Physiological, anatomical, and behavioral changes after acoustic trauma in Drosophila melanogaster

Significance

Noise-induced hearing loss (NIHL) is an important health issue, yet its mechanisms and potential treatments remain unclear. We present the first study of NIHL in the fruit fly Drosophila, which has many advantages as an NIHL model. We examined auditory function and structure after exposing flies to acoustic trauma. Trauma impairs auditory system function and changes neural mitochondria size, suggesting metabolic stress. In mutant flies with a reduced ability to cope with such stresses, the responses to trauma were more severe and recovery delayed or impaired.

Abstract

Noise-induced hearing loss (NIHL) is a growing health issue, with costly treatment and lost quality of life. Here we establish Drosophila melanogaster as an inexpensive, flexible, and powerful genetic model system for NIHL. We exposed flies to acoustic trauma and quantified physiological and anatomical effects. Trauma significantly reduced sound-evoked potential (SEP) amplitudes and increased SEP latencies in control genotypes. SEP amplitude but not latency effects recovered after 7 d. Although trauma produced no gross morphological changes in the auditory organ (Johnston’s organ), mitochondrial cross-sectional area was reduced 7 d after exposure. In nervana 3 heterozygous flies, which slightly compromise ion homeostasis, trauma had exaggerated effects on SEP amplitude and mitochondrial morphology, suggesting a key role for ion homeostasis in resistance to acoustic trauma. Thus, Drosophila exhibit acoustic trauma effects resembling those found in vertebrates, including inducing metabolic stress in sensory cells. This report of noise trauma in Drosophila is a foundation for studying molecular and genetic sequelae of NIHL.

Noise-induced hearing loss (NIHL) is a pervasive and growing health issue arising from occupational and recreational hazards, with significant costs in health care and personal quality of life. Despite this, the molecular and physiological mechanisms involved in the etiology or recovery from injury are not yet fully understood. Importantly, intense acoustic trauma can induce permanent damage—unlike other vertebrates, mammals cannot regenerate auditory hair cells (1, 2). NIHL associated with permanent changes in auditory sensitivity causes multiple consistent effects: stereocilia bundle disruption, inner (IHC) and outer hair cell (OHC) death or damage, supporting cell tissue disruption, and eventual spiral ganglion cell damage or loss (3–7). Most studies to date used mammalian model organisms such as mice (8, 9), rats (10), and guinea pigs (11–14). These animals have difficult access to the inner ear inside the temporal bone and high maintenance costs coupled with relatively long generation times.

Drosophila is a compelling alternative model system with strong genetic tools, inexpensive production of large numbers of animals, and an accessible auditory system that is becoming better understood genetically and physiologically. During courtship, Drosophila males vibrate their wings to produce a courtship song composed of pulse and sinusoidal components (15, 16). This song facilitates species identification and mate selection (16, 17). Drosophila males and females detect airborne vibrations via Johnston’s organ (JO) in the second antennal segment (18). The JO is an array of chordotonal mechanoreceptors (or scolopidia; Fig. 1 A–C). Via the aristae, acoustic energy is transformed to rotational movement of the third antennal segment, activating mechanosensitive channels on JO neuron dendrites. Like vertebrate hair cells, JO neurons are ciliated and respond to mechanical stimulation. Although JO has morphologically diverged from hair cells in the human inner ear, the genetic program for its development shares a strong homology (19, 20). For example, the Atoh1 gene required for vertebrate auditory hair cell specification was found by direct homology to the fly atonal gene required for JO specification and atonal/Atoh1 genes can be functionally exchanged between mice and flies (21, 22). The advantages of studying hearing in Drosophila are that the genome is fully sequenced, genetic tools for extensively manipulating the genome are at hand, genetic background effects can be effectively eliminated, and large numbers of individuals can be tested.

Fig. 1.

