Transcriptomic Analyses of Inner Ear Sensory Epithelia in Zebrafish

Abstract

Analysis of gene expression has the potential to assist in the understanding of multiple cellular processes including proliferation, cell-fate specification, senesence, and activity in both healthy and disease states. Zebrafish model has been increasingly used to understand the process of hearing and the development of the vertebrate auditory system. Within the zebrafish inner ear, there are three otolith organs, each containing a sensory macula of hair cells. The saccular macula is primarily involved in hearing, the utricular macula is primarily involved in balance and the function of the lagenar macula is not completely understood. The goal of this study is to understand the transcriptional differences in the sensory macula associated with different otolith organs with the intention of understanding the genetic mechanisms responsible for the distinct role each organ plays in sensory perception. The sensory maculae of the saccule, utricle, and lagena were dissected out of adult Et(krt4:GFP)^sqet4 zebrafish expressing green fluorescent protein in hair cells for transcriptional analysis. The total RNAs of the maculae were isolated and analyzed by RNA GeneChip microarray. Several of the differentially expressed genes are known to be involved in deafness, otolith development and balance. Gene expression among these otolith organs was very well conserved with less than 10% of genes showing differential expression. Data from this study will help to elucidate which genes are involved in hearing and balance. Furthermore, the findings of this study will assist in the development of the zebrafish model for human hearing and balance disorders.

INTRODUCTION

Microarray technology is a revolutionary tool used to query samples for analyzing thousands of genes simultaneously (Chen et al., 2002; Kierzek et al., 2015). By using this technology, we can get massive amounts of data including the information on the genes of interest and can get clues about the physiological function of genes (Serifi et al., 2016). Microarray is an ideal tool for biological and medical study (Streets et al., 2017; Keen and Sigmund, 2017). The traditional methods of gene detection include enzyme digestion, restriction fragment length polymorphism, and direct sequencing. These methods are time consuming, expensive, and they do not provide quantitative data. Furthermore, they are unable to detect multiple mutations in different genes at a given time.

Zebrafish has become an important model for studying human genetic diseases due to its easy maintenance, embryo transparency, in vitro fertilization, various mutant species, and well-established genetic study methods (Whitfield, 2002; Nicolson, 2005; Yariz et al., 2012; Grati et al., 2015). With the advent of new transgenic and imaging technologies that allow the analysis of phenotypes in astonishing detail, the zebrafish will remain at the forefront of hearing research for many years to come. CRISPR/ Cas9 can easily be used as a reverse genetics technique to verify gene function in zebrafish (Hwang et al., 2013; Jao et al., 2013; Zou et al., 2015; Shaw et al., 2017; Zou et al., 2019). Therefore, reverse genetics as well as forward genetics techniques can be used to deduce the genetic developmental pathway in zebrafish. There is a 70% similarity between the genomes of zebrafish and humans, which makes the zebrafish an ideal model for studying genes implicated in human hearing loss (Howe et al., 2013). However, there are only a few studies reported regarding transcriptomes of zebrafish inner ear which limits our ability to extrapolate the findings to humans (McDermott et al., 2007; Barta et al., 2018).

The inner ear of adult zebrafish has three otolith organs (the saccule, utricle, and lagena) that are very similar in structure, each of which is composed of a cell-packed sensory epithelium coupled with an otolith (Platt, 1993; Bang et al., 2001). There are only the saccule and utricle in the otic vesicle during the first week post-fertilization, and the lagena is formed after 7 days postfertilization (Haddon and Lewis, 1996; Bang et al., 2001; Bever and Fekete, 2002). It has been demonstrated that in zebrafish the saccule is a major hearing organ and the utricle is a key balance organ (Riley and Moorman, 2000; Yao et al., 2016). However, the function of lagena is still unknown. To understand the gene expression pattern of zebrafish inner ear, and to differentiate the genes expressed in hearing and balance sensory epithelia, we examined the transcriptomes of sensory epithelia of all three otolith organs using microarray analysis. Results from our analysis indicate that a large portion of the genes are expressed at similar levels across tissue types and that some genes are differentially expressed among these otolith organs. Quantifying gene expression among these different tissue types will allow us to understand the molecular mechanisms mediating specific physiological functions of otolith organs in zebrafish. This study will also further validate the use of zebrafish model to understand the biology of the human inner ear.

