Axonemal Dynein DNAH5 is Required for Sound Sensation in Drosophila Larvae

Abstract

Chordotonal neurons are responsible for sound sensation in Drosophila. However, little is known about how they respond to sound with high sensitivity. Using genetic labeling, we found one of the Drosophila axonemal dynein heavy chains, CG9492 (DNAH5), was specifically expressed in larval chordotonal neurons and showed a distribution restricted to proximal cilia. While DNAH5 mutation did not affect the cilium morphology or the trafficking of Inactive, a candidate auditory transduction channel, larvae with DNAH5 mutation had reduced startle responses to sound at low and medium intensities. Calcium imaging confirmed that DNAH5 functioned autonomously in chordotonal neurons for larval sound sensation. Furthermore, disrupting DNAH5 resulted in a decrease of spike firing responses to low-level sound in chordotonal neurons. Intriguingly, DNAH5 mutant larvae displayed an altered frequency tuning curve of the auditory organs. All together, our findings support a critical role of DNAH5 in tuning the frequency selectivity and the sound sensitivity of larval auditory neurons.

Supplementary material

The online version of this article (10.1007/s12264-021-00631-w) contains supplementary material, which is available to authorized users.

Introduction

The auditory system uses a sophisticated mechanism that amplifies the mechanical input to increase the acuity of sound detection. Hair cells, the mechanosensory cells in the vertebrate cochlea, are equipped with an active amplification process. These cells respond to mechanical vibrations with cell body contractions, which in turn provide positive feedback on the mechanical displacement, especially for vibrations induced by faint sound [1–5]. Such cochlear mechanics improves the ear’s sensitivity to faint sound and accounts for spontaneous otoacoustic emissions [6]. Recent studies have documented mechanical amplification in the Drosophila auditory system [7–9]. Mammalian hair cells rely on Prestin molecules for amplification, while the insects’ auditory neurons mechanically assist hearing by their specific ciliary structures [10–13].

The chordotonal (Cho) neurons on the body wall of Drosophila larvae act as mechanoreceptors for sensing sound and muscle stretch [14–18]. Due to the feasibility of physiological and morphological studies at single-cell resolution, larval Cho neurons serve as an attractive model system for investigating the key molecules and mechanisms involved in insect auditory transduction. At the tip of the dendrite, each Cho neuron bears a cilium, which displays a structural organization identical to its counterpart in adult flies [17, 19–21]. The cilium is divided into a proximal region extending from the ciliary base to the ciliary dilation and a distal region beyond the ciliary dilation. In the proximal region, the cilium has an axoneme comprising nine microtubular doublets with dynein arm-like protrusions [21]; however, these protrusions are absent from the distal ciliary region. The dynein arms have been implicated in ciliary motility [22–25], which is pivotal for mechanotransduction in Cho neurons.

The dynein arms can be categorized into inner and outer arms according to their locations within the cilia [26, 27]. Each outer and inner arm is a multi-protein complex formed by several axonemal dynein heavy chains, intermediate chains, light intermediate, and light chains. The heads of the dynein arms are formed by the heavy chains. The heavy chains possess a microtubule-binding domain and up to 6 ATPase domains [28]. The ATPase domains of the axonemal heavy chain are essential for the conversion of chemical energy into mechanical energy via ATP hydrolysis. Then, this energy generates a sliding of adjacent microtubules and gives rise to ciliary bending. These properties of the dynein arms demonstrate the functional significance of axonemal dynein heavy chains. In Drosophila, 11 genes encoding axonemal dynein heavy chains have been identified [29], several of which are promising candidates for the force-generating motors that drive the active amplification in the fly’s auditory organ [30]. Although a recent study has explored the role of one dynein heavy chain gene, dnah3, in auditory neuronal function in adult flies, it appears that this gene is not essential for the mechanical amplification, and the precise distribution of dynein heavy chains in Cho neurons is still unclear [31]. Moreover, the physiological function of axonemal dynein heavy chains in larval auditory neurons remains poorly understood. In addition, axonemal dynein heavy chains affect the beating frequency of motile cilia in Chlamydomonas [27], so it is worth studying whether dynein heavy chains participate in the frequency-tuning of Cho neurons in Drosophila.

Here, we report the role of the Drosophila gene dnah5 in auditory perception. The subcellular localization of DmDNAH5 in the proximal cilia of Cho neurons suggests a motor function of this protein. By combining behavioral analysis, Ca²⁺ imaging, and electrophysiological recording, we show that Dmdnah5 is required for both the amplification of sound responses and the frequency-tuning of Cho neurons.

