key: cord-0329961-o0s2gr4y authors: Prelic, Sinisa; Mahadevan, Venkatesh Pal; Venkateswaran, Vignesh; Lavista-Llanos, Sofia; Hansson, Bill S.; Wicher, Dieter title: Functional interaction between Drosophila olfactory sensory neurons and their support cells date: 2021-10-04 journal: bioRxiv DOI: 10.1101/2021.10.04.463011 sha: 4c7c19ae2bd329ad4a8fb707f5e39a7571fc15b0 doc_id: 329961 cord_uid: o0s2gr4y Insects detect volatile chemicals using antennae, which house a vast variety of olfactory sensory neurons (OSNs) that innervate hair-like structures called sensilla where odor detection takes place. In addition to OSNs, the antenna also hosts various support cell types. These include the triad of trichogen, tormogen and thecogen support cells that lie adjacent to their respective OSNs. The arrangement of OSN supporting cells occurs stereotypically for all sensilla and is widely conserved in evolution. While insect chemosensory neurons have received considerable attention, little is known about the functional significance of the cells that support them. For instance, it remains unknown whether support cells play an active role in odor detection, or only passively contribute to homeostasis, e.g. by maintaining sensillum lymph composition. To investigate the functional interaction between OSNs and support cells, we used optical and electrophysiological approaches in Drosophila. First, we characterized the distribution of various supporting cells using genetic markers. By means of an ex vivo antennal preparation and genetically-encoded Ca2+ and K+ indicators, we then studied the activation of these auxiliary cells during odor presentation in adult flies. We observed acute responses and distinct differences in Ca2+ and K+ fluxes between support cell types. Finally, we observed alterations in OSN responses upon thecogen cell ablation in mature adults. Upon inducible ablation of thecogen cells, we notice a gain in mechanical responsiveness to mechanical stimulations during single-sensillum recording, but a lack of change to neuronal resting activity. Taken together, these results demonstrate that support cells play a more active and responsive role during odor processing than previously thought. Our observations thus reveal that support cells functionally interact with OSNs and may be important for the extraordinary ability of insect olfactory systems to dynamically and sensitively discriminate between odors in the turbulent sensory landscape of insect flight. Insects detect volatile chemicals using antennae, which house a vast variety of olfactory sensory neurons (OSNs) 10 that innervate hair-like structures called sensilla where odor detection takes place. In addition to OSNs, the 11 antenna also hosts various support cell types. These include the triad of trichogen, tormogen and thecogen 12 support cells that lie adjacent to their respective OSNs. The arrangement of OSN supporting cells occurs 13 stereotypically for all sensilla and is widely conserved in evolution. While insect chemosensory neurons have 14 received considerable attention, little is known about the functional significance of the cells that support them. 15 For instance, it remains unknown whether support cells play an active role in odor detection, or only passively 16 contribute to homeostasis, e.g. by maintaining sensillum lymph composition. To investigate the functional 17 interaction between OSNs and support cells, we used optical and electrophysiological approaches in Drosophila. 18 First, we characterized the distribution of various supporting cells using genetic markers. By means of an ex vivo 19 antennal preparation and genetically-encoded Ca 2+ and K + indicators, we then studied the activation of these 20 auxiliary cells during odor presentation in adult flies. We observed acute responses and distinct differences in 21 Ca 2+ and K + fluxes between support cell types. Finally, we observed alterations in OSN responses upon thecogen 22 cell ablation in mature adults. Upon inducible ablation of thecogen cells, we notice a gain in mechanical 23 responsiveness to mechanical stimulations during single-sensillum recording, but a lack of change to neuronal 24 resting activity. Taken together, these results demonstrate that support cells play a more active and responsive 25 role during odor processing than previously thought. Our observations thus reveal that support cells functionally 26 interact with OSNs and may be important for the extraordinary ability of insect olfactory systems to dynamically 27 and sensitively discriminate between odors in the turbulent sensory landscape of insect flight. 28 Keywords 29 insect olfaction, Drosophila antenna, support cells, accessory cells, glia, cation imaging, thecogen cell, tormogen 30 cell 31 Olfaction is an ancient and critical sensory modality for all animals. Sensitivity to volatile chemicals underpins a 33 great variety of essential behaviors for survival and reproduction such as foraging for food, avoidance of biotic 34 and abiotic hazards, sexual mating, reception of inter-and intraspecific semiochemicals (Vosshall, 2000) , and is 35 a both ubiquitous and principal sense for metazoan life (Ache and Young, 2005) . The perception of airborne cues 36 begins by the detection of odors by dedicated, specialized sensory organs. Though the astounding variety of 37 smelling organs may seem diverse, the general features of olfactory systems are conserved and share several 38 invariable features which allow for specific and sensitive sampling of broad ranges of odors (Eisthen, 1997; 39 Krieger and Breer, 1999; Ache and Young, 2005; Eisthen and Polese, 2007; Ng et al., 2020) . It has long been noted 40 that even disparate olfactory tissues such as mammalian olfactory mucosa and arthropod sensilla display striking 41 similarities in olfactory transduction and structural constraint (Shirsat and Siddiqi, 1993 ; Abbas and Vinberg, 42 2021). With respect to cellular repertoire, olfactory organs are always composed of odorant receptor-equipped 43 sensory neurons innervating an epithelium, and a lesser-explored set of auxiliary cells that co-arise in 44 development, which remain closely apposed to their corresponding neurons, and are thought to play roles in 45 maintaining and potentiating the ability of neurons to perform their sensory function (Schmidt and Benton, 46 2020). 47 The various populations of support cell types and the functions of these "support networks" have been partially 48 elucidated across myriad organisms, which indicate that these cells fulfill many hitherto unknown or understated 49 tasks that many be endemic to sensory systems across disparate organisms (Charlton-Perkins et al., 2017). For 50 instance, mounting evidence points to the important role of many support cells in regulating sensory neuronal 51 activity, transmission and structural integrity. In the mouse auditory system, cochlear support cells reduce 52 neuron cell excitability by modulating extracellular space and the speed of K + redistribution through osmotic 53 shrinkage (Babola et al., 2020) . Ommatidial cone support cells in the Drosophila compound eye functionally 54 interact with photoreceptor neurons through means of altered metabolism and ion homeostasis (Charlton-55 Perkins et al., 2017). In the C. elegans peripheral chemosensory system, the amphid sheath glial cell (AMsh) is 56 able to autonomously respond to aversive chemicals and consequently suppresses its amphid ASH neuron 57 through GABA release to promote olfactory adaption (Duan et al., 2020) . In C. elegans mechanosensors, nose 58 touch receptors crucial for touch behaviors depend on ion homeostasis performed by supporting glial cells 59 harboring Na + /K + ATPases (Johnson et al., 2020) . Support cells have also garnered much attention following the 60 COVID-19 pandemic (Cooper et al., 2020) , with studies revealing the non-neuronal expression of SARS-CoV-2 61 entry genes in the sustentacular support cells of mammalian olfactory systems, which are implicated as central 62 players in the symptomatic anosmia following infection what temporal scale are support cells involved in the sensory process? Do animals show physiological or 66 behavioral differences contingent on the variability in support cell phenotype? And though sensory systems 67 show varying degrees of conservation and anatomical parallelisms, such as neuronal compartmentalization (Ng 68 et al., 2020), which functional elements beside neurons are selected or free to vary, and which remain stable in 69 evolution? 