Auditory robustness and resilience in the aging auditory system of the desert locust

After overexposure to loud music, we experience a decrease in our ability to hear (robustness), which usually recovers (resilience). Here, we exploited the amenable auditory system of the desert locust, Schistocerca gregaria , to measure how robustness and resilience depend on age . We found that gene expression changes are dominated by age as opposed to noise exposure. We measured sound-evoked nerve activity for young and aged locusts directly, after 24 hours and 48 hours after noise exposure. We found that both young and aged locusts recovered their auditory nerve function over 48 hours. We also measured the sound-evoked transduction current in individual auditory neurons, and although the transduction current magnitude recovered in the young locusts after noise exposure, it failed to recover in the aged locusts. A plastic mechanism compensates for the decreased transduction current in aged locusts. We suggest key genes upregulated in young noise-exposed locusts that mediate robustness to noise exposure and find potential candidates responsible for compensatory mechanisms in the auditory neurons of aged noise-exposed locusts.


Introduction
Life does not get easier or more forgiving; we get stronger and more resilient.Although this might prove true mentally, it is the opposite for our physiology.Both our robustness (the ability to resist change due to a stressor) and resilience (the ability of a system to recover after a stressor) decline with age (Ukraintseva et al., 2021).Both robustness and resilience are intricately associated with vulnerability to chronic disease and longevity (Kirkland et al., 2016).A useful conceptual framework to understand robustness and resilience is that of the network level built on hierarchies: from whole organisms to organs, cells, organelles, and then proteins (Crespi et al., 2021).At the whole organism-level, a large part of our understanding of the age-dependence of robustness and resilience is derived from human clinical data due to the intense monitoring of patients directly after and following recovery from trauma (Cauley et al., 2000;Niemeinen et al., 1981).For instance, middle-aged men are less frail with fewer disabilities than females during middle life but become less resilient in advanced age (Oksuzyan et al., 2008).
At the organ-level of the robustness-resilience hierarchy, the human nervous system (brain) remains the best studied due to the ease of cognitive tests on recovering patients.Patients >40 years are more likely to demonstrate higher levels of disability following traumatic brain injury than their younger counterparts, with the greatest improvement in survivors 16-26 years (Marquez de la Plata et al., 2008).Murine models also exhibit an age-dependence of recovery.Middle-aged rats have diminished recovery of motor function after moderate spinal cord injury compared to younger counterparts (von Leden et al., 2017).Dopaminergic neurons in the brain of aged mice recover poorly compared to younger animals in response to toxic injury (induced by the neurotoxin MPTP).In Caenorhabditis elegans, the regenerative ability of axons to recover decreases from 65% to 28% in young and old worms, respectively, mostly due to the inability of injured axons to form growth cones from severed axon stumps (Byrne et al., 2014;He and Jin, 2016).This all suggests that all nervous systems lose their resilience with age.
To bridge the robustness-resilience hierarchy between cells and organs, we require organs and experimental approaches with deep physiological penetrability and amenability.The brain is a multimodal integrator of information with countless cognitive measures of its performance, but auditory organs detect one sensory modality-sound-and have an almost singular performance output to detect sound-induced vibrations as sensitively as possible.The auditory system is an excellent physiological system to study robustness and resilience as a function of age.Its function can be intricately and rigorously quantified, and it can be stressed in a likewise quantitative and precise manner through noise exposure.Auditory system function can be repeatedly decreased, and the system can repeatedly recover.Proteins and organelles can be studied alongside whole-organ function.
There is large interanimal diversity in auditory organ robustness and resilience.For instance, the auditory systems of young and old rats are equally robust and resilient to 113 dB sound pressure level (SPL) noise exposure (Fraenkel et al., 2003), whereas pubescent or young mice lacked the resilience of 16-week-old mice (comparing Jensen et al. [2015] and Kujawa and Liberman [2009] and Ohlemiller et al. [2000]).Counterintuitively, aged mice were more robust to noise exposure (as measured from change in thresholds) (Kujawa and Liberman, 2006).Adding to the complexity is the intensity-dependence of resilience; the mouse's auditory system is resilient at low noise exposures but not so at loud intensities (Amanipour et al., 2018).The cellular basis of auditory robustness-or lack thereof-has been attributed to the loss of ribbon synapses in sensory inner hair cells in mice (Kujawa and Liberman, 2009;Wang and Ren, 2012) and integrity of the lateral wall (Ohlemiller et al., 2018).
A few pioneering studies have combined measures of robustness and resilience by measuring auditory responses directly after and weeks after noise exposure (Kujawa and Liberman, 2009).Kujawa and Liberman (2009) localized a lack of robustness and resilience to specialized ribbon synapses (at the base of the sensory inner hair cells) that decreased directly after noise exposure (decreased robustness) and failed to recover (decreased resilience).By contrast, outer hair cells lacked robustness but were resilient after noise exposure.The endocochlear potential, an electrochemical gradient across the sensory epithelium necessary for efficient transduction, appeared resilient (i.e., able to recover) but not robust to change (Ide and Morimitsu, 1990).
No study to date has measured the age-dependence of robustness and resilience in an auditory system.We exploited the experimentally versatile auditory organ of the desert locust.Auditory systems of insects capitulate fundamental aspects of hearing loss, including age-related (Austin et al., 2023;Blockley et al., 2022;Keder et al., 2020) and noise-induced (Boyd-Gibbins et al., 2021;Warren et al., 2020) auditory decline.Müller's organ of locust spans a small section of the inside surface of the tympanum.Locusts have 2 tympani embedded laterally in the first abdominal segment.Müller's organ is composed of 3 groups of ~80 auditory neurons, with ~46 group-III auditory neurons tuned to ~3 kHz (Jacobs et al., 1999;Warren and Matheson, 2018).We measured the effect of noise (or mock noise exposure) immediately after (robustness), 24 hours after, and over 48 hours (resilience) for young and aged locusts.At these three time points, we measured the electrophysiological properties of the auditory neurons, their transduction currents, and auditory nerve activity.We combined our detailed electrophysiological approach with gene expression analysis directly after noise exposure in young and aged locusts to understand changes in robustness at the molecular level of the hierarchy.

