Drosophila melanogaster mutants of btv used in this study are listed in Table 1. For P-induced male recombination, the transposase source was supplied by w⁺; Sp btv¹ pr rl cn/CyO; Dr P {Δ2-3}/TM6, Ubx. The Sternopleural (Sp) and purple (pr) genes are used as flanking markers to map btv. These flies were crossed to w/Y; P {GT1}BG01771 males. Dysgenic male offspring were crossed to w⁺; Df (2 L)TW12 Tft pr/CyO, pr females to recover Sp pr⁺ or Sp⁺ pr recombinants. Recombinants were subsequently tested for btv genotype by crossing to w; btv¹ 40AG13/CyO. Df (2 L)TW12 is unrelated to the btv region, while Tufted (Tft) is a dominant bristle marker. A total of 5,966 flies (250 pair matings) were screened and 9 male recombinants recovered (Supplementary Table S1), giving a recombination rate of 0.15%.

To recover putative deletions between the BG01771 insertion and the insertion sites of nearby P-element insertions KG02815 or KG08320, the BG01771 insertion was first crossed to flies containing the transposase source, w⁺; Sp/CyO; P {Δ2-3} Sb/TM6, Ubx. Resulting w⁺/w; BG01771/Sp; +/P {Δ2-3} Sb females were crossed to either w⁺/Y; KG02815 or w⁺/Y; KG08320 males. Dysgenic w/Y; BG01771/KG; +/P {Δ2-3} males were crossed to w; Sco/CyO females and the offspring screened for white eyes. About 330 putative deletions were crossed to w; btv¹ 40AG13/CyO to test for btv phenotype.

To measure sedentary behavior and associated loss of flight ability, we developed the drop zone assay (DZA; Supplementary Figure S2). A square plexiglass cover, with a hole in its center, was placed on top of a 4 liter glass beaker. Flies were introduced in groups of 10 through the hole into the beaker, observed for 1 min and assigned a score from 1 to 6 as follows. Score 1: the fly did not touch the beaker bottom but flew directly upwards and landed on the plexiglass ceiling. Score 2: the fly landed on the beaker wall and climbed to the top. Scores of 3, 4, 5, or 6: the fly fell to the beaker floor, and either climbed immediately more than halfway to the top (3), climbed less than mid-way to the top (4), did not climb, but wandering around the beaker floor (5), or remained at the landing position (6).

Extracellular recordings used to assay the sound-evoked potentials (SEPs) in the antennal nerve were performed as previously described (Eberl et al., 2000; Eberl and Kernan, 2011). Transepithelial potentials (TEP) and mechanoreceptor potentials (MRP) were recorded from adult anterior notopleural bristle organs essentially as previously described (Kernan et al., 1994), except that decapitated flies were mounted on a chlorided silver pin, which served as the basal electrode; transepithelial potentials were recorded with an EPC7 amplifier (List Medical) in current-clamp mode, and data was acquired and analysed with Powerlab/LabChart (AD Instruments). TEP values are relative to a zero obtained by inserting the bristle electrode into the body cavity, in cisepithelial configuration. The MRP was calculated as the maximal absolute change in TEP within 100 ms of the stimulus onset. Adaptation was calculated as the difference between the minimum TEP reached and the TEP averaged over the last 200 ms of a 1 s stimulus, as a percent of the MRP.

Drosophila heads, with proboscii removed to facilitate infiltration, were fixed by immersion overnight at 4°C in a fixative containing 2.5% glutaraldehyde, 2.0% paraformaldehyde and 0.04% CaCl₂ in 0.1 M phosphate buffer, pH 7.4 (PB). CaCl₂ provides increased membrane stabilization. Heads were washed in PB, post-fixed with OsO₄, dehydrated in an ethanol series and embedded in Polybed 812. Ultrathin sections (75 nm) were stained with aqueous uranyl acetate and lead citrate and examined with a Hitachi 7000 electron microscope.