Materials and Methods

Fly Strains

The UAS-GFP strain was a gift from Yongqing Zhang (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences); the UAS-GCaMP6m and UAS-tdTomato strains were gifts from Yuh-Nung Jan (University of California, San Francisco, USA). The w¹¹¹⁸ (BDSC_3605), Iav-Gal4 (BDSC_36360), y¹v¹P{nos-phiC31∖int.NLS}X; {CarryP}attP40 (BDSC_25709), y¹w^67c23P{Crey}1b; D*/TM³, Sb¹ (BDSC_851), UAS-dnah5-RNAi (BDSC_51725) lines were from the Bloomington Drosophila Stock Center (https://bdsc.indiana.edu/). Nos-Cas9 (TH00788.N) flies were from the Tsinghua Stock Center (http://fly.redbux.cn/). The dnah5¹, dnah5^KI, Dnah5::EGFP^Gal4, UAS-DNAH5 (attp40), IAV::RFP (attp40) lines were constructed in the laboratory using methods described in the following sections. w¹¹¹⁸ larvae were used as controls in all behavioral tests. The dnah5¹ mutant line was backcrossed to the w¹¹¹⁸ control for 5 generations. Flies were raised on standard medium at 25°C and 60% humidity under a 12 h/12 h light/dark cycle. See Table S1 for all fly genotypes used in experiments.

Generation of dnah5¹ Mutant Line

The dnah5¹ mutant line was generated using the CRISPR/ Cas9 system as previously described [32], with a gRNA (5′-GAAAATTCAATTCCGTACCG-3′) targeting the genomic region encoding the ATPase domain of Dmdnah5. The guide RNA was expressed under U6b promoter control in the presence of Nos-Cas9. The indels in offspring were confirmed by genomic DNA sequencing.

Generation of Dnah5::EGFP^Gal4 Knock-in Line

The 5’ and 3’ homologous arms of Dmdnah5 were cloned from nos-Cas9 flies by PCR amplification. To generate the targeting vector, the 5′ arm and 3′ arm flanking an attP-3P3-RFP-loxP cassette were cloned into pBSK+ vectors using a MultiS One Step Cloning Kit (Vazyme). The sgRNA sequences were as follows:

• sgRNA1: 5′-TAGTTGCCTAACAAGCGAAC-3′;

• sgRNA2: 5′-GCCGCTGCGACGATCCAGGG-3′.

A mixture of the targeting vector and two sgRNAs was injected into Nos-Cas9 embryos. F1 flies with RFP-positive eyes were selected as the dnah5^KI knock-in line and verified by genotyping. To generate the Dnah5::EGFP^Gal4 line, the replaced genomic region of dnah5 before the stop codon was amplified from w¹¹¹⁸ flies and cloned into the pBSK-attB-EGFP-T2A-Gal4 vector from Bowen Deng at the Chinese Institute for Brain Research [33], using the ClonExpress II One Step Cloning Kit (Vazyme).

These vectors were then injected into embryos from nos-phiC31 females crossed with dnah5^KI line males. The Dnah5::EGFP^Gal4 line was obtained from F1 flies with red eyes and verified by PCR. Finally, the Dnah5::EGFP^Gal4 line was crossed to hs-Cre flies to remove screening markers in the genome.

Generation of Transgenic Flies

We used the sequence of the Dnah5 transcript isoform G (NM_001300327) to design primers for Dnah5 gene cloning. The 14175-bp full-length Dnah5 cDNA was cloned from the total RNA of third-instar larvae by RT-PCR in three pieces using the primers:

• Forward primer1: 5′-ATGTTCGTGCAAAAGAAAAAGTTGGTGG-3′;

• Reverse primer1: 5′-GTGTGATAACCAAGCGATCCGTGC-3′.

• Forward primer2: 5′-GCACGGATCGCTTGGTTATCACAC-3′;

• Reverse primer2: 5′-GTGGTGAAGCGATCGGCGAATAC-3′.

• Forward primer3: 5′-GTATTCGCCGATCGCTTCACCAC-3′;

• Reverse primer3: 5′- TCACTTGATGTCACAGAGCAGGGCC-3′.

The three segments were assembled with a MultiS One Step Cloning Kit (Vazyme) and subcloned into the pACU2 vector [34] to generate the UAS-DNAH5 rescue line. The sequence of full-length Dnah5 was validated by sequencing.