Locust husbandry
Desert locusts, Schistocerca gregaria (males), were reared in crowded gregarious conditions 150-250 in 60 cm 3 cages in their fast-aging gregarious state, where they can live up to 2 months.They had a 12 hours light/dark cycle at 32/25 • C and were fed on a combination of fresh wheat and bran ad libitum.The gregarious state of the desert locust contrasts with their isolated solitarious state where they live for up to 9 months.The founding progeny of the Leicester Labs strain were solitary copulating adults collected at Akjoujt station ~250 km north-east from Nouakchott, Mauritania in May 2015.We performed a survival assay of our crowded gregarious locusts.We found that 21 days post last molt into adults, more than half of the locust population had died.Very few locusts continued to die over the next 12 days until we stopped counting.

Locust aging, noise exposure, and acoustic stimulation
After hatching, desert locusts go through 5 molts before they are adults with functional wings.Both control and noise-exposed locusts ~10 days post their last molt (as they change color from pink to gray/ yellow) were taken from their large 60 cm 3 cages and had wings clipped to expose their abdominal tympani.We do not find it likely that trauma from wing clipping will have a large effect on the auditory system.Similar transcriptomic responses to wound healing, which may influence hearing ability (e.g., serine protease response), have only been found local to the wound site in Drosophila (Patterson et al., 2013).Young (10 days post last molt) locusts were directly exposed to noise and then used as the young cohort.To generate the aged cohort, we took locusts as described above and transferred them to aquarium tubs (40 × 20 × 30 cm) with ab libitum wheat and milled bran for a further 14 days.Locusts were housed in aquarium tubs when not used directly after noise exposure (the 24 and 48 hours cohorts).All aquarium tubs were in identical environmental conditions to the large 60 cm 3 cages.
For noise exposure, between 10 and 20 locusts in both the noiseexposed group and the control group were placed in a cylindrical wire mesh cage (8 cm diameter, 11 cm height).Both cages were placed directly under a speaker (Visaton FR 10 HM 4 OHM, RS Components Ltd).For the noise-exposed group only, the speaker was driven by a function generator (Thurlby Thandar Instruments TG550, RS Components Ltd) and a sound amplifier (Monacor PA-702, Insight Direct Ltd) to produce a 3 kHz tone at 120 dB SPL, measured at the top of the cage where locusts tended to accumulate.Throughout the paper, we refer to the noise that the locusts are exposed to as a 3 kHz 120 dB SPL pure tone.This tone was played continuously for 12 hours overnight (21:00-09:00) for the noise-exposed group during their natural darkness period.The control group was housed in an identical cage with a silent speaker for 12 hours.
All recordings were performed within a 12 hours window during the day.Each locust was only experimented on once.SPLs were measured with a microphone (Pre-03 Audiomatica, DBS Audio) and amplifier (Clio Pre-01 preamp, DBS Audio).The microphone was calibrated with a B&K Sound Level Calibrator (CAL73, Mouser Electronics).For hook electrode and patch-clamp recordings, the locust ear was stimulated with the same speaker and amplifier as above with a 3 kHz pure tone duration of 0.5 seconds.For hook electrode recordings, the 3 kHz tone had a rise and fall time of 2 ms.Tones were played 3 times for each locust at each SPL and the average response was taken for each SPL.For intracellular recordings from individual auditory neurons, the speaker was driven by a custom-made amplifier controlled by an EPC10-USB patch-clamp amplifier (HEKA-Elektronik) controlled by the program Patchmaster (version 2x90.2,HEKA-Elektronik) running under Microsoft Windows (version 10).

In vivo hook electrode recordings from auditory nerve six
Locusts were secured ventral side up with their thorax wedged in a plasticine channel and their legs splayed and held down with plasticine.An advantage of using insects is that they do not need to be anesthetized, avoiding any deleterious effects of the anesthetic and more accurately representing in vivo conditions.Locusts show no behavior that suggests experience of pain, do not show protective behavior to damaged parts of the body, and continue to exhibit normal behavior despite traumatic injury (Burrows, 1997).A section of the second and third ventral thoracic segment was cut with a fine razor blade and removed with fine forceps.Tracheal air sacks were removed to expose nerve 6 and the metathoracic ganglia.This preparation left the abdomen, including the first segment where the ears reside, intact, thus maintaining the operation of the ear in vivo.Hook electrodes constructed from silver wire 18 µm in diameter (AG549311, Advent Research Materials Ltd) were hooked under the nerve, and the nerve was lifted out of the hemolymph.Signals were amplified 1000 times by a differential amplifier (Neurolog System), and then filtered with a 500-Hz high-pass filter and a 50-kHz low-pass filter.These amplified and filtered data were sampled at 25 kHz by Spike2 (version 8) software running on Windows (version 10).To compute the σ ratio, we rectified the nerve signal and removed any DC offset.We then divided the sound-evoked response by the averaged surrounding background neural activity.In some batches of recordings, 3-kHz pickup was detected for the largest sound amplitude tones.These recordings were adjusted using a consistent multiplication factor to remove the effect of this pickup from the data.For all locusts, the treatment was blinded to the experimenter until all data were collected and analyzed.