The fluorescently-tagged IFT proteins IFT88 [GFP-NompB (Han et al., 2003)] and IFT140 [RempA-VenusYFP (Lee et al., 2008)], expressed from transgenes, were imaged in situ by their native fluorescence, in combination with immunolabelling of the proteins Futsch or NompA. Pupal antennae or pupal or pharate adult abdomen and halteres were dissected in PBT (0.2% Triton-X in PBS) and fixed in 4% formaldehyde in PBT. After three 10 min washes in PBT, specimens were incubated in blocking solution (PBT with 5% normal goat serum) for 1 h at room temperature. Futsch was labelled by incubating with mAb 22C10 (Developmental Studies Hybridoma Bank, Iowa City, IA) at a 1:100 dilution for 2.5 h at room temperature, followed by incubation with Alexa Fluor 546-conjugated goat anti-mouse antibody (Invitrogen) at 1:500 dilution, for 2 h, also at room temperature. NompA was labelled with rabbit anti-NompA (Chung et al., 2001) at a 1:500 dilution, followed by incubation with Alexa Fluor 647-conjugated goat anti-rabbit (Invitrogen) at 1:500 dilution. Specimens were mounted with Vectashield mounting media (Vectorlabs, CA) on a microscope slide and examined with a confocal microscope (Leica SP5).

Genomic DNA was isolated from adults and digested with EcoRI, HindIII or XhoI, run on agarose gel, and transferred to Hybond membrane according to instructions provided with the membrane. Southern blot analysis was performed using DIG High Prime DNA Labeling and Detection Starter Kit I (Roche) according to manufacturer’s instructions. A dot blot was conducted to confirm labeling and determine the concentration of labeled probe.

Primers presented in figures and used in PCR reactions include:

Se13F: 5’-ACTTGTTATCGTCCAACACC-3′.

Dhc9R: 5’-GTGCCAGCAGAACTTGATGA-3′.

PS5F: 5’-CAGCAACATCATCTGCAGCA-3′.

PS7R: 5’-ATAAGAATGCGGCCGCAATCTACAGGCGAC-3′.

In a behavioral mutagenesis screen (Eberl et al., 1997), we isolated a mutant, 5P1, with reduced courtship song response. This mutant showed severely disrupted sound-evoked potentials (SEPs) in the antennal nerve (Eberl et al., 2000) and the corresponding gene was named beethoven (btv), with the btv^5P1 allele also referred to as btv¹ (Caldwell et al., 2003). Most btv¹ mutant flies retained a small SEP, especially if subjected to loud sound (Eberl et al., 2000), consistent with retention of a residual behavioral response to courtship song presentation (Eberl et al., 1997). This appears to be the null phenotype because overlapping deletions that remove the btv locus (such as Df (2 L)TW119/Df (2 L)TW201) also retain this residual SEP. To better understand the mutant phenotype, we examined the ciliary structure in more detail, and tested whether the residual response represented degeneration in progress. Thus, we compared electrophysiological phenotypes of btv mutants to controls 1–3 days after eclosion, after 9 days, and after 30 days. As btv⁺ control flies, we used the genetic background on which the btv¹ mutation was induced (Table 2; Eberl et al., 1997). This strain, like the Canton-S wild-type strain, shows a general age-dependent decline in SEPs (Supplementary Figure S1) though little change before 30 days of age.

Consistent with our previous results, we found that many, but not all, btv¹ flies showed small SEPs in response to a standard pulse song stimulus (Figure 1A). At a lower (20%) stimulus intensity, control flies showed SEPs of about 55% of their response to the standard amplitude regardless of age up to 30 days (Figures 1A,B). In contrast, average SEPs of btv¹ mutants were the same at loud and soft stimuli (Figure 1B). Therefore, we scored the percent of antennae that showed any detectable response. Compared to control flies, in which all antennae showed evoked responses, less than 40% of btv¹ antennae showed a response at 1–3 days, declining to less than 10% by 30 days (Figure 1C). Thus, the residual auditory function seen in btv¹ flies undergoes an early age-dependent decline.

To investigate in more detail the morphological effects on JO scolopidia resulting from loss of btv function, and to examine the degenerative effects at the morphological level, we used TEM to characterize JO scolopidia from control and mutant flies at 1–3 days, 9 days and 30 days old (Figures 2–4). The normal structure of an individual JO scolopidium is diagrammed in Figure 2, with TEM sections showing the inner and outer dendritic segments of the neuron, including the ciliary rootlet, the basal bodies, the proximal ciliary segment (bearing dynein arms), the ciliary dilation and the distal ciliary segment (lacking dynein arms). The scolopale cell, supported by actin-rich scolopale rods with embedded microtubules, encloses the sensory cilia within an extracellular cavity called the scolopale space, which likely contains a specialized receptor lymph to drive receptor potentials during mechanotransduction. The scolopale cell secretes NompA protein (Chung et al., 2001) into the extracellular matrix to form the dendritic cap, which connects the neuronal ciliary tips to the a2/a3 joint cuticle. A cap cell surrounds the distal end of the scolopale cell and enwraps the dendritic cap, but it is unclear whether it contributes proteins to the dendritic cap structure.