Organization of Drosophila JO and physiological response to sound. (A) Deconvolution micrograph of labeled scolopidia in JO. The actin scolopale rods are labeled with phalloidin (magenta), mitochondria in some JO neurons are labeled with mito-GFP (green), and nuclei are labeled with TOPRO-3 (blue). (B) Schematic diagram of an individual scolopidium, oriented and colored similarly to scolopidia in A. (C) Approximate longitudinal section of JO in an untraumatized control 40AG13 fly. bb, basal bodies; cap, dendritic cap; cd, ciliary dilation; m, membranous structure; mt, mitochondria; N, nuclei of JO neurons; ScN, nuclei of scolopale cells; t, trachiole. (Scale bar: 1 μm.) (D) Example of SEPs recorded in response to acoustic stimulation (stim). The top trace is the synthetic courtship song pulse stimulus; immediately below is the resulting SEP, with the analyzed amplitude and latency parameters indicated. The bottom trace shows multiple SEPs from a wild-type fly in response to a pulse train.

In this study, we establish Drosophila as an inexpensive and flexible model system for genetic and physiological study of NIHL. We exposed two control strains [Canton-S (CS) and 40AG13] to acute acoustic trauma and examined physiological, behavioral, and anatomical effects. Our findings show immediate effects on auditory function, with reduced and delayed evoked activity. Although evoked potential amplitudes were restored after 7 d, the latency of these potentials did not fully recover and we found significant changes in JO neural mitochondrial morphology. We also tested mutant flies with a reduced copy number of nervana 3 (nrv3) encoding a Na⁺/K⁺ ATPase β subunit expressed in JO neurons (23). We hypothesized that compromised JO ionic homeostasis would confer susceptibility to noise trauma. Indeed, nrv3 heterozygotes showed increased sensitivity to trauma and a significantly reduced auditory functional recovery.

Results

Physiological/Behavioral Responses to Trauma in Control Strains.

Electrophysiology of nontraumatized wild-type CS and 40AG13 flies (Fig. 2 A and B, dark bars) showed clear responses to the pulse stimuli, with sound-evoked potential (SEP) amplitudes similar to previous studies (18). We focused on test stimuli presented at 85 dB sound pressure level (SPL) [14.2 mm/s particle velocity (PV)] because this most closely approximates the levels heard naturally during courtship interactions (24). Responses increased monotonically with increasing stimulus SPL from 65 dB (0.5 mm/s PV) to 85 dB (Fig. S1 A–C, dark bars). Increasing the SPL to 95 dB (40.7 mm/s PV) resulted in similar or slightly decreased responses (Fig. S1D, dark bars), also matching previous behavioral results (25). Untreated CS flies showed larger SEP amplitudes than 40AG13 flies at all time points and for all stimulus levels (Fig. 2, Fig. S1, dark bars), reflecting differences in strain genetic backgrounds.

Fig. 2.

Acoustic trauma reduces SEPs and courtship behavior without affecting circadian locomotor activity. Average SEP amplitudes are shown for flies of two laboratory strains, either exposed (white bars) or not (black bars), to acoustic trauma at two times posttrauma. (A) SEPs of CS flies. n = 69 (sham), 72 (trauma) for 0 d; 42 (sham), 39 (trauma) for 7 d. (B) SEPs of 40AG13 flies. n = 66 (sham), 75 (trauma) for 0 d; 39 (sham), 37 (trauma) for 7 d. (C) Mean normalized courtship index (CI) scores for sham (black bars) and traumatized (white bars) 40AG13 male flies 0 and 7 d after treatment, n = 12 for sham and n = 16 for traumatized flies at each time point. (A–C) Horizontal bars and asterisks show significant differences (t test; *P < 0.05; **P < 0.01; ***P < 0.001) and error bars indicate SEM. (D) Average activity (measured in crossings/s) of sham (solid line) and trauma-exposed (dashed line) 40AG13 male flies at 1-min intervals for 1 h after cessation of trauma treatment. n = 17 flies per treatment.

SEPs of flies exposed to acoustic trauma showed highly significant decreases immediately after trauma cessation (Fig. 2 A and B, 0 d; P < 0.001, t test). SEP amplitudes for CS and 40AG13 flies were reduced 18.5% and 15.8% from nontraumatized groups, respectively. Reductions were significant across stimulus SPL range (Fig. S1 A–D, 0 d), with trauma-induced SEP reductions generally larger in response to lower SPL stimuli. At 7 d posttrauma, no significant differences were seen between treatment groups, regardless of genotype or SPL (Fig. 2 A and B, Fig. S1 A–D, 7 d; t test), indicating that JO can recover from traumatic sound.