Whole-cell patch-clamp recordings
Electrodes with tip resistances between 3 and 4 MΩ were fashioned from borosilicate glass (0.86 mm inner diameter, 1.5 mm outer diameter; GB150-8P, Science Products GmbH) with a vertical pipette puller (PC-100, Narishige).Recording pipettes were filled with intracellular saline containing the following (in mM): 170 K-aspartate, 4 NaCl, 2 MgCl2, 1 CaCl2, 10 HEPES, 10 EGTA, and 20 TEACl.Intracellular tetraethylammonium chloride was used to block K + channels necessary for the isolation of the transduction current.To further isolate and increase the transduction current, we also blocked voltage-gated sodium channels with 90 nM Tetrodotoxin in the extracellular saline.The addition of ATP to the intracellular saline did not alter the electrophysiology of the recordings so was omitted.During experiments, Müller's organs were perfused constantly with locust saline.The saline was adjusted to pH 7.2 using NaOH.The osmolality of the intracellular and extracellular salines' was 417 and 432 mOsm, respectively.
Whole-cell voltage-clamp recordings were performed with an EPC10-USB patch-clamp amplifier (HEKA-Elektronik) controlled by the program Patchmaster (version 2x90.2,HEKA-Elektronik) running under Microsoft Windows (version 7).Electrophysiological data were sampled at 50 kHz.Voltage-clamp recordings were low-pass filtered at 2.9 kHz with a 4-pole Bessel filter.Compensation of the offset potential was performed using the "automatic mode" of the EPC10 amplifier, and the capacitive current was compensated manually.The calculated liquid junction potential between the intracellular and extracellular solutions was also compensated (15.6 mV; calculated with Patcher's-PowerTools plug-in from www3.mpibpc.mpg.de/groups/neher/index.php?page=s oftware).Series resistance was compensated at 77% with a time constant of 100 μs.The resting potential was measured directly (within 10 seconds) after whole-cell recordings were established by changing the clamped voltage until the current was 0. We measured the amplitude of the discrete depolarisations by measuring the largest 3 discrete depolarisations at a − 100 mV holding potential.For all patch-clamp electrophysiological analysis, we used Igor Pro 9 (Wavemetrics Inc).

RNA extraction
Müller's organs from the tympanal ears of locusts were extracted by piercing the tympanum with fine forceps with the two forcep tips flanking the folded body, on which Müller's organ is anchored.The forceps were closed around the folded body to grasp Müller's organ, which was then pulled out back through the tympanal opening.Müller's organs were wiped onto a frozen pestle and submerged into a liquid nitrogen-surrounded Eppendorf tube.We pooled both ears of ~15 locusts into 4 biological cohorts for each of the 4 conditions (young control, young noise-exposed, aged control, and aged noise-exposed).One sample failed quality control and was discarded, leaving 15 samples in total.Our previous study found no statistical difference in cell number between young and aged locusts (Blockley et al., 2022), so we did not feel the need to adjust for cell number or RNA yield.The frozen samples were homogenized by hand with the pestle for 3 minutes while the Eppendorf tube remained in the liquid nitrogen.Trizol (10 μl) was added to the samples and homogenized for a further 3 minutes at room temperature.All further steps were at room temperature.Next, 490 μl of Trizol was added to the samples and mixed gently by pipetting.Samples were centrifuged for 10 minutes at 12,000 g.The supernatant of the sample was collected and left to incubate at room temperature for 5 minutes.Chloroform (100 μl) was added to the samples, and the sample was vortexed.Samples were incubated for 3 minutes at room temperature and centrifuged for a further 20 minutes at 12,000 g.The aqueous phase was separated into a fresh RNase-free tube with 1 μl of glycogen and 250 μl of isopropanol added before room-temperature incubation.
Samples were centrifuged for 10 minutes at 12,000 g.The supernatant was removed and the RNA was washed with 500 μl of 75% ethanol.The pellet was then dislodged via vortexing, before another centrifuge spin at 7500 g for 5 minutes.The supernatant was removed and the pellet was air dried until it began to turn clear (approximately 2 minutes).
Samples were suspended in 10 μl of RNase-free water and incubated for 5 minutes at 55 • C. RNA was quantified via nanodrop, and quality control was carried out on a Bioanalyzer (Agilent).Samples were stored at − 80 • C.

