTOF), an inner HC marker²¹, is exclusively present in inner HCs when HCs become mature. CALB2 encoding calretinin, a calcium-binding protein, is illustrated as localized in SGNs^22,23,24. All these immunolabeling experiments also exhibited similar cell-type-specific localization in both species (Fig. 1f). Furthermore, RNAscope^TM in situ hybridization also confirm the spatial expression patterns of ANXA1 in root cells and spindle cells as well as OTOGL in HCs and SCs (Supplementary Fig. 3a).

Furthermore, we were able to map the unique biological functions of each cell type via gene ontology (GO) enrichment analysis of these highly variable genes (Supplementary Data 1), revealing remarkably similar biological functions across species. For example, sensory processing of sound mapped to HCs, actin filament-based processes to SCs, modulation of chemical synaptic transmission to SGNs, myelination to Schwann cells, and cell junction organization to marginal stria cells. Collectively, we constructed an accessible cellular taxonomy of the juvenile and adult macaque cochlea, revealing an evolutionarily conserved, highly specialized sensory organ across mammalian species based on a clustering analysis for identifying cell types of cochlear cells using the expression of canonical marker genes in mice.

Auditory receptors in macaque cochlea

HCs with unique mechanical sensory properties in the auditory system play a crucial role in sound transduction. To gain comprehensive insights into their molecular identities compared to mouse species, we integrated our macaque and mouse single-nucleus data separately using the single-cell variational inference (scVI) method²⁵ (Supplementary Fig. 6) and subset the HC population based on the unique expression of the HC marker MYO7A/Myo7a (Fig. 2a and Supplementary Figs. 5 and 6). We observed a strong correspondence between the transcriptomic signatures of the HC lineages in macaques and mice, as mature/terminal marker genes^26,27,28 (OHC: SLC26A5/Slc26a5, KCNQ4/Kcnq4 and PCP4/Ocm; IHC: SLC17A8/Slc17a8, CALB2/Calb2, and ATP2A3/Atp2a3) are selectively enriched in these two subtypes (Fig. 2b and Supplementary Fig. 7). The motor protein PRESTIN was precisely localized in the OHC basolateral membrane in both species (Fig. 2b, c and Supplementary Fig. 7) and OHCs in macaque species exhibited the biophysical feature of motility evoked by electrical stimuli (Fig. 2d and Movie S1). Furthermore, well-characterized transcriptional regulators that specify and maintain the fate of OHCs and IHCs, including IKZF2⁹ (which encodes IKAROS family zinc finger 2) and TBX2^10,29 (which encodes T-box transcription factor 2), were strongly expressed in HCs across species (Fig. 2b and Supplementary Fig. 7). These findings suggest that HCs in the cochlea of both species might share conserved gene regulatory machinery.

**Fig. 2: Characterization of auditory receptors in the macaque cochlea.**

Glial cell diversity in the macaque cochlea

In the mouse cochlea, glial cells exhibit spatial and functional heterogeneity and can be categorized into three distinct subtypes (Supplementary Fig. 8): satellite glial cells (SGSs), myelinating Schwann cells (MSCs), and non-myelinating Schwann cells (NMSCs). Unique marker genes for these subtypes have been well-established in mice³⁰. To investigate this diversity in the macaque cochlea, we sub-clustered glial cells into three subtypes (Fig. 3a and Supplementary Fig. 8c). Similar to murine glial types, macaque glial cells displayed region-specific differences in the spiral ganglion region, including cells enveloping sensory neurons (SGCs), cells wrapping around axons forming myelin (MSCs), and Remak bundles (NMSCs), as demonstrated by immunostaining of the pan-glial marker MBP for glial localization (Fig. 3b, c). However, we were unable to map macaque glial subtypes using several defined markers (SCN7A, encoding an atypical Na-sensitive Na-channel that marks NMSCs^30,31; NCAMP, encoding a non-compact myelin-associated protein that marks MSCs^30,32; and IGFBP6, encoding the insulin-like growth factor binding protein 6 that marks SGCs³⁰) that have been validated in mice³⁰ (Fig. 3d and Supplementary Fig. 8b and d). To further harmonize both species’ glial subtypes, we changed different highly variable genes (hvgs) imputed for anchoring and found that varying the number of hvgs did not affect the species or cell-type mixings in the manifold embedding spaces, and we still obtained a relatively stable high-level structure for the data manifold (Supplementary Fig. 9a–c). This might suggest that there is relatively high transcriptomic divergence in peripheral glial cell types between macaques and mice. Indeed, these cluster pairs showed relatively low classification accuracy with various cells showing overlaps (Supplementary Fig. 9d) as well as cell mixing upon multiple integrative method we used for cross-species integration (Supplementary Fig. 10). For example, we found that the pan-astrocyte marker GFAP showed region-specific localization in mouse MSCs but was more diverse in macaque glia cells, including cells engulfing SGNs and cells parallel to axons (Fig. 3e).

**Fig. 3: Diversity of glial cells in the primate auditory peripheral system.**

In the central nervous system (CNS), astrocytes play auxiliary roles in neuron development, synapse formation and elimination, potassium homeostasis, and blood-brain barrier function^33,34,35. A recent single-cell RNA sequencing analysis of the peripheral sensory system revealed that glial cells in the mouse cochlea share many core glial signature genes with radial glial cells and astrocytes in the CNS³⁰. To explore potential functional implications, we plotted genes associated with astrocyte function and compared the molecular profiles of glial cells in both species (Fig. 3f). Our analysis revealed several apparent species-specific gene expression patterns related to astrocyte functions. For instance, glial cells in mice showed predominant expression of Na + -K + -ATPase channels (higher levels of ATP1B1, ATP1A1, and ATP1A2), while glial subtypes in macaques displayed higher levels of glutamate receptors (GRIA2 and GR1K2), potentially associated with potassium buffering. Regarding debris clearance, CLU³⁶ and LRP1³⁷ were more prominently expressed in macaque glial cells, while MEGF1037 and MERTK37 were more abundant in mouse glial cells.

Consistent with these findings, our RNAscope^TM results suggest that peripheral glial cells in macaque could be divided into two main types (Fig. 3j): GRIA2-positive SGSs predominantly located in the modiolus region (Fig. 3g), and CLU-/GRIA2-positive or CLU-positive Schwann cells distributed along the outer modiolar region as neuronal fibers project into both cochlear epithelium (Fig. 3h) and cochlear nuclei (Fig. 3i). These spatial distributions were generally consistent with the sequencing data showing widespread GRIA2 expression in macaque glial cells and selective CLU enrichment in GC_1/GC_3 clusters (Fig. 3f).

While we observed differences in gene expression patterns between species, these findings should be interpreted with caution given the technical limitations of snRNA sequencing. Further validation would be needed to determine whether these molecular differences reflect genuine functional divergence in cochlear glial cells across species.

