Cell-type–specific roles of Na⁺/K⁺ ATPase subunits in Drosophila auditory mechanosensation

Abstract

Ion homeostasis is a fundamental cellular process particularly important in excitable cell activities such as hearing. It relies on the Na⁺/K⁺ ATPase (also referred to as the Na pump), which is composed of a catalytic α subunit and a β subunit required for its transport to the plasma membrane and for regulating its activity. We show that α and β subunits are expressed in Johnston's organ (JO), the Drosophila auditory organ. We knocked down expression of α subunits (ATPα and α-like) and β subunits (nrv1, nrv2, and nrv3) individually in JO with UAS/Gal4-mediated RNAi. ATPα shows elevated expression in the ablumenal membrane of scolopale cells, which enwrap JO neuronal dendrites in endolymph-like compartments. Knocking down ATPα, but not α-like, in the entire JO or only in scolopale cells using specific drivers, resulted in complete deafness. Among β subunits, nrv2 is expressed in scolopale cells and nrv3 in JO neurons. Knocking down nrv2 in scolopale cells blocked Nrv2 expression, reduced ATPα expression in the scolopale cells, and caused almost complete deafness. Furthermore, knockdown of either nrv2 or ATPα specifically in scolopale cells causes abnormal, electron-dense material accumulation in the scolopale space. Similarly, nrv3 functions in JO but not in scolopale cells, suggesting neuron specificity that parallels nrv2 scolopale cell–specific support of the catalytic ATPα. Our studies provide an amenable model to investigate generation of endolymph-like extracellular compartments.

Using the energy of ATP hydrolysis, the Na pump extrudes cytoplasmic Na⁺ (out) and extracellular K⁺ (in) in a 3:2 ratio and maintains the gradient of these cations across the membrane (1, 2), thus controlling the electrolytic and fluid balance in the cells and organs throughout the body (1). Among its other functions, the Na pump helps maintain the resting potential of cells, regulates cellular volume, and facilitates transport of solutes in and out of cells. Ion homeostasis of most biological systems depends on the Na pump. In the auditory system, this pump has been linked to the maintenance of the inner ear osmotic balance (3). The scala media of the inner ear is filled with a K⁺-rich extracellular fluid known as endolymph, which is essential for preserving the sensory structures and supporting transduction. Maintaining the endolymph homeostasis is critical to sustain auditory functions. Loss of endolymphatic balance causes collapse of the endolymphatic compartment, leading to hearing loss in mammals (4). K⁺ channels and pumps, including the Na pump, ensure proper cycling and secretion of K⁺ ions in the stria vascularis cells of the cochlea. The Na pump has also been linked to age-related hearing loss (5) and Ménière disease (6). A detailed functional analysis of this pump is therefore necessary to gain insight into the molecular physiology of hearing loss resulting from loss of auditory ionic homeostasis.

Although vertebrate and invertebrate auditory systems differ structurally, they evolved from the same primitive mechanosensors (7, 8), and there are striking developmental genetic similarities between the two lineages. The fly auditory organ, Johnston's organ (JO), is a chordotonal organ (cho) housed in the second antennal segment (9). The JO comprises an array of ∼250 auditory units or scolopidia. Each scolopidium comprises two to three ciliated sensory neurons associated with several support cells. These bipolar neurons are monodendritic with a single distal cilium and a proximal axon (Fig. 1A). The scolopale cell, a principal support cell, encloses the neuronal dendrites in a fluid-filled lumen, the scolopale space. This fluid, the receptor lymph, resembles cochlear endolymph and, like the endolymph, is believed to be rich in K⁺ ions (7). The scolopale cells are structurally enforced with actin-based scolopale rods. Auditory mechanosensation involves the transduction of the mechanical sound stimulus (10) through the rotation of distal antennal segments into a neuronal response in JO (11). Using electrophysiological techniques we can record sound-evoked potentials (SEPs) from the auditory nerve (9, 12). The JO also mediates gravity and wind detection, in addition to auditory mechanosensation (13–15).

Fig. 1.

