A burden shared: The evolutionary case for studying human deafness in Drosophila

studying deafness. By exploiting more forcefully the molecular-genetic conservation between human hearing and hearing in morphologically distinct models, such as the fruit fly Drosophila melanogaster , we believe, a deeper understanding of hearing and deafness can be achieved. An understanding that moves beyond the surface of the ‘deafness genes ’ to probe the underlying bedrock of hearing, which is shared across taxa, and partly shared across modalities. When it comes to understanding the workings (and failings) of human sensory function, a simple fruit fly has a lot to offer and a fly eye might sometimes be a powerful model for a human ear. Particularly the use of fly avatars , in which specific molecular (genetic or proteomic) states of humans (e.g. specific patients) are experimentally reproduced, in order to study the corresponding molecular mechanisms (e


Hearing and deafness -the backdrop
Amongst the various sensory ailments that afflict humankind, the widespread loss of hearing probably cuts the deepest.The socioemotional fabric of human existence is intimately linked to our ability to hear -and listen to -a multiplicity of sounds, most notably languagebased communication and music.It may therefore have come as no surprise that an impaired sense of hearing (much more so than a loss of vision (Deal and Rojas, 2022), was identified as one of the major risk factors for the wider neurological decline summarily referred to as dementia (Livingston et al., 2017).One could posit that for humans, Descartes' famous dictum 'cogito ergo sum' (I think, therefore I am), often transposes to 'audio ergo sum' (I hear, therefore I am).
The central role that the sense of hearing plays in human identity is perhaps best exemplified by the fact that its loss can be culture-forming in its own right.The sensory void caused by (mostly congenital) deafness has proven a powerful force for the creation of an entirely new language (sign language), novel cultural traditions and communities, collectively referred to as Deaf Culture.While other forms of sensory loss have also led to distinct technical or cultural innovations (e.g. the tactile writing system Braille), none of them seems to have reached the levels seen for deafness.All of this goes to demonstrate not only the cultural prolificacy of humans, but also the central role of our sense of hearingand deafnesstherein.Both in its presence and absence, hearing lies close to human nature; as such it has catalysed and fuelled cultural richness and complex societal debate (Belohrad, 2023;Cooper, 2007;Dennis, 2004;Freeman et al., 2022;Mand et al., 2009;Sparrow, 2005Sparrow, , 2010;;Tucker, 1998;Wallis, 2020).
Notwithstanding the crucial contributions that audition makes to our sense of identity, neither the sense of hearing nor its loss, are unique to humans.In fact, much of what we know about the function (or loss of function) of the human ear has been learned from research across a wide range of other animal species, including African clawed frogs (Xenopus laevis), domesticated chicken (Gallus gallus domesticus), Mongolian gerbils (Meriones unguiculatus), common house mice (Mus musculus), and others.
Across evolution, hearing organs have developed in a vast number of animal taxa.Yet one group can arguably claim the prize for having been particularly prolific in this regard.In insects, ears are thought to have evolved independently more than 20 times (Yack, 2004).Insect ears can be divided into two functionally distinct categories.On the one hand there are the pressure-sensitive 'tympanal ears', which capture sound-generated forces over the area of specialized receiver membranes (tympanal membranes), thus operating in a similar fashion to the ears of mammals, which employ 'ear drums' (or tympanic membranes) for the same purpose.On the other hand, there are 'antennal ears', which couple antennal receiver structures to the sound-generated motion of air particles.These antennal sound receivers operate like inverted pendulums that oscillate in response to friction with the surrounding air.An example for such an antennal ear can be found in the fruit fly Drosophila melanogaster, one of the standard models in today's, molecular-oriented life science (Albert and Kozlov, 2016).
But despite all the dissimilarities, both in operational mechanisms and anatomical presentation, that are evident not only between the ears of humans and insects (see Fig. 1), but also between the different types of insect ears themselves (or between the different ears of vertebrates for that matter) there is a common ground, a bedrock one might say, upon which all animal hearing rests.This bedrock of hearing is of a distinctly molecular, microscopic and internal nature; it defies the stark divergence that manifests itself in macroscopically visible external traits, which together form vastly distinct ears.Within that bedrock and the molecular machinery it is composed of, ears do not only become increasingly more similar across species, they partly become virtually identical, as the corresponding genes -and gene productshave retained their sequence identities and are able to mutually rescue each other's functions!