Trauma also affected song-induced courtship behavior in male 40AG13 flies, as revealed in a male auditory courtship assay (25). This strain was chosen for behavioral study for its robust activity in this assay. Immediately after trauma, males showed significantly reduced courtship behavior compared with control flies (Fig. 2C, 0 d). At 7 d after trauma, courtship responses were not significantly different between control and traumatized flies (Fig. 2C, 7 d). To ensure that directed courtship activity differences were not due to underlying changes in baseline locomotor behavior, we quantified locomotor activity for 1 h after cessation of trauma (a time span much longer than the duration of courtship behavioral experiments). Trauma had no noticeable effect on overall locomotor behavior (Fig. 2D).

Acoustic trauma also affected timing of SEP responses to song pulses. Immediately posttrauma, SEPs to the second stimulus pulse (Fig. 3A, pulse 2, gray box) of nontraumatized flies had the lowest latencies. In traumatized flies, these latencies depended similarly on pulse number, but were significantly delayed compared with controls by ∼44–52 μs (Fig. 3A, white boxes; P < 0.0001, Welch’s t test). At 7 d after trauma (Fig. 3B), latencies also increased with pulse number. Increased latencies were still seen in traumatized flies compared with controls (Fig. 3B, white boxes; P = 0.0337–0.0066, Welch’s t test), but latency differences were reduced to ∼29–37 μs for subsequent pulses. Importantly, this incomplete recovery of latency is unlike SEP amplitudes, which recover to levels indistinguishable from controls after 7 d.

Fig. 3.

Poststimulus SEP response latency increases with acoustic trauma. Plots show distributions of SEP peak latencies (measured from the first pulse stimulus peak to the large negative peak of the subsequent SEP; Fig. 1D) for the second to fifth SEPs. Box and whisker plots show medians as central vertical lines, boxes represent 25–75% quartiles, and whiskers show 5–95 percentiles. Gray bars represent nontraumatized flies, white bars traumatized. (A) SEP latencies immediately after treatment (0 d). For sham flies, n = 133 (pulse 2), 134 (pulses 3–5). For trauma flies, n = 140 (pulse 2), 139 (pulse 3), 140 (pulse 4), and 141 (pulse 5). (B) Latencies 7 d posttrauma. For sham flies, n = 78 (pulse 2) and 79 (pulses 3–5); for trauma flies, n = 76 (pulses 2–5). Vertical brackets and asterisks between each sham/trauma pair indicate significant differences (Welch’s t test; *P < 0.05; **P < 0.01; ****P < 0.0001). Note: x axis absolute values are irrelevant because they depend on tubing length and sound propagation time during stimulus delivery, but this was constant for all measurements; relative values are informative.

Physiological Response to Trauma in nrv3 Mutants.

In mammals, acoustic trauma induces metabolic stress in cochlear hair cells and spiral ganglion neurons, changing mitochondrial fission/fusion dynamics; with sufficiently intense trauma, this induces apoptosis pathways (5, 26, 27). To test whether flies with slightly compromised physiological components would be sensitized to acoustic trauma, we exposed flies heterozygous for a mutation in the nrv3 gene, encoding a Na⁺/K⁺ ATPase β subunit specifically expressed in auditory neurons (23; Fig. S2), to acoustic trauma and measured SEPs. Na⁺/K⁺ ATPase is expressed in JO scolopidia where it likely regulates ion homeostasis (23). We hypothesized that mutant heterozygote JO neurons would be more severely affected by constitutive activity driven by the trauma stimulus, and demonstrate a higher sensitivity to trauma-induced metabolic stress.

Compared with 40AG13/CyO sibling control flies, untreated nrv3/40AG13 heterozygotes had similar SEP amplitudes at all time points (Fig. 4 A and B, dark bars). Immediately after trauma, however, control and mutant flies had significant SEP reductions (Fig. 4 A and B, light bars, 0 d; 40AG13/CyO: P < 0.0001, nrv3/40AG13: P < 0.0001, t test), with control and nrv3 heterozygotes showing 44.4% and 48.4% reductions, respectively. One day posttrauma, treated 40AG13/CyO and nrv3/40AG13 SEPs had partially recovered, but were still significantly reduced—with the nrv3 heterozygotes showing a larger deficit—from controls (by 20.7% and 29.8%, respectively). Seven days after trauma, control fly SEPs were not significantly different from traumatized flies (Fig. 4A, 7 d; t test), whereas nrv3 heterozygotes (Fig. 4B, 7 d) had not fully recovered, with SEP values significantly lower than nontraumatized flies (P = 0.0311, t test; a 16.4% reduction).