ATPα is expressed in JO, with highest expression in scolopale cells. (A) Diagram of a single JO scolopidium showing sensory neurons and support cells with important structural features. (B) Confocal image showing ATPα protein expression in JO, stained with α5 monoclonal antibody (green). ATPα is specifically expressed in the neuron (N) cell body and the scolopale cell (S) as indicated by the white brackets. (C) Magnified image of JO scolopidia showing ATPα expression specifically in scolopale cells (white arrows) and in the plasma membrane of neurons (white arrowhead). α5 (green) counterstained with Texas Red phalloidin (magenta) labeling the actin in scolopale rods. (Scale bars: B and C, 10 and 5 µm, respectively.)

We used the auditory mechanosensory system of Drosophila melanogaster, with its molecular genetic and electrophysiological techniques, to understand the role of the Na pump in maintaining auditory ion homeostasis. Our hypothesis is that the Na pump is important in maintaining the ion homeostasis of the auditory receptor lymph. We show that ATPα is the sole Na pump α subunit in JO and that it has elevated expression in scolopale cells. The β subunits show cell-type specific expression and functions in JO, with nrv2 being specific to scolopale cells and nrv3 specific to neurons. We also show that ATPα preferentially localizes to the scolopale cell ablumenal membrane. Such functional pump localization is consistent with a role in pumping K⁺ ions into the scolopale cell en route to the receptor lymph, a role that resembles its contribution to generating vertebrate inner ear endolymph.

Results

ATPα Is Expressed in JO Scolopidia and Required for Hearing.

Drosophila has three α subunit genes, ATPα, JYalpha, and CG3701. ATPα encodes at least nine mRNA isoforms (16). JYalpha is testes-specific and has been linked to the mechanism for hybrid sterility between D. melanogaster and Drosophila simulans (17) and CG3701 is an α-like subunit with low expression in adult and pupal stages but moderate expression in testis (18). In adult flies, ATPα is expressed in the eye and brain (Fig. S1A), consistent with other studies (19), and in JO (Fig. 1B).

ATPα is expressed in the plasma membrane of JO neurons and much more abundantly in scolopale cells (Fig. 1 B and C). To determine if ATPα is functionally important in JO scolopidia, we wanted to test hearing in flies carrying ATPα mutations. However, ATPα mutants are homozygous lethal at early larval stages (20, 21). To circumvent these limitations, we used RNAi to knock down ATPα using the Gal4/UAS system (22, 23). We used ato-Gal4 (Fig. S2A) to drive expression of double-stranded RNA under UAS control in the JO sense organ precursor cells to knock down ATPα only in these cells and their progeny. Knockdown animals were deaf, with complete loss of SEPs (Fig. 2).

Fig. 2.

ATPα, but not α-like, is required in JO for hearing. Histogram of SEPs from the antennal nerves of α subunit knockdown (black bars) and control animals (white bars). The control genotypes are ato-Gal4/UAS-Dicer2 and NompA-Gal4/FM4; +/CyO, whereas the knockdown animals are ato-Gal4/UAS-ATPα (or α-like)-RNAi; UAS-Dicer2/+ and nompA-Gal4; UAS-ATPα (or α-like)-RNAi/CyO;+. Knockdown experiments are shown for two α subunit genes (ATPα and α-like) with ato-Gal4 (drives expression in entire JO lineage) and nompA-Gal4 (drives expression in scolopale cells). Hearing loss in ATPα knockdown flies is highly significant with both drivers (two-tailed t test with Welch’s correction for ato-Gal4 (P < 0.001, n = 5 per genotype) and for nompA-Gal4 (P < 0.0001, n = 15 per genotype.)

Next we used UAS-RNAi against the α-like subunit (CG3701) using the same ato-Gal4 driver. Driving CG3701 knockdown in the JO sense organ precursor cell lineage had no effect on hearing (Fig. 2). This is consistent with CG3701 expression primarily in testis. JYalpha expression is also testis-specific (18). Therefore, ATPα is likely the only α subunit gene required for hearing in Drosophila.

ATPα Is Necessary for Scolopale Cell Function in Hearing.