The IAV::RFP construct was generated by in-frame fusing the coding sequence of RFP before the stop codon of the iav genomic sequence (including a 2-kb promoter and 3372-bp IAV coding sequence). The two fragments were assembled with the MultiS One Step Cloning Kit (Vazyme) and subcloned into the NheI and BamHI digested pBSK-attB plasmid.

The UAS-DNAH5 and IAV::RFP constructs were injected and integrated into the attP40 site on the second chromosome through phiC31-mediated gene integration. Transgenic flies were obtained and confirmed by PCR.

Locomotion Behavioral Assay

Behavioral assays are widely used in studies on the physiological functions of Drosophila genes [35, 36]. The locomotion assay was performed as previously described [37]. Third-instar larvae were rinsed with distilled water and transferred to the center of a 2% agarose plate with a diameter of 90 mm. Larvae were allowed to crawl freely at room temperature and videotaped for 90 s. Then a single larva was tracked and analyzed for the locomotor speed and movement trajectories in a 90-s period.

Startle Behavioral Assay

Flies were reared at room temperature in an incubator with 12-h light/dark cycles and humidity control. Based on previous studies [18, 38], we placed a 2% agar plate (100 mm) on top of a speaker and adjusted the parameters to achieve a 70–90 dB SPL (dB of Sound Pressure Level using 2 × 10⁻⁵ Pa as a reference) pure tone (500 Hz). Third-instar larvae were gently collected, washed twice with PBS, and transferred to the agar plate. After they began to crawl freely, their behavioral responses to sound stimuli were analyzed and recorded. To evaluate the sound response, the larvae were stimulated with a 1-s pure tone, and repeated 10 times. Stimulation was applied only when each larva was freely crawling. Larvae that did not respond to sound were scored 0. A score of 1 was given to larvae that showed startle responses such as retracting the mouthhook or retraction with excessive turning. The sum of responses in 10 trials served as the response score. All the behavior tests were conducted at room temperature.

Drosophila Larval Dissection and Electrophysiological Recording

As previously described [18, 39], fillet preparations were made by dissecting 3^rd-instar larvae in hemolymph-like saline (bath solution) containing (in mmol/L): 103 NaCl, 3 KCl, 5 N-[Tris(hydroxymethyl)methyl]-2-aminoethanesulfonic acid (TES), 10 trehalose, 10 glucose, 7 sucrose, 26 NaHCO₃, 1 NaH₂PO₄, and 4 MgCl₂, adjusted to pH 7.20 and 270–275 mOsm. Before use, 2 mmol/L Ca²⁺ was added to the saline. The muscles covering the lch1 Cho neurons were gently removed with fine forceps to expose the lch1 neurons. The Cho neurons were visualized and identified by fluorescent markers driven by Iav-Gal4. Glass electrodes were pulled with a P-97 puller (Sutter Instruments) from thick-walled borosilicate glass. The recording electrodes were pulled to a diameter of ~3 μm and filled with external saline solution. Action potentials were recorded extracellularly at a sample rate of 20 kHz and then low-pass filtered at 2 kHz. A Multiclamp 200B amplifier, Digidata 1550A, and Clampex 10.5 software (Molecular Devices) were used to acquire and process the data.

In Vivo Ca²⁺ Imaging

Ca²⁺ signals of larval Cho neurons induced by sound stimulation were recorded from third-instar larvae as previously described [18, 38]. We expressed the genetically-encoded Ca²⁺ indicator GCaMP6 [40] under the control of Iav-Gal4. A freely-moving larva was gently collected, rinsed twice with PBS, and then immobilized between a glass slide and a glass coverslip. The imaging data were acquired using an Olympus FV1200 confocal microscope with 20 × objective lens. To determine the effect of sound stimulation on Ca²⁺ dynamics, we applied a 500-Hz pure tone at 80 dB SPL to each larva. GCaMP6 and red fluorescent proteins (as references) were excited by a 488-nm and a 543-nm laser, respectively. The GCaMP6 fluorescence significantly increased upon sound stimulation, whereas the red fluorescent proteins displayed no change. All the imaging data were recorded from lch5 chordotonal neurons. F₀ was the average GCaMP6 intensity from the first 5 s before sound stimulation, and ΔF/F₀ was calculated for each data point. Ca²⁺ levels from the soma or the distal dendritic tips were used for analysis.

Statistical Analysis

GraphPad Prism 8 (https://www.graphpad.com/) was used for statistical analysis and graph generation. The data are shown as the mean ± SEM. Statistical significance is indicated by NS for no significance, *P < 0.05, **P < 0.01, and ***P < 0.001.