70 Particularly across the range of insect taxa, sensory systems are generally conserved, and take the form of 71 sensilla, chitinized hair-like protrusions from the cuticle on insect bodies, most often acting in chemo-, mechano-72 , hygro-and thermo-sensing with similar underlying cytological organization (Steinbrecht, 1996; Chai et al., 73 2019). Sensilla of the insect model organism Drosophila melanogaster are typically innervated by one-or-few 74 sensory neurons, and are classified based on morphological shape as well as identity of sensory neurons that 75 innervate them. These sensory neurons individually express distinct receptors from a wide range of receptor 76 families such as the odorant receptor (OR), gustatory receptor (GR), ionotropic receptor (IR), pickpocket (Ppk) 77 and transient receptor potential (TRP) protein families (Gallio et al., 2011; Joseph and Carlson, 2015) , as well as 78 non-canonical transporter-receptors recently described such as Amt (Vulpe et al., 2021) . Drosophila specifically 79 possess many appendages with chemosensory sensilla, such as maxillary palps, proboscis, wings, sexual organs 80 (e.g. ovipositor), bodily bristles, and tarsi. However, the most particularly tractable and broadly studied system 81 is that of the antenna, a bilaterally-occurring appendage on insect heads. D. melanogaster antennae are 82 segmented into three parts, named scape, pedicel and funiculus, referring to first, second and third antennal 83 segments, respectively. The funiculus is characterized by an arista, a large modified bristle that arises from the 84 proximal part of the funiculus and extends laterally. The segment also houses the largest collection of olfactory 85 sensilla in Drosophila. 86 This olfactory system in particular presents an amenable model to study cellular and molecular underpinnings 87 of olfaction. To date, concerted efforts have led to a near-complete description of the antennal architecture, 88 namely, an anatomical atlas of the distribution and identification of all sensillum types, as well as an exhaustive 89 understanding of the number, identity and odor-tuning properties of ORs and their respective olfactory sensory 90 neurons (OSNs) innervating each sensillum (Montell, 2021 a phenomenon known to occur widely beyond sensory neurons (Holsbeeks et al., 2004) . Despite this, the 131 apparent diversity of auxiliary cells, and the extent of their participation in odor reception and transduction 132 remains largely unaddressed. This is surprising, given widespread conservation of sensillum cell architecture 133 across insect genera (Steinbrecht, 1996) , the shared terminal differentiation and developmental origin with 134 sensory neurons (Ghysen and Dambly-Chaudiere, 1989; Hartenstein and Posakony, 1989, 1990; Chai et al., 135 2019), as well as the structural homology with various other non-chemosensory organs such as chordotonal 136 organ scolopidia and the arista (Keil and Steinbrecht, 1984; Foelix et al., 1989; Yack, 2004) . Moreover, auxiliary 137 cells form numerous seals and contacts around the sensory neuron by virtue of septate junctions, as well as 138 maintain closed compartments such as the sensillum lymph and perineuronal lumen between neuron and 139 thecogen cell (Steinbrecht, 1980; Keil and Steinbrecht, 1983, 1987 The repertoire of Drosophila support cells occurs in a highly stereotyped fashion. Aside from sensory neuron(s), 144 whose axons may be enveloped by one or more glial cells, each sensillum is populated by a thecogen, tormogen 145 and trichogen cell. These have been termed sheath, socket and shaft/bristle/hair cells, respectively, and are 146 collectively referred to as accessory, auxiliary or supporting cells. Though largely treated as functionally 147 equivalent due to their poor molecular characterization and an inability to experimentally manipulate them 148 separately from one another, the triad of support cell types has been dissociated partially based on physiological 149 identity and unique features evident during development. For instance, tormogen cells in particular have been 150 reported to express specific cytochrome P450 enzymes at their apical poles (Willingham and Keil, 2004 ) and 151 display features of active transport across the apical portion, which is covered in microvilli. In Calliphora blow 152 flies, this apical portion features an enrichment in mitochondria, particles and vesicles, and exhibits non-specific 153 alkaline phosphatase and Mg 2+ -activated ATPase activity (Gnatzy and Weber, 1978) . In the silk moth, two 154 tormogen cells arise during development, where the inner tormogen cell degenerates within 2 days post-155 apolysis (Keil and Steiner, 1990, 1991) . Atypically to the general rule of one support cell type per sensillum, 156 Drosophila coeloconic sensilla are populated with two tormogen cells (Shanbhag et al., 2000) . A mature 157 tormogen cell is distinct from the trichogen cell in that it extends a characteristically long, stalk-like protrusion 158 that terminates below the level of the OSN soma, at least in Drosophila (Nava Gonzales et al., 2021). Though the 159 validity of the marking is somewhat unclear, tormogen cells have been historically tagged using a promoter 160 found upstream of the Suppressor of Hairless gene (Su(H)) termed ASE5 (Barolo et al., 2000) , a DNA-binding 161 protein component of the Notch signaling pathway (Bray and Furriols, 2001) which reportedly contributes to the 162 terminal differentiation of the tormogen socket and trichogen shaft cells (Schweisguth and Posakony, 1992, 163 1994; Gho et al., 1996) . Its counterpart, the trichogen cell, is uniquely involved in the olfaction-essential 164 formation of nanopores on the sensillum surface by modifying the cuticular envelope during metamorphosis in 165 an Osiris gene-dependent fashion (Ando et al., 2019) . Moreover, in Drosophila embryo chemosensory organs, 166 the trichogen cell has been shown to expresses artichoke (atk), a gene required for correct morphogenesis of 167 ciliated sensory organs like sensilla, without which larvae exhibit impaired chemotaxis (Andrés et al., 2014) . 168 Currently, no adult cell type-specific markers are known to mark the trichogen cell. As a result, studies have been 169 limited in their ability to dissociate tormogen cells from trichogen cells without the use of the high resolution 170 approach of electron microscopy. Finally, thecogen cells are characterized by their tight, innermost association 171 with their OSN(s). Their plasma membranes are closely apposed, whereby the thecogen cell envelops the soma 172 and inner dendrites of the neuron, terminating basally below the soma and apically at the root of the sensillum 173 shaft (Keil, 1997 which are equivalent to the triad found in olfactory tissues. Though named differently depending on which organ 185 they are found in, this collection of supporting cells is both ubiquitous across sensory modalities, disparate body 186 parts and insect genera, which suggests these cells perform vital roles with regard to sensory perception 187 (Kaissling, 1986; Schmidt and Benton, 2020) . 