The material carriers of this identity are evolutionarily conserved molecular modules, which have retained their specific gene and protein sequences, functional affiliations and immediate interaction networks.They represent, in other words, living molecular fossils that have remained largely unchanged on their journey through deep time.The common ancestry of the fundamental building blocks of life, is arguably the central epistemic rationale for much of today's life-scientific efforts, from evolutionary medicine to research using the genetic model organism Drosophila melanogaster, aka 'the fruit fly'.
An impressive experimental demonstration of such a building block also originated from Drosophila.It concerns one of the major proneural genes of the animal ear, which was named atonal, as flies deficient for the gene failed to develop chordotonal organs (ChOs) (Jarman et al., 1993), the fly's equivalent of vertebrate inner ear hair cells.atonal belongs to the class of basic-helix-loop-helix (bHLH) transcription factors (Baker and Brown, 2018), it is a key developmental regulator of the Drosophila antennal ear and the mechanosensory cells of Johnston's Organ (JO), the Drosophila 'inner ear'.atonal was later shown to be equally essential for the formation of inner ear hair cells in the vertebrate cochlea (Bermingham et al., 1999).
Further work, which explored the functional interchangeability of fly and mammalian atonal genes could show that the Drosophila gene and its mouse orthologue Math1 can mutually rescue each other's null mutant phenotypes (Wang et al., 2002) demonstrating a remarkable functional conversation, especially when considering that their last common ancestor will have died almost 700 million years ago.Going one step further, Weinberger et al. (2017) (Weinberger et al., 2017) probed the functional roles, and conservation, of atonal coding sequences across a wide range of taxa and demonstrated that for each taxon, from sponges, annelids and cephalochordates to the mouse, the atonal orthologues (i) were able to produce a functional Drosophila ear and (ii) led to distinct phenotypes, which suggested the rescue of specific modules of auditory function.Numerous review articles have compared the specific modes of operation of insect and vertebrate (focus: human) ears (Albert and Kozlov, 2016;Warren and Eberl, 2023;Warren and Nowotny, 2021) and identified multiple ways, by which the study of insects can make vital contributions to our scientific knowledge, and clinical application of that knowledge.The antennal ears of the fruit fly could take a leading role in this effort.The hearing mechanisms of Drosophila (Albert and Göpfert, 2015) are poised at a serendipitous junction between an operative simplicity -granting easy experimental access -and an informative complexity -harbouring value for the translation to mammals.To give two brief examples of these settings: (i) Drosophila auditory transducer function can be studied quantitatively in vivo, and in intact animals (Albert et al., 2007); (ii) these data allow for an unparalleled quantification of auditory ageing and homeostasis (Keder et al., 2020); (iii) Drosophila auditory amplification is a local, transducer-based process, independent from efferent synaptic feedback (Kamikouchi et al., 2010) and Prestin-mediated somatic motility (Kavlie et al., 2015).A powerful model inevitably needs to be simpler than the phenomenon it is meant to represent.This is also true for models of human hearing (Albert et al., 2020).Thus, it is both the similarities and dissimilarities to the mammalian condition that provide value and together set the stage for using Drosophila in radically new, and more focused, ways.This review is meant to describe the underlying rationale and possible implementations.Specifically, we will use atonal as a benchmark gene (see Fig. 2) to propose the generation of fly avatars (Bangi et al., 2021) for probing the specific functional (cellular and subcellular) roles of identified human genes, and groups of human genes, including those of patients.Fly avatars have emerged previously as a promising tool towards personalized medicine for cancer research (Jiang et al., 2022).