Sensory neuron diversity in the macaque cochlea

We next investigated the diversity of SGNs. Defined by canonical markers in mice^23,24 (Supplementary Fig. 11), 955 SGNs in macaques can be re-clustered into type I/II subclasses and subdivided into four subtypes with distinct expression patterns (Fig. 4a, b and Supplementary Fig. 12). However, subtle differences were observed in macaque sensory neurons (Fig. 4c, d and Supplementary Fig. 12c). TH expression was preferentially enriched only in mouse type II neurons, and the type IC marker genes POU4F1 and LYPD1 were specifically expressed in mouse type IC neurons. Interestingly, we found that, unlike mouse SGNs, expression of the type IB marker CALB1 was enriched in macaque type II neurons (Fig. 4b, c and Supplementary Fig. 13). Immunolabeling of CALB1 with type II markers, including GATA3 and PRPH proteins, confirmed that CALB1 protein is co-localized with these two type II SGN markers, indicating that CALB1 is a macaque-specific marker for type II sensory neurons (Fig. 4d).

**Fig. 4: Detailed characterization of the macaque spiral ganglion neurons.**

Differentially expressed genes (DEGs) analysis identified a cohort of conserved transcription factors (TFs) that were share either in type I or type II SGNs in both species (Fig. 4e, g). Among them, some have been previously reported, such as PROX1³⁸, RUNX1³⁹, and SMAD1²². For example, RUNX1 is essential for type IC SGN specification. Immunolabeling and RNASCOPE assay confirmed these TFs selectively expressed in SGN subtypes (Fig. 4f, h). Additionally, we found primate-specific TFs PBX3 particularly expressed in type II SGNs (Fig. 4i). Gene set enrichment analysis showed that the key TF PBX3 is involved in GO terms related to neuron development (Fig. 4j). Immunofluorescence staining confirmed that the PBX3 protein is localized in the cell nucleus of macaque SGNs, probably as a specific subtype, but not in mouse SGNs (Fig. 4k). However, future experiments should be conducted to determine if PBX3 plays primate-specific roles in SGNs.

The similarity and specificity of expression profiles in the cochlear epithelium

We focused on the cochlear epithelium, also known as the organ of Corti, which is a highly specialized structure for sound detection and transduction^40,41,42. To explore cross-species similarities and differences in the cochlear epithelium at different taxonomic levels, we employed mutiple label-transfer algorithms (scVI, CCA, and Symphany) to integrate our in-house mouse cochlear epithelium clusters as references and projected the 10, 870 macaque single-nucleus transcriptomes onto them based on their distances in their joint-embedding spaces (Fig. 5a, b and Supplementary Fig. 14; see “Methods”), resulting in highly overlapped anchoring identity (Fig. 5c). Additionally, label transfer-based integrated cell types exhibited highly similar transcriptomes (Fig. 5d, e) though we found CC_ISC_OSC and IBC_IPh_HeC varied across species, suggesting their species-specificity.

**Fig. 5: Comparative analysis reveals conserved and divergent features across macaque and mouse cochlear epithelium.**

Further analysis of the expression of gene families across cluster pairs indicated that, as demonstrated in other systems^43,44,45, certain families, such as G protein-coupled receptors and TFs (especially the zinc finger types), displayed cell type-specific expression patterns in the cochlear epithelium (Fig. 5f, g). This suggests that zinc finger type TFs might function as markers of cochlear epithelium identity in both species. Indeed, our immunolabeling or RNAscope^TM assay of selective TFs (GATA3, IKZF2, PBX3, and NR2F2) showed similar expression patterns across species (Fig. 5h, i). However, a supervised learning task for testing the ability of gene-sets related to signaling, ion channels, and cell adhesion to distinguish cochlear epithelium consensus types showed similar classification performance when trained and tested within species (r-value greater than 1.12) but declined sharply across species (60% in macaques and mice; Fig. 5j). Altogether, these results indicate that evolutionary conservation might be maintained by combined TF codes, while species-specific adaptation might be driven by species-enriched genes (either TFs or effector genes).

Mapping deafness-associated genes across species

To further examine our dataset and gain deep insights into the cross-species expression profile of genes related to hereditary hearing loss (HHL), we mapped HHL-related genes (https://hereditaryhearingloss.org, updated on 02/17/2022) categorized into four types: autosomal dominant nonsyndromic HL (48 genes), autosomal recessive nonsyndromic HL (74 genes), syndromic HL (41 genes), and X-linked nonsyndromic/age/noise/cisplatin-related HL (46 genes) in our databases for systematic characterization of their expression profiles. It is important to note that genes linked to age, noise, and cisplatin-related HL by genome-wide association studies (GWAS) have not been extensively investigated in the inner ear (GWAS Catalog (ebi.ac.uk)). Through the overall examination of HL-related gene expression in the macaque, we observed that the majority of these genes exhibit cell-type-specific expression patterns (Fig. 6 and Supplementary Fig. 15). For instance, POU4F3—a well-established deafness gene—is predominantly expressed in HCs, with its mutations known to cause DFNA15, a common autosomal dominant hearing loss disorder. The gene Gjb2, which encodes connexin 26 for gap junctions, is broadly expressed in SCs and cells of the lateral wall. Mutations in Gjb2 are associated with both congenital HL and later-onset deafness. Notably, our data demonstrate highly conserved expression patterns across 1-, 5-, and 11-year-old macaques, further validating the robustness of our dataset. Cross-species comparative analysis reveals that deafness-associated genes are evolutionarily conserved between macaques and mice, sharing analogous expression profiles. Together, this analysis revealed the cross-species distributions of HL-related genes within the whole cochlear map. Given the close evolutionary proximity of macaques to humans, establishing this atlas holds significant implications for advancing research on human auditory pathologies.

**Fig. 6: Mapping deafness-related genes in macaque cochlea.**

Discussion

The cochlea is of particular interest for comparative characterization due to its highly specialized architecture for auditory processing^2,46. With the aid of single-cell transcriptomic technologies that enable the systematic profiling of the cellular makeup of many organs^47,48 and tissues⁴⁹, we first obtained a comprehensive single-nucleus transcriptomic atlas of the cochlea in the non-human primate Macaca fascicularis, also known as the cynomolgus, crab-eating, or long-tailed monkey. We developed a broad cell map from four major structures, including the sensory epithelium, spiral ganglion, stria vascularis, and spiral ligament in both juvenile and adult macaques and identified 20 cell types based on their unique expression profiles. This atlas will enrich our understanding of the cell type-specific signatures in the non-human primate cochlea and will allow researchers to uncover potential molecular changes under both pathological and physiological conditions, such as aging.

A common theme that emerged from our analyses was the general evolutionary conservation of the cochlea across non-human primates and mice. The sensory epithelium was particularly conserved in mammals, with similar structures and cell types in both species. However, our results also indicated that, compared to shared HC-enriched DEGs, the number of DEGs shared in SCs was dramatically reduced. This divergence in SCs may contribute to frequency tuning adaptation because the highest hearing range in mice can be up to 80 kHz^50,51 (ultrasonic hearing), while the hearing range in macaques is approximately 50 Hz to 32 kHz⁵², which is quite similar to that of humans (20 Hz to 20 kHz). We hypothesized that one possible reason for the SC diversity across species is the potential for SCs as precursors for mitotic HC regeneration after damage, as Mutai et al.⁵³ recently showed that LGR5 expression was relatively higher in the cochlea of a 5-year-old macaque. In contrast to the cochlear epithelium, glial cells exhibited robust heterogeneity in their subtype identities, although glial cells in macaques still showed classical morphologies and interacted with sensory neurons. The crosstalk between glial cells and neurons in the non-human primate peripheral nervous system remains largely unknown. However, in the mouse cochlea, Tasdemir-Yilmaz et al.³⁰ found that the IGFBP6-PHB2 gene pair might serve as the ligand-receptor coupling between SGCs and neurons, but its function remains to be explored. Therefore, further work utilizing the powerful advances in spatial transcriptomic technology is needed to fill this knowledge gap.