188 In this study, we characterize all known ways to access these cells for experimental manipulation by 189 systematically exploring the distribution and reliability of specific cell markers of the heterogeneous types of 190 non-neuronal cells in the antenna. Subsequently, we use live cation imaging in an ex vivo antennal preparation 191 to characterize whether major supporting cell classes detectably respond to odor presentation events, and find 192 concomitant ion fluxes immediately following neuronal activation by the odor proxy and synthetic agonist 193 VUAA1. We observe ion-specific physiological differences between thecogen and tormogen auxiliary cell 194 responses, which indicate that these cells are functionally distinct yet both coupled to OSN activity. Last, by way 195 of in vivo single sensillum recording (SSR) between adult flies with intact and ablated thecogen cells, we find 196 differences in responses to a panel of ecologically relevant odorants, without change to OSN resting activity in 197 flies lacking thecogen cells. Curiously, we find a gain in mechanosensitivity in thecogen cell-free flies, as well as 198 hints of sensillum-specific effects. Given a long research history but relative scarcity of insight, we also speculate 199 on the potential action of support cell types with respect to the structure and function of insect olfactory tissues. 200 Altogether, our live cation imaging and electrophysiological examinations indicate that support cells acutely 201 respond to chemical cues or neuronal activity in real time, and can no longer be viewed as passive or stimulus-202 acquiescent elements of the Drosophila olfactory system. This renewed consideration may have broad 203 implications for our understanding of odor processing in insect sensory apparatuses within and beyond 204 chemosensory sensilla, as well as complex multicellular sensory compartments in other species. 205 Lines used 207 A list of transgenic fly lines used in this study can be found in Table 1 where otherwise noted. ASE5-GAL4 and nompA-GAL4 lines were kindly provided by Craig Montell (University of 215 California, Santa Barbara). The ASE5-GAL4 enhancer fragment was originally constructed from an in vivo 216 expression assay using lacZ reporter gene analysis; ASE5 refers to a 372bp 5′ subfragment of an enhancer 217 containing five Suppressor of Hairless (Su(H)) binding sites, termed "ASE5", that drove high and specific 218 expression in tormogen (socket) cells in a variety of adult bristles (Barolo et al., 2000) . The exact extent of the 219 enhancer fragment used to construct nompA-GAL4 is to our knowledge unknown but has originally been used 220 to drive expression in scolopale cells of the Johnston's organ and is derived from the promoter of no 221 mechanoreceptor potential A (nompA) gene (Chung et al., 2001; Todi et al., 2004 Todi et al., , 2005 . 222 DNA vector construction of pUASTattB-GINKO1 vector and generation of UAS-GINKO1 flies 223 The genetically-encoded fluorescent potassium indicator GINKO1 (Shen et onto the slide and incubated overnight at 4°C. Next, the slides were washed 3x for 10 minutes using PT buffer 270 on a shaker and blocked in PTS buffer for 30 minutes. Slides were subsequently incubated with secondary 271 antibodies for 2h at room temperature in the dark. Slides were washed 3x for 5 minutes using PT buffer, and 272 finally mounted using 60 µl Vectashield (Vector, Burlingame, CA, USA) under a coverslip. 273 For staining nompA>RFP & nompA::GFP or Orco>GFP, we used the primary antibodies chicken anti-GFP and 274 rabbit anti-RFP (Invitrogen, Carlsbad, CA, USA), at the relative dilutions of 1:1000 and 1:500, respectively, and 275 the secondary antibodies goat anti-chicken-A488 and goat anti-rabbit-A546 (Invitrogen) at the relative dilutions 276 of 1:250 and 1:500, respectively. For staining nompA>UAS-GAL4>GFP, we used the primary antibodies rabbit 277 anti-RFP 1:1000 (Invitrogen), and the secondary antibodies goat anti-rabbit 1:100 or 1:250 (Invitrogen). For all 278 DNA stainings, we used 1:1000 Hoechst dye staining. 279 Micrographs were captured using a cLSM 880 (Carl Zeiss, Oberkochen, Germany) using 10×, 20×, 40× or 63× 281 water immersion objectives (C-Apochromat, NA: 1.2, Carl Zeiss). Where Airyscan is noted, images were obtained 282 using the Airyscan detector and mode on the cLSM 880 (Huff, 2015) . Where linear unmixing is noted, images 283 were obtained using linear unmixing mode in ZEN software (Carl Zeiss, www.zeiss.com). Z-stack maximum 284 intensity projection images were obtained at 1 μm intervals for whole antennal overviews and 0.5 μm intervals 285 for detailed sections and close-ups. protocol was programmed to sample images every 5 seconds over 180 cycles, allowing for 15 minutes of 294 continuous specimen imaging. Each sampling event follows a 50 ms exposure to 475 nm light generated by the 295 monochromator. All chemical applications to the sample were performed by pipetting a volume of 100 µl of 100 296 µM VUAA1 and/or 100 µM CdCl2 in Ringer solution onto the immersed objective for advection and diffusion over 297 the submerged antenna. Afterward, a background region was marked along with observed cells, which were 298 marked as regions of interest (ROIs). Where ROIs were subsetted for parallel analysis, each region was 299 qualitatively judged as responding (responder) or non-responding (non-responder) to the treatment based on 300 whether a change in signal was noticeable upon visual inspection, and were labeled as such for subsequent 301 processing. TILLVision software was used to generate a matrix of average fluorescence values for background 302 region and ROIs; this matrix was exported for data analysis using R. for 3 seconds pre-and 10 seconds post-stimulus. Stimuli were delivered for 500 ms and were added to pre-336 humidified air being constantly delivered onto the fly at a rate of 0.6 LPM. Stimuli were prepared by pipetting 337 10 μl of desired compound, dissolved in hexane (10 -4 ), onto a filter paper of diameter 10 mm. No more than 5 338 sensilla were recorded from each fly and odors were used a maximum of 5 times with a gap of 30 minutes for 339 re-equilibrating pipette headspace. 340 Quantitative analysis of peak frequency responses and area-under-curve of treatment responses 341 Response frequency plots were generated in AutoSpike by selecting recordings and creating frequency time 342 courses using 25 ms bins. All time courses were saved with a timestamp of when odorant presentations were 343 applied, such that they could be aligned in post-processing. Response frequency plots were charted using 344 GraphPad Prism 9.0.1 based on n=4-9 flies for each of the 16 treatments, 3 fly genotypes, and 2 heatshock 345 conditions. For air gust corrections, we used a three-step pipeline. We initially plot the raw response trace (step 346 1). The mean response trace for air gust treatments was then subtracted from other traces, and the resulting 347 time courses and their mean traces are shown labeled as 'air gust corrected' (step 2). These average-corrected 348 traces were then smoothed with a 400 ms rolling average (step 3) using the rollmean function within the zoo R 349 package to (i) remove leftover response artefacts due to micro-timing mismatches in stimulus onsets left over 350 from subtracting the air gust responses previously, and (ii) to leave only large effects behind that would more 351 easily be judged qualitatively. A 400 ms duration for rolling average was selected beforehand based on typical 352 response durations of approximately 400 ms, which we deemed a conservative (response-removing) approach. 353 All resulting traces of the three-step processing workflow were subsequently mined for peak (maximum) 354 response and area-under-curve using a custom R script specified to only survey data points within a predefined 355 'response window' of stimulus onset to 1 second after stimulus onset, with the exception of the smoothed data 356 where the 'response window' was defined as 0.5 seconds prior to stimulus onset to 1 second after stimulus 357 onset, due to the shifting of the responses by the smoothing transformation. Evaluation of Drosophila support cell markers for the third antennal segment 378 The third antennal segment of D. melanogaster is populated by a plethora of neuronal and non-neuronal cell 379 types, including a discrete set of auxiliary cells, which form stereotypical multicellular arrangements and 380 compartments sealing the sensillum lymph and enveloping the OSNs which project their dendrites, often 381 branched, into the lymph space for chemical sensing ( Figure 1A ). More distantly from the OSN are epithelial cells 382 that flank sensillum compartments entirely, and glial cells that insulate OSN axons projecting into the olfactory 383 lobe via an olfactory nerve ( Figure 1A ). Sensilla cover the entire surface of the funiculus, including the protruding 384 arista and sacculus pits, which also host the same cellular repertoire as sensilla ( Figure 1B) . Given the scarcity of 385 characterization of non-neuronal funiculus cell markers, we began by exploring and validating all available ways 386 to label each support cell type by use of Drosophila binary expression systems (Viktorinová and Wimmer, 2007) . 