Human hearing and deafnessa 'fly over'
Hearing loss is ubiquitous and multicausal.Currently more than 1.5 billion people, i.e. nearly 20 % of the world's population, live with hearing loss; the WHO projects this to rise to 2.5 billion by 2050 (WHO World Report on Hearing).Congenital hearing loss occurs in approximately 1-3 out of 1000 live births and thus accounts for only ~1 % of the total load of global hearing impairment.Roughly half of all congenital hearing loss is considered to be hereditary and ~70 % of it due to isolated, non-syndromic, monogenic causes.In sum, monogenic forms of congenital hearing loss thus represent less than 0.5 % of all human hearing impairment.Recent researchand current clinical efforts have nonetheless -rightly and successfully -concentrated on this cohort, as it offered both clear avenues for therapeutic intervention and unprecedented insights into the function of our ears (Ahmed et al., 2017;Akil et al., 2019Akil et al., , 2012;;Al-Moyed et al., 2019;Petit et al., 2023).The fact remains however that we are still staring at the tip of an iceberg, or a mountain for that matter …and a high mountain it is.
Age-related decline, exposure to harmful levels of noise, as well as infections and ototoxic drugs all contribute to this Mount Deafness.But they affect different people differently and our ability to predict, and possibly prevent, the loss of hearing is still rather poor.But despite the as yet unexplained, and clinically unmanageable, complexity of hearing loss, there is strong evidence that much of it (an estimated 60-70 %) is due to genetic factors (Raviv et al., 2010).
It is against this background that we here propose to make better, more targeted and more concerted use of existing scientific models for research into auditory genetics (both in vivo and in silico).Only this, we feel, will enable to understand the molecular bedrock that underlies the function, and malfunction, of our ears.To explore, and illustrate, this case we have conducted a cursory 'fly over' of the corresponding orthology landscape between humans, fruit flies (Drosophila melanogaster) and common house mice (M.musculus).In a nutshell, we compiled two comprehensive (yet short of exhaustive) lists of genes known to be associated with hearing loss, one for humans and one for the mouse (see supplemental spreadsheet).We then compared these gene lists to each other and to their respective fly orthologues.
The compilation of genes associated with hearing loss in both human and mouse was achieved through thorough analysis using various databases and literature sources in September 2023.Initially, the human hearing loss gene list was constructed by querying the term "hearing impairment" in the advanced search feature of the Online Mendelian Inheritance in Man (OMIM; see https://omim.org/)website.The search criteria included the presence of "hearing impairment" in the text, records featuring clinical synopses containing hearing impairment, and Mendelian Inheritance in Man number prefixes associated with a phenotype description for which a molecular basis is known.Furthermore, a meticulous curation process was undertaken to assemble a gene list manually, incorporating genes from the Hereditary Hearing Loss Homepage (see https://hereditaryhearingloss.org/), the Deafness Variation Database version 9 (Azaiez et al., 2018), and a selection of genome-wide association studies (GWASs) on age-related hearing loss from the literature (Trpchevska et al., 2022;Wells et al., 2019).Conventional literature search using PubMed was also conducted to curate additional genes implicated in the human auditory system.
Mouse genes were curated using the Mouse Genome Informatics (MGI; see https://www.informatics.jax.org/)Mammalian Phenotype Browser to select "hearing/vestibular/ear phenotype."Entries contained in the sub-trees of "abnormal ear physiology" and "abnormal ear morphology" were further assessed to extract genes associated with abnormal hearing.Otitis media was excluded, as well as entries without a gene position or hearing phenotype.The International Mouse Phenotyping Consortium (IMPC) (https://www.mousephenotype.org/)release 19 was accessed to download a list of genes associated with hearing loss within the "hearing/vestibular/ear phenotype" category to further cross reference content.All mouse genes were cross-referenced against literature describing large mouse screens for hearing loss (Bowl et al., 2017;Ingham et al., 2019;Lewis et al., 2022).Finally, the list of mouse genes was cross-referenced against the curated human gene list to determine differences between the two species.