MATERIALS AND METHODS

Dissection and Isolation of Zebrafish Otolith Organs

Adult transgenic Et(krt4:GFP)^sqet4 zebrafish (Parinov et al., 2004; Gleason et al., 2009; Go et al., 2010; Lu and DeSmidt, 2013; Zamora and Lu, 2013) at 2 years old were used for the experiments in this study. The animal care protocol for all procedures used in this study was approved by the University of Miami Animal Care and Use Committee and complies with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. After zebrafish were anesthetized with 0.01% buffered MS-222 solution, individual otolith organs such as saccules, utricles, and lagenas were carefully dissected out from the brain cavity as described by Liang and Burgess (Liang and Burgess, 2009). Under a dissecting fluorescence microscope (Zeiss SteReo Discovery V20), the otolith was removed from each otolith organ, and the nerve bundle attached to the sensory epithelium was peeled away with a pair of fine forceps. Finally, the membrane surrounding the neurosensory epithelium was trimmed off using a pair of Vannas spring scissors. Eight sensory epithelia from each otolith organ were collected.

RNA Extraction and Purification

Total RNA, including small RNAs (more than 18 nt), from saccular, utricular, and lagenar epithelia separately suspended in RNAlater were extracted and purified using the Qiagen miRNeasy Mini Kit. On-column DNase digestion was performed to further eliminate DNA contamination in the collected RNA. Quality and quantity of RNA were determined using an Agilent 2100 BioAnalyzer. The experiment was repeated twice for two separate biological replicates.

GeneChip Microarray

An amount of 15-ng total RNA per sample was prepared using the Ovation Pico WTA System V2 (part#3302–12) and yielded cDNA product that was fragmented and labeled using the Encore Biotin Module according to the manufacturer’s protocol. Affymetrix Zebrafish Gene 1.0 ST Array (part# 902007) were scanned using GeneChip Scanner 3000 7G system. The Affymetrix GeneChip Command Console Software was used to perform background subtraction, normalization, and assess quality control metrics before being passed on for analysis.

Data Analysis

Data analysis followed a detailed procedure specific for zebrafish 1.0 ST array (Musso et al., 2015). In particular, raw CEL files were processed using the Oligo package (Carvalho and Irizarry, 2010) as part of the Bioconductor suite (www.bioconductor.org) in the R statistical framework (www.r-project.org). The background subtraction and normalization were performed using the robust multiarray average method implemented in the Oligo package. Boxplots of intensity values were compared for all chips before and after normalization to visualize the corresponding effects on mean and quartile values. Following normalization, probeset IDs were matched to corresponding transcript IDs. Specifically, the zebrafish 1.0 ST array NetAffx annotation file was downloaded in CSV format from the Affymetrix website (www.affymetrix.com), and transcript/gene IDs corresponding to given probe IDs were extracted. All transcript and gene IDs were mapped to corresponding Ensemble gene IDs. For Ensemble transcript IDs, corresponding Ensemble gene IDs were obtained using the BioMart community portal (Smedley et al., 2015). The Synergizer web application (Berriz and Roth, 2008) was used to convert gene IDs from other annotation frameworks to Ensemble gene IDs. Batch effects have been discovered by principle component analysis (PCA) plot and removed using function removeBatchEffect from Bioconductor package limma (Law et al., 2016; Koper et al., 2017). Differential expression analysis was tested under default parameters (P < 0.05, False Discovery Rate (FDR) < 0.05).