387 Broadly, we looked to systematically examine expression patterns and suitability as marking devices. 388 First, we performed immunofluorescence co-imaging of both nompA protein and nompA promoter-driven 389 expression of a RFP fluorescent reporter, to check whether protein and promoter reliably labeled the same cells. 390 Due to a notable homology between chordotonal organs and olfactory sensilla, we were aware that scolopale were positive for GFP-tagged nompA ( Figure 1C ). Next, we looked at the localization of nompA protein across a 396 longitudinal cross-section of both second and third antennal segments, finding nompA protein across the 397 entirety of the funiculus, including within the characteristic sacculus pits ( Figure 1D ). Upon closer inspection with 398 super-resolution imaging, clear tube-like sheathing structures of nompA protein in all observed sensilla were 399 apparent ( Figure 1E ). To better understand whether the protein localization conformed or overlapped with 400 thecogen cellular shape, we found that the nompA-GAL4 driven cytoplasmic reporter expression did not overlap 401 with that of nompA protein ( Figure 1F ), suggesting that nompA is excreted or exported from cells with an active 402 nompA promoter, and that nompA protein is not suitable as a cellular marker. We additionally checked whether 403 nompA is found within the apical sensillum lumen, for the possibility that it would colocalize with Orco or coat 404 the outer dendrites. We found nompA exclusively at the base of the sensillum (Supplementary Figure S1A) , 405 suggesting that nompA does not interact at the odor-receptor interface but rather acts as an extracellular 406 scaffold or matrix component likely holding OSN dendrites in place. Subsequently, we co-stained OSNs and 407 thecogen cells through use of the OR-coreceptor (Orco)-GAL4 and nompA-GAL4 drivers, respectively. Here, we 408 observed hallmark features of thecogen cells, as described in morphological EM studies of Drosophila sensilla 409 (Shanbhag et al., 2000) , namely a closely and thinly OSN-enveloping cell, sheathing the inner dendrites, with a 410 nucleus at the reported distance from the OSN, of an apposed, sheathing cell in close proximity with the OSN 411 ( Figure 1G ). To confirm our expectation that nompA-GAL4 faithfully labels thecogen cells across a broad range 412 of different types and morphological classes of sensilla, we asked whether this thecogen cell marking approach 413 would label predicted locales. Specifically, we asked whether the marker would occur in ab3 sensilla, in different 414 morphological types of sensilla, as well as whether they would occur in Orco-negative sensory neurons 415 expressing IRs, such as those found in coeloconic sensilla of the sacculus. Indeed, thecogen cells marked with 416 nompA co-occurred in all three cases: in ab3 (Or22a-immunopositive) sensilla ( Figure 1H ), in large basiconics 417 negative for Or22a (i.e. ab1 and ab2) as well as trichoid sensilla characterized by their thick, rounded bases, 418 termed basal drums (Shanbhag et al., 1999) (Figure 1I ), and finally in Orco-negative coeloconic sensilla contained 419 within the sacculus ( Figure 1I and 1J) . Interestingly, we also found thecogen cells in distal portions of the 420 funiculus, without attributable Orco-positive neurons ( Figure 1K, Supplementary Figure S1B ). We also located 421 the inverse case, of neurons without attributable thecogen cells ( Figure 1L ), as well as the proximomedial region 422 of the funiculus opposite the arista where nompA-GAL4 seemed not to drive expression ( Figure 1M ) Figure S1C) . 433 We also surveyed the use of ASE5-GAL4 to mark tormogen cells by a similar immunofluorescence approach. We 434 found a more widespread and uniform reporter expression across the entire funiculus ( Figure 1N ). ASE5-GAL4 435 marked the cytoplasm of more globular cells unalike to those of thecogen cells, indicating a marking of a different 436 cell type ( Figure 1O ). To our knowledge, there have been no attempts to determine whether the GAL4 marker 437 strictly marks tormogen cells, or perhaps whether it may in fact partially mark its sister trichogen cell. Using 438 confocal imaging one cannot differentiate between the two, but we observe very even and confluent cellular 439 tiling between the cells without overlap ( Figure 1O ). Given that tormogen and trichogen cells often fold and 440 extend over each other (Shanbhag et al., 2000) , this expected lack of overlap indicates that ASE5-GAL4 likely 441 truly marks the tormogen cells of the funiculus. This is supported by the fact that such ASE5 promoters 442 conjugated with lacZ and GFP reporters have been shown to mark tormogen but not trichogen cell types (Barolo 443 et al., 2000) . 444 Lastly, we attempted to compare and contrast available glial cell markers for the third antennal segment by use 445 of the GAL4 expression system. We first drove the expression of nuclear GFP using the promoter of repo, a 446 classical Drosophila glial homeodomain transcription factor expressed exclusively in glial cells, and observed the 447 absence of any glial nuclei near sensilla ( Figure 1P GCaMP6f as an intracellular probe for Ca 2+ (Figure 2A) . Surprisingly, we found an abrupt increase in intracellular 460 Ca 2+ following exposure of the antennal preparation to the odorant proxy VUAA1, a synthetic, non-competitive 461 allosteric Orco agonist (Jones et al., 2011) , with a complete absence of response to its solvent, DMSO ( Figure 462 2B). Upon repetition, we additionally noticed in recordings that some tormogen cells were responding to VUAA1 463 pulses, while others were entirely non-responding. We reasoned that this may simply come as a result of not all 464 sensilla being VUAA1-sensitive (e.g. Orco-negative sensilla), but perhaps also due to differential expression 465 across age. Prior labeling attempts with ASE5-GAL4 have previously been suggested to be developmentally 466 regulated and decrease in transcriptional activity with age (Larter et al., 2016) . To test this possibility, we 467 replicated the experiment with a batch of freshly eclosed flies (<12h after eclosion) and found decidedly fewer 468 responders than with our typical batches of older flies of age 2-12 days (Supplementary Figure S2A) , excluding 469 the explanation that older flies experience deteriorating ASE5-GAL4 driven reporter expression with age. 470 Given that tormogen cells are also present in sensilla equipped with OSNs that do not express Orco, such as 471 those broadly found in coeloconic sensilla which instead express ionotropic receptors (IRs) sensitive to carboxylic 472 acids and amines, we repeated the experiment with balsamic vinegar, which has been shown to elicit responses 473 in IR-positive, Orco-negative OSNs (Jain et al., 2021) . Moreover, we were prompted to test balsamic vinegar due 474 to the fact that tormogen cells also occur in duplets in Drosophila coeloconic sensilla (Shanbhag et al., 2000; 475 Nava Gonzales et al., 2021). Here we found no significant intracellular Ca 2+ responses to vinegar as compared to 476 VUAA1 stimulations ( Figure 2C ). Next, we qualitatively subset these cells into responding and non-responding 477 groups for parallel analysis, based on whether Ca 2+ rises were elicited upon either VUAA1 or vinegar stimulation 478 during the experiment on a cell-by-cell basis. Thereafter we observed significant VUAA1-responding cells in the 479 responding subgroup, and inversely, significant balsamic vinegar-responding cells within the non-responding 480 subgroup ( Figure 2D ). Subsequently, using a similar approach to better understand the contribution of voltage-481 gated Ca 2+ channels to the intracellular Ca 2+ rise in tormogen cells, we used cadmium blocking of Ca 2+ channels 482 at a concentration of 100 µM CdCl2 (Wicher and Penzlin, 1997) and discovered that Ca 2+ fluxes were evidently 483 maintained in the presence of Cd 2+ -blocked Ca 2+ channels within the responding subpopulation of tormogen 484 cells ( Figure 2E ). 