To investigate if the elevated expression of ATPα in the scolopale cell has physiological relevance for auditory function, we wanted to remove it only from the scolopale cells. We used nompA-Gal4 (24) (Fig. S2B) to drive ATPα RNAi only in JO scolopale cells. Staining with an ATPα antibody showed that the RNAi knocks down ATPα protein almost completely in JO scolopale cells (Fig. 3B) compared with controls (Fig. 3A). However, ATPα expression is retained in other cell types, including cap cells and neurons. Electrophysiological recordings revealed these knockdown animals to be deaf (Fig. 2), indicating that ATPα function is required in the scolopale cells. To distinguish whether this requirement is developmental or physiological, we used Gal80^ts, a temperature-sensitive repressor of Gal4, for temporal control. Conditions that prevent RNAi knockdown during development (18 °C) but allow it for 3 d at the adult stage (30 °C) resulted in significant hearing reduction compared with genotypically identical flies raised and maintained at 18 °C (Fig. S3). Thus, ATPα function is required at the adult stage after development is complete. ATPα is also required during development as flies raised at 30 °C and switched to 18 °C at adulthood are not rescued (Fig. S3).

Fig. 3.

ATPα knockdown in JO scolopale cells disrupts the scolopale space. Confocal images showing ATPα (α-5 mAb; green) and Nrv3 (αNrv3; magenta) expression in adult antenna in control animals (A) and in ATPα knockdown animals (B) in which ATPα expression is abolished in scolopale cells. Genotypes as in Fig. 2. (Scale bars: A and B, 10 µm). EM cross-sectional view showing three adult JO scolopidia in control (C) and in scolopale cell ATPα knockdown animals (D). (E) Diagram of a scolopidium, with dotted line indicating the level of cross sections (C and D). EM showing longitudinal-sections of adult JO control (F) and scolopale cell ATPα knockdown animals (G). The block arrows with black border show the electron light area in the scolopale space (SS) (C and F); the white block arrows show the electron-dense material (D and G). Abnormal mitochondria present in the scolopale space as shown in (D) (asterisk). (Scale bar: C, D, F, and G, 1 µm.)

EM indicated abnormal accumulation of electron-dense material in the JO scolopale space in knockdown animals (Fig. 3 D and G). This material included membrane-bound organelles such as mitochondria. To determine if such morphological defects affected extracellular proteins known to be present in the scolopale space, we examined Eyeshut/Spacemaker (Eys) protein (25, 26), which localizes in the scolopale space (27), but found no differences compared with control (Fig. S4).

Functional Importance of ATPα.

Campaniform and bristle organ receptor lymph, and likely that of cho organs, is thought to be rich in K⁺, resembling vertebrate inner ear endolymph (7). Therefore, our model is that the Na pump actively transports K⁺ ions toward the JO scolopale space. However, the Na pump is also essential for septate junction formation in a pump-independent manner (28, 29). Septate junctions are critical to maintain luminal integrity in embryonic salivary glands and trachea. The scolopale cell has extensive septate junctions (30, 31), likely to seal the scolopale space. Morphological defects such as accumulation of organelles including mitochondria may indicate putative defects at the septate junctions between scolopale cell and cap cell or between the scolopale cell and the neuronal cell bodies. Accumulation of mitochondria in the scolopale cell could also imply metabolic stress. When we stained knockdown (NompAGal4/+; UASRNAi-α/UASDicer2) and control (NompAGal4/+; +/CyO) JO with Lucifer Yellow (32) in conjunction with Texas Red Phalloidin, a marker for the actin-rich scolopale rods, we found that the Lucifer Yellow remained excluded from the scolopale space to the same extent in knockdown and control animals (Fig. S5). This result suggests that ATPα may not be a major contributor to junctional integrity of scolopale cells. Alternatively, the RNAi knockdown effect may be insufficient to compromise the junctional complex while still disrupting hearing physiology.

Expression of β Subunits in JO.

Flies have three Na-pump β-subunit genes encoded by nervana genes nrv1, nrv2, and nrv3. All three nrv genes have tissue-specific expression. nrv1 is present in the eye, muscle, heart, and fat body as well as digestive and excretory tissues (18), whereas nrv2 is important in the tracheal system, where it is required for pump-independent septate junction integrity (28, 29). The nrv3 gene is expressed in the brain, the eye, and the JO in adults (Fig. S1B) and in the embryonic CNS and cho sensory cells (Fig. S1D) (29). In addition, nrv3 is the principal β subunit in adult photoreceptor cells (19).