Measurement of gene expression in sensory tissues by quantitative reverse transcription PCR

Tissues from the inner ear were dissected as described earlier, and RNA from the utricle, saccule, and lagena was extracted using Quick-RNA microprep kit (Zymo Research, Irvine, CA), and further purified using RNA Clean and Concentrator kit (Zymo Research, Irvine, CA). cDNA was synthesized using Superscript III First-Strand Synthesis kit of reverse transcription PCR (RT-PCR; Life Technologies Inc., Carlsbad, CA) from 1 μg of total RNA. The synthesized cDNA from individual tissues was used as template for quantitative RT-PCR (qRT-PCR) using gene-specific primers. The qRT-PCR was performed using 1X Platinum SYBR Green Quantitative PCR SuperMix-UDG (with Carboxy-X-Rhodamine (ROX) reference dye at a final concentration of 50 nM) and 0.3 μM each of the forward and reverse primers. The gene expression was quantified using amplification and dissociation curves following 2^−ΔΔCt method, and values from the utricle were used for normalization to compare the gene expression in the saccule and lagena.

RESULTS AND DISCUSSION

In order to ensure the accuracy and quality of sensory epithelia that we dissected for this study, we used transgenic Et(krt4:GFP)^sqet4 zebrafish to isolate the otolith organs. The hair cells of transgenic Et(krt4:GFP)^sqet4 zebrafish express green fluorescent protein (GFP) (Parinov et al., 2004; Gleason et al., 2009; Go et al., 2010; Lu and DeSmidt, 2013; Zamora and Lu, 2013; Grati et al., 2015). The GFP fluorescent marker facilitates morphological observations of live hair cells and quantification of hair cells in the sensory epithelium of the zebrafish inner ear. The experiment was repeated twice to provide two biological replicates for the analysis. The RNA expression profile of three otolith organ sensory epithelia was determined by microarray analysis. Figure 1 shows the PCA plot after removing batch effects. Samples of the same otolith organs are clustered together, indicating the samples are separated by the biological variation of interest. About 86.9% (PC1, 59.1% plus PC2, 27.8%) of variance can be explained by the biological variation of the three different types of otolith organs. As demonstrated in Figure 1, data of the saccule and utricle are located far apart from each other, with those of the lagena lying in the middle of these two otolith organs.

Fig. 1.

PCA of microarray expression levels across three different otolith organs in zebrafish. The x-axis represents the greatest source of variance across all samples and the y-axis represents the second greatest. Each symbol represents the results of one microarray hybridization, as indicated in the legend. Red symbols represent RNA samples from the first batch and blue symbols represent RNA samples from the second batch. Six samples are well grouped by different otolith organs, indicating that there is less variance between samples of the same otolith organs than between two different batches. The two greatest principal components account for 86.9% (59.1% + 27.8%) of the variance.

Analysis of the transcripts obtained from this experiment allowed us to determine which genes are expressed in each otolith organ. The expression levels of the top 200 genes in the saccule, utricle, and lagena, are shown in Figures 2–4, respectively. Expression levels and abundance rankings for the top 200 genes in all the three otolith organs are also illustrated. These figures demonstrate that gene expression is largely conserved among all otolith organs. Some of the genes were abundantly expressed in all the three otoliths organs including runt-related transcription factor 1 (runx1), mitochondrial cytochrome oxidases including mt-co1, and mt-co3 as well as mitochondrial encoded NADH dehydrogenases such as mt-nd1–4.

Fig. 2.

Expression levels of top 200 genes in the saccule. The bars show the fluorescent density of each gene after normalization. Numbers in green, red, and blue signify the abundance rank of the genes in the saccule, utricle, and lagena, respectively. In this and all subsequent figures, the last six digits of ENSMUST identification numbers are used.

Fig. 4.