485 Next, we generated flies expressing the novel genetically-encoded K + indicator GINKO1 (Shen et al., 2019) ( Figure 486 2F) to determine whether OSNs, as well as tormogen cells, respond to odor presentation with intracellular K + 487 flux ( Figure 2G ). To validate the use of the indicator for the first time in Drosophila, we first tested the GINKO1 488 indicator driven by Orco-GAL4 expression in OSNs, as a test case where we would expect to see K + efflux in OSNs 489 during VUAA1 stimulation. Here we found a steady and substantial K + efflux in OSNs ( Figure 2H , Supplementary 490 Figure S2B ). Inversely, we found a complete lack of change or response to baseline K + levels within tormogen 491 cells following identical stimulations with VUAA1 ( Figure 2I , Supplementary Figure S2C ). Taken together, 492 tormogen cells show strong Ca 2+ influx upon odorant presentation, which does not seem to be dependent on 493 Ca 2+ channel ion flow, as well as no flux with respect to K + . 494 It is unknown whether thecogen cells, like tormogen cells, respond acutely during odor presentation. Using the 496 same ex vivo antennal preparation and set up as in the previous experiment, we expressed the Ca 2+ and K + 497 indicators CaMP6f and GINKO1 serially in the thecogen cells of the antenna using nompA-GAL4 ( Figure 3A ). 498 Strikingly, and unlike the results in tormogen cells, we observe a complete lack of Ca 2+ response to VUAA1 499 stimulation ( Figure 3B ). This is not the case however with thecogen intracellular K + dynamics, which show a 500 marginal increase following VUAA1 stimulation ( Figure 3C ). Because thecogen cells are relatively smaller in 501 volume and size compared to other cells within the antenna (Shanbhag et al., 2000) , and due to the small rise in 502 intracellular K + in thecogen cells following VUAA1 stimulation, we decided to validate the observed K + rise and 503 rule out the possibility of it being an experimental artefact. By lowering the concentration of K + in the 504 physiological medium five-fold, from 5 mM to 1 mM, we hypothesized that a K + influx into thecogen cells should 505 be enhanced following VUAA1 stimulation, as a result of increased K + efflux into the extracellular lymph from 506 neurons repolarizing at a lowered ambient K + concentration (Contreras et al., 2021) . Indeed, we observed a 507 marked increase in peak response to VUAA1 stimulation in thecogen cells ( Figure 3D ). The results are suggestive 508 of an excess extracellular K + clearance mechanism, which has often been attributed to glial cells of the tripartite 509 synapse as a homeostatic means to regulate the excitability of neurons, such as to prevent neuronal 510 hyperexcitability (Walz, 2000; Sibille et al., 2015) . 511 To determine whether this explanation could hold, we looked at the effect of varying extracellular K + 512 concentrations on the intracellular K + concentrations measured using the GINKO1 indicator expressed via the 513 GAL4/UAS system ( Figure 3E ). First, we surveyed the neuronal dynamic with respect to K + efflux at an 514 extracellular [K + ] of 1 mM, with the expectation of a larger efflux than with the conventional medium of 5 mM 515 that approximates a physiological K + concentration of the sensillum (Reinert et al., 2011) (see Figure 2H and 516 Supplementary Figure S2B ). As expected, we saw a longer and more prominent efflux of K + at the relatively low 517 extracellular [K + ] of 1 mM ( Figure 3F ). We performed dose-response experiments for both OSN and thecogen 518 cell with varying, physiologically relevant concentrations of K + , across 2 orders of magnitude between antennal-519 and sensillum-relevant concentrations of 1-150 mM (Reinert et al., 2011) , and found a general, opposite flux 520 pattern between OSNs and thecogen cells in peak K + response, decreasing at higher ambient K + concentrations 521 ( Figure 3G) . A generally mirrored trend is observed upon plotting a dose-response curve at the average peak K + 522 influx time point of 85 seconds post-VUAA1 stimulation for thecogen cells ( Figure 3H ) and Orco-positive OSNs 523 ( Figure 3I ). 524 In conclusion, we find that thecogen cells demonstrate no Ca 2+ flux in response to antennae stimulated by 525 VUAA1, in spite of concomitant OSN activation by VUAA1, but seem to respond to local extracellular rises in K + 526 by absorbing ambient K + temporarily. These features are unlike the tormogen cell, which shows the exact 527 opposite trend. The results with respect to cation movement within OSNs as well as thecogen cells are 528 summarized in Table 2 . 529 Table 2 . Summary of intracellular cation dynamics following odor presentation in neurons and support cells. Tormogen cells respond to VUAA1 application by influx of cytoplasmic Ca 2+ while thecogen cells exhibit a 531 cytoplasmic K + influx. Asterisk (*) denotes a possible K + buffering mechanism coupling OSNs and their thecogen 532 cell. Ca would thus be expressed upon heatshocking of flies at the GAL80 ts -restrictive temperature of 32°C ( Figure 4A ). 547 All flies were therefore reared at the GAL80 ts -permissive temperature of 18°C from egg to eclosion to prevent 548 rpr expression in thecogen cells. After eclosion, flies were separated into a control cohort kept at 18°C, and an 549 ablation cohort that underwent 24h of heat-induced thecogen cell apoptosis at the GAL80 ts -restrictive 550 temperature of 32°C, wherein GAL80 ts loses its GAL4-repressive function and inhibition of rpr expression is lifted. 551 The rearing and heatshocking schema for both experimental flies and parental controls is summarized in Figure 552 4B. 553 First, we checked whether heat induction would remove thecogen cells fully without trace. Here, we 554 concurrently used a UAS-GFP construct to act as a label to follow the presence or absence of thecogen cells 555 following induced apoptosis. We screened antenna to look for remaining GFP signal in the event of thecogen 556 cells being leftover. Antennae were inspected closely, and no GFP signal across the entire depth of antenna was 557 found following both 24h and 48h heatshocking periods during confocal imaging using long exposure and high 558 excitation laser intensity, indicating a complete loss of thecogen cells ( Figure 4C ). 559 Next, to assess the neurophysiological properties of fly sensilla and neurons with and without intact thecogen 560 cells, we used single sensillum recording (SSR) in female flies to screen the responses of 3 sensilla subtypes (ab1, 561 ab2 and ab3) to a panel of 16 treatments (n = 4-9 female flies). As controls, we also tested both parental lines 562 used to generate the experimental fly line where thecogen cell-specific apoptosis could be induced. We excluded 563 the smallest neurons ab1D, ab2B and ab3B from most analysis due to poor signal-to-noise ratio rendering them 564 difficult to uniquely identify among neighboring neurons and recording noise. First, we obtained a response 565 profile to the panel of 16 treatments, including no treatment, a blank gust of air, and 14 ecologically-relevant 566 odorants, of which some are diagnostic odors, i.e. best known ligands for the particular sensilla that were 567 recorded from. For the gust of air and odorant presentation treatments, we used a 0.5 second stimulus duration. 