To determine which of these three genes participate in JO function, we first tested their expression in JO. We stained cryosections of adult head with attached antennae using pan-Nrv monoclonal antibody Nrv5F7 (33) and a polyclonal antibody against Nrv3. Staining with these antibodies largely overlaps in the brain, eye, and the JO neurons. However, in the scolopale cell, there is specific staining only with Nrv5F7, but not with Nrv3 antibody (Fig. S6), indicating either or both Nrv1 and Nrv2 but not Nrv3 are present in scolopale cells. Subsequent immunostaining with a polyclonal antibody against Nrv1 (29) revealed no Nrv1 protein expression in the JO neurons and scolopale cells, only a low level of Nrv1 expression in the cap cells (Fig. S7). Previously, Nrv1 protein was found to express at low levels in the R7–R8 adult photoreceptor cells, but to be absent in brain tissue (19). From the differential staining pattern of Nrv5F7 and Nrv3 (Fig. S6) and the Nrv1 pattern (Fig. S7), it is likely that only the nrv2 gene has a scolopale cell–specific expression. Indeed, staining JO with a polyclonal antibody against Nrv2 protein (29) indicates strong scolopale cell expression (Fig. 4A), although some staining is also seen in cap cells and ligament cells. Thus, among Na-pump β-subunit genes, there is a cell type–specific expression pattern with nrv3 expressed only in JO neurons, nrv2 expressed primarily in scolopale cells, and nrv1 expressed at a low level in cap cells.

Fig. 4.

Scolopale cell-specific nrv2 gene knockdown abolishes Nrv2 protein and reduces ATPα protein. (A) Control JO section showing colocalization of ATPα subunit (α-5 mAb; green) and Nrv2 (α-Nrv2 Ab; red) in the scolopale cell. The scolopale cell and the neuronal cell body regions are indicated with white brackets. N or S represents neuronal or scolopale cell region (A and B), as indicated by the white boxes from which fluorescent intensities were measured. Genotypes as in Fig. 5. (B) Scolopale cell-specific nrv2 RNAi knockdown shows a complete absence of Nrv2 protein from the scolopale cell region in JOs processed side by side with the control JOs and imaged with identical settings. ATPα also appears reduced in these cells. (C) Bar graph showing fluorescent intensity ratio (F_S/F_N) for Nrv2 and ATPα protein expression. F_S represents the fluorescent intensity of the subunit proteins under investigation in the scolopale cell region (box labeled S in A and B) and F_N represents ATPα expression in neurons (box labeled N in A and B). Data are from five control antennae and six nrv2 knockdown antennae. Using t tests, P = 0.0008 for Nrv2 protein expression, with Welch’s correction, and P < 0.0001 for ATPα expression. (Scale bar: A and B, 10 µm.)

β Subunit Functional Requirements in the JO.

We wanted to test if lack of any of the nrv genes also had a physiological effect on hearing. Although we generated two intragenic nrv3 deletion mutants, nrv3¹⁵ and nrv3⁴⁷, they are lethal at early larval stages. Similarly, nrv2 mutants are homozygous lethal (28), whereas no nrv1 mutant alleles are available. Therefore, to conduct adult hearing studies, we used RNAi to knock down expression of these β subunit genes in the JO sense organ precursors using the ato-Gal4 driver. Knocking down expression of nrv1 or nrv2 with this driver had no significant effect on hearing (Fig. 5). However, knocking down nrv3 with ato-Gal4 resulted in nearly complete deafness (Fig. 5). Thus, nrv3 is required in the JO, most likely in the JO neurons based on its expression pattern.

Fig. 5.