Here, a significant disparity emerged on the most basic level when comparing hearing loss associated genes in humans and mice.As of end of September 2023, at least 579 genes are associated with syndromic and isolated hearing loss in humans; however, the number of associated genes in the mouse (989 genes) is already nearly double what we know in humans and only 239 genes are currently recognized as in common between the two species.Remarkably, a forward-genetics mouse screen published in 2019 estimated as many as 1000 genes to be involved in hearing (Ingham et al., 2019).In just four years, our understanding of mouse genetics in relation to hearing is rapidly approaching this projection, with the pace of novel mouse models showing no signs of abating.This discrepancy highlights a significant gap in our knowledge on several levels, evidenced by the disparity between common human and mouse genes.This suggests that our current understanding of the genetic contributions to hearing loss, particularly in humans, is likely significantly underestimated.In an attempt to start resolving this disparity we also produced two new 'pan-mammalian' deafness gene lists, (i) one list combined the mouse deafness genes with the mouse orthologues of the human deafness genes (list name: MHC, length: 1324 genes) and (ii) another list combined the human deafness genes with the human orthologues of the mouse deafness genes (name: HHC, length: 1384 genes).These lists are available as supplemental resource (Supplemental File 1) but were not used for our analyses.
Table 1 provides an overview of the core data we compiled, and compared, for this conceptual review.The detailed lists are available as Supplemental File 1.The data set contains (i) global lists of human and mouse protein coding genes, (ii) the above introduced lists of human and mouse deafness genes, as well as (iii) a list of human cilia genes.The inclusion of cilia genes emerged from the observation that genes expressed in Johnston's Organ (JO), i.e. the fly's inner ear (auditory transduction by JO neurons -in contrast to the hair cells of the human cochlea -operates with ciliated cellular compartments), shared a vast overlap with genes expressed in the fly's pupal eye (whichmuch like the human cochlea operates with non-ciliated, microvillar sensory cells).To provide only one quantitative example: 12,557, i.e. 94.2 % of all protein coding genes detected in JO (13,324), are also expressed in the fly's pupal eye.The strong overlap between the auditory and visual transcriptomes was evident throughout all our analyses.We conclude that these cross-modal similarities represent a large, and largely unexplored, resource for our understanding of human auditory function and disease.This aspect becomes even more potent when considering that the cilium (kinocilium), although not involved in their auditory transduction, is an essential cellular structure for the development of mammalian hair cells.The fly's eye, in turn, could be a powerful model for the study of microvillar organization and function, especially as it is an experimentally easily accessible structure, for which numerous neurogenetic tools and experimental techniques have been created.

Humans (mice) and flies: genomic snapshots of a distant relationship
When it comes to the most widely used, and arguably most powerful, mammalian model of human hearing and deafness, the mouse remains uncontested (Bowl and Dawson, 2015;Friedman et al., 2007;Jones and Jones, 2014;Ohlemiller et al., 2016;Schaette, 2014).In order to illustrate, and explore, the molecular basis of this current consensus, we have subjected the protein coding genomes of humans, mice and fruit flies to a cursory analysis (Fig. 3).The last common ancestor of the two mammals and Drosophila is thought to have died ~686 million years ago (Mya; median separation time).The last ancestor of mice and humans, in contrast, was still alive ~87 Mya (see Fig. 3A).
Our data set (see table 1 for a numerical overview) shows 19,786 protein coding genes in humans.For 15,989 (≙ ~80.8 %) of these genes at least one DIOPT tool supports the existence of at least one orthologous (human-fly) gene-pair (simple DIOPT score ≥1).For mice, our data contained 22,610 protein coding genes.For 16,467 (≙ ~72.8 %) of these, at least one DIOPT tool supports the existence of at least one orthologous (mouse-fly) gene-pair (simple DIOPT score ≥1).The total number of predicted orthologous gene pairs is 33,938 (human-fly) and 31,572 (mouse-fly), respectively.The overlap between the two sets of orthologue pairs is substantial, with 30,938 pairs being shared (Fig. 3B).