568 The concentrations of the odorants used are listed in the Materials & Methods section. A response profile for all 569 treatments was calculated such that a count of all neuron spikes within 1 second prior to stimulus onset was 570 subtracted from the count of all neuron spikes within 1 second following stimulus onset. In all fly groups, we 571 observed generally conserved best ligand responses and unaltered odor tuning profiles with respect to tested 572 sensillum or neuron odor responses, comparable to those in control flies ( Figure 4D , Supplementary Figure S3A-573 B). However, we also observed broad non-specific responses in the heat-induced condition of experimental flies 574 where thecogen cells were ablated. We reasoned that this could be explained by changes to the resting activity 575 of the neurons, or that OSNs that had lost their thecogen cell support were gaining non-specific odor response 576 profiles. 577 To address both possibilities, we noticed that blank gusts of air and hexane stimulations were eliciting responses 578 in a manner restricted to only the experimental cohort of heat-shocked flies, most evident in ab1ABC and ab3A 579 neurons (Supplementary Figure S3C) . On this front, we mined the SSR data to estimate the resting activity of 580 neurons prior to any treatment. Here, we took an average spike count per second on active SSR recording during 581 the 'no treatment' recording ( Figure 4E ). We noticed a weakly significant decrease in resting activity in ab1ABC 582 neurons between non-heatshocked and heatshocked flies, though evidently not different enough to those of 583 baseline levels shared between parental and experimental cohorts. Moreover, there was no change in resting 584 activity in ab2A and ab3A neurons, indicating that thecogen cell ablation has no effect on resting spontaneous 585 activity in these neurons ( Figure 4E ). However, given that all three sensilla host multiple neurons, we additionally 586 surveyed the resting activity of the small B neurons in the ab2 and ab3 sensilla which have been reported to be 587 ephaptically coupled (Zhang et al., 2019) . Here, we noticed a significant decrease with heatshocking, a 588 phenomenon restricted only to the B neurons' resting activities, though not in a meaningful manner relative to 589 control flies ( Figure 4F ). 590 Figure S4) . Here, we wanted to verify whether any responses would remain that 600 could be attributed to odor sensing, rather than the gain in mechanosensitivity resulting from thecogen cell 601 ablation. Following a three-step analysis of all odor response traces in all cohorts, we found some remaining 602 olfactory component in the ab1 sensillum, and no remaining olfactory component in ab2 and ab3 sensilla upon 603 correcting traces for their mechanoresponse component (Supplementary Figure S5) . 604 Aside from this, we also noticed several trends when comparing between heatshocked and unheatshocked flies, 605 which were plotted using a bubble chart showing differences between heatshocked and unheatshocked cohorts. 606 This was done for peak response frequency (Supplementary Figure S4C ) as well as total response area-under-607 curve (Supplementary Figure S4D) . We found no major effects restricted to odors specific to those sensilla 608 between control flies with intact and ablated thecogen cells (dashed boxes, Supplementary Figure S4C-D) . We 609 additionally found no differences in neuron resting activity prior to stimulus onset, in line with our previous 610 results ( Figure 2E-F) . However, we noted that the ab1 sensillum generally exhibits higher responses in thecogen 611 cell ablated flies, while ab2 sensilla show no changes between heat treatment cohorts. Somewhat tentatively, 612 for both peak frequency and area-under-curve comparisons, the ab3 sensillum exhibits lower responses in 613 thecogen cell ablated flies in terms of area-under-curve as well as peak response frequencies (Supplementary 614 Figure S4C -D), which may indicate different sensillum-specific tolerances of thecogen cell ablation. Nonetheless, 615 from this data analysis we can only find tentative evidence for remaining olfactory sensitivity to odor 616 presentations, whereby most of the contribution to treatment responses was found to come from gained 617 mechanoresponses. We leave the data open to interpretation and as a reference for future studies. 618 Recap of study 620 In this study we have taken three broad approaches to begin to understand the role of Drosophila antennal 621 support cells in odor perception. Initially, we set out to address the scarcity of systematic descriptions available 622 prior to further investigation. First, we evaluated a variety of support cell type-specific genetic immunolabeling 623 techniques to identify the validity, suitability and limitations of each as tools to target support cell types 624 specifically, and characterized the cellular distribution of each across the funiculus. Of particular interest was 625 the thecogen cell-specifying genetic driver nompA-GAL4, which we showed labeling thecogen cells across responses of tormogen and thecogen cells to OSN stimulations with the odor proxy VUAA1. We found that a 634 subset of tormogen cells undergo an acute and steep cytoplasmic Ca 2+ influx immediately following stimulation, 635 without concomitant K + influx, indicating the quick activation of tormogen cells during odor presentation events. 636 The opposite trend was observed in thecogen cells, over a range of concentrations of ambient K + , pointing to 637 the potential role of thecogen cells as ionic sinks involved in K + buffering of the sensillum or perineuronal lymph. 638 Third, by way of removing thecogen cells in adult flies using inducible apoptosis, we assayed three distinct 639 basiconic sensillum subtypes electrophysiologically using SSR for response profile changes following thecogen 640 cell ablation. We firstly observed a broad loss of specificity to odorants, despite the generally conserved best-641 ligand property in thecogen cell-free sensilla in all three tested sensilla. We also noted the lack of any change in 642 OSN resting activity following thecogen cell apoptosis, as well as in heatshocked controls. The broadly observed 643 response to both mechanical air gust treatments and odorant pulses was attributable to a gain in 644 mechanosensitivity, likely as a result of a loss of cellular or dendritic integrity within the sensillum architecture 645 that depended on the thecogen sheath. This result is distinct from a previous report of odor insensitivity in 646 nompA-null mutants, where sensillum biogenesis did not proceed correctly and where sensory dendrites were 647 unable to innervate the sensillum (Chung et al., 2001) . In our study, we induced the removal of thecogen cells 648 at a post-development stage, and still found that tested neurons retained some degree of best ligand specificity, 649 while observing a functional gain of mechanical responsiveness to gusts of air. These effects of thecogen cell 650 ablation are in our view not attributable to the loss of support cell OBPs, given that our study has used 651 comparable pulse durations and odor concentrations as previously in studies showing robust olfactory responses 652 in the absence of basiconic OBPs (Xiao et al., 2019) , and also that thecogen cells themselves have been shown 653 only to express Obp28a in ab1-3 sensilla in a patchy manner (Larter et al., 2016) . Hence, we posit that the effects 654 of thecogen cell ablation are not attributable to the absence of any nascent OBPs which would otherwise exist 655 in sensilla with intact support cells. Lastly, we attempted to remove the mechanoresponsive component of all 656 responses, and found evidence of impaired odor sensing in OSNs, which becomes mostly but not entirely absent. 657 We also tentatively suggest that thecogen cell loss may be tolerated differently by OSNs with respect to odor 658 detection in a sensillum-specific fashion. 