β subunit genes nrv2 and nrv3 are required in specific JO cell types for hearing. Histogram of SEPs of β subunit knockdown (black bars) and control animals (white bars). RNAi knockdown of the three β subunit genes (nrv1, nrv2, and nrv3) with ato-Gal4 and nompA-Gal4 drivers for the entire JO lineage and for the scolopale cell only, respectively, are shown. Genotypes as in Fig. 2 except that the UAS-RNAi constructs are against β subunits, and the ato-Gal4 genotypes also include UAS-Dicer2. Hearing loss is highly significant in ato-Gal4–mediated knockdown of nrv3 (P < 0.0001, n = 15 per genotype) and in nompA-Gal4–mediated knockdown of nrv2 (P < 0.0001, n = 10). All P values are based on two-tailed t tests with Welch’s correction.

To reconcile the enriched Nrv5F7 antibody staining of scolopale cells and Nrv2 staining with the lack of functional defects upon RNAi-mediated knockdown of nrv1 or nrv2 in the JO sense organ lineage, we used nompA-Gal4 to test for scolopale cell–specific effects. Knockdown of nrv2, but not nrv1, resulted in almost complete hearing loss (Fig. 5). Thus, nrv2 has a strong requirement in scolopale cells. Failure of ato-Gal4 to reveal this function may reflect differences in expression level and timing compared with nompA-Gal4. Knocking down nrv1 specifically in the cap cells with the cap cell–specific pyx-Gal4 driver (13) had no effect on hearing (Fig. S7).

nrv2 Mediates Scolopale Cell-Specific Expression of ATPα.

To confirm that hearing loss resulting from nrv2 knockdown is coupled with loss or decrease of Nrv2 protein from scolopale cells and to determine the effect on ATPα, we used the Nrv2-specific polyclonal antibody (29) and costained with α-5, a monoclonal antibody against ATPα (Fig. 4, Fig. S8). In control animals, Nrv2 and ATPα colocalized in the scolopale cell. A partial overlap was also observed in the ligament cells, another type of JO support cell. As seen previously, ATPα expressed most robustly in the scolopale cells, but also in the JO neurons and the antennal nerves (Fig. 4A). The Nrv2 protein stained only in the support cells with highest expression in the scolopale cells. In nrv2 knockdown animals, Nrv2 staining was completely lost from the scolopale cells, but the remaining supporting cells retained staining (Fig. 4B, Fig. S8). We also observed reduced ATPα expression in the scolopale cells with an almost complete loss in the cell body of these cells. To quantify the decrease in ATPα and Nrv2 protein levels in the scolopale cells, we measured their fluorescent intensity ratios in scolopale cells versus neurons (Fig. 4) under identical imaging conditions using the neuronal signal from ATPα as the universal reference. Relative Nrv2 expression is greatly reduced in scolopale cells in nrv2 knockdown animals. Similarly, relative scolopale cell ATPα expression is also strongly reduced in nrv2 knockdown animals, although still detectable (Fig. 4). Thus, nrv2 is required for ATPα expression, localization, or stability in scolopale cells.

EM analysis of JO in nrv2 knockdown animals revealed abnormal electron-dense material accumulation in the scolopale space and presence of organelles, particularly mitochondria (Fig. 6), similar to our observations in ATPα knockdown animals (Fig. 3 D and G). Furthermore, scolopidia in nrv2 knockdown animals appear narrower than those in controls. These findings support Nrv2 as the specific β subunit partner of ATPα in scolopale cells.

Fig. 6.

Knocking down nrv2 in JO scolopale cells disrupts the scolopale space. (A and C) Control sections. (A) Cross- and (C) longitudinal sections showing the normal structure of scolopidia with electron-light scolopale space in a scolopidium as shown by the white block arrow with black border. The control image in C is the same as in Fig. 3F. (B and D) Experimental sections. (B) Cross- and (D) longitudinal sections showing abnormal accumulation of electron-dense material, shown by the white block arrow, including membrane bound bodies such as mitochondria (asterisks in B) in the scolopale space. (Scale bar: 1 µm.) Genotypes as in Fig. 5.

Model for Subcellular Localization of ATPα.