Bioinformatic analysis
Samples were sequenced on the Illumina platform by Novogene sequencing (Cambridge).Quality trimmed samples read quality was checked using FastQC (v0.11.5) (Andrews, 2010).Transcriptomic reads were mapped to the iqSchGreg1.2reference genome (NCBI) using the STAR 2-pass method (v2.7.9a) (Dobin and Gingeras, 2015).Gene count number was generated using HTSeq (Anders et al., 2015).Sample normalization and differential expression analysis were performed using DESeq2 (v1.28.1) (Love et al., 2014) on Rstudio (4.2.1) using the model: design = ~ condition (with condition: young control, young-deaf, aged control, and aged-deaf).To calculate p-values, we used DESeq2's inbuilt Wald test and p-values were adjusted for multiple hypothesis testing using DESeq2's inbuilt Benjamini and Hochberg false discovery rate method.Statistically significant genes were those with p-adj <0.05.We did not set a fold change as a threshold.Gene ontology (GO) terms were generated for the locust genome as in our previous study (Austin et al., 2023).GO term enrichment was performed using topGO (Alexa and Rahnenfuhrer, 2022) with an false discovery rate <0.05.When listing our top 10 GO terms (Fig. 4), only terms which contained more than 1 gene were included.

Power analysis
We designed experiments that formed (Figs. 2, 3) to have a power above 95%, which gives an ability to detect a difference between young and aged or control and conditioned locusts or control and experimental Müller's organs of 95%.(If these series of experiments were run an infinite number of times.)Our false negative rate, or type II error probability, is <5% (1-power) (probability of not finding a difference that is there).Our false positive rate, or type II error (probability of finding a difference that is not there), is determined by our p-values, which was set at 0.05.In order to calculate the power, we used the raw data and effect size reported in Blockley et al. (2022) for patch-clamp recordings of the transduction current 48 hours after recovery from noise exposure (Blockley et al., 2022, Fig. 6Diii).There exists no analytical methodology for conducting power calculations on linear mixed-effect models (LMEM).Therefore, we generated a dataset simulated from the raw data of Blockley et al. (2022), fitted an LMEM, and then ran repeated simulations of the LMEM 500 times.We used the proportion of times that the LMEM reported a difference to calculate the power.For this paper, the experiments that measured differences in transduction current all had at least 95% power when using 12 individual control locusts and 12 individual noise-exposed locusts.This simulated power analysis also only applies for the effect size reported in Blockley et al. (2022) and may be lower if the actual effect size, in this paper, is reduced.Models were fitted in R (version 2.4.3), on a Windows PC running Windows 10 using the package LME4 (Bates et al., 2015), and simulations were run with the package simr (Green and MacLeod, 2016).

Statistical analysis
Throughout the manuscript, n refers to the number of recorded neurons and N refers to the number of Müller's organ preparations used to achieve these recordings (i.e., n = 23, N = 13 means that 23 neurons were recorded from 13 Müller's organs).Only 1 Müller's organ was used for electrophysiology experiments.All n numbers are displayed on the figures for clarity.The spread of the data is indicated by 1 standard deviation as the standard deviation indicates the spread of the data, unlike standard error.Median and Q1 and Q3 are displayed by bars when individual measurements are plotted.For all hook electrode recordings, 80% of patch-clamp recordings, the treatment of the locust (control or noise-exposed) was blinded to the experimenter; lone working conditions, due to COVID restrictions, made complete blinding impossible.All data either remained blinded or were recoded to be completely blind when analyzing the data to avoid unconscious bias.
To test for differences and interactions between control, noiseexposed, and aged locusts, we used either a linear model (LM) or LMEM, with treatment and age as fixed effects and locust identity and SPL as a random intercept, when repeated measurements are reported.Models were fitted in R (version 3.4.3)with the package LME4 (Bates et al., 2015).The test statistic for these analyses (t) is reported with the degrees of freedom (in subscript) and p-value, which are approximated using the Satterthwaite equation (lmerTest package) (Kuznetsova et al., 2017).Curves were fitted to the data using the drm package in R for patch-clamp and hook electrode recordings (Ritz and Strebig, 2015).The drm package was also used to compute t and p-values when comparing control and noise-exposed 4-part log-linear models.F statistics of the log-linear model fits were computed by excluding treatment (control or noise-exposed) as a factor.Higher F statistics denote a stronger effect of treatment.
In order to compare responses between control and noise-exposed locusts across SPLs, we adopted an approach first implemented in pharmacology research.In our work, the "dose-response curves" are equivalent to SPL-auditory response curves.This allowed us to maximize the information contained in each dataset and to quantitatively compare model parameters such as hill coefficient (steepness of slope), maximal asymptote (maximum σ ratio), and inflexion point (σ ratio at the steepest and noise exposure for young and aged locusts.(D) σ ratio in young control locusts and young noise-exposed directly after noise exposure.Black and red dots and gray and pink shaded regions represent the average and positive standard deviation for control and noise-exposed locusts, respectively.Thin gray and pink lines are individual plots for each auditory nerve for control and noise-exposed locusts, respectively.The inset F statistic gives a quantitative comparison of the effect of noise exposure.The hill coefficient, maximum asymptote, and inflection point are compared between the average σ ratio of control and noise-exposed locusts with respect to SPL. (E) σ ratio for young control and young noise-exposed locusts 24 hours after noise exposure.(F) σ ratio for young control locusts and young noise-exposed 48 hours after noise exposure.(G) σ ratio for aged control and aged noise-exposed locusts directly after noise exposure.(H) σ ratio for control and aged noise-exposed locusts 24 hours after noise exposure.(I) σ ratio for aged control and aged noise-exposed locusts 48 hours after noise exposure.part of the slope).We did this using the drm function of drc package (Version 3.1-1, Ritz and Strebig, 2015).
We fitted 4-part log-linear models to each individual locust's auditory response, with auditory nerve responses (σ ratio) as the dependent variable with treatment (control or noise-exposed) and SPL as the independent variables.The t and p-values are reported for each model parameter: hill coefficient, maximal asymptote, and inflexion point-shown on each graph in Figs. 2 and 3.The equation of the 4-parameter log-linear fits is: where Y is the σ ratio, b is the slope at the inflexion point, c is the lower asymptote, d is the higher asymptote, and e is the SPL (or X value) producing a response halfway between b and c.
To test whether the factor of treatment (control or noise-exposed) significantly affected auditory nerve response, we compared the above model to a model in which treatment was omitted as an independent variable, using the ANOVA function (Ritz et al., 2015).This gave an F statistic labeled on each graph in Figs. 2 and 3.The model failed to predict realistic values for the maximum asymptote, especially for noise-exposed locusts.Therefore, to statistically compare the maximum transduction current, we performed an LM on the transduction current at 110 dB SPL with noise exposure as the independent variable for each locust (hook electrode recordings) or neuron (patch-clamp recordings).