659 The coupling of support cells to OSNs remains enigmatic. Here we have demonstrated two separate kinds of 661 response, of two distinct support cell types, to odor presentation events. This indicates a coupling in real time 662 between the OSN and the supporting cells, and some yet unknown coupling mechanisms must exist to explain When activated, the support cells experience a cytoplasmic Ca 2+ influx from intracellular stores, followed by K + 698 efflux from the support cells which in turn elicit local neuronal burst firing (Babola et al., 2020) . Interestingly, the 699 release of K + in this model leads to a rapid increase in extracellular space due to support cell crenation, which 700 may have consequences in subsequent K + redistribution and the termination of neuronal firing (Babola et al., 701 2020). Though this is a case study from an immature and developing sensory system, it may be an indication of 702 potential mechanisms for how support cells may influence sensory neurons by purely ionic and physical 703 responses within a contained compartment such as that of the sensillum lymph, by means of swelling and 704 shrinking and modulating the ionic milieu of the environment. For one, the morphological shape of Drosophila 705 supporting cells, with respect to the aqueous lymph, seems to freely allow such mechanisms. Indeed, swelling 706 has been suggested in older studies where moth thecogen cells have been observed to manifest signs of swelling 707 with increasing incubation time with lanthanum ion solutions, as well as unusually close apposition of thecogen 708 and neuronal membranes following freeze substitution (Steinbrecht, 1980; Keil and Steinbrecht, 1987) , which 709 may provisionally indicate some intrinsic capacity of support cells to deform lymph spaces. 710 More yet, in the mouse olfactory epithelium, it has also been demonstrated that olfactory ensheathing cells 711 (OECs) are sensitive to neighboring neuronal activity, as observed in cytosolic increases of Ca 2+ , and that such 712 coupling is dependent on ATP and glutamate release from the neuron (Rieger et al., 2007) . Release and uptake of K + is entangled with several dynamic processes, namely the generation of action potentials 721 and dendritic potentials, as well as the further concentration and release of K + from the same or neighboring 722 cells (Ransom et al., 1986; Syková, 1991; Ransom, 1992; Ransom and Ye, 1995; Walz, 2000) . Though difficult to 723 compare in our data due to differing shapes and genetically-encoded cation indicators used, tormogen cells 724 seem to exhibit a much quicker Ca 2+ response than the K + sequestering occurring in thecogen cells, perhaps due 725 to aforementioned changes in shape or due to quicker mobilization of intracellular Ca 2+ stores, which may hint 726 at the active role of tormogen cells in maintaining properties of the sensillum lymph space through mechanisms 727 such as export of accessory proteins. And though often overlooked, the possible existence of an isolated 728 perineuronal cleft between the thecogen cell and OSNs in Drosophila must also be taken into regard, which may 729 differ in ionic composition from the sensillum lymph, a notable feature described in moths (Steinbrecht, 1980; 730 Keil and Steinbrecht, 1987) . At least in some moth species, the thecogen cell and neuron have been purported 731 to contribute jointly to the electrical properties of the sensillum (De Kramer et al., 1984; De Kramer, 1985; 732 Kaissling, 1986) . 733 Moreover, it is worthy of note that the glia-like thecogen cells are reminiscent of glial cells of the tripartite 734 synapse model, where glia are thought to modulate synaptic communication through K + ion buffering, dispersion 735 and redistribution (Beckner, 2020) . This is not a novel comparison. To some degree, chemosensory organs like 736 sensilla have unmistakable similarities with neuronal synapses, and have thus been compared and used to model 737 synaptic clefts, which also feature arrangements of receiving neurons and non-neuronal players which evolve 738 together to respond to small, transient chemical signals (Shaham, 2010) . In the same vein, perhaps we can also 739 apply concepts from the role of the glial cell in the tripartite synapse to chemosensory systems in efforts to 740 generate novel hypotheses. For instance, does the thecogen cell share features with the astroglial cradle 741 (Nedergaard and Verkhratsky, 2012) , shielding the neuron from the known multitude of external factors such as 742 an ever-changing sensillum lymph chockful with accessory proteins and debris? Are ensheathing thecogen cells 743 perhaps playing an essential role in sensilla that are innervated with multiple neurons experiencing interneuron 744 ephaptic inhibition (Zhang et al., 2019) ? Furthermore, recent evidence in C. elegans has shown that peripheral 745 glial support cells engulf, phagocytose and prune thermosensory neuronal endings depending on their activity 746 load (Raiders et al., 2021) . Upon disruption of this pruning, a behavioral temperature preference is lost (Raiders 747 et al., 2021) . Do similar processes exist in insect sensillum repertoires? This is stated in light of the fact that 748 compartmentalization of sensory neurons has independently evolved many times, which act as means to 749 integrate sensory inputs at the earliest stages of odor processing (Ng et al., 2020) . However, neither presence 750 nor potential contributions of support cells are stressed, even though they may be foundational. 751 The K + concentration ranges of insect antennae are also particularly notable. A particle-induced X-ray emission 752 study of the Drosophila antenna indicates that the [K + ] at the center of the funiculus is 50 mM, while at the 753 sensillum edges it is 5 mM, a 10-fold lower concentration (Reinert et al., 2011) . At face value, these results can 754 be interpreted as vastly differing ion concentrations between the hemolymph and sensillum lymph. The 755 difference may be explained by separately maintained, isolated compartments, physically sheltered from the 756 OSN by the triad of support cells, in a manner similar to the perilymph and endolymph of mammalian cochlea 757 (Zdebik et al., 2009 ). Analogously, sensillum lymph space as well as the perineuronal lumen between thecogen 758 and OSN cells would be maintained separately from the hemolymph, and would thus exhibit drastically different 759 ambient [K + ] by virtue of enclosure or by concentrating K + from the surroundings. Counterintuitively, sensillum 760 lymph in insects has been reported to be high in K + (Thurm and Kuppers, 1980; Steinbrecht, 1989) alike to the 761 vertebrate endolymph (Corey and Hudspeth, 1979) . We speculate that similar principles may apply to the 762 olfactory sensilla of Drosophila, namely that the perineural cleft between thecogen and OSN are functionally 763 distinct from the sensillum lymph, an idea long suggested (Keil and Steinbrecht, 1987) . Tangential support for 764 such a hypothesis has been observed in the nearby Drosophila auditory system, within the Johnston's organ of 765 the second antennal segment. Scolopale cells, a cell type homologous to that of thecogen cells, have been found 766 to express the Na + /K + ATPase pump preferentially localized on the side facing the perineuronal lumen, and have 767 been suggested to pump K + ions to maintain a K + -rich ionic presence for the sensory neuron (Roy et al., 2013) . It 768 is reasoned that sensory transduction events deplete K + and thus require active replenishment. The knockdown 769 of an ion pump subunit via RNA interference hence resulted in deafness, loss of scolopale cell integrity, and 770 additional morphological defects such as the presence of swollen cilia implying an ionic imbalance in the 771 scolopale space (Roy et al., 2013) . In our experiments we similarly observe the loss of cellular integrity following 772 thecogen cell ablation, with similar sensory detriments to odor detection, namely a loss of specificity to 773 odorants, and speculate that similar explanations may at least partially account for the observed olfactory 774 dysfunction. 