Expression levels of top 200 genes in the lagena. The bars show the fluorescent density of each gene after normalization. Numbers in green, red, and blue represent the abundance rank of the genes in the saccule, utricle, and lagena respectively.

Next, key differences in gene expression among the saccule, utricle and lagena were analyzed in order to understand the molecular basis underlying functional and developmental differences in these otolith organs. Transcript abundance in the saccule was compared with those of the utricle and lagena. Figure 5 illustrates the most differentially expressed genes in the saccule compared to the lagena (A) and utricle (B). We observed higher expression levels of wnt11r, cep41, prox1a, raraa, sema3e, gdf10a, otol1a, ctgfa, nr2f1a, and vwa2 in the saccule. These genes have also been demonstrated to be expressed in the otic vesicle in zebrafish in previous studies which serves to verify their relevance in hearing function of the saccule in zebrafish (Glasgow and Tomarev, 1998; Lee et al., 2012; Lu and DeSmidt, 2013; Pistocchi et al., 2008; Maier et al., 2014; http://zfin.org/). The functions of cep41, aldh1a2, otol1a, ctgfa, and nr2f1 (mouse orthologous genes) have been related to ear morphology and otic capsule development in mouse models (http://www.informatics.jax.org/). prox1a is widely expressed in the mouse inner ear, especially in the cochlea (Gray et al., 2004; Hartman et al., 2007; Hume et al., 2007; Dominguez-Frutos et al., 2009). Raraa has shown to be expressed in a wide variety of mouse ear compartments, such as in the vestibular component, the middle ear labyrinth, the organ of Corti, the limbus lamina spiralis, the spiral ligament, and the stria vascularis (Romand et al., 2002; Gray et al., 2004; Visel et al., 2004). Knockout mouse model for raraa orthologue showed hearing abnormalities (Lufkin et al., 1993). Sema3e, gdf10, and ctgf expression have been observed in the cochlea, otic capsule, pharyngotympanic tube, and tubotympanic recess of mice (Diez-Roux et al., 2011). Knockout mouse models for ctgf orthologue showed abnormalities in hearing and vestibular functions (Ivkovic et al., 2003; Doherty et al., 2010). Similar phenotypes were observed in a knockout mouse model (Qiu et al., 1997) for the nr2f1a orthologue that has been shown to be expressed in the cochlea (Jonk et al., 1994). In a patient with CHARGE syndrome and carrying a de novo balanced translocation involving Chromosomes 2 and 7, translocation breakpoints were mapped and Sema3e orthologue was identified within 200 kb of the breakpoint on 7q21.11 (Martin et al., 2001; Lalani et al., 2004). The chromosomal locations of human orthologues of cep41, cbln4, wnt11, masp1, aldh1a2, pla2g4c, sema3e, c1qtnf5, entpd5a, and apcdd1l have been associated with hereditary hearing loss in human patients (http://hereditaryhearingloss.org/). Figure 6 shows the differentially expressed genes in the utricle compared to the saccule (A) and lagena (B). Figure 7 shows the differentially expressed genes in the lagena compared to the saccule (A) and utricle (B). The comparisons show quite different gene expression profiles in three otolith organs, indicating different functional roles of these otolith organs in the zebrafish ear. The congruity between gene expression patterns in our study and others demonstrates the quality of the samples we collected from the three otolith organs in this study.

Fig. 5.

Differentially expressed genes in Saccule. The numerical values represent the fold difference in expression in the saccule versus the lagena (A) or the saccule versus the utricle (B).

Fig. 6.

Differentially expressed genes in the utricle. The numerical values represent the fold difference in expression in the utricle versus the saccule (A) or the utricle versus the lagena (B).

Fig. 7.

Differentially expressed genes in the lagena. The numerical values represent the fold difference in expression in the lagena versus the saccule (A) or the lagena versus the utricle (B).