Net transcriptomic changes and auditory nerve response between young and aged locusts
We extracted ears from young and aged male locusts either directly after a 12 hours noise exposure or after a 12 hours mock noise exposure.RNA was extracted, sequenced, and reads aligned to the iqSchGreg1.2reference genome.We performed a principle component plot on normalized reads and found that the samples were bioinformatically well separated based on age (Fig. 1A) but not so for noise exposure.Far more genes changed in expression (p-adj < 0.05) as a function of age alone, with 1172 differentially expressed genes (DEGs) between young and aged locusts not exposed to noise (Fig. 1B).In this group, the soundevoked auditory nerve response was higher in young locusts compared to aged locusts (Fig. 1E and F, stat = 16.58,p ≤ 0.0001).Nine hundred thirteen genes were differentially expressed between young and aged noise-exposed locusts, with a majority (647) of DEGs shared between young and aged locusts not exposed to noise (Fig. 1B).For these noiseexposed locusts, the auditory nerve response in young locusts directly after noise exposure was higher than aged locusts after noise exposure (Fig. 1F and F, stat =20.65, p ≤ 0.0001).
In the differential expression analysis for auditory conditioning (noise-exposed versus non-noise-exposed control), there were fewer DEGs when compared to aging.Exposure to noise in either young or aged locusts only led to 160 DEGs (Fig. 1C).Young locusts differentially expressed 154 genes after noise exposure.In aged locusts, only 9 genes were differentially expressed after noise exposure, 3 of which were found to be differentially expressed in young and old locusts (Fig. 1C).

Auditory nerve response in young and aged locusts
We recorded directly from auditory nerve six, using hook electrodes while stimulating the intact auditory system with a 3 kHz tone from a loudspeaker (Fig. 1D).We recorded young and aged locusts that were either noise-exposed or non-noise-exposed (control).Recordings were carried out directly after 24 hours and 48 hours after noise exposure (example raw data for each day shown in Fig. 2A-C).To compare the auditory nerve response, we fitted our data with a 4-part log-linear model and computed the F statistic, which gives an objective measure of the difference; an F statistic of 0 means there is no difference in the nerve response.
There was a clear difference in the recovery of the transduction current magnitude between young and aged noise-exposed locusts (Fig. 3A-D).After noise exposure in the young locusts, the transduction current reduced (maximum asymptote: t (59) = − 2.89, p = 0.005), compared to control locusts (F statistic of 27.3) (Fig. 3E).The transduction current magnitude in the noise-exposed young locusts then recovered over the next 48 hours (maximum asymptote at 24 hours, LM: t (95) = − 1.43, p = 0.15 and 48 hours, LM: t (40) = − 0.84, p = 0.41), which resulted in an F statistic that was reduced to 7.74 and 4.05 at 24 and 48 hours (Fig. 3E-G).Noise exposure in the aged locusts led to a greater difference compared to their non-noise-exposed counterparts (maximum asymptote, LM: t (47) = − 4.86, p = 1.3 × 10 − 5 ) directly after noise exposure (F statistic = 39.4).The transduction current maximum asymptote of the aged noise-exposed locusts remained below that of control locusts at 24 and 48 hours after noise exposure (24 hours, LM: t (84) = − 4.95, p = 3.8 × 10 − 6 ; 48 hours, LM: t (41) = − 3.54, p = 1 × 10 − 3 ).Other parameters, such as the inflection point of the loglinear fit to the transduction current recovered in the aged locusts, resulted in an F statistic that decreased to 19.8 over 48 hours (Fig. 3H-J).The transduction current of aged locusts failed to recover to the same extent as the young locusts after 48 hours (F statistic at 48 hours after noise exposure 4.05 vs. 19.8,young vs. aged) (Fig. 5B).