Using a comparative analysis strategy, we identified essential functionally relevant molecular and regulatory mechanisms⁵⁴ that provide insights into cell type determination. For instance, through comparative analysis, we demonstrated that cochlear cell types share a core set of molecular features, including the expression of TFs, but also exhibit differences such as the enrichment of PBX3 observed in macaque sensory neurons.

Currently, our study also holds some limitations. Due to limited sample availability, our single-cell atlas does not encompass critical embryonic developmental stages of crab-eating macaque. Furthermore, integrating spatial and epigenetic multi-omics data along with establishing functional validation models using macaque cochlear organoids would enable deeper exploration into the molecular mechanisms underlying the establishment of the tonotopic gradient in the basilar membrane and the regenerative mechanisms of HCs and SGNs. These areas will naturally become a key focus of our future work.

Methods

Animals and ethical compliance

All experiments involving macaques (Macaca fascicularis) were conducted in strict accordance with the guidelines set forth by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) for the ethical treatment of non-human primates. The study protocol was reviewed and approved by the ethical committee of the State Key Laboratory of Primate Biomedical Research. For the purpose of single-nucleus RNA sequencing (snRNA-seq), cochleae from one 1-year-old male, two 5-year-old male, and one 11-year-old female macaque were utilized. Immunostaining experiments were performed using cochleae from three 5-year-old male and one 11-year-old male macaque. Electromotility recordings were obtained from one 1-year-old male macaque. All animals were housed in facilities that meet the standards of the AAALAC, and their well-being was closely monitored throughout the study.

C57BL/6J mice used in this study were purchased from GenePharmtech Co., Ltd. and housed in a standard specific pathogen free environment. Mice of either sex aged postnatal days 21 to 25 were used in this study with no obvious sex phenotype observed. All mice were housed under controlled conditions, including room temperature (23 ± 3 °C), relative humidity (55 ± 5%), and a 12-h light/dark cycle. During the feeding process, all mice were provided with sterile water and nutrient-balanced feed. Additionally, each cage was equipped with non-toxic and harmless bedding.

Macaque RNA-seq, quality control, and clustering

Sample collection

For single-nucleus RNA sequencing (snRNA-seq), juvenile and adult macaques were deeply anesthetized with ketamine (10 mg/kg) and perfused with cold artificial cerebrospinal fluid (ACSF) following the cessation of the pedal withdrawal reflex. The entire inner ear was rapidly extracted, and the cochlea was carefully dissected and embedded in chilled DMEM medium. During dissection, the outer bones encasing the cochlear duct were meticulously removed using surgical forceps under a dissection microscope. To obtain a comprehensive representation of cell types, care was taken to slightly remove the outer bones attached to the lateral wall of the cochlea. The entire cochlear duct was then transferred into a 1.5 ml tube, quickly frozen in liquid nitrogen, and stored at −80 °C until nuclei isolation. Both cochleae from each animal were collected.

Purification of nuclei

Nuclei isolation was performed following the 10× Genomics protocols (CG000393_Rev A). Briefly, the frozen cochlear ducts were homogenized in chilled 1 ml lysis buffer containing 0.1% NP40, 10 mM Tris-HCl, 146 mM NaCl, 1 mM CaCl2, 21 mM MgCl2, and 40 U/ml RNase inhibitor. The tissues were gently pipetted up and down with a 1 ml tip until no visible clumps remained. Homogenates were incubated on ice for 7 min before adding an equal volume of ST Wash buffer containing 10 mM Tris-HCl, 146 mM NaCl, 1 mM CaCl2, 21 mM MgCl2, 0.01% BSA, and 40 U/ml RNase inhibitor. Cell debris and clumps were filtered through a 40-μm strainer, and the supernatant was centrifuged at 500 × g and 4 °C for 5 min. The washing steps were repeated twice, and nuclei were resuspended in a buffer containing 1× PBS, 1% BSA, and 40 U/ml RNase inhibitor. DAPI dye was added, and nuclei were manually counted under a fluorescence microscope (BX53, Olympus).

10× Chromium RNA sequencing

For loading onto the 10× Chromium instrument, nuclei were diluted to 700–1200/μl. The guidelines for nuclei loading on the 10× Genomics Chromium Next GEM Chip, nuclear barcoding, cDNA amplification, and library construction were followed according to the 10× Genomics protocols (CG000204_Rev D). Libraries were sequenced on the Illumina NovaSeq 6000 System at Oebiotech Co., Ltd. in Shanghai, China. Reads were aligned to the Macaca_fascicularis_5.0 genome, and the gene expression matrix was generated and quantified using the default 10× Cell Ranger v3 pipeline (v5.0.0). Filtered count matrices from the Cell Ranger pipeline were used for downstream analysis.

Quality control of RNA-seq data

The filtered count matrices were analyzed using Scanpy (v1.9.8) following their recommended standard workflow. Cells expressing more than 5000 genes and less than 500 counts were excluded from further analysis.

Clustering of RNA-seq data

After quality control, 5619 nuclei of 1-year-old macaque, 32,506 nuclei of 5-year-old macaques, and 5661 nuclei of 11-year-old macaque were used for clustering. Initially, the entire cochlea of each animal was analyzed separately following a standard pipeline, which included normalization (sc.pp.normalize_total), log-transformation (sc.pp.log1p), identification of highly variable genes for clustering (sc.pp.highly_variable_genes), regression by total counts (sc.pp.regress_out), principal component analysis (PCA), neighbor identification, Leiden clustering, and uniform manifold approximation and projection (UMAP) visualization. For cell annotation of the macaque cochlea, the markers of the mouse cochlea listed in Supplementary Fig. 4 were referenced. To integrate the data from the 1, 5, and 11-year-old macaques, single-cell variational inference (scVI-tools, V.1.1.2) was employed. A total of 2000 highly variable genes were identified using the Cell Ranger flavor for correcting batch effects and facilitating data integration.

Mouse RNA-seq, quality control, and clustering

Sample collection

snRNA-seq was conducted using eight cochleae obtained from mice at P21-22. These were pooled together to form one replicate. Euthanasia was performed using carbon dioxide (CO₂) inhalation. Briefly, each mouse was placed in a chamber prefilled with 5% CO₂. The gas concentration was then increased at a controlled flow rate (30% chamber volume/min) to over 30% and maintained for at least 5 min after breathing ceased. Death was confirmed by subsequent cervical dislocation to ensure a humane endpoint. The cochleae were quickly and meticulously extracted using a cold DMEM medium for dissection. The calcified bony labyrinth exterior was gently removed using forceps while ensuring maximal tissue preservation. The entire cochlear duct was quickly transferred into a 1.5 ml tube before being rapidly frozen in liquid nitrogen. The samples were stored at −80 °C until nuclear isolation was necessary. It should be noted that all dissection procedures were conducted under a dissection microscope. All dissections were completed within one hour to ensure the acquisition of high-quality cells.