Results

Generation of Dmdnah5-Null Mutants

Human and mouse mutations in axonemal dynein have been shown to cause ciliary dysfunction diseases, characterized by symptoms such as randomization of left-right asymmetry, recurrent respiratory infections, and infertility [41–46]. The Drosophila gene CG9492 shares the highest protein sequence identity of 54.2% with human dnah5 [47], thus we named this gene Dmdnah5.

To explore the role of Dmdnah5, we first established a mutant line (dnah5¹) using the CRISPR/Cas9 system [32]. Because of the bulky 23.5-kb size of the gene span, we targeted the gene regions encoding the essential functional domains of Dmdnah5. Genomic DNA sequencing of dnah5¹ mutants confirmed a 5-bp deletion within the coding region of the predicted ATPase domain, resulting in a frame-shift and premature termination of the DNAH5 protein (Fig. 1A, B). These results indicated that dnah5¹ is likely a null mutant for the Dmdnah5 gene.

Fig. 1.

Generation of the DNAH5 mutant allele. A Schematic of the dnah5 gene span and position of the sgRNA target site. The sequence alteration in the dnah5¹ allele is shown below. B Verification of the indel in the dnah5 locus by direct sequencing of PCR products amplified from homozygous control and mutant flies. Dashed box in the chromatogram indicates the 5 bases deleted in the dnah5¹ mutant allele. The PAM sequence is underlined in magenta related to A.

Drosophila DNAH5 is Expressed in Chordotonal Neurons

Given the high sequence similarity between Drosophila dnah5 and its mammalian orthologs, we speculated that Dmdnah5 may be functionally conserved and might play a role in ciliated cells as with human dnah5. In Drosophila, motile cilia have been reported only in spermatozoa and chordotonal neurons [23]. To gain first insight into the roles of Dmdnah5, we examined whether dnah5 is expressed in chordotonal neurons and spermatozoa by the Gal4/UAS system. Using homologous recombination facilitated by the CRISPR/Cas9 system [33], we generated a knock-in allele of Dmdnah5, in which the last six exons were replaced by the gene-targeting construct. Taking advantage of this knock-in allele, we then obtained a Dnah5::EGFP^Gal4 line with an in-frame fusion EGFP-tag and a self-cleavable T2A-Gal4 following the last exon (Fig. S1).

The reporter UAS-GFP driven by Dnah5::EGFP^Gal4 (this knock-in strain allowed us to visualize the expression pattern of dnah5 without disrupting the gene function) revealed that Dmdnah5 is exclusively expressed in chordotonal neurons, including a cluster of five pentascolopidial neurons, lch5, and three singlet neurons lch1, vchA, and vchB, in each abdominal segment of the larva (Fig. 2A, B).

Fig. 2.

Expression pattern of Drosophila dnah5. A Dnah5^Gal4-driven GFP restricted to chordotonal neurons on the larval body wall. The Dnah5^Gal4 knock-in strain enables us to visualize the expression of dnah5 faithfully without affecting the function and endogenous expression of this gene. Blue dashed line outlines the larval body. White dashed line denotes one abdominal segment. Scale bar, 100 µm. B Dnah5^Gal4 labels all types of chordotonal neurons including lch1, lch5, vchA, and vchB neurons. White arrow marks the dendritic cilia of lch5 neurons, as shown at high magnification in the dashed box. Scale bars, 10 µm.

Although previous studies have reported the expression of Drosophila axonemal dyneins in spermatozoa [31], we found that the expression of Dmdnah5 was not detectable in the larval reproductive system (Fig. 2A). This result is consistent with the expression profile of dnah5 in diverse species [30, 46, 48].

Drosophila DNAH5 is Localized to the Proximal Cilia Region and is not Required for Cilia Morphogenesis or Trafficking of Inactive Channels

To analyze the endogenous expression pattern of DNAH5 protein, we crossed the Dnah5::EGFP^Gal4 line to a fusion reporter line of Inactive (IAV::RFP). Inactive (IAV) is a member of Drosophila transient receptor potential (TRP) channel subfamily V, which is known to exist as a heteromer with another TRPV channel, Nanchung, and has a restricted distribution in the proximal cilia region in Cho neurons [21, 49, 50]. Moreover, the expression and function of Inactive depend on Nanchung, and vice versa [49]. Notably, the fluorescence of GFP and RFP exhibited co-labeling in the proximal region of the cilia, indicating that DNAH5 is abundant in this region equipped with dynein arms (Figs 3A, B and S3A). In addition, DNAH5 had a low expression level in the soma and non-cilia dendritic region, indicating the existence of dispersed dynein heavy chains apart from those assembled into dynein arms. These results raise the possibility that Drosophila DNAH5 acts as a component of the dynein arm in the proximal cilia of Cho neurons.