Fig. 4.

Reduced Na⁺/K⁺ ATPase β subunit gene dosage sensitizes flies to acoustic trauma. Mean SEPs for control (black bars) and traumatized (white bars) flies. (A) 40AG13/CyO control SEPs. n = 20–22 for 0 d, 23–28 for 1 d, and 16–18 for 7 d. (B) nrv3/40AG13 heterozygote SEPs. n = 16–23 for 0 d, 23–25 for 1 d, and 21 for 7 d. Horizontal bars and asterisks show significant differences (t test; *P < 0.05; ***P < 0.001; ****P < 0.0001). Error bars indicate SEM.

Anatomical Effects of Trauma in Control and nrv3 Mutants.

Using transmission electron microscopy (TEM), we examined the JO scolopidia of traumatized and control flies (40AG13 and nrv3 heterozygotes) for anatomical differences at 0, 1, and 7 d posttrauma (for scolopidial morphology, see Fig. 1 A–C). No consistent gross morphological differences were found resulting from genotype or trauma condition (Figs. S3–S5). All scolopidia showed normal structure in both longitudinal and cross-sections. These results suggest robust homeostatic mechanisms that maintain the auditory organ’s physiological competence even under high-duty-cycle metabolic stress.

Compared with untreated controls, traumatized 40AG13/CyO and their sibling nrv3/40AG13 mutant heterozygous flies also displayed no gross JO anatomical differences at 0, 1, or 7 d posttreatment. At the subcellular level, mitochondria in JO neurons appeared smaller 7 d posttrauma. To quantify this effect, JO neuron mitochondrial cross-sectional areas were measured for each genotype/treatment group 0, 1, and 7 d after trauma cessation (Fig. S6). Immediately and 1 d after trauma, 40AG13/CyO control mitochondria did not show significant reduction in the area (Fig. 5A, 0 d, 1 d), whereas 7 d posttrauma, a 13.8% decrease in the area was found (Fig. 5A, 7 d; P < 0.001, Kruskal-Wallis). Nonexposed nrv3/40AG13 mutant heterozygotes (Fig. 5B, gray bars) generally had smaller cross-sectional areas than untreated 40AG13/CyO control flies. Acoustic trauma had a similar but more immediate and pronounced effect on nrv3/40AG13 flies as on the 40AG13/CyO controls. Neurons showed marked reductions in mitochondrial cross-sectional areas at all three time points (Fig. 5B, 0 d: ∼28% decrease, P < 0.001; 1 d: 14% decrease, P < 0.01; 7 d: 25.1% decrease, P < 0.001, Kruskal-Wallis). These results support the hypothesis that acoustic trauma induces metabolic stress on the fly’s auditory system, thereby affecting mitochondrial function.

Fig. 5.

Acoustic trauma reduces mitochondrial size. Natural log of mitochondrial cross-sectional areas calculated from (A) control flies (40AG13/CyO), and (B) nrv3/40AG13 heterozygous mutants 0, 1, and 7 d after acoustic trauma. Gray bars correspond to nontraumatized groups; white bars represent flies exposed to trauma. For control flies, N (no trauma) = 616 for 0 d, 504 for 1 d, and 581 for 7 d; N (trauma) = 474 for 0 d, 473 for 1 d, and 529 for 7 d. For mutants, N (no trauma) = 650 for 0 d, 810 for 1 d, and 810 for 7 d; N (trauma) = 772 for 0 d, 865 for 1 d, and 601 for 7 d. Horizontal bars and asterisks indicate significantly different pairs (Kruskal-Wallis; ns, not significant, **P < 0.01, ***P < 0.001). Boxes represent 25–75 percentiles from the median value (middle line). Whiskers indicate 5–95 percentiles.

Effect of Acoustic Trauma on Circadian Locomotor Behavior.