Differential gene expression and GO term annotation
We analyzed DEGs between cohorts of locust in four groups: young control versus aged control, young noise-exposed versus aged noiseexposed, young control versus young noise-exposed, and aged control versus aged noise-exposed, as shown in Fig. 1A and B. We focused first on aging alone, where 1172 genes changed (Fig. 1B), with our findings resembling those of our previous study (Austin et al., 2023).Control-aged locusts had GO terms and genes associated with innate immune response whereas young control locusts had increased GO terms for cuticular development and metabolic processes (Fig. 4Aii, E).
The GO terms generated by differential expression between the young noise-exposed and aged noise-exposed locusts strongly resemble those found in their young and aged control counterparts (Fig. 4B).Of the 913 genes found to be differentially expressed between young noiseexposed and aged noise-exposed locusts (Fig. 1B), just over half of these DEGs were upregulated in young locusts (Fig. 4Bi).Genes differentially expressed between young and aged locusts, regardless of whether they have been noise-exposed or not, include 2 genes for cuticle protein and 2 further genes that determine tissue structural properties (elastin and reelin) (Fig. 4E).We found several biological processes outside of the top 10 GO terms that may provide an explanation for our electrophysiological observations (Figs. 1,2,3,5).GO terms enriched between young noise-exposed and aged noise-exposed locusts include positive regulation of neuronal action potential, termination of signal transduction, axonal defasciculation, motor neuron axon guidance, neuron cellular homeostasis, neurotransmitter metabolic process, ion transport, synaptic target recognition, and numerous terms associated with calcium ion binding (Supplementary Information).Most of these GO terms were also found between young control and aged control locusts.
In comparison to the effects of aging, fewer genes changed in response to noise in young locusts, and even fewer in aged locusts (Figs.1B, 4Ci, 4D).More genes are upregulated than downregulated after noise exposure in young locusts (Fig. 4Ci).The top 10 GO terms upregulated after noise exposure in young locusts are dominated by neural processes (Fig. 4Cii), unlike in age-related comparisons.As well as these top 10 GO terms, we also observed GO terms associated with positive regulation of neuron projection development, ion-gated channel activity, and protein heterodimerization activity.Between aged control and aged noise-exposed locusts, there were not enough DEGs to generate GO terms (Fig. 4D).The 9 genes that are differentially expressed are related to sugar metabolism/detection, cytosolic transport, and gene expression (Fig. 4D and E).A full list of all DEGs is available in the Supplementary Information.

Age dominates over noise
We measured the ability of young and aged auditory systems to recover after noise exposure.We measured both robustness (the extent of the immediate change) and resilience (the ability to recover).Age has a large effect on the auditory nerve response, and the number of DEGs as a function of age is nearly 10-fold higher.A total of 1172 genes changed as a function of age alone between 10 and 24 days in Müller's organ (Figs.1B, 4A).Only 154 genes were differentially expressed as an immediate response to noise exposure in young locusts, and only 9 in old locusts (Figs.1B, 1C, 4).Although age is the best predictor of disease (Niccoli and Partridge, 2012), neurological degeneration (Hou et al., 2019), and physiological performance (Donato et al., 2003;Rittweger et al., 2008), we were surprised that such a prolonged, loud, and specific insult of noise on a dedicated auditory system produced so few DEGs.Age-related deterioration versus toxic insults in other organs paint a mixed picture.Age dominates the decline in kidney function twice as accurately as life-long smokers (Toyama et al., 2020), whereas the effect of ethanol on markers of liver function outweighs most aging effects in mice (Giavarotti et al., 2002).Obviously, if the stressor is severe or specific enough, age may dominate less.An auditory-relevant example was found in mice exposed to noise at 100 or 91 dB SPL; only the louder noise exposure resulted in permanent impairment.At 91 dB SPL, the mice recovered and their auditory response was indistinguishable even a year later (Fernandez et al., 2015).For hearing, even though repeated noise exposure reduces hearing thresholds, age-related auditory decline comes to dominate in the later stages of life in both humans and the auditory model of the desert locust (Blockley et al., 2022;Corso, 1992).A major target of noise exposure in the mouse auditory system is the ribbon synapses at the base of the auditory receptors that transfer information to the spiral ganglion neurons of the auditory nerve (Fernandez et al., 2015;Kujawa and Liberman, 2009).In the desert locust, the number of auditory receptors decreases steadily with age, and noise exposure compounds this effect.Noise also decreases the receptor lymph potential and auditory nerve width (Blockley et al., 2022).