Fig. 3.

Subcellular localization of Drosophila DNAH5 and trafficking of Inactive channel. A Endogenous expression of Drosophila DNAH5 in chordotonal neurons revealed by the knock-in allele Dnah5::EGFP with a fused EGFP tag. Counterstaining with IAV::RFP reveals the enrichment of DNAH5 within the proximal ciliary region. Insets: magnification of the ciliary region. Scale bar, 10 µm. B Sketch depicting the localization of Inactive (IAV) channels in the chordotonal organ. C Ciliary localization of IAV channels in lch5 revealed by IAV::RFP staining. Bright field images show the scolopale space wrapping the cilia. Scale bars, 10 µm.

Previous studies have shown that a dynein heavy chain encoding gene, btv, is responsible for retrograde transport toward the base of the cilium [21]. In btv mutants, the cilia become disorganized and the distal cilia show a leaked distribution of the Inactive channel, which is proposed to mediate the auditory transduction in Cho neurons. Accordingly, the btv mutant line shows largely diminished sound-evoked potentials [51]. To determine whether DNAH5 is also involved in the intraflagellar transport process, we first labeled the Cho neurons of dnah5¹ mutants with the UAS-GFP transgene, under the control of a Cho neuron-specific driver, Iav-Gal4. The morphology of Cho neurons in dnah5¹ mutants displayed no significant abnormality compared with that in control larvae (Fig. S2). Besides, we introduced the IAV::RFP fusion transgene into dnah5¹ mutant larvae. Judging from the area occupied by the RFP signal in the scolopale space, we found that IAV protein was normally trafficked to the proper ciliary region and showed a confined localization along the cilia in dnah5¹ mutant larvae (Figs 3C and S3B). These results suggest that DNAH5 is unlikely to be associated with the intraflagellar trafficking of the IAV ion channel or the morphological development of ciliated neurons.

Larval Locomotor Pattern and Sound-evoked Startle Response of dnah5¹ Mutants

Cho neurons have been reported to function as proprioceptors that provide feedback on locomotion; alternatively, they constitute the larval auditory receptor organ in which sound waves are transduced into electrical signals [14, 15]. To test whether DNAH5 contributes to the physiological functions of Cho neurons, we next analyzed the behavioral responses of dnah5¹ mutant larvae.

The crawling trajectories of dnah5¹ mutant larvae were similar to those of controls (Fig. 4A). With detailed analysis, we found that neither the form of movements (Fig. 4B), as revealed by the distance traveled in a certain period, nor the locomotor speed (Fig. 4C) was influenced by the loss of dnah5 gene function. These results suggest that the ability of Cho neurons to sense muscle stretch changes during locomotion does not require the participation of DNAH5.

Fig. 4.

DNAH5 is required for the sound-induced startle response in Drosophila larvae. A Crawling trajectories of Ctrl (w¹¹¹⁸) and dnah5¹-null mutant larvae. Scale bar, 1 cm. Gray arrows represent the travelling distance. B, C Travelling distance and average speed (in 90 s) are comparable in Ctrl and dnah5¹-null mutant larvae. Two-tailed unpaired Student’s t test was used to test the difference between Ctrl and dnah5¹ larvae. n = 24/group. Error bars indicate SEMs. NS, not statistically significant. D Schematic of the behavioral assay in wild-type larvae. Left: larvae crawl freely without sound stimuli. Gray arrows denote the crawling direction. Right: larvae show a startle responses to sound (head contraction and turning indicated by gray arrows). E Summary of startle responses to increasing sound intensity. The score is the sum of 10 tests for each larva. Compared to wild-type larvae, the behavior scores of dnah5 rescue larvae showed no significant differences at each intensity. n = 10/group. One-way analysis of variance followed by Holm–Sidak post hoc analysis was used for comparison among multiple groups at each sound intensity; *P < 0.05, **P < 0.01, NS, not significant vs Controls. All error bars denote ± SEM. The sound stimulus is a 500-Hz pure tone.