Acoustic trauma, in addition to sensorineural effects, may affect the animals’ gross behavior and activity levels, which in turn may manifest as changes to mitochondrial morphology. To control for this indirect effect, we exposed several fly lines to sham and noise trauma and monitored circadian locomotor behavior (SI Materials and Methods). In the first experiment (Fig. S7), we found no obvious differences in locomotion or circadian effects for flies of CS or 40AG13 control strains at 1-h resolution (Fig. S7 A, 1–2), and both sham and noise-treated flies of both genotypes showed expected crepuscular activity peaks during light/dark (LD) conditions (Fig. S7A, days 1–5, alternating light/shaded regions). Decreasing temporal resolution to 12 and 24 h (Fig. S7 B, 1–2 and C, 1–2) also revealed no significant acoustic trauma effects.

In the second experiment, we tested 40AG13, 40AG13/CyO, and nrv3/40AG13 flies (Fig. S8). Again, no qualitative differences in hourly locomotion resulting from noise trauma and no significant trauma effects at the 12- or 24-h resolution (Fig. S8 B, 1–3 and C, 1–3) were observed in any genotype (Fig. S8 A, 1–3). Our results are consistent with the idea that the decreased mitochondrial areas seen in 40AG13/CyO and nrv3/40AG13 require normal auditory function, rather than arising as a by-product of altered locomotor behavior posttrauma treatment.

Discussion

In this study, we establish that Drosophila react to acute acoustic trauma with significant changes in auditory physiological response magnitude and latency, auditory behavior, and mitochondrial size. In studies using cats (28), gerbils (29), and guinea pigs (12, 13), auditory nerve (AN) compound action potential (CAP) reductions are on the same order as the Drosophila SEP attenuation. The acoustic test stimulus and primary test level (85 dB SPL, 14.2 mm/s PV) was chosen for ethological relevance (15, 16) and to ensure a robust physiological response. Most acoustic trauma studies focus on changes in auditory thresholds and CAP/single-fiber tuning curves (6, 14, 29–31) by systematically varying test stimulus SPL and frequency; similar studies in Drosophila may be needed to better compare trauma effects by expressing changes in auditory sensitivity in terms of acoustic energy.

Importantly, our results show that this physiological effect has significant consequences on whole organism behavior. That traumatized 40AG13 males show reduced courtship behavior (Fig. 2C) suggests that the reduction in auditory system function has a real effect on mating behavior. We currently cannot determine whether the reduction in courtship is attributable only to a sensory deficit in the auditory system or whether there are also effects of trauma on courtship circuits downstream of hearing. However, noise trauma did not significantly change fly locomotor behavior immediately after (Fig. 2D) or 1 wk posttrauma (Figs. S7 and S8), suggesting there is no general effect of trauma on behavior.

CS SEPs showed larger nontraumatized responses across all SPLs used (Fig. 2 and Fig. S1) and demonstrated slightly smaller reductions in SEP magnitudes in response to trauma than did 40AG13, suggesting that genetic background differences affect hearing sensitivity and response to physiological perturbation. Genomic differences have significant effects on the susceptibility and response to interventions in NIHL and age-related hearing loss in mice (32, 33) and humans (34). A strength of Drosophila is that rapid generation times, large number of lines, and diverse molecular and genomic tools enable large-scale and fine-resolution genetic studies, making it an excellent system to study interactions between genetic background effects and NIHL.

Increased SEP latencies with trauma (∼40–50 µs for SEP pulses two through five) resemble delays observed in other systems. CAP latencies (measured at the N₁ peak) similarly increased in cats after acute acoustic trauma (28). Gunshot noise trauma delayed guinea pig CAPs by ∼50–200 µs compared with controls (13). Conversely, other studies found reduced latencies of single AN fibers in response to acute noise (28, 30, 31, 35). Latency reductions are probably due to the particular mammalian cochlea mechanical tonotopy and tuning properties. Increasing probe SPLs to elicit a response in a tonotopic region with trauma-induced threshold shift may activate fibers with a much higher center frequency, which are more basal in the cochlea and have a smaller traveling wave delay and shorter latency (11, 36). Basilar membrane mechanical property changes from OHC damage are also implicated in latency reduction (35, 36). Because the JO lacks the cochlear structure and traveling wave dependence, it is unsurprising that this response to noise trauma is absent.