Furthermore, we analyzed the differential expression of 96 transcripts to investigate their potential structural and functional contributions in the auditory system. We observed the expression of actn1, anxa6, arf1, argap17a, calm2a, calm2b, cdh23, dpysl2b, chd3, flnb, fscn2b, pls1, twf2b, ush1c, xirp2a, ywhaqa, and ywhaqb genes (Fig. 8). There are no studies available regarding the functional characterization and physiological relevance of these genes in the auditory system of zebrafish. However, some of these genes have been shown to be expressed in the mouse auditory system and implicated in maintaining the structure of stereocilia as well as in hearing. The expressions of calm2, cdh23, dpysl2b, flnb, pls1, twf2b, and ush1c orthologues have been observed in the mouse auditory system (Kamata et al., 1998; Di Palma et al., 2001; Ficker et al., 2004; Johnston et al., 2004; Visel et al., 2004; Lelli et al., 2009; Diez-Roux et al., 2011; Kamiya et al., 2014). Inner ear hair cells of Pls1-null mice have smaller and thinner stereocilia bundles and the mice develop progressive hearing loss at all frequencies (Taylor et al., 2015). Another gene USH1C was mapped to the PDZ73 gene to Chromosome 11p15.4-p15.1 by genomic sequence analysis and FISH (Scanlan et al., 1999). Mice with spontaneous hypomorphic mutations and knockout for USH1C showed behavioral, hearing, and vestibular abnormalities (Johnson et al., 2003; Lefevre et al., 2008; Grillet et al., 2009; Tian et al., 2010). Similarly, several phenotypes such as age-related hearing loss, degeneration of stereocilia, hair cell death, behavioral changes, and loss of harmonin in apex of hair cells have been observed in mouse models generated by chemical induction, knockouts, and spontaneous methods for cdh23 gene (Noben-Trauth et al., 1997; Di Palma et al., 2001; Zheng et al., 2005; Schwander et al., 2007; Bahloul et al., 2010; Manji et al., 2011; Han et al., 2014). In human studies, USH1C mutations were determined to be the cause of Usher syndrome type I in seven Acadians, one Pakistani, and one Canadian homozygous for the Acadian alleles (Ouyang et al., 2003), in two siblings from a Caucasian British family with hearing loss diagnosed at 4 years of age and retinitis pigmentosa (Saihan et al., 2011), and in 12 patients from 8 Israeli families of Yemenite Jewish origin with retinitis pigmentosa and late-onset hearing loss (Khateb et al., 2012). In a Cuban family, two different CDH23 mutations were identified (Bolz et al., 2001). Two nonsense and two frameshift mutations in the CDH23 gene in four families with USH1D (Bork et al., 2001), three novel mutations in 33 patients with type I Usher syndrome in whom USH1B and USH1C had been excluded (von Brederlow et al., 2002), a comprehensive catalog of 33 novel CDH23 mutations with recessive nonsyndromic deafness or Usher syndrome type I (Astuto et al., 2002), three families with Usher syndrome type I in which affected members carrying mutations in both CDH23 and PCDH15 (Zheng et al., 2005) have been observed.

Fig. 8.

Expression level of genes associated with to stereocilia structure in the saccule, utricle and lagena.