In the btv¹ mutant, we saw no significant changes to the inner dendritic segment at any of the ages tested, compared to control. The inner dendritic segments are tightly associated by desmosomes, both with each other and with scolopale rods in the apposed scolopale cell, and the base of the cilium always contains proximal and distal basal bodies and a ciliary root (Figure 3B). Near the basal bodies, the axoneme is usually complete with the expected nine microtubule doublets (Figure 3B). Beyond the ciliary base, the sensory cilia of btv¹ mutants show some segments constructed reasonably well, while other segments are fragmented and disorganized (Figure 3D) compared to btv⁺ controls (Figures 3A,C). At the mid-scolopale level of btv JO at all ages examined, we found frequent loss of microtubule doublets and loss of ciliary membrane (Figures 3E–J), often resulting in very deformed ciliary profiles (Figures 3G,J). Despite these defects, we often see clear dynein arms present on the axonemes (Figure 3G). At the expected location of the ciliary dilation, about three fourths of the distance from the basal bodies to ciliary tip (Figure 4A), we never see the organized grid-shaped matrix of the ciliary dilation in btv flies. Instead, we sometimes observe a complex of additional membranes, vacuoles and microtubules embedded in electron dense material (Figure 4B).

We were surprised to find, in control animals, excess membranes accumulating in the scolopale space with age. The appearance of these membranes varies from scolopidium to scolopidium, from loosely organized membranous material floating in the scolopale space (Figure 4A) to robust vacuoles whose expansion occludes the cilia to the edge of the space, as in btv (Figure 3E). To ensure that this is not specific to our control strain, 40AG13, we checked the Canton S wild-type strain and found similar levels of membrane accumulation in the scolopale space with age (not shown). These membranes are similar in btv mutants and control flies; however, in btv flies 9 days and older, the ciliary occlusions are sometimes accompanied by deformation of the ring of scolopale rods, and even invagination of membranes between scolopale rods, resulting in crescent-shaped scolopidial profiles (Figure 3H). In addition, only in the btv mutants, starting at the 9 day time-point, we begin to see additional elongated tubular structures (Figures 3D,I,J) that resemble strings of sausages. These accumulations may correlate with fewer ciliary profiles at 30 days (Figure 4B), and therefore may represent degenerating ciliary material. Alternatively, the appearance of membranous vesicles in the scolopale space may be caused by ciliary vesicle shedding (Ojeda Naharros and Nachury, 2022), which could be enhanced upon loss of retrograde IFT. The effects of age on control and mutant scolopidia are summarized in Table 2. To test whether the degenerative effects could result from activation of apoptosis, we stained sections of 14 day old btv mutant and control antennae with the TUNEL technique, and found no difference in labeling (not shown), suggesting that the consequences of age on btv mutants are not apoptotic in origin.

To understand the molecular basis of the btv phenotype, we pursued identification of the btv gene. Therefore, we carried out a series of genetic mapping studies and isolated several new alleles. As we shall describe, btv is identical with Dhc36D (also called DHC1b and CG15148), and encodes the retrograde IFT dynein motor, Dync2h1. Henceforth, we will use Dync2h1, according to the nomenclature proposed by Braschi et al. (2022). Because the Dync2h1 gene is strongly mutually nested with the CG5674 gene, it was critical that we used multiple approaches to conclusively determine whether btv corresponds to Dync2h1. This analysis is summarized below and in Figure 5 and described in additional detail in the Supplementary Material.

First, we mapped the btv¹ allele isolated in an ethyl methanesulfonate (EMS) mutagenesis screen (Eberl et al., 1997) to the 36E1-3 polytene chromosome region to the left of the rdo locus by deficiency mapping (Eberl et al., 2000; Caldwell et al., 2007) (Figure 5). Using P-induced male recombination (PIMR) (Chen et al., 1998), we refined the btv map position to the right of the KG08320 P-element (Figure 5 and Supplementary Table S1). Next, while the P {GT1}BG01771 and PBac {WH}f06878 insertions (Figure 5C) do not show a btv phenotype, btv⁵, the PBac {WH}f06884 insertion in exon 23 of Dync2h1 (Figure 5C and Table 1), fails to complement other btv alleles. Using a variety of strategies (see details in Supplementary Material), we generated several additional mutations, including btv⁴, del#1, and del#2, which likely affected both Dync2h1 and CG5674.