As with SEP amplitudes, it is difficult to compare recovery time courses between studies using different species and experimental parameters. Nonetheless, Drosophila hearing recovery resembles vertebrate acoustic trauma responses in some aspects. At 7 d posttrauma, SEP amplitudes for both CS and 40AG13 flies were not significantly different from unexposed controls, whereas SEP latencies remained elevated. Previous studies show that not all noise-induced physiological changes recover at identical rates. In gerbils exposed to noise trauma, CAP (N₁) amplitudes had not returned to pretrauma levels, even if CAP thresholds recovered (29). In guinea pigs (14) and mice (9), large losses in AN fibers or IHCs lead to long-lasting reductions in CAP or auditory brainstem response, which remain even after behavioral or physiological thresholds return to normal. Often the nominal thresholds “mask” permanent damage caused by trauma, which may involve ongoing degeneration of neuronal elements (9).

One striking result of our study is the lack of JO gross morphological changes after trauma. In vertebrates, acoustic trauma often correlates with stereocilia disruption (37), OHC and IHC damage or loss (4, 6, 7, 26, 37), spiral ganglion cell loss (6, 7, 14, 37), and damage to supporting tissue and nonsensory cochlear cell types (6, 7). Although nompC is an important mechanosensitive channel in Drosophila auditory function (18, 38, 39), the identity, arrangement and structure of other mechanotransduction system elements (e.g., additional mechanosensitive ion channels, support and connector proteins) in JO sensory neuron cilia are unclear (40). Thus, we cannot rule out that noise trauma mechanically disrupts this system as observed in vertebrate IHC stereocilia (41). Because models of Drosophila transducer function suggest similar properties as vertebrate hair cells (40), similarity in trauma-induced damage patterns would not be surprising. Previous studies showed changes in auditory epithelial cell–cell junctions after trauma, suggesting these junctions (and their molecular components) as noise-induced damage targets (42, 43). In the alligator, noise trauma sufficient to cause only temporary threshold changes induced transient microlesions in hair cell plasma membranes, allowing abnormal Lucifer Yellow dye diffusion into hair cell cytoplasm (4). Posttrauma, hair cell appearance and CAP responses were grossly normal. JO scolopidial integrity may also be transiently compromised during trauma, and thus appear normal on subsequent histological analysis.

Anatomical structures and functions of auditory systems undergoing acoustic trauma in Drosophila and vertebrates are sufficiently different that the mechanisms responsible for SEP or CAP decreases may not be shared. It is currently thought that reduced CAP after intense sound trauma is caused by reduced AN fiber recruitment (resulting from damage or death of hair cells and/or spiral ganglion neurons) and from a “broadening” of the response from desynchronization of afferents with similar center frequencies (11, 36). Also, hair cells or AN fiber death or degeneration may occur over days to months (6, 9). We investigated physiological and anatomical status immediately after, 1 d, and 7 d posttrauma. Although 7 d is a significant portion (∼10%) of average Drosophila melanogaster lifespan, molecular and cellular changes responsible for cellular damage and degeneration may still require longer posttrauma survival times to manifest in sensory cell loss.

In contrast to gross morphology, we found subcellular trauma-induced changes. Traumatized flies showed consistent and robust decreases in mitochondrial cross-sectional area in JO neurons; these changes were not associated with any changes in circadian locomotor behavior (Figs. S7 and S8). After trauma, mitochondrial dysfunction has been implicated as a cause of auditory pathology (44, 45) and a consequence of acoustic trauma via oxidative stress and reactive-oxygen species production (27, 46–48). Reduced mitochondrial size is often associated with disrupted balance of ongoing mitochondrial fission/fusion, biasing the system toward fission. Increased fission is also associated with cellular metabolic stress and may activate apoptotic pathways (49, 50). Elevated metabolic stress may arise from excitotoxicity (46, 47) and/or disruption of ionic homeostasis (27, 46). The latter is supported by nrv3 heterozygote responses to noise trauma. The reduced Na⁺/K⁺-ATPase β subunit in JO neurons (Fig. S2) is sufficient to maintain normal sensory function (Fig. 4B) under nontraumatized conditions, but not with increased demands under noise trauma. This manifests in an exaggerated immediate hearing detriment, lack of full recovery 7 d after trauma (Fig. 4B), and exaggerated mitochondrial size reduction in mutant heterozygotes (Fig. 5B). Thus, the acoustic trauma protocol can be used to reveal cryptic or dominant phenotypes in animals with compromised genotypes whose hearing appears normal under unstressed conditions. The result is particularly striking because the mutants still had one functional nrv3 gene copy.