Nuclear purification

The nuclear isolation procedure utilized followed the 10× Genomics protocols (CG000393_Rev A). The frozen tissues were homogenized in a cold 1 ml lysis buffer, which included 0.1% NP40, 10 mM Tris-HCL, 146 mM NaCl, 1 mM CaCl2, 21 mM MgCl2, and 40 U/ml RNase inhibitor. This was done using careful pipetting techniques until no visible clumps were observed. Homogenates were then allowed to maintain on ice for 7 min before an equal volume of ST Wash buffer was added. The buffer consisted of 10 mM Tris-HCL, 146 mM NaCl, 1 mM CaCl2, 21 mM MgCl2, 0.01% BSA and 40 U/ml RNase inhibitor. Any remaining cell debris and clumps were removed using a 40-μm strainer, which allowed the supernatant to be released. It was then centrifuged at 500 × g and 4 °C for 5 min. This washing procedure was repeated twice. The nuclei were then re-suspended in a buffer that comprised of 1× PBS, 1% BSA, and 40 U/ml RNase inhibitor. To finalize the process, a DAPI dye was added, and the nuclei were manually counted under a fluorescence microscope (BX53, Olympus).

10× Chromium RNA sequencing procedure

The purified nuclei were diluted to a density of 700–1200/μl for utilization in the 10x Genomics Chromium system. The processes of nuclei loading onto the Chromium Next GEM Chip, nuclear barcoding, cDNA amplification, and library preparation were all conducted according to the 10× Genomics protocols (CG000204_Rev D). Subsequent sequencing of libraries was performed on the Illumina Nova 6000 System at the Oebiotech Company, Ltd., based in Shanghai, China. For read alignment, we employed the Mm10_3.0.0 genome. The gene expression matrix was generated and quantified using the default settings of the 10× Cell Ranger v3 pipeline (version 5.0.0). Filtered count matrices obtained from the Cell Ranger pipeline were subsequently used for downstream analysis.

Quality control of RNA-seq data

We utilized Scanpy (v1.9.8), adhering to the recommended parameters set by the developers. Briefly, cells expressing more than 5000 genes or producing fewer than 500 counts were systematically excluded from downstream analysis. This action served as a vital quality control measure, ensuring data accuracy and reliability.

Clustering of RNA-seq data

After cell filtering, a total of 6,775 nuclei were subjected to separate analytical clustering. A comprehensive downstream analysis was conducted, inclusive of normalization (as outlined by sc.pp.normalize_total), log-transformation (in accordance with sc.pp.log1p), high variable genes extraction (utilizing sc.pp.highly_variable_genes), and regression by total counts (prescribed by sc.pp.regress_out). Supplementary techniques, including PCA, neighbor identification, Leiden clustering, and UMAP visualization were also employed. Established marker-sets were applied to facilitate cell-type annotations for these analyzed clusters (refer to gene lists in Supplementary Fig. 4). scVI integration was implemented for the combination of the two replicates, with 2000 highly variable genes as input and Cell Ranger mode as the parameter choice for batch effect correction.

Integration and clustering

Label transfer

Our initial approach involved using SeuratDisk to convert AnnData objects (based on scVI-integrated data) into Seurat objects. Pre-processed transcriptomic profiles from macaques and mice were rendered into raw expression matrices, which were subsequently converted into human genes using the biomaRt software package. Both macaque and mouse genes were then limited to those one-to-one orthologs that exhibit expression in both species, aligning with the orthology annotation provided by NCBI Homologene (https://www.ncbi.nlm.nih.gov/homologene). In an effort to map macaque sensory epithelium cell-types onto corresponding mouse data, we employed the label transfer embedding feature available in the Seurat pipeline⁴⁹. Data extracted from the cochlear epithelium of both species underwent normalization and scaling using the SCTransform function and was further regressed with species identity and with UMI counts (nCount_RNA). We utilized the top 30 canonical components to identify integrative anchors (by deploying the FindTransferAnchors function with the parameters: normalization.method = “SCT”, reference.assay = “SCT”, query.assay = “SCT”, reduction = “cca”, dims = 1:30). These anchors were subsequently used for projecting macaque cells (query data) onto the mouse data set via the TransferData function (weight.reduction = “pca”, dims = 1:30). A macaque cell mapping rate exceeding 25% was established as the threshold, as quantified by the projection score, defined as the percentage of query cells of a specific clustering mapping onto a respective reference cluster.

DEG calculation

To pinpoint cell-types of DEGs exhibiting both shared and divergent traits across species, we employed the label transfer-based Seurat objects derived from the species integration process. Positive DEGs were identified through the deployment of the FindAllMarkers function in Seurat. This was executed by conducting the Receiver Operating Characteristic (ROC) test with preset parameters (min.pct = 0.1, AUROC threshold = 0.7).

Consequently, each cell-type within a species was compared against all other remaining nuclei within that respective class. To evaluate differential expression, SCT-normalized counts were utilized. The relationships between DEGs across species within a specific cell-type were visualized on a Venn diagram, generated using the eulerr package (v6.1.1). Heatmaps of DEGs for all subclasses within the cochlear epithelium were established by down-sampling each subclass to 50 random nuclei per species. These SCT-transformed counts were subsequently scaled and visualized with the DoHeatmap function.

To elicit type II SGN-enriched DEGs in comparison to type I SGNs and vice versa, we utilized the original Seurat objects derived from both species. Positive DEGs were determined using the FindAllMarkers function coupled with the ROC test to compare type II and type I SGNs across both species. Venn diagrams, made using the eulerr package, were implemented to illustrate the relationship between type II/type I-enriched DEGs across species. The relationship of conserved DEGs (inclusive of transcription factors) were demonstrated through the generation of scatter plots using the ggplot software package.

Conservation of gene families

To elucidate the conservation and divergence of co-expression patterns within gene families between primates and mice, we performed a MetaNeighbor analysis (Crow et al., 2018) using gene groups curated by the HUGO Gene Nomenclature Committee (HGNC, RRID SCR_002827) at the European Bioinformatics Institute (https://www.genenames.org) and the Synaptic Gene Ontology (SynGO, RRID SCR_017330). Group hierarchy was provided to propagate HGNC annotations, ensuring comprehensive parent annotations and including only sets with more than five groups in the analysis. Briefly, the MetaNeighbor algorithm was used to compare data at the cluster level using labels from the cross-species integration in Seurat. Two sequential steps were then performed at the phylogeny level for cross-species comparisons between primates and mice. In these comparisons, the data from the two species were used as the testing and training sets across two rounds of cross-validation, reporting the average performance (AUROC) across folds. The results were then compared to the average within-species performance.