Clic4, slc17a8, tmc1, tmc2a, tmc2b, and trpv4 have been demonstrated to be highly expressed in zebrafish in saccular, lagenar, utricular, crista, and neuromast hair cells, which is in agreement with the findings of the present study (Fig. 9) (Amato et al., 2012; Einhorn et al., 2012; Gabashvili et al., 2007; R. Maeda et al., 2014; Mangos et al., 2007; Obholzer et al., 2008). In case of mice, Clic5 has been shown to express in saccular and utricular maculae (Gagnon et al., 2006). Mice homozygous for a spontaneous mutation in Clic5 exhibit head bobbing and circling behavior, inability to swim, and complete deafness by 7–8 months of age caused by dysmorphic stereocilia and progressive hair cell degeneration. Slc17a8 is not expressed in the auditory system; however, studies on mouse models show changes in behavior, hearing, vestibular ear, and nervous system, (Gras et al., 2008; Seal et al., 2008). Tmc1 expression has been shown in otic capsule, macula of utricle, utricle, and cochlea of mice by in situ hybridization (ISH) (Diez-Roux et al., 2011; Kawashima et al., 2011). Spontaneous mutation and knockout of Tmc1 in mouse models have shown hearing defects. In the mouse, tmc2 is expressed in the utricular macula, utricle, and cochlea. Reverse genetic experiments reveal that deletion of Tmc1 in mice caused deafness, whereas Tmc2 deletion did not cause any distinct phenotype. Interestingly, targeted deletion of both genes yielded mice with auditory and vestibular malfunction due to loss of mechanotransduction activity of inner ear hair cells (Kawashima et al., 2011).

Fig. 9.

Expression level of genes associated with to Ion channels in the saccule, utricle and lagena.

In humans, a truncating mutation in the CLIC5 was found to be the cause of autosomal recessive deafness-103 in two siblings of Turkish decent (Seco et al., 2016). By genomic sequence analysis, tmc1 gene was mapped to the Chromosome 9q13-q21 (Kurima et al., 2002). A Caucasian family in North American was also found to have hearing loss associated with the 9q13-q21 chromosome location. Ten Pakistani families were found to have mutations in tmc1 and were diagnosed with DFNB7/11 (Kitajiri et al., 2007), as were seven Turkish families (Hilgert et al., 2008) and members of a Chinese family that were diagnosed with DFNA36 (Zhao et al., 2014). By genomic sequence analysis, the tmc2 gene was mapped to Chromosome 20p13 (Kurima et al., 2002) and TRPV4 gene to Chromosome 12q24.1 (Liedtke et al., 2000).

The information about the genes involved in regulating cell cycle can provide important clues about the development and maintenance of the zebrafish inner ear. The genes related to cell cycle in the saccule, utricle, and lagena are shown in Figure 10. We observed the expression of brca2, ccnb2, cdkn1ba, cdkn1bb, cdkn1ca, gadd45aa, mad2l1, mcm4, notch2, and pmp22b in all the three otolith organs. The expression of cdkn1b has been seen in the cochlea, cochlear duct epithelium, and organ of Corti of mice (Nagahama et al., 2001; Chen et al., 2002; Hartman et al., 2007), and Notch2 in the mouse otocyst (Hamada et al., 1999). Ccnb2 has been mapped to the Chromosome 15q22.2. Cdkn1c has been associated with Beckwith–Wiedemann Syndrome (BWS) in nine unrelated Japanese patients and detected mutations in two BWS patients (Hatada et al., 1996). A study of patients with BWS identified multiple mutations in the Cdkn1c gene which presumably caused this syndrome (Lam et al., 1999). By sequencing this gene, they were able to identify five probands of this gene. Three probands were found among 7 familial cases where 2 probands were found among 54 sporadic cases. Patients with BWS have a reduced expression of Cdkn1c which indicates that the disease is caused by haploinsufficiency. (Algar et al., 1999) Another study examined 7 families with hereditary BWS identified 32 individuals with BWS and observed mutations in Cdkn1c (Algar et al., 2000). Another study (Romanelli et al., 2010) identified 7 novel CDKN1C mutations in 8 out of the 50 BWS patients. The MAD2L1 gene has been mapped to 5q23-q31 by fluorescence ISH (Xu et al., 1997).

Fig. 10.

Expression level of genes associated with cell cycle in the saccule, utricle and lagena.