Our results indicate that ATPα is present in the scolopale cell membrane (Fig. 1C). We propose a model in which the Na pump is present preferentially in the ablumenal (outer) plasma membrane of the scolopale cell, where it pumps K⁺ ions into the scolopale cell cytoplasm in exchange for Na⁺ ions (Fig. 7A). This position will ensure the K⁺ enrichment of the scolopale cell cytoplasm with subsequent K⁺ movement to the scolopale space receptor lymph by an unknown mechanism. In support of this model, we found in high-magnification deconvolution images that ATPα is found primarily outside the scolopale rods (Fig. 7B). Furthermore, a transgenic fly line that has a chimeric ATPα–GFP fusion protein (34) shows similar localization. This protein (which may not reflect expression of all ATPα splice forms) accumulates primarily around the junction between scolopale cell and cap cell or between the scolopale cell and the neuronal inner dendritic segments (Fig. 7C). This also implies a putative septate junction function of the Na pump, although our dye-exclusion approach revealed no junctional compromise. Although further experiments will be required to fully address this question, our data are most consistent with a role for the Na pump in receptor lymph homeostasis.

Fig. 7.

Model of ablumenal localization of ATPα in the scolopale cell. (A) Diagram showing a single scolopidium with predicted locations for Na pump. Red arrows represent the direction of K⁺ transport through the Na pump (green bar) in the plasma membrane. (B) Cross-section of JO scolopidia showing presence of ATPα protein stained with α5 mAb (green), indicated by white arrows on the scolopale cell ablumenal plasma membrane surrounding the scolopale rods (magenta). Texas Red phalloidin, which stains the scolopale cell actin rods, is used as a marker for the boundary between the luminal and ablumenal plasma membranes of the scolopale cell. (C) Longitudinal sectional view of the scolopidium shows ATPα-GFP fusion protein at the scolopale cell-cap cell junctions (white brackets), in the apical epithelium (asterisk) and in neuron cell bodies (N). ATPα-GFP expression in the scolopale cell outside the scolopale rods (brackets) is consistent with preferential ablumenal localization.

Discussion

The Na pump is important for sustaining JO auditory transduction by mediating ion homeostasis. Our study revealed that the Na pump is localized preferentially to the scolopale cell ablumenal plasma membrane, from where it likely pumps K⁺ ions into the scolopale cell cytoplasm en route to the scolopale space. Several lines of evidence support this conclusion. First, we observed that ATPα, the principal JO α subunit, has a strikingly high expression in the scolopale cell. This suggested that the ATPα gene has a scolopale cell–specific specialized role. Second, most ATPα protein localizes outside the scolopale rods, supporting an ablumenal plasma membrane localization. Third, ATPα knockdown in the scolopale cell resulted in deafness, loss of scolopale cell integrity, and morphological defects such as presence of distended cilia, implying ionic imbalance in the scolopale space. Taken together, the Na pump is likely to be involved in maintaining JO receptor lymph ion homeostasis. Other molecular players must work in conjunction with the Na pump to maintain the ion homeostasis of the system. However, their identification must await further study.

The Na pump may also have alternative functions that are not dependent on pump activity. Our results show subcellular localization of the ATPα–GFP fusion protein accumulating near the scolopale cell–cap cell junction and the scolopale cell–neuron inner dendritic segment junction (Fig. 7B). Septate junctions are known to be present between these cell types. Insect septate junctions form a transepithelial diffusion barrier that limits solute passage through the spaces between adjacent cells in an epithelium (35). The Na pump has a pump-independent cell junctional activity responsible for maintaining the epithelial barrier function in the Drosophila tracheal system, mediated by the Nrv2 β subunit (28, 29). In the fly auditory system, failure to preserve junctional integrity of the scolopale cells because of a lack of functional pumps may also cause fluid retention inside the scolopale space, which could manifest as the morphological abnormalities we saw when either ATPα or nrv2 were knocked down in the scolopale cell. However, our Lucifer Yellow dye exclusion assay, in animals in which ATPα has been knocked down with nompA-Gal4, argues against such a junctional role of the Na pump in scolopale cells, although one cannot absolutely rule out a junctional role of the pump because the Lucifer Yellow molecules may be too large to detect a mild junctional compromise, or RNAi may not completely inhibit the pump. Furthermore, septate junctions require numerous other components, so loss of only one component may not completely dismantle the septate junctions in these cells. Future genetic rescue experiments using constructs with inactivated pump function in an ATPα knockdown background would indicate whether the observed knockdown phenotype is a pump-independent function of the Na pump. However, because of the early requirement of the Na pump during development and cross-reactivity of the RNAi to both the endogenous and rescue construct gene copies, such experiments are currently not feasible.