Plasticity within the locust's auditory dendrite
In young locusts exposed to noise, their Müller's organs recover their electrophysiological function over 48 hours.Resilience was measured at both the level of the transduction currents in auditory neurons and the action potential-propagating auditory nerve.After noise exposure in older locusts, the transduction currents of auditory neurons are not only less robust but less resilient compared to noise-exposed younger locusts (Fig. 5).Although the auditory nerve response of older noise-exposed locusts was less robust than younger noise-exposed animals, it was no less resilient.This level of auditory resilience in the nerve is maintained despite the reduction in resilience of the transduction current with age.Axons of the auditory neurons, which comprise the auditory nerve, are downstream and action potentials are triggered by the transduction current (Warren and Matheson, 2018).We suggest a compensatory plastic mechanism that mitigates for the lack of transduction current to Fig. 5. Auditory nerve response recovers, but the transduction current remains lower than controls over 48.The transduction current is generated in the cilium, and the nerve response is generated at the axon hillock and is measured from the axon bundles that form the auditory nerve.F statistics of comparisons of the 4-part log-linear model fit were taken from Figs. 2 and 3 for (Ai) the transduction current and (Aii) nerve response (σ ratio) of noise-exposed and control locusts directly after, 24 hours after, or 48 hours after mock noise exposure (black) or noise exposure (red).The F statistic gives a quantitative measure of the difference between noise-exposed locusts and locusts not exposed to noise (control), such that an F statistic of 0 represents no difference.(B) Schematic of the auditory neuron where these electrophysiological responses are recorded.The transduction current is less robust and resilient in response to noise exposure compared to the nerve response.deliver a similar auditory nerve response to sound.The textbook site of neural plasticity is the synapse (Kandel et al., 2000), but here compensatory changes take place between the cilium (where transduction takes place) and the dendritic spike initiation site (assumed to be the dendrite dilation), located ~5 µm proximal to the cilium (Fig. 5).In auditory neurons of fruit flies, there is an increased concentration of voltage-gated sodium channels in the bulbous ciliary dilation (Ravenscroft et al., 2023).This suggests that the ciliary dilation is the site where the transduction current is integrated to trigger the active (voltage-gated channel-mediated) propagation of transduction potentials in the form of discrete depolarisations.We believe that the dendrite dilation is the dendritic spike initiation point.Regardless of where they are generated, dendritic spikes travel to the soma where axonal spikes are triggered, presumably in the axon hillock (Hill, 1983;Warren and Matheson, 2018).
Synapses are the typical site associated with compensatory plasticity (Turrigiano, 2012).Such plasticity in the mammalian auditory system is suggested to maintain similar levels of auditory input.If auditory input decreases (due to noise overexposure or age deterioration), the gain of auditory signals can be increased in a compensatory fashion.A by-product of such excessive gain is hyperactivity, which leads to tinnitus, the perception of noise in the absence of auditory input (Schaette and Kempter, 2006).Previously, in the locust's auditory system, we found no increase in the background spiking rate of the auditory nerve in aged locusts with decreased auditory performance (Austin et al., 2023).This suggests that whatever mechanism compensates for decreased input of the transduction current avoids the problematic side-effect of increased spontaneous activity.
A potential mechanism to compensate for decreased transduction current and still deliver sensitive auditory nerve response is a decrease of the action potential threshold, such that less transduction current is required to trigger a spike.This is supported by our GO terms "positive regulation of action potential" differentially expressed between young noise-exposed and aged noise-exposed locusts.It appears that this GO term is not dependent on noise exposure but develops regardless as a function of age.This would allow the auditory neurons of older locusts to compensate for reduced transduction currents.At the gene expression level, we find many voltage-gated ion channel genes change their expression after noise exposure such as the β− 2 subunit of the voltagegated potassium channel and the α− 1 of voltage-dependent calcium channel.It is also possible that changes occur in the cilium itself to compensate against a decreased transduction current; when comparing young control and young noise-exposed locusts, we find GO terms associated with ciliary protein localization and organization (Fig. 5).Mature auditory receptors of mammals do not spike after development, and single channel conductances appear as binary steps between open and closed states of the transduction channel (Beurg et al., 2015), unlike the discrete depolarisations of insect auditory receptors (Hill, 1983;Warren and Matheson, 2018).The absence of action potentials and no active mechanism to conduct the transduction current suggest that compensatory mechanisms of mammalian hair cells are likely to be restricted to the hair cell's ribbon synapse, such as increasing the Ca 2+ dependence of vesicle exocytosis (Boero et al., 2021).

Gene transcription changes suggest parallel dysfunction in the auditory nerve and the cilium
The auditory nerve is a common site of dysfunction in mammals and insects (Blockley et al., 2022;Pilati et al., 2012;Tagoe et al., 2014;Wan and Corfas, 2017).Schwann cells, that envelop the axons that form the auditory nerve, change their phenotype upon nerve injury (Jessen and Mirsky, 2005).In young noise-exposed locusts, we find a suite of neuronal GO terms (in the top 10) upregulated compared to their non-noise-exposed counterparts, including "central nervous system myelin formation" and 2 further axon-specific GO terms (Fig. 4C).Although the locust's auditory nerve is not myelinated, Schwann cells layer around multiple axons to serve the same insulating and supporting function (Gray, 1960).The neuronal genes differentially expressed after noise exposure include genes related to fasciculation.Fasciculation is the process whereby growth cones advance along the axons of already established nerves (Honig et al., 1998).After the axons are formed, cell-cell adhesion molecules maintain the nerves integrity (Denda and Reichardt, 2007;Wu and Reddy, 2012).The collection of fasciculation-related genes differentially expressed after noise exposure includes "tenurin-m," "laminin subunit alpha," subunit gamma-1," and "integrin alpha-PS2-Like."Loss of ten-m causes aberrations in fasciculation and incorrect nerve localization in the neuromuscular junction (Zheng et al., 2011).Integrin binds to other proteins in the extracellular matrix, such as laminin and fibrillin, also found in to be upregulated after noise exposure (Arimori et al., 2021;Jovanović et al., 2008).We speculate that there is dysregulation of fascicular homeostasis after noise exposure, and these genes are expressed in response.Of the fasciculation genes differentially expressed in the locust Müller's organ, only the mouse ortholog of "integrin alpha-PS2-Like," Itga2b is differentially expressed in the modiolar tissue of young mice (day 1 and day 7) after noise exposure (Newton and Forsythe, 2017).The modiolar tissue contains the myelinated spiral ganglion neurons that form the auditory nerve.

Conclusion
The transduction current of auditory neurons is resilient in young locusts but loses its resilience in older locusts.The extent of the noisemediated decrease in transduction current, robustness, is also exacerbated in older locusts.The sound-evoked auditory nerve response, however, is just as resilient in young and aged locusts exposed to noise.To explain this discrepancy, we propose a compensatory mechanism integrated into the auditory neuron's distal dendrite that increases the gain of the transduction current.A change in the action potential threshold of dendritic spikes is the most intuitive explanation.Gene expression changes support coincident dysfunction in the auditory nerve.