Because the btv¹ mutation was generated by EMS, we expected to find a point mutation. Further characterization, however, revealed a 401 bp deletion along with a 6 bp insertion (Figure 6C). The deletion removes 125 bp of Dync2h1 exon 13 and part of the intron between exons 12 and 13. Because this deleted region excludes any CG5674 sequences, the btv¹ lesion firmly establishes Dync2h1 as btv. This is supported by our discovery in btv², using Mismatch Endonuclease Arrays (MENA) (Comeron et al., 2016), of a single nucleotide deletion in Dync2h1 exon 22 resulting in a frameshift that introduces an early stop codon (Comeron et al., 2016). This position is in the middle of a large intron in the overlapping gene CG5674 on the opposite strand, and therefore unlikely to affect its function (Figure 5C). Finally, we found numerous sequence polymorphisms in different lab strains compared to the sequence used for the genome annotation (Supplementary Table S1), emphasizing the need to compare sequence deviations relative to the background strain on which the mutation was induced.

We find no differences in the electrophysiological phenotypes between btv alleles that also disrupt CG5674 and those that do not. Thus there is no evidence that CG5674 contributes to the btv phenotype.

Taken together, all these observations demonstrate that the btv gene corresponds to Dync2h1.

All our evidence indicates that btv corresponds to CG15148, which encodes a dynein heavy chain most similar to members of the Dync2h1 isoform. The annotated intron-exon structure and protein sequence evolved significantly over the course of this project. We sequenced PCR amplicons from cDNA of adult heads from 40AG13 and w¹¹¹⁸ control strains to define the intron-exon structure. Each amplicon was generated at least two independent times to distinguish sequence changes from PCR-induced changes. The current annotation (D. melanogaster r6.50; Gramates et al., 2022) resembles our empirical data most closely. Exons 2–24 (Figure 5C) match our cDNA sequencing results. We did not confirm exons 1 and 25, reported in the annotation. However, protein sequences predicted in exon 1 retain considerable conservation with Dync2h1 isoforms from other insects, and some modest conservation with vertebrate homologs. Upstream of the putative starting methionine in exon 1 (Figure 5D), we found a putative TATA box at about 635 bp upstream, and a perfect match to a Rfx transcription factor binding site consensus at about 100 bp upstream and to a Fd3F consensus site about 90 bp upstream (Laurençon et al., 2007; Cachero et al., 2011; Newton et al., 2012). This is consistent with Rfx-dependent (Laurençon et al., 2007) and Fd3F-dependent (Newton et al., 2012) regulation of btv. While the starting methionine in exon 1 is more likely correct, a second possible starting methionine is present in exon 2 (Figure 5D; Supplementary Figure S4). Protein sequences predicted from exon 25 share similarity with closely related insect homologs but not with vertebrate homologs. However, until the splicing structure is confirmed experimentally, we cannot be certain of the structure at the C-terminus. Assuming the longest prediction, the protein would consist of 4,237 amino acids, for a predicted molecular weight of 481.45 kDa, barring post-translational modifications (Supplementary Figure S4). Consistent with the structure of other dynein heavy chains, this Dync2h1 contains 4 conserved ATP-binding domains, called P-loops (Holzbaur and Vallee, 1994), as highlighted in the predicted protein sequence (Supplementary Figure S4) and in alignment with the Dync2h1 sequences from several species across the ciliated taxa (Supplementary Figure S5). The sequence of P-loop 1, GPAGTGKT, is identical to that in all other dyneins. P-loops 2 and 3 show intermediate levels of conservation, while P-loop 4 shows the least similarity between organisms (Supplementary Figure S5). To examine the localization of the Btv protein, we generated a monoclonal antibody against a Btv peptide in collaboration with the Developmental Studies Hybridoma Bank. Unfortunately, this antibody showed no signal on Western blots or in tissues (data not shown).