Figure 11 shows the expression of over 80 genes that have been implicated in human deafness or related to hearing in animal models. col11a1a, col11ab, col11a2, col4a5, dfna5b, eya1, eya2, grhl2a, grhl2b, kcnq4, otofa, otofb, and slc26a5 are highly expressed in the zebrafish otic vesicle, hair cells, and semicircular canal (Croushore et al., 2005; Blasiole et al., 2006; Albert et al., 2007; Xiao and Baier, 2007; Baas et al., 2009; Yokoi et al., 2009; Fang et al., 2010; Hoffman et al., 2010; Janicke et al., 2010; Landgraf et al., 2010; Wu et al., 2014; Chatterjee et al., 2015; Wang et al., 2015). col11a1 has been shown to express in the inner ear and otic capsule of mice (Yoshioka et al., 1995; Diez-Roux et al., 2011) and spontaneous mutation mouse models show both auditory and vestibular abnormalities. Although col11a2 expression has not been observed in the auditory system of mice, a knockout model still seems to show vestibular abnormalities (Li et al., 2001). Abundant Dfna5 expression has been observed in the cochlea and inner ear of mice (Maeda et al., 2001) and Dfna5 knockout mice develop hearing abnormalities (Van Laer et al., 2005). Eya1 is expressed in otocyst, otocyst epithelium, inner ear, otic capsule, and vestibule (Abdelhak et al., 1997; Lillevali et al., 2004; Ozaki et al., 2004). Spontaneous Eya1 knockout and targeted mouse models show hearing abnormalities (Johnson et al., 1999; Xu et al., 1999; Ahmed et al., 2012). Grhl2 expression in mice has been observed in the inner ear, cochlea, middle ear, pharyngotympanic tube (Visel et al., 2004; Diez-Roux et al., 2011). Similarly, Kcnq4 expression has been observed in the utricle (Holt et al., 2007), and knockout mouse models for this gene show hearing abnormalities (Kharkovets et al., 2006). Otof expression has shown to be expressed in the inner ear of mice (Wilson et al., 2005; Schwander et al., 2007). Otoferlin deficient mice (Otof ^−/−) are profoundly deaf demonstrating complete abrogation of exocytosis in inner hair cells (Roux et al., 2006). Slc26a5 expression has been observed in the cochlea of mice (Lelli et al., 2009) and radiation, targeted, and knockout models show hearing abnormalities (Liberman et al., 2002; Gao et al., 2007; Palos et al., 2008). The mutations in SLC26A5 cause deafness in 2 Japanese siblings (Mutai et al., 2013), col11a2 in American, Dutch, Iranian, Tunisian, and Turkish families (Chakchouk et al., 2015), DFNA5 in 2 Dutch families, 2 Chinese families, 1 Iranian family, and a 328-member cohort study (Van Laer et al., 1997; Van Camp et al., 2002; Yu et al., 2004; Van Laer et al., 2008). An autosomal dominant form of progressive, non-syndromic sensorineural hearing loss was identified in an American family in which a specific GRHL2 mutation cosegregated with affected patients (Peters et al., 2002). Similarly, in a large five-generation family, the GRHL2 mutation was shown to cause deafness (Vona et al., 2013). A heterozygous mutation in KCNQ4 has been identified in a large deaf family, and it was concluded that KCNQ4 related hearing loss is intrinsic to outer hair cells and KCNQ4 forms heteromeric channels with KCNQ3 (Kubisch et al., 1999). A study on three families from the Netherlands and Belgium (Van Camp et al., 2002) and two from Indonesia and the United States (Coucke et al., 1994) found missense mutations in KCNQ4 gene. KCNQ4 was also shown to be the cause of deafness in a five-generation American family (Talebizadeh et al., 1999), Dutch family (Van Hauwe et al., 2000), and four-generation Spanish family (Mencía et al., 2008). Three Indian siblings with OTOF mutation suffered profound hearing loss (Yasunaga et al., 2000), and one out of four Druze families had a novel OTOF mutation (Adato et al., 2000). In one Cuban family, two Spanish families, and eight sporadic Spanish patients, an OTOF mutation was associated with nonsyndromic sensorineural hearing loss (Migliosi et al., 2002). The screening for the same OTOF mutation was extended to 289 additional unrelated families, finding 15 new cases, 9 of which were homozygous and 6 of which were heterozygous (Rodríguez-Ballesteros et al., 2003). Thus, the deafness genes that we identified in zebrafish have implications in human hearing loss validating the use of zebrafish models to identify and characterize human deafness genes.