Cell-type replicability across species

MetaNeighbor analysis was employed to quantify the replicability of cochlear cell types within and across species, as described by Crow et al. (2018). The get_variable_genes function was first used to identify highly variable genes for each cell type. These genes were then used as input for the MetaNeighborUS function, running the fast_version and one_vs_best programs with parameters set to TRUE to obtain the reported area under the receiver operator characteristic curve (AUROC) scores. These scores were used to determine the similarity of the two closest neighboring clusters to the training cluster compared to the second closest cluster. AUROC scores for each pair of labels were plotted as a heat map to visualize cell-type replicability across species.

GO analysis of conserved cluster markers

Conserved marker genes from integrated clusters were calculated using the FindConservedMarkers function (group.var = “species”, assay = “SCT”, only.pos = TRUE) for Gene Ontology (GO) analysis. Genes that were not detected in at least 20% of a single cluster in both species were removed from the marker gene list. The gprofiler package (Raudvere et al., 2019) was used to compute the enrichment of GO terms (gost: user_threshold = 0.01, correction_method = “g_SCS”).

Cumulative distribution plot of the number of gene families

For the cumulative distribution plot, genes expressed in at least 20% of the cells within a given cluster were included. Gene family annotations for Homo sapiens were acquired from the KEGG BRITE database (Kanehisa et al., 2021).

Inferring the cellular localization of HL genes

A list of hereditary deafness-related genes was obtained from the Hereditary Hearing Loss Homepage (https://hereditaryhearingloss.org/), and a list of acquired hearing loss genes was acquired from the NHGRI-EBI GWAS Catalog (https://www.ebi.ac.uk/gwas/). Normalized counts were used to generate heat maps of these disease-related genes for all cell-types.

RNAscope^TM plus immunofluorescence

Mice were anesthetized with pentobarbital sodium (10 mg/kg body weight, intraperitoneal) and intravascularly perfused with cold 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS, pH 7.2). 5-year-old and 11-year-old macaques were deeply anesthetized with ketamine (10 mg/kg) and perfused with cold ACSF, followed by the removal of the pedal withdrawal reflex. For cochlear preparations around postnatal day 21 (P21) to P25, the oval and round windows were opened to permit flushing with PBS (pH 7.2), replacement with lymphatic fluid through the scala, and post-fixation with 4% PFA overnight at 4 °C. Cochleae were then decalcified with 0.5 M EDTA (pH 8.0) for 1–6 h until the bone softened. After thorough washing with PBS (pH 7.2), the cochleae were dissected out. For macaque cochlear preparations, the same procedure was followed, but decalcification with 0.5 M EDTA (pH 8.0) was extended to one week.

The treated cochleae described above were incubated in 30% sucrose (wt/vol) at 4 °C overnight, subsequently embedded in OCT compound, and the embedded blocks were stored at −80 °C. Cochlear blocks were sectioned into 14 µm-thick slices using a cryostat and stored at −20 °C for subsequent RNAscope^TM and immunofluorescence use.

For RNAscope^TM, the RNAscope^TM Multiplex Fluorescent Detection Kit v2 (catalog no. 323100) purchased from Advanced Cell Diagnostics (Bio-Techne SAS, Rennes) was used strictly according to the manufacturer’s instructions. After OCT removal, sections were dried at 60 °C for 30 min and fixed in 10% neutral buffered formalin on ice for 15 min. They were then dehydrated in a graded ethanol series and dried, incubated with hydrogen peroxide at room temperature (RT) for 10 min, and washed in H₂O. Subsequently, sections were immersed in boiling target retrieval solution for 5 min for antigen retrieval, washed in H₂O and PBS, and incubated with Protease III solution in a hybridization oven at 40 °C for 30 min to permeabilize tissues and expose target RNA. Target-specific probes were then hybridized to the sections at 40 °C for 2 h (each target-specific probe contains a mixture of short oligonucleotides designed to bind to a specific target mRNA and detectable in one of three fluorescence channels, C1, C2, or C3, Supplementary Data 2). After washing the sections several times, they were sequentially incubated with Amp1, Amp2, Amp3, and horseradish peroxidase for cascade signal amplification. Fluorophores were added using the corresponding TSA dyes (TSA520, TSA570, or TSA650, 1/1000 dilution) according to the channel selected. When performing co-staining of probes and antibodies, tissues were incubated with primary antibodies overnight at 4 °C before Protease III treatment, while secondary antibody incubation was conducted following completion of the RNAscope^TM protocol using identical procedures to those described for the immunofluorescence method below.

For immunofluorescence labeling, the cryosections were incubated overnight at 4 °C in a blocking solution (Co-Detection Antibody Diluent, 323160) that contained the primary antibody, followed by the secondary antibody for 2 h at room temperature. After washing with PBS three times, DAPI (1:1000 dilution, Roche, 10236276) was added for 10 min at room temperature. Slides were mounted using Fluoroshield histology mounting medium (Sigma, F6182). Confocal images of cochlear cryosections were obtained using an LSM700/900 microscope, with images captured as 1024 × 1024-pixel rasters at 20×/63× magnification.

Immunofluorescence and imaging

Mice were anesthetized with pentobarbital sodium (10 mg/kg body weight, intraperitoneal) and intravascularly perfused with cold 4% PFA in phosphate-buffered saline (PBS, pH 7.2). 5-year-old and 11-year-old macaques were deeply anesthetized with ketamine (10 mg/kg) and perfused with cold ACSF, followed by the removal of the pedal withdrawal reflex. For cochlear preparations around postnatal day 21 (P21) to P25, the oval and round windows were opened to permit flushing with PBS (pH 7.2), replacement with lymphatic fluid through the scala, and post-fixation with 4% PFA overnight at 4 °C. Cochleae were then decalcified with 0.5 M EDTA (pH 8.0) for 1 to 6 h until the bone softened. After thorough washing with PBS (pH 7.2), the cochleae were dissected out. For macaque cochlear preparations, the same procedure was followed, but decalcification with 0.5 M EDTA (pH 8.0) was extended to one week.

The microdissected cochlear pieces were blocked with a solution containing 10% goat serum, 1% bovine serum albumin, 0.3% TritonX-100, and 0.01% sodium azide (NaN3) in PBS (pH 7.2) for 1 to 2 h at room temperature. The tissues were then incubated with primary antibodies (see below) diluted in the blocking solution overnight at 4 °C. After three washes in PBS, the tissues were incubated with secondary antibodies (1:1000 dilution in blocking buffer, Thermo Fisher Scientific, see below) for 2 h at room temperature. Following secondary antibody incubation, the tissues were washed three times in PBS and incubated with DAPI (1:5000 dilution, Roche, 10236276). The samples were then washed three times with PBS and mounted using Fluoroshield histology mounting medium (Sigma, F6182). Confocal images of whole-mount cochleae were obtained using an LSM700/900 microscope, with images captured as 1024 × 1024-pixel rasters at 63× magnification.

Primary antibodies were as follows: anti-PBX3 (1:500 dilution, Thermo Fisher, PA5-40482); anti-TUJ1 (1:500 dilution, Biolegend, MMS-435P); anti-MBP (1:200 dilution, SIGMA, M3821); anti-GFAP (1:800 dilution, Abcam, ab7260); anti-CALB2 (1:200 dilution, Millipore, MAB1568); anti-CALB1 (1:100 dilution, New England Biolabs, 13176); Anti-PRESTIN (1:50000 dilution, abcam, ab242128); Anti-OTOF(1:200 dilution, abcam, ab53233); Anti-GATA3(1:250 dilution, abcam, ab199428); Anti-PRPH (1:100 dilution, abcam, ab246502); Anti-MYO7A(1:500 dilution, Proteus Bioscience, 25-6790).