Avoidance behavior and startle responses allow animals to escape from potential harm [18, 52]. As reported in a previous study [18], Drosophila larvae showed evident startle responses with head retraction or head turning upon sound stimulation (Fig. 4D). As the intensity of pure tone stimuli (500 Hz) increased from 70 dB to 90 dB SPL, the startle behavior scores of wild-type larvae showed moderate increments. In contrast, the startle behavior scores of dnah5¹ mutants increased sharply with increasing sound intensity. Moreover, disruption of the dnah5 gene largely reduced behavioral responses to sound at low and medium intensities (70 dB and 80 dB SPL), while leaving the startle response to the 90 dB SPL pure tone intact (70 dB: P < 0.05 vs Control; 80 dB: P < 0.01 vs Control; 90 dB: P > 0.05 vs Control; Fig. 4E). The startle responses to low and medium sound levels were restored by overexpression of DNAH5 in Cho neurons lacking dnah5 gene function (70–90 dB: P > 0.05 vs Control; Fig. 4E). These results indicate that Drosophila DNAH5 is required for the sound-induced larval startle response, especially for the amplification of response to low-level sound.

Sound-evoked Ca²⁺ Responses Are Impaired in Cho Neurons of dnah5¹ Mutants

To further investigate how DNAH5 plays a role in sound sensation, we live-imaged intact third-instar larvae expressing GCaMP6 in Cho neurons. The Ca²⁺ dynamics was monitored as the fluorescence signal of GCaMP6 (Fig. 5A). When exposed to an 80 dB SPL pure tone for 2 s, the Ca²⁺ signal in lch5 Cho neurons of control larvae reached a peak at the end of the sound stimulus and fell back to the baseline within 6 s (Fig. 5A, B). Strikingly, the same stimulus elicited a dramatically decreased Ca²⁺ response in Cho neurons of the dnah5¹ mutant larvae (Ctrl: 141.40 ± 12.64%; dnah5¹: 15.75 ± 8.64%; P < 0.001; Fig. 5C). Consistent with the behavioral response of dnah5¹ mutants to an 80 dB SPL sound, the Ca²⁺ increase at this intensity was reduced but not eliminated, suggesting that DNAH5 is associated with sound perception mechanisms other than auditory transduction. Moreover, specific knockdown of the dnah5 gene in Cho neurons led to a similar although less severe defect in the Ca²⁺ response (66.65 ± 6.12%, P < 0.001 vs Control; Fig. 5B, C). Taken together, these findings support the conclusion that DNAH5 acts in the process that increases the sensitivity of Cho neurons to lower sound levels.

Fig. 5.

Calcium responses of Cho neurons are reduced in DNAH5 mutants. A Sound-induced Ca²⁺ increases in lch5 Cho neurons of wild-type larvae (Ctrl) and dnah5-null mutant larvae (dnah5¹) at different time points monitored using GCaMP6. Sound stimuli are 500-Hz pure tones at 80 dB SPL. White dashed lines denote the borders of lch5 Cho neurons (rainbow color range: 0–4095; scale bars, 10 μm). B Ca²⁺ dynamics over time in lch5 neurons of control, Dnah5-RNAi, and dnah5¹ larvae. Black bar indicates sound stimulation (80 dB SPL). n = 9 for Ctrl, n = 6 for Dnah5-RNAi, n = 7 for dnah5¹; shaded areas represent SEM. C Statistical analysis of Ca²⁺ responses of lch5 neurons with different genotypes. Sound stimuli are 500-Hz pure tones (80 dB SPL). n = 9 for Ctrl, n = 6 for Dnah5-RNAi, n = 7 for dnah5¹; one-way analysis of variance followed by Holm–Sidak post hoc analysis was used for comparison among the groups; ***P < 0.001; mean ± SEM.

Sound-induced Action Potential Firing in Cho Neurons of dnah5¹ Mutants

The high sensitivity and temporal resolution of extracellular electrophysiological recordings in single neurons allowed us to characterize the role of DNAH5 in Cho neurons in Drosophila larvae. As described previously [15, 18], Cho neurons showed clear increases in spike frequency to pure tones. This physiological response increased in a stimulus-dependent manner as sound intensity gradually increased. We found that the lch1 Cho neurons of the dnah5¹ mutant did not respond to a low sound level (60 dB SPL) that was sufficient to trigger firing in Cho neurons of control larvae; however, higher sound levels elicited an impaired response in Cho neurons lacking functional DNAH5 (60 dB: P < 0.05 vs Control; 70 dB: P < 0.01 vs Control; 80 dB: P < 0.01 vs Control; Figs 6A, B, D and S4A, B). Moreover, overexpressing DNAH5 in Cho neurons driven by Iav-Gal4 fully rescued the defect in dnah5¹ mutant larvae (50–90 dB: P > 0.05 vs Controls; Figs 6C, D and S4C). These phenotypes revealed by extracellular recordings are in accord with the results of behavioral assays, reflecting that DNAH5 is required for the physiological response to low-intensity sound in Drosophila larvae.