Our finding of organelles such as mitochondria in the scolopale space, often devoid of plasma membrane enclosure, raises the question of their origin. The most likely source of these is the scolopale cell itself. One possibility is that concomitantly compromised ion homeostasis and osmotic balance results in scolopale cell membrane rupture, releasing cellular contents. Torn membranes can rapidly reseal themselves through a Ca²⁺-dependent process (36). In Drosophila embryos, cells undergoing such a cell membrane tear form a membrane plug to reseal the gap in the lipid bilayer through a coordinated activity of the cell membrane and the cytoskeleton (37). A second possibility is that the extraneous material in the scolopale space results primarily from a developmental requirement of the Na pump. The cho cell lineage for each scolopidium comes from a single sense organ–precursor cell, specified in the imaginal disk epithelium. Lineage cell division occurs early, before massive changes in cell shape, with enormous subsequent elongation. Loss of ion homeostasis during this process may prevent the high developmental fidelity required for these cell shape changes. A third possible explanation may be partial apoptosis of the scolopale cell. Ultrastructurally, it is clear that the scolopale cell is still alive in the knockdown animals because the nuclei are not heteropyknotic and we see no cell shrinkage; both of which are hallmarks of apoptotic cells. Nevertheless, ionic or osmotic imbalances in the cell may trigger a subset of apoptotic features such as blebbing of the plasma membrane. Such blebs into the scolopale space would initially be membrane-bound, but this membrane may be unstable and break down in the context of the scolopale space. The mitochondria found in the scolopale space also frequently display swelling or disrupted cristae, observations that are also consistent with apoptotic features in Drosophila (38).

Neuronal receptor currents resulting from auditory transduction likely cause depletion of ions in the receptor lymph; therefore, a mechanism must exist by which the receptor lymph is replenished with a constant supply of K⁺ ions. The receptor lymph filling the scolopale space in JO is likely to be highly enriched in K⁺ ions in analogy to the bristle receptors and campaniform sensilla in insects that have been shown to be rich in K⁺ ions (39, 40). In addition, vertebrate endolymph is K⁺ enriched (41). Our model of the Na pump in the fly auditory system is that it is present in the ablumenal plasma membrane of the scolopale cell where it actively transports K⁺ ions into the scolopale cell en route to the scolopale space to help maintain its K⁺-rich ionic composition (Fig. 7A). RNAi-mediated knockdown of Na pump subunits results in deafness and loss of morphological integrity. All of these findings are consistent with our model. However, to absolutely confirm the relevance of this model, additional experiments are required. First, it would be informative to measure the receptor lymph ionic concentration and directly demonstrate the high K⁺ concentration within the scolopale space. However, such experiments may prove to be technically challenging because of the small size of the scolopidium. We also need to identify other molecules that work in combination with the Na pump to maintain the receptor lymph for efficient auditory transduction.

In the vertebrate inner ear, the endocochlear potential is achieved by maintaining the ionic composition of the endolymph in the fluid compartment into which the stereocilia project (42). Deregulation of the ion concentration or fluid volume in the endolymph, mediated by cells in the stria vascularis and the lateral walls of the organ of Corti, is thought to underlie hearing disorders such as Ménière disease (43), and may contribute to age-related hearing loss (6). Although the Na pump is thought to participate in enriching K⁺ in the endolymph and to contribute to fluid volume homeostasis in the endolymph, the precise mechanisms of Ménière and related diseases are not well understood. In this article, we have taken advantage of our rapidly advancing understanding of the Drosophila auditory system to systematically investigate the expression and functional roles of each Na pump subunit in the auditory organ. This study defines the cell-type specificity of Na pump subunits as well as the functional and morphological consequences of cell type–specific loss of function of these subunits. It also sets the stage for future studies to elucidate the detailed pathways and mediators of ion transport and fluid regulation. Our model of the Na pump subcellular localization in the ablumenal membrane of the scolopale cell provides a useful system to investigate endocochlear potential generation in endolymph-like extracellular compartments and its malfunctions in connection to inner ear disorders such as age-related hearing loss and Ménière disease.