To confirm that this Dync2h1 dynein motor performs the retrograde IFT function in Drosophila, we examined the morphological phenotype of sensory neurons in the auditory organ, JO. We previously showed that the kinesin II motor, including the Klp64D heavy chain and DmKAP, mediates the anterograde IFT motor function (Sarpal et al., 2003). Loss of anterograde function causes complete failure to extend an axoneme from the JO basal bodies. With loss of retrograde motor function, we expect anterograde transport to be largely intact, resulting in partial axoneme assembly. However, retrograde transport is required for clearing the growing cilium of assembly byproducts, so we expect that as growth and assembly proceed, proteins and complexes that would normally be cleared will accumulate, leading to impaired assembly. Consistent with this, we previously reported morphological defects that include partial and variable assembly of the axoneme and abnormal swellings of the ciliary membrane (Eberl et al., 2000). Our further ultrastructural analysis (Figures 2–4) shows ciliary defects that are fully consistent with loss of retrograde IFT.

If retrograde IFT is disrupted, we should expect accumulation of IFT components in the cilium due to failure of the retrograde transport clearing function. Thus, we examined the expression and distribution of the GFP-NompB protein (Han et al., 2003), which encodes the IFT complex B protein homologous to Chlamydomonas IFT88, in btv mutants. In the mature JO, GFP-NompB normally localizes to the cilium, with most fluorescence near the ciliary tip, in the vicinity of the ciliary dilation (Figure 7A). In JO cilia of similarly staged in btv¹ mutants, imaged under identical conditions, the GFP-NompB protein correctly localizes to the cilium. However, localization within the cilium is disrupted, showing excess fluorescence both in proximal cilium close to the basal bodies, as well as ectopic fluorescence distally (Figure 7B). This distal fluorescence likely represents ciliary material sloughed off in a “trail of crumbs” fashion along the length of the tubular dendritic cap, where it is left behind developmentally rather than being cleared toward the cell body by retrograde IFT motor activity. Unlike growth of unattached cilia such as Chlamydomonas flagella, the JO cilia, embraced by the dendritic cap, extend in concert with extensive developmental stretching of the scolopidium (Figure 7C). The dendritic cap forms early in antennal development (Todi et al., 2008), and in the btv mutant, developmental stretching culminates in the cilium stretching to its normal length despite failure of proper axonemal growth. Thus, finding distal fragments of cilium indicates a failure of the cilium to remain intact during scolopidium elongation, rather than excess extension of the cilium. Therefore, the delocalization of GFP-NompB in btv mutants supports a retrograde IFT transport function for the btv-encoded dynein.

Cachero et al. (2011) reported that Dync2h1 transcripts in embryos show a “chordotonal-enriched pattern,” with some additional lower level expression in es organs. To test whether btv mutant phenotypes are confined to ch organs, we examined the localization of RempA-YFP, a labelled version of the IFT140 protein (Lee et al., 2008) in btv mutant campaniform organs and bristle organs. In ch organs, we had previously shown that RempA-YFP is de-localized in btv¹ mutants, extending beyond the normal location of the ciliary tip (Lee et al., 2008). In campaniform organs of the haltere, RempA-YFP is normally confined near the proximal connecting cilium and proximal to the distal, non-axonemal region of the dendrite, but in btv mutants RempA is released distally, extending to the dome and filling the blade-shaped distal tip of the dendrite (Figures 8A–C). Similarly, in btv mutant bristle organs of the adult abdomen, RempA colocalizes with the distal dendritic cap marker NompA, in contrast with their discrete localization patterns in the wild type (Figures 8D,E).

To test for electrophysiological effects of btv mutations on bristle organs, we recorded transepithelial potentials (TEPs) and mechanoreceptor potentials (MRPs) (Figures 9A,B) as previously described (Kernan et al., 1994). We found that, although TEPs do not differ significantly between btv mutants and control flies (Figure 9C), MRPs are reduced by about 50% in btv mutants (8.1 ± 0.9 mV) compared to controls (17.4 ± 1.3 mV; t-test with Welch’s correction, p < 0.0001) (Figure 9D). Adaptation of the MRP, (ratio of peak and end-stimulus MRP amplitudes) is the same in btv and controls (Figure 9E).

Regarding age effects on bristle organ function, although MRP amplitudes are slightly reduced in old (15.5 ± 1.7 mV) vs. young (20.9 ± 1.7 mV) control flies (t-test, p < 0.04) (Supplementary Figure S6), this is not significant in btv mutants (7.6 ± 1.6 mV vs. 8.5 ± 1.1 mV; t-test, p = 0.66). Six out of 38 old bristles were non-responders, but these were divided between both control (3 of 21) and btv mutants (3 of 17). These data do not support accelerated degeneration in btv mutant bristles.