Fig. 11.

Expression level of genes associated with deafness in the saccule, utricle and lagena.

To further confirm our microarray data, we performed real-time quantitative PCR (qRT-PCR). We used the top differentially expressed genes in three otolith organs from our microarray data. We computed the log2-fold difference between the expression of genes in saccule and lagena using utricle as the normalizer. As shown in Figure 12, we observed that the differentially expressed genes identified by the two techniques, qRT-PCR and microarray, are highly consistent. These differentially expressed genes may provide valuable information to understand different biological properties (such as structural and functional differences) of three otolith organs.

Fig. 12.

Expression levels of genes in the saccule and lagena using qRT-PCR; values from the utricle were used for normalization.

Remarkably, zebrafish and mammalian hair cells share many fundamental features at the cellular and molecular level (Coffin et al., 2004). Several genetic studies have demonstrated that molecules required by the auditory and vestibular system of zebrafish are also important in mice and humans (Nicolson, 2005; Yariz et al., 2012; Diaz-Horta et al., 2014; Grati et al., 2015; Bensaid et al., 2016; Varshney et al., 2016; DeSmidt et al., 2019; Zou et al., 2019). The inner ear of mammals is encased in the bony labyrinth deep inside the temporal bone, which prevents the sensory hair cells from being easily accessed. Contrarily, the hair cells of zebrafish larvae are found not only in the inner ear, but also within a lateral-line system that runs along exterior surface of the fish. Although the zebrafish uses hair cells in the lateral line system to detect water currents, the hair cells are remarkably similar in structure and function to the hair cells of the inner ear of mammals. Furthermore, zebrafish larvae are optically transparent. The lateral line contains a series of neuromasts, mechanoreceptors that form an interconnected network along the head and body of the fish. Because the hair cells of the lateral line are located on the external surface of the zebrafish, treatment and subsequent observation of live hair cells make the lateral line system particularly amenable to hair cell research. This access holds the promise of providing relatively straightforward pharmacodynamic assays and hair cell structure and in vivo functional studies. In addition, hair cells of zebrafish can mature within 24 hr (Kindt et al., 2012; Dow et al., 2015), whereas mammalian hair cells develop over several weeks (Kraus and Aulbach-Kraus, 1981; Romand and Varela-Nieto, 2003). The rapid development of zebrafish hair cells allows the entire maturation process of the hair cells to be recorded within one imaging session. Furthermore, several genetic techniques have been well established using the zebrafish model. For example, transgenic modifications of the zebrafish genome can allow certain gene products such as fluorescent markers to be expressed in a cell- or tissue-specific manner (Kwan et al., 2007). The fluorescent markers can easily be used in the transparent larvae, which enables visualization of hair cell structures in vivo and allows for cellular functions to be dynamically imaged in living cells.

In summary, the present investigation is the first transcriptomic study investigating the quantitative expression differences among the saccule, utricle and lagena of adult zebrafish at the genomic level. Although the expression of all the genes investigated is conserved among three otolith organs, there are few genes that are highly differentially expressed in the saccule, utricle and lagena. These differential expressed genes may impart unique structures and physiological functions to these otolith organs. This data set provides a unique tool to the auditory and neuroscience community to explore the functional roles of these genes in future studies employing CRISPR/Cas9 genome editing.

Fig. 3.

Expression levels of top 200 genes in the utricle. The bars show the fluorescent density of each gene after normalization. Numbers in green, red, and blue represent the abundance rank of the genes in the saccule, utricle, and lagena respectively.