Notably, the processes underlying the physiological and anatomical deficits seem to be somewhat “decoupled,” or acting on different time scales. In both 40AG13/CyO and nrv3/40AG13 flies, the physiological deficit from trauma is apparent immediately but after 7 d, control flies show recovery, whereas nrv3 heterozygotes still have reduced function. Trauma-induced differences in mitochondrial morphology show a different pattern: in control flies, they are only detectable after 7 d (by which time SEP amplitudes have recovered), whereas in nrv3 heterozygotes they are significant at all time points. This makes the relationship between physiological and anatomical effects of trauma unclear. Overstimulating the JO neurons could result in multiple metabolic stressors (e.g., reactive-oxygen species generation, molecular depletion, ionic imbalances) any of which could initiate reduction (and subsequent recovery) of stimulus-driven activity and could also initiate mitochondrial morphological change. The different time scales of mitochondrial effects in nrv3 heterozygotes may reflect chronic subphenotypic stress that sensitizes mitochondria to the effects of noise trauma. Genetic and molecular signaling cascades that may be involved in trauma-initiated physiological and anatomical changes have yet to be determined, and further gene transcriptional analysis is required.

We have shown that several acoustic trauma effects observed in vertebrates manifest in a Drosophila model, including reductions in evoked response magnitude, increased response latency, and mitochondrial changes indicative of cellular stress. This suggests similarity in trauma-induced changes in cellular physiology and gene expression in Drosophila and vertebrate models.

Materials and Methods

Genetic Strains.

Adult wild-type CS and control 40AG13 (described in ref. 25) flies were used. In addition, we crossed nrv3¹⁵/CyO (23) with 40AG13 flies; the resultant nrv3¹⁵/40AG13 heterozygous flies were tested vs. 40AG13/CyO sibling controls. Flies were reared on a yeast–cornmeal–agar medium at 22–25 °C and age-matched at 0–4 d posteclosion.

Acoustic Trauma.

Flies anesthetized with CO₂ were placed into custom-made plastic vials made by fusing the top halves of two 15-mL centrifuge tubes (Corning). The tube was coated with a strip of agarose/sucrose/apple juice gel running a tube’s inside length to provide nutrition and humidity. A small square of fine nylon mesh placed over one end of the tube was held in place by screwing a fitted cap with a large hole in the center. The other end was screwed into one of six caps embedded in a thick plastic plug inserted into the narrow end of a plastic funnel (20.3-cm base and 9.53 cm in height). Each of these caps also had central holes covered by nylon mesh. As Drosophila respond to the air PV component, both mesh-covered open tube ends allowed sound-driven air to pass longitudinally through the tube with minimal restriction. The funnel base was coupled to an 8-inch (20.3-cm) 8Ω speaker (Radio Shack) so that air PV was efficiently transmitted through the tubes.

Trauma stimulus was a continuous 250-Hz tone, approximating the D. melanogaster pulse-song carrier frequency (15) and that consistently produces maximal SEP amplitudes. The tone, stored on a G3 iMac computer, was fed to an Optimus MPA-40 amplifier (Radio Shack) driving the speaker. The speaker and tube apparatus was installed in a modified incubator (68.9-cm × 46.7-cm × 45.7-cm interior volume) lined with acoustic foam (Auralex Acoustics) at 22–25 °C and humidified with a water pan on the incubator floor.

Flies were exposed to trauma stimulus for 24 h at ∼120 dB SPL, with PV values ranging from 340 to 336 mm/s within the tubes. Nontraumatized control flies were placed in identical tubes and kept for 24 h at the same temperature as exposed flies. After each treatment period, flies from each trauma condition (trauma or sham) were placed in food vials until used in neurophysiological recordings.

Electrophysiology.

Electrophysiology was performed as described previously (18) (SI Materials and Methods). Synthetic acoustic stimuli for electrophysiology resembled the pulse components of Drosophila courtship song (15): five pulses, each 5 ms in duration, with an interpulse interval of 35 ms, for a total stimulus duration of 165 ms. Pulse trains were repeated 10 times, with an interstimulus interval of 425 ms. Stimulus trains were presented to the fly at four sound pressure levels: 65, 75, 85, and 95 dB SPL, with corresponding PVs of 0.5, 4.2, 14.2, and 40.7 mm/s. All acoustic stimuli (trauma and neurophysiological) were calibrated with a digital SPL meter (CEL-240, Casella CEL Inc.) and a custom-built calibrated PV microphone with preamplifier [gift of M. Göpfert (University of Göttingen, Göttingen, Germany)].