Several lines of evidence support the conclusion that the btv-encoded Dync2h1 carries out the retrograde IFT function. First, as in all other organisms characterized, there is only a single Dync2h1 gene in the Drosophila genome. Ciliary function for Dync2h1 is supported by studies that report btv gene expression in ciliated cells. The btv gene expression is regulated by the Rfx transcription factor (Laurençon et al., 2007), as would be expected for a central component of IFT. In that study, btv gene expression was reduced by 124 fold in dRfx mutants. By in situ hybridization, btv gene expression was seen in embryonic ch organs, and in embryonic cell-sorting microarray experiments, btv gene expression showed 8-fold enrichment in cell lineages associated with the chordotonal proneural gene atonal (Cachero et al., 2011). Furthermore, the btv-encoded Dync2h1 is directly regulated by the forkhead transcription factor Fd3F (Newton et al., 2012), a regulator of ciliary genes. Binding sites for these two transcription factors are identified in the promoter region of the DCH1b gene [Figure 5D, (Laurençon et al., 2007; Newton et al., 2012)].

Secondly, the profile of chordotonal defects we see in the btv mutant, including axonemal disruption and loss of ciliary integrity that worsens with distance from the basal bodies, is entirely consistent with loss of retrograde IFT. The defects are remarkably similar to those we documented for rempA mutants (Lee et al., 2008), encoding IFT140, an IFT-A particle protein that should be transported by Dync2h1 in retrograde IFT. This similarity is consistent with the idea that the Btv-mediated assembly of cilia is carried out through transport of IFT-A complexes including RempA.

Additional evidence supporting a retrograde IFT function for the Btv protein involves the mislocalization of other components in btv mutants. Here we show that NompB, encoding the IFT-B complex protein IFT88, is abnormally distributed in defective cilia of btv mutant JO. This agrees with the finding of Lee et al. (2010) that both IFT88 and the TRPN ion channel NompC are mislocalized in btv mutants. Also, in embryonic ch organs, Kwon et al. (2020) reported that the transport requirements of NompC and TRPV ion channel Iav for Dync2h1 are different, where Dync2h1 is required for both pre-ciliary trafficking and retrograde transport of NompC, primarily in the distal compartment, while Iav only requires Dync2h1 for the former. Furthermore, Lee et al. (2018) used time-lapse imaging to show that retrograde movement of GFP-NompB is absent in btv¹ mutants. NompC is normally localized only to the distal ciliary segment, beyond the ciliary dilation, and NompB and Iav are delimited to the proximal segment, but btv mutants lack a ciliary dilation and lose the sequestered localization of these proteins. Furthermore, we previously showed that RempA, normally restricted to the ciliary dilation by the adult stage, and the Iav, normally delimited to the proximal ciliary segment, are also mislocalized in btv mutants (Lee et al., 2008). Mutation of btv also results in redistribution of the Eys/Spam extracellular matrix protein that is normally deposited into the scolopale space, and that provides protection against desiccation at high temperature conditions (Cook et al., 2008). This protein is normally distributed into a major mid-scolopale space region and a minor proximal region immediately adjacent to the inner dendritic segment; in btv mutants this collapses into a single larger proximal cluster (Lee et al., 2008). It is not clear whether this redistribution arises from loss of a positioning cue provided by the sensory cilium, or whether distal ciliary bloating or membrane leakage forces the scolopale space material more proximally. On the other hand, the Drosophila Tubby-like protein dTulp is not mislocalized in btv mutants (Park et al., 2013). In summary, in the current study we show detailed ultrastructural evidence of the btv mutant, along with mislocalization of ciliary proteins in the btv mutant background, including NompB and RempA. Overall, the btv mutant morphological defects combined with the spectrum of mislocalized proteins demonstrate clearly that Dync2h1 encoded by btv is centrally involved in retrograde IFT in ch organs.