Although this starting analysis confirmed a vast amount of orthologue pairs between mammals and fruit flies, their respective degrees of conservation differ vastly (ranging from 1 to 18 for the simple DIOPT scores; the highest possible simple score is 19).We thus inspected the density distributions of the corresponding weighted DIOPT scores (see Fig. 3C); weighted DIOPT scores take high-quality GO annotations into account to also assess functional conservations.For both human-fly and mouse-fly comparisons, violin plots of the distributions assume a characteristic 'umbrella-mushroom' shape: a first, major density of low-score gene pairs corresponding to genes with low degrees of conservation (weighted scores <4), sits at the top of the plot; a second, minor density of high-score gene pairs, corresponding to genes with high degrees of conservation (weighted scores >12), sits at the bottom.Across the global genome, the message is clear: mammals and flies share a common root, but their respective orthology space is dominated by distant relatives.The story (and corresponding plot), is inverted, however, when comparing humans and mice.The distribution now resembles an upsidedown mushroom.A major density of gene pairs with high scores (weighted scores >12, corresponding to high degrees of conservation) sits at the bottom of the plots, and a second, minor density with low scores, corresponding to low degrees of conservation (weighted scores <4), sits at the top.Another way to illustrate these relations is to create a circular plot (Fig. 3D), in which a 'divergence score' can be calculated for each pair of orthologues ([divergence score] = [maximum possible DIOPT scoreactual DIOPT score] / [highest possible score]).After sorting the resulting list in ascending order, the data is projected onto a 360 • circular plot, where the lowest divergence scores start at the origin (0 • ) and the highest divergence scores end at a full circle (360 • ); here, radial values indicate the divergence scores themselves (ranging from 0 to 1).In a graph organised like that, a linear divergence model, where all theoretically possible divergence scores are equally represented, would look like the dashed spiral in Fig. 3D.Every pair that is lying on the inside of that 'linear spiral' (=centripetal deformation) indicates a

Table 1
Quantification of ortholog pairs between species/organs/categories. Orthologous parameters were calculated with the DIOPT browser.higher degree of conservation than a linear model would suggest; every pair that is lying on the outside (centrifugal deformation), indicates a lower degree of conservation.Also here, one can clearly see that the mammalian-fly comparison is dominated by centrifugal deformation and only the inner core of gene pairs (near the origin, and virtually invisible in the graph) shows the centripetal signature of higher-thanexpected conservation.To take a glimpse at the functions of the most conserved, and most divergent, genes between humans and flies, we ranked them according to their weighted DIOPT scores and divided them into three groups (Low-, Middle-, High-Score).Each group was subjected to an analysis with the Gene Ontology enRIchment anaLysis and visualization (GOrilla) tool (Eden et al., 2009).The most highly enriched terms can be seen in Fig. 3E.In a nutshell, the most divergent (i.e.most derived/most apomorphic) gene pairs, are related to gene regulatory or biosynthesis functions; the middle tercile houses many classic 'effector' genes, linked to ion channel or transmembrane transport function; the tercile with the most conserved (i.e.most ancestral/most plesiomorphic) pairs, in turn, is dominated by genes involved in various forms of metabolic function.Especially for the most highly conserved group of gene-pairs, the fly ear (or eye) might be a powerful model of human hearing and deafness (see Fig. 3F for the respective interrelations).

The orthology space of deafness: when hearing is lost, kinship is gained
We then repeated our previous mammal-fly orthology analysis, this time focusing on the two lists of mammalian deafness genes we had created, one for humans (579 genes) and one for mice (989 genes).An initial, eyeballing comparison of the DIOPT score distribution heatmaps for the global gene lists (including all protein coding genes, Fig. 4A left) to the deafness-specific gene lists (including a comprehensive list of genes currently linked to hearing impairment, Fig. 4A right) shows a clear increase of more highly conserved gene-pairs (DIOPT scores >12) in the deafness cohorts.This was true for all three comparisons, humanmouse, human-fly, and mouse-fly.Violin plots of the corresponding DIOPT score distributions (Fig. 4B) show a pronounced increase of the highly conserved second density (DIOPT scores >12) in both the humanfly (Fig. 4B top) and the mouse-fly (Fig. 4B middle) comparison.The previous umbrella mushroom shape now gives way to a more dumbbelllike appearance with two almost equally strong populations of genepairs with low (DIOPT <4) and high (DIOPT >12) degrees of conservation, respectively.In the human-mouse comparison (Fig. 4B bottom), the population with low conservation is almost completely lost.Also across the deafness genome, the message is clear: the genes linked to hearing loss in humans (and mammals) are more closely related to, and more conserved in, flies than their global genomes (and evolutionary distance) would suggest.Genetically speaking, deafness appears to be an old disease, and its genetic fabric widely shared across extant taxa (see Fig. 4C-E for further illustrations).Here, also the fruit fly Drosophila melanogaster holds a considerable share that validates its usefulness as a scientific model for human hearing loss.At the same time, however, our analysis also highlights the value of the mouse model, as almost all known mammalian deafness genes are highly conserved between humans and mice.In fact, one conclusion that can be drawn from these relations is to more closely linkand align -the mouse and fly model.Combined, these two systems could propel each other.