Secondary antibodies were as follows: donkey anti-mouse IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor 555 (1:1000 dilution, Thermo Fisher, A-31570); donkey anti-rabbit IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor 488 (1:1000 dilution, Thermo Fisher, A-21206); goat anti-mouse IgG1 cross-adsorbed secondary antibody, Alexa Fluor 647 (1:1000 dilution, Thermo Fisher, A-21240); and phalloidin Alexa Fluor 488 (1:1000 dilution, Thermo Fisher, A12379).

Electrophysiological recording

Patch-clamp recordings were conducted on outer hair cells (OHCs) of 1-year-old macaques. Cochlear tissues were dissected in a dissection solution containing 5.36 mM KCl, 141.7 mM NaCl, 1 mM MgCl2, 0.5 mM MgSO4, 0.1 mM CaCl2, 10 mM HEPES, 3.4mM L-glutamine, and 10 mM D-glucose (325 mOsm, pH = 7.25). During recording, the tissues were transferred to a recording chamber followed by the application of an extracellular recording solution composed of 144 mM NaCl, 0.7 mM NaH2PO4, 5.8 mM KCl, 1.3 mM CaCl2, 0.9 mM MgCl2, 10 mM HEPES, and 5.6 mM D-glucose (325 mOsm, pH = 7.35). The organ of Corti was visualized using an upright microscope (Olympus, BX51WI) with a 60× objective. Whole-cell recordings were performed to examine the electromotile activity in 1-year-old OHCs by applying a depolarizing voltage step from a holding potential of −64 to 164 mV. Changes in cell length were observed and recorded using Micromanager.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and the Supplementary Materials, or contact the authors for access. The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2025)^55,56 in the National Genomics Data Center (Nucleic Acids Res 2025), China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA032615) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa.

Code availability

The custom codes for producing figure results are available at GitHub (https://github.com/Shinnyee/Cell-atlas-of-non-human-primate-cochlea).