Our finding of altered relative distribution of RempA-YFP and NompA in btv mutants is consistent with our finding of functional defects as reflected by reduced MRPs. Nevertheless, the functional consequences of btv dysfunction in es organs is relatively mild compared to those in ch organs. In es organs, the axonemal segment of the cilium is very short, while the tubular bundle of the distal dendrite is non-axonemal. In contrast to the defective ch cilia tips, the distinctive tip shape of campaniform sensilla is retained, but accumulates RempA (Figure 8). If Dync2h1 requires axonemal structure as a substrate to move cargo, then it may be possible for some cargoes to diffuse along such a short distance sufficiently well to move into the dendrite. Thus, despite de-localization of RempA and NompA, assembly of sensory structures in es organs is sufficient to provide some mechanosensory function. Even in ch organs, where the ciliary segments are relatively long, some residual sensory function is maintained in btv null mutants. Alternatively, it is possible that in es organs, the Dync2h1 function is not specifically IFT per se, but a related transport function for certain cargo proteins, including RempA. Dync2h1 does not appear to have a role in establishing the TEP, suggesting that the ion transport mechanisms to establish and maintain the receptor lymph properties do not depend on this dynein motor. Although some forms of mechanosensory adaptation are motor-driven, loss of Dync2h1 does not affect the MRP adaptation, suggesting that adaptation does not require this dynein.

Until recently, ciliary assembly was thought to be universally dependent upon IFT. Consistent with this, we find that retrograde IFT function mediated by Dync2h1 in Drosophila is essential for normal ciliary assembly in ch organ sensory cilia, including JO. In es organs, which have short ciliary segments, the Dync2h1 has a more modest requirement. Dync2h1 is dispensable in sperm flagella: btv mutants are male-fertile (Eberl et al., 2000). Other exceptions to IFT-dependence have been found. This supports the idea that in Drosophila, contrary to many other taxa, assembly of the sperm flagellum axoneme employs an IFT-independent mechanism (Han et al., 2003; Sarpal et al., 2003; Lee et al., 2008). While the Rfx transcription factor is transiently expressed during spermatid development (Vandaele et al., 2001), there is no evidence available that Rfx activates ciliary genes for spermiogenesis. Furthermore, in a survey of ciliary motility machinery, it was found that a surprising number of axonemal components are encoded by separate genes in sperm compared to mechanosensory cilia (Zur Lage et al., 2019). While IFT is highly conserved in Drosophila, it is activated only in sensory cilia, not in the male germline. In mammals, however, IFT is essential for spermiogenesis, though not for maintenance of sperm tail flagella (San Agustin et al., 2015).

In Tetrahymena, Dync2h1 (dhc-2) appears to be dispensable for ciliary integrity, though it regulates cilia length such that in mutants, ciliary length is variable and cilia are fewer in number (Rajagopalan et al., 2009). Furthermore, in this ciliate, IFT122A, an IFT-A complex protein, participates in retrograde trafficking of IFT88 and IFT172 proteins; these proteins accumulate in the cilia but ciliary assembly is not significantly disrupted (Tsao and Gorovsky, 2008). On the other hand, Dync2h1 phenotypes in Chlamydomonas (Pazour et al., 1999) and in Caenorhabditis (Signor et al., 1999; Wicks et al., 2000) are more severe, while the phenotypes we see in Drosophila ch organs are intermediate, allowing a small amount of residual sensory function. Thus, there is both taxonomic and cell type variation in the degree to which ciliary assembly can proceed with compromised retrograde IFT dynein function.

Our finding that btv mutants, while suffering strong hearing loss, are not completely deaf in contrast to other auditory mutants such as tilB (Kavlie et al., 2010), iav and nan (Kim et al., 2003; Gong et al., 2004) and the kinesin II mutants klp64D and DmKAP (Sarpal et al., 2003). This allowed us to test for progressive hearing loss, leading to our finding that by 30 days of age, btv mutant flies have a significant reduction in auditory capacity, evident primarily in the reduced fraction of flies retaining any residual auditory response. Morphologically, we see strong degeneration in many mutant cilia. However, we see no evidence that the sensory neurons die during this interval, and the lack of TUNEL labeling is consistent with degeneration and resorption of material only in the ciliary compartment.

Even in wild-type flies we see some hearing loss with age, consistent with other studies (Keder et al., 2020). We also see morphological changes with age that likely reflect the fact that, as in vertebrates, the auditory periphery encounters strong metabolic challenges. These changes include the deposition of excess membranes and increasing signs of putative disruption of ionic homeostasis. The morphological changes correlate with age-dependent reductions in hearing, but at present we cannot determine whether there is a causal relationship between these factors. The answer to this question must await further study, and may require the ability to record from single auditory neurons.