Further encouragement comes from a recent study that spot-sampled the expressionand some aspects of functionof a selection of fly orthologues of human and mouse deafness genes (Sutton et al., 2024); the study found that ~70 % (38 of 54) of tested orthologues were expressed in one of the different cell types of the fly's ear.
For highly conserved genes, such as atonal, however, the experimental proof-of-principle has been provided that restoration of their function can happen across taxa (see also Fig. 2): Fly orthologues can rescue the function of mouse orthologues and vice versa, and it is all but certain that this also applies to the near-identical human orthologues.
One thus might use atonal homologues as evidence-based cut-off points, and concentrate the analyses along a three-stage fly-mouse-human model pipeline on gene-pairs with higher degrees of conservation than the respective atonal pairs (ATOH1/ato for human-fly and ATOH1/Atoh1 for human-mouse).In mouse, this would mean focussing on gene pairs with higher weighted DIOPT scores than 18.29; in Drosophila, this would mean focussing on gene pairs with higher weighted DIOPT scores than 7.87.As the shaded area in Figure panels 4C and 4D illustrates, this would still leave ~40-60 % of all currently known deafness genes within the scope of such a focused exploration.Precisely, these would be 227 (≙~40 %) deafness gene pairs in the human-mouse cohort and 373 (≙~64 %) deafness genes in the human-fly cohort.Notably, the overlap of those two cohorts is substantial, with 158 genes occurring in both (see Supplemental File 1).These 158 genes could form the starting platform for an unprecedented experimental exploration.'Humanized' fly avatars, in which these strongly conserved human deafness genes, or groups of those genes, have been reconstituted would serve as living petri dishes to help unravel the molecular interaction networks involved and their role in hearing and hearing loss (see Fig. 5 for a conceptual overview).To this end, the vast toolbox of fly genetics and existing (and publicly available) genetic libraries can be exploited.In a nutshell, for each identified human hearing loss (hHL) gene, which meets the criteria of Fig. 4. 'Deafness Genes'-the topical orthology space between flies, mice and humans.(A) Heatmap comparison between total, genomic (left) and topical, deafnessrelated genes (right).Note that deafness related genes show a substantially larger proportion of more highly-conserved, high DIOPT score genes.(B) Violin plot comparisons between total, genomic (left half of violins) and topical, deafness-related genes (right half of violins).Note that for all three comparisons, the deafnessrelated cohort assumes the signature patterns of more closely related gene sets (also see Fig. 3C).(C + D) Quantitative comparison of human deafness gene conservation in two model organisms of hearing research, mouse (C) and fly (D).To create the graph all pan-genomic orthologue pairs were ranked according to their DIOPT scores and then divided into 10 % wide subsections.The darker points show the percentage of deafness genes found in each subsection (left y-axes).The distribution was fitted with sigmoid curves (solid line).Averages (means) of DIOPT weighted scores of each 10 %-wide segment are also colour-coded.The lighter points (connected by dotted line) show the cumulative percentage.The vertical dashed line mark the DIOPT score positions, at which the respective atonal orthologues occur; the highlighted area to the left of that line hosts the genes we recommend for a deeper analysis through 'fly avatars'.