Materials and Methods

Drosophila Stocks.

The following fly stocks were used: nompA-Gal4, Sp/CyO, and pyx-Gal4 are previously described in (13) and (24), ato-Gal4 was a gift from B. Hassan (Katholieke Universiteit, Leuven, Belgium) via G. Boekhoff-Falk (University of Wisconsin, Madison, WI). UAS-ATPα-RNAi (v12330), UAS-αlike-RNAi (v10737), UAS-nrv1-RNAi (v46542), UAS-nrv2-RNAi (v2660), UAS-nrv3-RNAi (v44486), and UAS-Dicer2 were supplied by the Vienna Drosophila RNAi Centre. The GFP fusion line ATPα (34) was donated by W. Chia (National University of Singapore, Singapore). w¹¹¹⁸ was used for normal expression of Na pump subunits in JO. UAS-GFP and tub-Gal80^ts were obtained from the Drosophila Stock Center in Bloomington, IN.

RNAi and UAS/Gal4 System.

RNAi was used to knock down Na pump subunit genes in specific spatiotemporal patterns using the Gal4/UAS system (22, 23). The Gal4 drivers ato-Gal4 and nompA-Gal4 have JO-specific and scolopale cell–specific expression patterns (Fig. S7). UAS-Dicer2 was used in conjugation with the Gal4 driver in certain cases to enhance the knockdown effect.

Electrophysiology.

Auditory recordings were conducted in experimental and control flies as described elsewhere (9, 12). The fly was mounted in a 200-µL pipette tip trimmed such that only the head protruded. The neck was immobilized with plasticine. The computer-generated pulse component of the Drosophila courtship song was played through a speaker and the sound was transported through a Tygon tube (Fisher Scientific) placed at a distance of 1 mm from the fly’s head. The sound stimulus intensity was measured at 5.3 mm/s at the position of the antennae, using a calibrated Emkay NR3158 particle velocity microphone (Knowles). Two tungsten electrodes were used; the recording electrode was inserted at the joint between the first and second antennal segment from a dorsofrontal direction and the reference electrode was inserted in the head cuticle. The signals were amplified by a DAM50 differential amplifier (WPI) and digitized and normalized using Superscope II software (GW Instruments). Details about this assay are available elsewhere (9, 12).

EM.

The heads of nompA-Gal4; UAS-ATPα-RNAi and nompA-Gal4; UAS-nrv2-RNAi and corresponding controls were fixed, processed, and embedded in Epon resin according to the protocol described in (44). Transmission EM was conducted with a JEOL 1230 instrument.

Immunohistochemistry and Microscopy.

Antennae were dissected in PBS, fixed in 4% (vol/vol) paraformaldehyde in PBS for 30 min, embedded in OCT (Ted Pella) and then cut into 25-µm sections in a cryostat. The antennal cryosections were stained with primary polyclonal antibodies against Nrv2 generated in rabbit and Nrv1 and Nrv3 generated in guinea pig, respectively (29), generous gifts from G. Beitel (Northwestern University, Chicago, IL). Monoclonal antibodies α5 recognizing ATPα [1:100 diluted in PBS + 0.1% (vol/vol) Triton X-100 (PBT) + BSA], 5F7 recognizing all three Nrv isoforms and 21A6 recognizing Eys/Spam (1:200 diluted in PBT+BSA) were obtained from The University of Iowa Developmental Studies Hybridoma Bank. Secondary antibodies labeled with Alexa Fluor-488 or TRITC were obtained from Sigma (1:200 diluted in PBT+BSA). Texas Red phalloidin (1:200 diluted in PBT+BSA) was used to stain the scolopale rods for 1 h. All confocal images were taken using a Leica SP2 Confocal Microscope except for the ATPα ablumenal localization images, for which a DeltaVision Deconvolution microscope was used.

Statistics.

Statistical analyses of electrophysiology data and fluorescence ratio data were performed in GraphPad Prism software using the Student t test, with Welch’s correction for unequal variances where needed. The relative fluorescent signals for Nrv2 and ATPα in the scolopale cells were calculated using ATPα signal in neurons as the reference.