Fig. 6.

DNAH5 regulates electrophysiological responses of Cho neurons to sound stimuli. A–C Representative recordings showing the action potential firing responses in lch1 neurons of control (A), dnah5¹-null mutant (B), and dnah5 rescue (C) to 80 dB SPL sound (500-Hz pure tone). D Intensity-dependent curve of sound-induced action potentials recorded from control, dnah5¹, and dnah5 rescue larvae. Δno. APs denotes increase in the frequency of action potentials in 1 s after sound stimulus onset (500-Hz pure tone) compared to 1 s before stimulus onset. Curves are fitted with a Boltzmann equation (n = 7, 5, and 6 for group Ctrl, dnah5¹, and dnah5 rescue, respectively; *P < 0.05; **P < 0.01; NS, not significant vs Controls). E Group data of the spontaneous spike frequency of lch1 neurons in Control (n = 9), dnah5¹ mutant (n = 6), and dnah5 rescue larvae (n = 6). NS, not significant vs Control. F Exponential fits of numbers of sound-induced action potentials in different genotypes. Bin width, 100 ms (500-Hz pure tones at 80 dB SPL; n = 9, 5, and 6 for Ctrl, dnah5¹, and dnah5 rescue, respectively). G, H Representative recordings of lch1 neurons from wild-type (upper) and dnah5¹ mutant (lower) larvae in response to 80 dB SPL sound at 300 Hz (G) and 400 Hz (H). I Averaged tuning curves of action potential responses to pure tone sound ranging from 200 Hz to 800 Hz at 80 dB SPL. n = 5/group; *P < 0.05; NS, not significant. We used the two-tailed unpaired t test for comparison between two groups and one-way analysis of variance followed by Holm–Sidak post hoc analysis for comparison among three groups. All error bars denote ± SEM.

We also found that the rates of spontaneous action potentials in lch1 Cho neurons were comparable among wild-type, dnah5¹ mutant, and dnah5 rescue larvae (dnah5¹ vs Control: P > 0.05; dnah5 rescue vs Control: P > 0.05; Fig. 6E). Moreover, we noted that the mutation in Dmdnah5 slightly altered the adaptation kinetics of the action potential response to the 80 dB SPL pure tone. The sound response of control larvae was rapidly-adapting with an adaptation time constant of 200.16 ± 23.91 ms, while dnah5¹ mutant larvae exhibited a prolonged adaptation time constant of 420.88 ± 161.81 ms, which was rescued to 243.43 ± 55.92 ms by the introduction of a wild-type dnah5 transcript into the dnah5¹ mutant background (Fig. 6F).

In addition, axonemal dyneins have also been reported to influence the beating frequency of motile cilia [27, 53]. Chlamydomonas has been a particularly powerful model organism for the investigation of mechanisms underlying ciliary assembly and motility. In light of the ciliary beating defects in Chlamydomonas mutants that are unable to assemble inner arms or outer arms, previous studies have shown that outer arm dyneins control ciliary beat frequency whereas inner arm dyneins determine the amplitude of the beat waveform [54–56].

Given that DNAH5 is a component of the outer arm [53], we further investigated whether mutation in this gene affects the frequency tuning of Cho neurons. We evaluated the Cho neuronal firing response to different frequencies of sound at the same intensity (80 dB SPL). In the frequency range tested (200–800 Hz), control larvae showed the highest sensitivity to a pure tone of 500 Hz. Interestingly, the tuning curve of dnah5¹ mutants displayed two prominent peaks instead of only one, with the maximal responses at ~300 Hz and 500 Hz (300–500 Hz: P < 0.05 vs Controls; 600–800 Hz: P > 0.05 vs Controls; Figs 6A, B, G–I and S5A, B). The alteration of the tuning curve by dnah5 mutation suggests that DNAH5 plays roles not only in the amplification of sound responses but also in frequency tuning in Cho neurons.

Discussion

In the current work, we provided multiple lines of evidence to demonstrate the role of one axonemal dynein heavy chain, CG9492 (DNAH5), in sound sensation in Drosophila larvae. Using genetic labeling, we first found that DNAH5 was expressed in Cho neurons and localized to the proximal cilia of Cho neurons. Then, based on the startle behavioral assays, we noted that the dnah5¹ mutant showed decreased startle responses to lower sound levels and this phenotype was rescued by expression of DNAH5 in Cho neurons. Moreover, using in vivo real-time Ca²⁺ imaging system in Cho neurons, we found that the Ca²⁺ response to sound was reduced in dnah5¹ larvae, compared with that in wild-type. In addition, by means of extracellular electrophysiological recordings in Cho neurons, we found that the dnah5¹ mutant showed lower sensitivity and different frequency tuning to pure tones. On the basis of these results, we concluded that DNAH5 resides in Cho neurons and participates in the sound response of Drosophila larvae.