(E) Divergence Spiral comparison of deafness-related and pan-genomic orthologue pairs.Note that for all deafness genes (more intensely coloured lines) a centripetal deformation is observed as compared to the pan-genomic data sets (paler colours), indicating a stronger conservation of the deafness-related genes.Color and style coding as in Fig. 2. being (i) conserved in Drosophila, (ii) predicted (or reported) to be expressed in the Drosophila ear and (iii) predicted (or reported) to contribute to fly hearing, existing Drosophila loss of function (LoF) mutants can be used for a rapid, confirmatory phenotyping.After successful confirmation of the hHL orthologue's suitability, novel 'null-mutant' fly lines can be generated, which are subsequently used for a heterologous rescue of the Drosophila 'null mutant' phenotype with its human gene orthologue.These 'humanised' Drosophila Avatars can then be used to test the human genes' specific roles in hearing (or the loss thereof).Avatars can also be employed to probe a human gene's (or gene-product's) suitability as novel biomarker for the different forms, or stages, of human auditory decline.A combination of several hHL genes is also possible to reproduce, and study, more complex human pathologies.Equally, Drosophila Avatars can be created in different fly lines, and genomic environments, to explore the nature/nurture dependencies of hearing loss.In this context, a specifically powerful, but yet underexploited, resource of the Drosophila research community, is provided by the Drosophila Genetic Reference Panel 2 (DGRP2) (Huang et al., 2014;Mackay et al., 2012).The DGRP2 hosts populations of more than 200 heavily inbred, and isogenic, Drosophila lines, each of which constitutes, one might say, a living library of identical twins.Using DGRP2 lines to create (e.g. via CRISPR/Cas9 or other gene-editing strategies) humanised genetic microenvironments of key auditory pathways in the fly, will we proposeenable a large-scale nature/nurture analysis of human hearing loss.Within each line, the influence of external environmental factors could be studied and between different lines the influence of internal, genetic predispositions.This may be nothing short of the world's largest twin study in the fight against deafness.It promises to reveal many of the factors that can mask, or unmask, forms of hereditary hearing loss and it would help understand what causes the numerous subclinical, or clinical, stages of our auditory decline.

Declaration of competing interest
The authors declare no competing interests.

Fig. 1 .
Fig. 1.A tale of two ears: the hearing organs of Drosophila (left) and humans (right).(A) Macroscopic level: (left) The Drosophila ear is formed by an antennal appendage of the head capsule.A feathery component (arista), which is attached to the third antennal segment (A3) captures sound-evoked particle motion and induces a rotation of A3 about its longitudinal axis.The Drosophila antennal ear is thus a sound velocity detector.(right) In humans, an 'ear drum' (tympanum) captures sound-evoked air-borne pressure waves and transmits their forces via a middle ear into the fluid-filled cochlea.The human ear is a sound pressure detector.(B) Mesoscopic level: (left) The 'inner ear' of Drosophila is housed by the second antennal segment (A2) and formed by the ~500 chordotonal neurons of Johnston's Organ (JO).JO neurons are stretch-activated and respond to the rotation of A3. (right) The inner ear of humans is housed by the cochlea and its auditory cells proper are formed by ~15,000 hair cells in the Organ of Corti.(C) Microscopic level: (left) In Drosophila, specialized, multicellular scolopidia form the cellular substrate of both transduction and amplification of auditory stimuli.Scolopidial (or chordotonal) neurons are primary, ciliated sensory neurons, which send trains of action potentials to the fly's brain.(right) In humans, sound-evoked forces are detected by mechanosensory (non-ciliary, but microvillar) bundles atop the hair cells.Hair cells do not produce action potentials themselves but form synapses with afferent neurons, which send 'spike trains' to the brain.Both hair cells and scolopidia use highly similar sensory transduction, adaptation and amplification mechanisms; they express vastly overlapping gene sets and can be traced back to a common evolutionary lineage.

Fig. 2 .