References

Kelly, M. C. & Chen, P. Development of form and function in the mammalian cochlea. Curr. Opin. Neurobiol. 19, 395–401 (2009).
Article CAS PubMed PubMed Central Google Scholar
Driver, E. C. & Kelley, M. W. Development of the cochlea. Development https://doi.org/10.1242/dev.162263 (2020).
Ehret, G. Quantitative analysis of nerve fibre densities in the cochlea of the house mouse (Mus musculus). J. Comp. Neurol. 183, 73–88 (1979).
Article CAS PubMed Google Scholar
Chardin, S. & Romand, R. Regeneration and mammalian auditory hair cells. Science 267, 707–711 (1995).
Article ADS CAS PubMed Google Scholar
Lefebvre, P. P., Malgrange, B., Staecker, H., Moonen, G. & Van de Water, T. R. Retinoic acid stimulates regeneration of mammalian auditory hair cells. Science 260, 692–695 (1993).
Article ADS CAS PubMed Google Scholar
Li, Q. et al. The pathogenesis of common Gjb2 mutations associated with human hereditary deafness in mice. Cell Mol. Life Sci. 80, 148 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Ahmed, M. et al. Eya1-Six1 interaction is sufficient to induce hair cell fate in the cochlea by activating Atoh1 expression in cooperation with Sox2. Dev. Cell 22, 377–390 (2012).
Article CAS PubMed PubMed Central Google Scholar
Bok, J. et al. Transient retinoic acid signaling confers anterior-posterior polarity to the inner ear. Proc. Natl. Acad. Sci. USA 108, 161–166 (2011).
Article ADS CAS PubMed Google Scholar
Chessum, L. et al. Helios is a key transcriptional regulator of outer hair cell maturation. Nature 563, 696–700 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Kaiser, M. et al. TBX2 specifies and maintains inner hair and supporting cell fate in the Organ of Corti. Nat. Commun. 13, 7628 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Giese, A. P. J. et al. CIB2 interacts with TMC1 and TMC2 and is essential for mechanotransduction in auditory hair cells. Nat. Commun. 8, 43 (2017).
Article ADS PubMed PubMed Central Google Scholar
Jia, Y. et al. TMC1 and TMC2 proteins are pore-forming subunits of mechanosensitive ion channels. Neuron 105, 310–321 e313 (2020).
Article CAS PubMed Google Scholar
Walters, B. J. & Zuo, J. Postnatal development, maturation and aging in the mouse cochlea and their effects on hair cell regeneration. Hear Res. 297, 68–83 (2013).
Article PubMed Google Scholar
Lim, D. J. & Anniko, M. Developmental morphology of the mouse inner ear. A scanning electron microscopic observation. Acta Otolaryngol. Suppl. 422, 1–69 (1985).
CAS PubMed Google Scholar
Rybak, L. P., Whitworth, C. & Scott, V. Development of endocochlear potential and compound action potential in the rat. Hear Res. 59, 189–194 (1992).
Article CAS PubMed Google Scholar
Li, Y., Liu, H., Zhao, X. & He, D. Z. Endolymphatic potential measured from developing and adult mouse inner ear. Front. Cell Neurosci. 14, 584928 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kolla, L. et al. Characterization of the development of the mouse cochlear epithelium at the single cell level. Nat. Commun. 11, 2389 (2020).
Article ADS PubMed PubMed Central Google Scholar
Sun, G. et al. Single-cell transcriptomic atlas of mouse cochlear aging. Protein Cell 14, 180–201 (2023).
CAS PubMed PubMed Central Google Scholar
Jean, P. et al. Single-cell transcriptomic profiling of the mouse cochlea: an atlas for targeted therapies. Proc. Natl. Acad. Sci. USA 120, e2221744120 (2023).
Article PubMed PubMed Central Google Scholar
Han, L. et al. Cell transcriptomic atlas of the non-human primate Macaca fascicularis. Nature 604, 723–731 (2022).
Article ADS CAS PubMed Google Scholar
Roux, I. et al. Otoferlin, defective in a human deafness form, is essential for exocytosis at the auditory ribbon synapse. Cell 127, 277–289 (2006).
Article CAS PubMed Google Scholar
Petitpre, C. et al. Neuronal heterogeneity and stereotyped connectivity in the auditory afferent system. Nat. Commun. 9, 3691 (2018).
Article ADS PubMed PubMed Central Google Scholar
Sun, S. et al. Hair cell mechanotransduction regulates spontaneous activity and spiral ganglion subtype specification in the auditory system. Cell 174, 1247–1263 e1215 (2018).
Article CAS PubMed PubMed Central Google Scholar
Shrestha, B. R. et al. Sensory neuron diversity in the inner ear is shaped by activity. Cell 174, 1229–1246 e1217 (2018).
Article CAS PubMed PubMed Central Google Scholar
Lopez, R., Regier, J., Cole, M. B., Jordan, M. I. & Yosef, N. Deep generative modeling for single-cell transcriptomics. Nat. Methods 15, 1053–1058 (2018).
Article CAS PubMed PubMed Central Google Scholar
Zheng, J. et al. Prestin is the motor protein of cochlear outer hair cells. Nature 405, 149–155 (2000).
Article ADS CAS PubMed Google Scholar
Peng, Z. et al. Temporospatial expression and cellular localization of VGLUT3 in the rat cochlea. Brain Res. 1537, 100–110 (2013).
Article CAS PubMed Google Scholar
Seal, R. P. et al. Sensorineural deafness and seizures in mice lacking vesicular glutamate transporter 3. Neuron 57, 263–275 (2008).
Article CAS PubMed PubMed Central Google Scholar
Garcia-Anoveros, J. et al. Tbx2 is a master regulator of inner versus outer hair cell differentiation. Nature 605, 298–303 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Tasdemir-Yilmaz, O. E. et al. Diversity of developing peripheral glia revealed by single-cell RNA sequencing. Dev. Cell 56, 2516–2535 e2518 (2021).
Article CAS PubMed PubMed Central Google Scholar
Hiyama, T. Y. et al. Na(x) channel involved in CNS sodium-level sensing. Nat. Neurosci. 5, 511–512 (2002).
Article CAS PubMed Google Scholar
Ryu, E. J. et al. Analysis of peripheral nerve expression profiles identifies a novel myelin glycoprotein, MP11. J. Neurosci. 28, 7563–7573 (2008).
Article CAS PubMed PubMed Central Google Scholar
Allen, N. J. Astrocyte regulation of synaptic behavior. Annu. Rev. Cell Dev. Biol. 30, 439–463 (2014).
Article CAS PubMed Google Scholar
Sancho, L., Contreras, M. & Allen, N. J. Glia as sculptors of synaptic plasticity. Neurosci. Res. 167, 17–29 (2021).
Article CAS PubMed Google Scholar
Allen, N. J. Role of glia in developmental synapse formation. Curr. Opin. Neurobiol. 23, 1027–1033 (2013).
Article CAS PubMed Google Scholar
Cunin, P. et al. Clusterin facilitates apoptotic cell clearance and prevents apoptotic cell-induced autoimmune responses. Cell Death Dis. 7, e2215 (2016).
Article CAS PubMed PubMed Central Google Scholar
Liu, C. C. et al. Astrocytic LRP1 mediates brain abeta clearance and impacts amyloid deposition. J. Neurosci. 37, 4023–4031 (2017).
Article CAS PubMed PubMed Central Google Scholar
Nishimura, K., Noda, T. & Dabdoub, A. Dynamic expression of Sox2, Gata3, and Prox1 during primary auditory neuron development in the mammalian cochlea. PLoS ONE 12, e0170568 (2017).
Article PubMed PubMed Central Google Scholar
Shrestha, B. R., Wu, L. & Goodrich, L. V. Runx1 controls auditory sensory neuron diversity in mice. Dev. Cell 58, 306–319 e305 (2023).
Article CAS PubMed PubMed Central Google Scholar
Fettiplace, R. & Hackney, C. M. The sensory and motor roles of auditory hair cells. Nat. Rev. Neurosci. 7, 19–29 (2006).
Article CAS PubMed Google Scholar
Gillespie, P. G. & Muller, U. Mechanotransduction by hair cells: models, molecules, and mechanisms. Cell 139, 33–44 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Fettiplace, R. Hair cell transduction, tuning, and synaptic transmission in the mammalian cochlea. Compr. Physiol. 7, 1197–1227 (2017).
Article PubMed PubMed Central Google Scholar
Paul, A. et al. Transcriptional architecture of synaptic communication delineates GABAergic neuron identity. Cell 171, 522–539 e520 (2017).
Article CAS PubMed PubMed Central Google Scholar
Hain, D. et al. Molecular diversity and evolution of neuron types in the amniote brain. Science 377, eabp8202 (2022).
Article CAS PubMed Google Scholar
Taylor, S. R. et al. Molecular topography of an entire nervous system. Cell 184, 4329–4347 e4323 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, S., Wang, S., Zou, L. & Xiong, W. Mechanisms in cochlear hair cell mechano-electrical transduction for acquisition of sound frequency and intensity. Cell Mol. Life Sci. 78, 5083–5094 (2021).
Article CAS PubMed PubMed Central Google Scholar
Yu, B. et al. Molecular and cellular evolution of the amygdala across species analyzed by single-nucleus transcriptome profiling. Cell Discov. 9, 19 (2023).
Article CAS PubMed PubMed Central Google Scholar
Peng, Y. R. et al. Molecular classification and comparative taxonomics of foveal and peripheral cells in primate retina. Cell 176, 1222–1237 e1222 (2019).
Article CAS PubMed PubMed Central Google Scholar
Yang, L. et al. Human and mouse trigeminal ganglia cell atlas implicates multiple cell types in migraine. Neuron 110, 1806–1821 e1808 (2022).
Article CAS PubMed PubMed Central Google Scholar
Li, J. et al. PIEZO2 mediates ultrasonic hearing via cochlear outer hair cells in mice. Proc. Natl. Acad. Sci. USA https://doi.org/10.1073/pnas.2101207118 (2021).
Heffner, H. E. & Heffner, R. S. Hearing ranges of laboratory animals. J. Am. Assoc. Lab. Anim. Sci. 46, 20–22 (2007).
CAS PubMed Google Scholar
Coleman, M. N. & Ross, C. F. Primate auditory diversity and its influence on hearing performance. Anat. Rec. A Discov. Mol. Cell Evol. Biol. 281, 1123–1137 (2004).
Article PubMed Google Scholar
Mutai, H. et al. Gene expression dataset for whole cochlea of Macaca fascicularis. Sci. Rep. 8, 15554 (2018).
Article ADS PubMed PubMed Central Google Scholar
Arendt, D. et al. The origin and evolution of cell types. Nat. Rev. Genet. 17, 744–757 (2016).
Article CAS PubMed Google Scholar
CNCB-NGDC Members and Partners Database resources of the national genomics data center, China National Center for Bioinformation in 2025. Nucleic Acids Res. 53, D30–D44 (2025).
Article Google Scholar
Zhang 张思思, S. et al. The GSA family in 2025: a broadened sharing platform for multi-omics and multimodal data. Genomics Proteomics Bioinformatics https://doi.org/10.1093/gpbjnl/qzaf072 (2025).