Funding sources
Ben Warren was funded by a Royal Society University Research Fellowship URF\R1\180022 and Enhancement Award FR\ERE210220.Thomas Austin and Christian Thomas were funded by a Royal Society Enhancement Award, awarded to Ben Warren.

CRediT authorship contribution statement
BW performed patch-clamp electrophysiology and analyzed the data.TA performed hook electrode electrophysiology and analyzed the data.TA also RNA extracted.CT performed bioinformatic analysis.All authors composed the figures, wrote, and approved the manuscript.
• Originality and plagiarism: This work is original, and all analyses were performed uniquely for this paper.Similar methods were used in a previous paper by the authors, and sufficient reference for this work has been used.• Data access and retention: Public access will be made for bioinformatic files by uploading FASTAs to the SRA database.Some data have already been submitted to the SRA database.• Multiple, redundant, or concurrent publication: This work will not be submitted elsewhere unless we receive rejection.The work will be submitted as a concurrent preprint on bioxriv.

Fig. 1 .Fig. 2 .
Fig. 1.Changes in gene expression and sound-evoked auditory nerve response as a function of age.(A) Principal component analysis of four cohorts.(B) Number of differentially expressed genes as a function of age in control and noise-exposed locusts.(C) Number of differentially expressed genes as a function of noise exposure in young and aged locusts.(Di) Schematic of experimental setup for hook electrode recordings from auditory nerve six.(Dii)The σ ratio is calculated by rectifying the nerve potential, integrating the area underneath the curve, and then dividing the sound-evoked nerve response by the background signal.(E) σ ratio of the auditory nerve response to a 3 kHz tone for young control and aged control locusts at different sound amplitudes.Black and red dots and gray and pink shaded regions represent the average and positive standard deviation for young and aged locusts, respectively.Thin gray and pink lines are individual plots for each auditory nerve for young and aged locusts.The inset F statistic compares the 4-part log-linear fit to the σ ratio from each individual nerve between young and aged locusts.The hill coefficient, maximum asymptote, and inflection point are compared between the average σ ratio of young and aged locusts with respect to SPL. (F) σ ratio of the auditory nerve response for young and aged locusts directly after noise exposure at different sound amplitudes.

Fig. 3 .
Fig. 3. Transduction current after noise exposure in young and aged locusts.(A) Schematic of experimental setup for patch-clamp recordings from individual auditory neurons during airborne acoustic stimulation.(B) Example sound-evoked transduction currents in response to a 3 kHz 110 dB SPL tone when voltageclamped at − 100 mV for control and noise-exposed and young and aged locusts directly after noise exposure.(C) Example sound-evoked transduction currents for control and noise-exposed and young and aged locusts 24 hours after noise exposure.(D) Example sound-evoked transduction currents for control and noiseexposed and young and aged locusts 48 hours after noise exposure.(E) Transduction current for noise-exposed and control young locusts at different sound amplitudes directly after noise exposure.Black and red dots and gray and pink shaded regions represent the average and positive standard deviation at each SPL for control and noise-exposed locusts, respectively.Thin gray and pink lines are individual plots for each auditory neuron for control and noise-exposed locusts, respectively.The inset F statistic gives a quantitative measure of the effect of noise exposure.The hill coefficient, maximum asymptote, and inflection point are compared with the average transduction current of control and noise-exposed locusts with respect to SPL. (F) Transduction current for noise-exposed and control young locusts at different sound amplitudes 24 hours after noise exposure.(G) Transduction current for noise-exposed and control young locusts at different sound amplitudes 48 hours after noise exposure.(H) Transduction current for noise-exposed and control-aged locusts at different sound amplitudes directly after noise exposure.(I) Transduction current for noise-exposed and control-aged locusts at different sound amplitudes 24 hours after noise exposure.(J) Transduction current for noise-exposed and control-aged locusts at different sound amplitudes 48 hours after noise exposure.

Fig.
Fig. Volcano plots the 10 most significant differentially expressed genes (DEGs) and enrichment plots with the 10 most enriched gene ontology (GO) terms.GO terms have been color-coded to denote cuticular, neural, and immune processes.(Ai) Volcano plot of the 10 most DEGs in young control and old control locusts and (Aii) top 10 enriched GO terms.(Bi) Volcano plot of the 10 most DEGs in young noise-exposed and aged noise-exposed and (Bii) top 10 enriched GO terms.(Ci) Volcano plot of the 10 most DEGs in young control and young noise-exposed locusts and (Cii) top 10 enriched GO terms.(D) Volcano plot of the 10 most DEGs in aged control and aged noise-exposed locusts.(E) Top 10 most DEGs for each plot in (Ai-Di), color donates in which group the gene is upregulated.

Use of patient images or case details:
• Acknowledgement of sources: We have adequately cited previous sources which have contributed to this research • Disclosure and conflicts of interest: We are not aware of any conflicts of interest.• Fundamental errors in published works: We agree to promptly notify the publisher if we are made aware of any errors in the work.• Reporting standards: We believe we have presented an objective account of the research performed and discussed its significance in the wider field.• Hazards and human or animal subjects: This work possesses no research performed on humans or protected animals.• No patient or case details are included in this work.