Recordings were performed a few hours after cessation of acoustic trauma (0 d time posttrauma) for some flies, and 1 d and 7 d later for others. The genotype/trauma treatment group identities were blinded to the experimenter and used in random order. Equal numbers of male and female flies were used in each recording session. Averaged SEPs were calculated and analyzed offline using numerical analysis software (GNU Octave). For SEP magnitudes, differences between control and traumatized groups were tested at all time points and SPLs using the Student t test (two-tailed; α = 0.05), with Welch’s correction if data were heteroscedastic. SEP latencies were defined as the time elapsed between the first positive peak of each stimulus pulse to the first large negative peak of the associated SEP (Fig. 1D). Because these times incorporate the sound PV travel time from the speaker to the fly and because the zero point is arbitrary, the absolute values of the calculated SEP latencies are less important than the values relative to each other. Timing data were similar for both genotypes (determined using histograms and Bartlett's test for homogeneity of variance), so SEP latency data were pooled for CS and 40AG13 flies. Poststimulus peak latencies were determined for pulses two through five in the train. Analysis of the first pulse was omitted because of hardware-induced variability in the pulse shape. Statistical differences were tested using Welch’s modified t test (two-tailed; α = 0.05) because of sample heteroscedasticity.

Courtship Behavior Analysis.

The courtship behavior of traumatized and control 40AG13 flies was examined using established protocols (25), with the following modifications. Synthetic pulse stimuli ranged from 76 to 78 dB SPL (0.9–2.0 mm/s PV) within the behavior chamber. Courtship behavior was scored during 1 min of silence followed by 1 min of synthetic pulse song, and normalized courtship index scores were calculated as percentage of increase in directed courtship behavior by dividing the courtship index scores for males presented song to those during the preceding silence. Significance was tested with Student t test (two-tailed; α = 0.05) after verifying normality and homoscedasticity.

Circadian Locomotor Analysis.

Circadian rhythm analysis was performed as described (51). Locomotor activity of control and noise-exposed adult male 40AG13 flies (3–5 d posteclosion) was monitored with Trikinetics Activity Monitors for 8 d of LD (12 h/12 h; Fig. S8 A, 1–3) at 25 °C. Data were analyzed offline using numerical analysis software (GNU Octave) and statistical software (Prism, GraphPad). Activity levels were quantified as averaged number of crossings per unit time (Figs. S7 and S8). Differences between traumatized and control flies were tested via Student t test (two-tailed; α = 0.05) (SI Materials and Methods).

Transmission Electron Microscopy.

Heads of flies were dissected 0, 1, and 7 d after trauma and fixed overnight at 4 °C in 2.5% (vol/vol) paraformaldehyde and 2% (vol/vol) glutaraldehyde in 0.1 M phosphate buffer (PB), pH 7.4, then rinsed in PB. They were postfixed in 1% (wt/vol) OsO₄ in PB for 1 h, dehydrated, and processed for embedding in Epon 812. Ultrathin sections (75 nm) were stained with aqueous uranyl acetate and lead citrate and examined with a Jeol 1230 electron microscope.

Mitochondrial Measurements.

Mitochondrial cross-sectional areas were measured from 10 neurons in each of three male nrv3/40AG13 heterozygotes and 40AG13/CyO sibling controls at 0, 1, and 7 d after sham or trauma treatment. Total mitochondria for each genotype/treatment group ranged from 473 to 865, with a total of 7,685 measurements. To ensure comparable results between genotypes and treatments, we restricted our analysis to longitudinal TEM sections of JO sensory neurons containing a large and well-defined nucleus profile (Fig. 1C) and measured all visible mitochondria. Measurements were made using a Bamboo tablet (Wacom Inc.) and ImageJ (NIH) on calibrated digital TEM images (Fig. S6). Because of nonnormal shapes of the data distributions, we used the natural log of cross-sectional area, resulting in quasi-normal data amenable to statistical testing. Differences were tested using the Kruskal-Wallis test (α= 0.05), followed by Dunn's test of multiple comparisons.