Download references

Acknowledgements

We would like to thank Oebiotech Company (Shanghai, China, http://www.oebiotech.com/) for providing the snRNA-seq platform. This research was supported by funds from the National Key Research and Development Program of China (2021YFA1101300 to R.C., 2021YFA1101800 to R.C. and X.C., 2020YFA0112503 to R.C., and 2024YFC2511100/1103 to L.L. and X.C.), the National Natural Science Foundation of China (82330033 to R.C., 82030029 to R.C., 92468302 to R.C., 92149304 to R.C., 82101236 to X.C., 82071059 to L.L. and 82471185 to L.L), the STI2030-Major Projects (2022ZD0205400 to J.Q.), the Taishan Scholars Program (tsqn202408320 to J.Q.), the Shandong Provincial Natural Science Foundation (ZR2024YQ049 to J.Q.), the Shenzhen Science and Technology Program (JCYJ20240813161801003 to R.C.), the China Postdoctoral Science Foundation (2021T140113 to X.C.), the Guangdong Major Project of Basic Research (2025B0303000009 to J.Q.), the Natural Science Foundation of Jiangsu Province (BK20232007 to R.C.), the 2022 Open Project Fund of Guangdong Academy of Medical Sciences (YKY-KF202201 to R.C.), the Jiangsu Provincial Scientific Research Center of Applied Mathematics (BK20233002 to R.C.), the Nanjing Medical Science and Technology Development Project (YKK19072 to L.L.), the Research Personnel Cultivation Program of Zhongda Hospital Southeast University (CZXM-GSP-RC164 to L.L.), the Fundamental Research Funds for the Central Universities (2242021R20050 to X.C.), and the Guangdong Basic and Applied Basic Research Foundation (2024KCXTD052 to R.C.).

Author information

These authors contributed equally: Xin Chen, Yuwei Che, Jieyu Qi, Ming Cen, Shan Gao.

Authors and Affiliations

Department of Otolaryngology Head and Neck Surgery, Zhongda Hospital, State Key Laboratory of Digital Medical Engineering, Jiangsu Provincial Key Laboratory of Critical Care Medicine, School of Life Sciences and Technology, School of Medicine, Advanced Institute for Life and Health, Southeast University, Nanjing, China
Xin Chen, Yuwei Che, Ming Cen, Shan Gao, Bin Zhu, Yiheng Ao, Xiangyu Ma, Xinlin Wang, Jiheng Wu, Fanliang Kong, Ling Lu & Renjie Chai
Co-innovation Center of Neuroregeneration, Nantong University, Nantong, China
Xin Chen & Renjie Chai
Department of Radiology, Zhuhai People’s Hospital, The Affiliated Hospital of Beijing Institute of Technology, Advanced Technology Research Institute, School of Life Science, Beijing Institute of Technology, Beijing, China
Jieyu Qi & Renjie Chai
Advanced Technology Research Institute, Beijing Institute of Technology, No. 6666 Furong Road, 250307, Jinan, China
Jieyu Qi
Department of Otolaryngology-Head and Neck Surgery, the Affiliated Drum Tower Hospital of Nanjing University Medical School, Jiangsu Provincial Key Medical Discipline (Laboratory), Nanjing, China
Cheng Cheng
Research Institute of Otorhinolaryngology, Nanjing, China
Cheng Cheng
State Key Laboratory of Primate Biomedical Research, Institute of Primate Translational Medicine, Kunming University of Science and Technology, Kunming, Yunnan, China
Rongrong Guo, Wei Si, Yongchang Chen & Tao Tan
Yunnan Key Laboratory of Primate Biomedical Research, Kunming, Yunnan, China
Rongrong Guo, Wei Si, Yongchang Chen & Tao Tan
Department of Otolaryngology-Head & Neck Surgery, West China Hospital, Sichuan University, Chengdu, China
Tian Shen
Department of Otolaryngology Head and Neck Surgery, Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China, Chengdu, China
Renjie Chai
Southeast University Shenzhen Research Institute, Shenzhen, China
Renjie Chai

Authors

Xin Chen
View author publications
Search author on:PubMed Google Scholar
Yuwei Che
View author publications
Search author on:PubMed Google Scholar
Jieyu Qi
View author publications
Search author on:PubMed Google Scholar
Ming Cen
View author publications
Search author on:PubMed Google Scholar
Shan Gao
View author publications
Search author on:PubMed Google Scholar
Bin Zhu
View author publications
Search author on:PubMed Google Scholar
Yiheng Ao
View author publications
Search author on:PubMed Google Scholar
Xiangyu Ma
View author publications
Search author on:PubMed Google Scholar
Cheng Cheng
View author publications
Search author on:PubMed Google Scholar
Rongrong Guo
View author publications
Search author on:PubMed Google Scholar
Xinlin Wang
View author publications
Search author on:PubMed Google Scholar
Tian Shen
View author publications
Search author on:PubMed Google Scholar
Jiheng Wu
View author publications
Search author on:PubMed Google Scholar
Fanliang Kong
View author publications
Search author on:PubMed Google Scholar
Wei Si
View author publications
Search author on:PubMed Google Scholar
Yongchang Chen
View author publications
Search author on:PubMed Google Scholar
Tao Tan
View author publications
Search author on:PubMed Google Scholar
Ling Lu
View author publications
Search author on:PubMed Google Scholar
Renjie Chai
View author publications
Search author on:PubMed Google Scholar

Contributions

X.C., M.C., S.G., and J.Q. produced the snRNA-seq data. X.C., Y.Che, and X.M. analyzed the scRNA-seq data. Y.Che, J.W., F.K., M.C., T.S., B.Z., and R.G. analyzed the anatomy and histology. X.C., J.Q., M.C., S.G., X.W., Y.Che., J.W., F.K., and Y.A. wrote the manuscript. C.C., W.S., Y.Chen, T.T., L.L., and R.C. edited the manuscript. R.C. and L.L. provided project management and supervised the study.

Corresponding authors

Correspondence to Ling Lu or Renjie Chai.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information (download PDF )

Description of Additional Supplementary Files (download PDF )

Supplementary Data 1 (download XLSX )

Supplementary Data 2 (download XLSX )

Reporting Summary (download PDF )

Transparent Peer Review file (download PDF )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Chen, X., Che, Y., Qi, J. et al. Molecular heterogeneity of the non-human primate cochlea. Nat Commun 17, 1633 (2026). https://doi.org/10.1038/s41467-026-68350-2

Download citation

Received: 11 August 2023
Accepted: 05 January 2026
Published: 17 January 2026
Version of record: 13 February 2026
DOI: https://doi.org/10.1038/s41467-026-68350-2

Subjects

Abstract

Similar content being viewed by others

Introduction

Results

A cell type atlas of the macaque cochlea

Auditory receptors in macaque cochlea

Glial cell diversity in the macaque cochlea

Sensory neuron diversity in the macaque cochlea

The similarity and specificity of expression profiles in the cochlear epithelium

Mapping deafness-associated genes across species

Discussion

Methods

Animals and ethical compliance

Macaque RNA-seq, quality control, and clustering

Sample collection

Purification of nuclei

10× Chromium RNA sequencing

Quality control of RNA-seq data

Clustering of RNA-seq data

Mouse RNA-seq, quality control, and clustering

Sample collection

Nuclear purification

10× Chromium RNA sequencing procedure

Quality control of RNA-seq data

Clustering of RNA-seq data

Integration and clustering

Label transfer

DEG calculation

Conservation of gene families

Cell-type replicability across species

GO analysis of conserved cluster markers

Cumulative distribution plot of the number of gene families

Inferring the cellular localization of HL genes

RNAscopeTM plus immunofluorescence

Immunofluorescence and imaging

Electrophysiological recording

Reporting summary

Data availability

Code availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Supplementary information

Rights and permissions

About this article

Cite this article

Share this article

Search

Quick links

RNAscope^TM plus immunofluorescence