In Drosophila, the cilium of auditory neurons is thought to be the place where sound signal transduction takes place [20]. Two transient receptor potential (TRP) ion channels, heteromer Nanchung-Inactive and NompC, are localized to the proximal and distal cilium, respectively [39, 49, 57]. They are candidate mechanotransduction channels in sound sensation and proposed as the “Nanchung-Inactive” model and the “NompC” model. In the “Nanchung-Inactive” model, the compound action potential of nanchung and inactive mutants is not elicited by sound stimulation that suffices to evoke the electrical response in wild-type flies [49, 57]. The sound-induced generator currents of giant fiber neurons, which are electrically coupled to auditory receptor neurons, are abolished in nanchung and inactive mutants. However, nompC mutants only display reduced sound-evoked generator currents [58]. In the “NompC” model, the nonlinear mechanical amplification of the nompC mutant, which is coupled to auditory neurons, is abolished. However, the nonlinear mechanical amplification is increased in nanchung or inactive mutants [59].

In the current study, we found that DNAH5 is localized to the proximal cilium of Cho neurons in larvae. In addition, DNAH5 co-localized with Inactive (Fig. 3A). This interesting characteristic suggests that DNAH5 affects the function of Nanchung-Inactive to modulate the sound response of Drosophila larvae. Another gene encoding dynein heavy chain, btv, is responsible for retrograde transport toward the base of the cilium [21]. In btv mutants, the cilium becomes disorganized and its distal part shows a leaky distribution of Inactive. In contrast to btv, we found that dnah5¹ exhibited normal organization of Cho neurons and subcellular localization of Inactive. These results demonstrate that DNAH5 is neither involved in the intraflagellar transport process nor morphological development. On the other hand, previous studies have shown dynein heavy chains form the head of the dynein arms, and further bind to microtubules [27, 60]. Each dynein heavy chain contains up to 6 ATPase domains and is essential for energy generation via ATP hydrolysis. This could give rise to a sliding of adjacent microtubules and ciliary bending. Since mammalian TRPV channels have been shown to bind to microtubules [61], the TRPV channels in Drosophila, Nanchung-Inactive, might have a similar association with microtubules in Cho neurons as well. Based on these findings, it is possible that DNAH5 participates in regulation of the sound-induced ciliary oscillation and then alters the activity of Nanchung-Inactive in Cho neurons of Drosophila larvae. Therefore, the mutation of DNAH5 affects larval responses to sound stimulation.

We also noted that the phenotype of the dnah5 mutant showed strong similarity to that of the nompC mutant [18]. Previous studies have shown that they both bind to microtubules [60, 62]. These results suggest that there is a connection between the functions of DNAH5 and NompC. However, the distribution of these two proteins is different (Fig. 3A) [39]. We speculate that the force that emerges from DNAH5 might be conducted along microtubules to distal cilium where NompC localized. This process might further result in modifications of both the mechanotransduction and amplification in auditory perception.

Frequency tuning is one of the most important characteristics of sound perception. For instance, the locust auditory organ displays a frequency tuning curve with the maximal response at 3 kHz [63], while the chordotonal neurons of Drosophila larvae were shown here to be most sensitive to a 500-Hz pure tone (Fig. 6I). Interestingly, the mutation of DNAH5 gave rise to changes in frequency tuning of the firing response to pure tones. The dnah5¹ larvae showed two prominent peaks with the maximum responses at 500 Hz and 300 Hz (Fig. 6I). This phenotype indicates that Drosophila DNAH5 plays an important role in frequency tuning and might further affect the behavioral responses to sound. Although mechanotransduction in the mammalian hair cell depends on transmembrane channel-like (TMC) genes rather than TRP channels, force-generating motor proteins, such as myosins, also participate in hearing [64–67], raising the question of whether myosin proteins are involved in the frequency tuning of hair cells.

Taken together, we demonstrated the expression pattern and subcellular localization of DNAH5 and its physiological functions in sound sensation in Drosophila. Our results show clearly that DNAH5 plays an important role in the sound response of Cho neurons in Drosophila larvae. Furthermore, this study provides insights into the functional mechanism of dyneins in hearing.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Conflict of interest

The authors declare that they have no conflict of interest.