Fig. 2. The 'fly avatar' approach: atonal as a benchmark example.Shown are rescues of a Drosophila atonal null mutant using three different transgenic constructs: The Drosophila atonal (ato) coding sequence, the coding sequence of the mouse atonal homologue 1 (MmAth1) and the mouse atonal homologue 5 (MmAth5).(A) Simple harmonic oscillator fits (solid lines) to the antennal receivers of the three rescues.The antennal receivers of atonal null mutant flies (not shown) fail to form during development, their antennal sound receivers are immobile and do not conform to a simple harmonic oscillator model.All three rescues show mobile receivers, which conform to a simple harmonic oscillator model.(top) The passive receivers (reflective of the 'passive mechanics' of the antennal joints) of the three rescues show little differences.(bottom) But the active receivers (reflective of the 'active mechanics' of the coupled sensory neurons) differ in their best frequencies and vibration magnitudes.(B) The complete loss of energy gain characteristic of the sound receivers of atonal null mutants (lower dashed line), is fully rescued by both Ato and MmAth1 (upper dashed line demarcates the wildtype level), MmAth5 leads to a 50 % rescue only.(C) Johnston's Organ (JO) does not form in atonal null mutant flies, the ear is void of neurons and no mechanically-evoked compound action potential (CAP) responses can be recorded from the fly's auditory nerve (not shown).In all three rescues, CAP responses reappear.(D) The working range and sensitivity of mechanically evoked nerve responses is near fully restored in MmAth1 rescues, whereas MmAth5 rescues show a partial rescue only.(E) Gating compliances, i.e. drops in stiffness over those ranges of displacements where transducer gating occurs (absent in atonal null mutants) are restored in all three rescue constructs, yet their shape is different in each, indicating changes on the level of the coupled neurons and their transducer channels.(F) Quantification of transducer properties calculated from gating compliance analysis.The three different coding sequences affect the different transducer populations (auditory and non-auditory) differently.All data modified from(Weinberger et al., 2017).

Fig. 3 .
Fig. 3. Gauging the distance -an exploration of the global orthology space between flies, mice and humans.(A) Evolutionary time tree reconciliation of the three species.Black and grey nodes indicate estimated time of separation from last common ancestor.(B) Top: colour legend for orthology comparisons between pairs of species.Bottom: Venn diagrams depicting the number of genome wide orthologue pairs (and shared orthologue pairs) between human+fly and mouse+fly.Note that the overall number of pairs of orthologues is highly similar.(C) Violin plots showing density distribution for different (whole genome) orthologue pairs as a function of their degree of conservation using DIOPT scores.Note the signature shapes of score distributions for more distantly related species as compared to more closely related species.The more distantly related pairs (human/fly and mouse/fly) show a major distribution peak at low DIOPT scores (<4) and a second minor distribution peak at high DIOPT scores (>12).This is inverted for the closely related pair (human/mouse).(D) 'Divergence Spiral' Plots illustrating the density distributions of ortholgue pairs as a function of their divergence.For each comparison, orthologue pairs were ranked in an ascending fashion according to their respective 'divergence scores' (=maximum possible DIOPT scoreactual DIOPT score).Spiraling curves then run counter-clockwise from 0 • to 360 • to complete a full circle for the given number of orthologue pairs, with ortholgue pairs of low divergence scores (i.e.highest degree of conservation) starting at the origin.Radius lengths encode divergence scores.The longer a spiral remains near the origin (centripetally deformed spirals) the more conserved the list of orthologue pairs is.The quicker the spiral reaches the circle's circumference (centrifugally deformed spirals), the less conserved the genomes are.A linear model (assuming a linear distribution of divergence scores across all orthologue pairs) is shown to guide the view.Spiral sections on the inside of the linear model show an enrichment of conserved gene pairs.(E) Orthologue heatmaps of protein-coding genes between human and mouse (left), mouse and fly (middle), and human and fly (right) are depicted using color gradients based on their DIOPT weighted scores.The higher the DIOPT scores are, the more functionally similar the orthologues are assumed to be.The top predicted biological processes (GO terms) of corresponding genes are shown for human-fly orthologs.DIOPT weighted scores are colour coded (see legend bars).(F) Ciliary and nonciliary sensory organelles support vision and hearing in both flies and humans.A more complete understanding of a specific sensory modality in one taxon may thus benefit from comparisons across taxa and across modalities.