775 A tremendous breadth of questions remains to be answered. Functionally, do Drosophila support cell activities 777 influence classic neuronal properties such as adaptation and sensitization (Wicher and Miazzi, 2021) , for 778 example in their ability to discriminate transient, repeated, sustained or excessive cues, perhaps by switching 779 rates of odor clearance from the sensillum lymph through adjustable release of enzymes or pinocytosis? Do 780 support cells vary heterogeneously and phenotypically between sensilla within or between organisms, beyond 781 that of known differential expression of OBPs? Curiously, to our knowledge a single study exists where support 782 cell heterogeneity has been found: thecogen cells in moths are relatively enlarged particularly in hygro-and 783 thermo-sensitive sensilla . Though we show responding and non-responding 784 subpopulations of tormogen cells in this study, further diversifying the class of auxiliary cells, we hypothesize 785 that many other support cell heterogeneities exist. 786 By the same token, do support cells found within larval olfactory systems (Hartenstein, 1988) Practically, we may get a glimpse into the set of underlying molecular players such as junction proteins, 797 purinergic receptors and ion channels, and to which tissues, sensilla, modalities, sexes, internal states and life 798 stages of the fly these can be attributed. Future research must take into account percipient evidence of the 799 varieties of response modes between support cell types. Namely, we theorize that Ca 2+ flux as observed in 800 tormogen cells may relate to yet unknown (intra)cellular signaling processes, while K + flux specific to thecogen 801 cells may relate to homeostatic feedback mechanisms that may be subject to modulation (Walz, 2000) . 802 Last, we may one day answer the question of whether support cells experience natural variation, and whether 803 they can be a locus for selection in evolution in light of olfactory performance. Transduction and Adaptation Mechanisms in the Cilium or Microvilli of Photoreceptors and Olfactory 919 Olfaction: Diverse species, conserved principles Nanopore Formation in the Cuticle of an Insect Olfactory 923 The extracellular matrix protein artichoke is required for 925 integrity of ciliated mechanosensory and chemosensory organs in Drosophila embryos Purinergic signaling in cochlear supporting cells reduces hair cell 928 excitability by increasing the extracellular space Evaluation of a high-throughput 930 deorphanization strategy to identify cytochrome p450s important for odor degradation in Drosophila Transcriptional profiling of olfactory system development 933 identifies distal antenna as a regulator of subset of neuronal fates A notch-independent activity of suppressor of 935 hairless is required for normal mechanoreceptor physiology A roadmap for potassium buffering/dispersion via the glial network of the CNS Supporting cells eliminate dying sensory hair cells to maintain epithelial 939 integrity in the avian inner ear Non-neuronal expression of SARS-CoV-2 941 entry genes in the olfactory system suggests mechanisms underlying COVID-19-associated anosmia Notch pathway: Making sense of suppressor of hairless Supporting cells remove and replace sensory 946 receptor hair cells in a balance organ of adult mice The role of SNMPs in insect olfaction Sensory neuron lineage mapping and manipulation in the Drosophila olfactory 949 system Multifunctional glial support by Semper cells in the 951 Drosophila retina nompA encodes a PNS-specific, ZP domain protein required to connect 953 mechanosensory dendrites to sensory structures Activity-mediated accumulation of potassium induces a switch 955 in firing pattern and neuronal excitability type COVID-19 and the Chemical Senses: 957 Supporting Players Take Center Stage Ionic basis of the receptor potential in a vertebrate hair cell Microglia monitor and protect neuronal function through 961 specialized somatic purinergic junctions The Electrical Circuitry of an Olfactory Sensillum in Antheraea polyphemus Passive electrical properties of insect olfactory sensilla may produce the biphasic shape 965 of spikes The taste response to ammonia in Drosophila Sensory Glia Detect Repulsive Odorants and Drive Olfactory Adaptation Evolution of vertebrate olfactory systems 2.17 Evolution of Vertebrate Olfactory Subsystems Olfactory neuron turnover in adult Drosophila Fine structure of a sensory organ in the arista of Drosophila melanogaster and 976 some other dipterans The coding of temperature in the Drosophila brain Subcellular localization of Suppressor of Hairless Drosophilasense organ cells during Notch signalling Genesis of the Drosophila peripheral nervous system Tormogen cell and receptor-lymph space in insect olfactory sensilla -Fine structure and 984 histochemical properties in Calliphora Hearing in Insects Fundamental principles of the olfactory code Contribution of odorant binding proteins to olfactory detection of (Z)-990 11-hexadecenal in Helicoverpa armigera The molecular basis of odor coding in the Drosophila antenna The drosophila melanogaster Na+/Ca2+ exchanger CALX controls the Ca2+ level in 994 olfactory sensory neurons at rest and after odorant receptor activation Two novel DEG/ENaC channel subunits expressed in glia are 997 needed for nose-touch sensitivity in Caenorhabditis elegans Development of Drosophila larval sensory organs: Spatiotemporal pattern of sensory neurones, peripheral axonal 999 pathways and sensilla differentiation Development of adult sensilla on the wing and notum of Drosophila melanogaster Sensillum development in the absence of cell division: The sensillum phenotype of the 1003 Drosophila mutant string Atonal regulates neurite arborization but does not 1005 act as a proneural gene in the Drosophila brain The eukaryotic plasma membrane as a nutrient-1007 sensing device The Airyscan detector from ZEISS: confocal imaging with improved signal-to-noise ratio and super-resolution Calmodulin regulates the olfactory performance 1011 in Drosophila melanogaster An odorant-binding protein required for suppression of 1013 sweet taste by bitter chemicals The Na+-K+-ATPase is needed in glia of touch receptors 1015 for responses to touch in C. elegans Functional agonism of insect odorant receptor ion channels Drosophila chemoreceptors: A molecular interface between the chemical world and the brain Chemo-electrical transduction in insect olfactory receptors The neural basis of Drosophila gravity-sensing 1023 and hearing Comparative morphogenesis of sensilla: A review Interrelations of sensory, enveloping and glial cells in epidermal mechano-and chemoreceptors 1027 of insects Mechanosensitive and Olfactory Sensilla of Insects Diffusion barriers in silkmoth sensory epithelia: application of lanthanum tracer to olfactory 1031 sensilla of Antheraea polyphemus and Bombyx mori Morphogenesis of the antenna of the male silkmoth, Antheraea polyphemus. II. Differential mitoses of 1033 "dark" precursor cells create the Anlagen of sensilla Morphogenesis of the antenna of the male silkmoth. Antheraea polyphemus, III. Development of 1035 olfactory sensilla and the properties of hair-forming cells Olfactory reception in invertebrates Organization and function of Drosophila odorant binding proteins Requirement for Drosophila SNMP1 for Rapid Activation and Termination of Pheromone-1040 Induced Activity Single-cell transcriptomes of developing and adult 1042 olfactory receptor neurons in Drosophila An RNA-Seq Screen of the Drosophila Antenna Identifies a Transporter 1044 Necessary for Ammonia Detection Odor-induced cAMP production in Drosophila melanogaster olfactory sensory neurons Drosophila sensory receptors-a set of molecular Swiss Army Knives Calmodulin modulates insect odorant receptor function Calmodulin affects sensitization of drosophila melanogaster 1051 odorant receptors Systematic morphological and 1053 morphometric analysis of identified olfactory receptor neurons in Drosophila melanogaster Artifact versus reality-How astrocytes contribute to synaptic events Neuronal Compartmentalization: A Means to Integrate Sensory Input at the Earliest Stage of 1057 Information Processing Gamma-aminobutyric acid (GABA)-mediated neural connections in the Drosophila antennal 1059 lobe Patterns of transcriptional parallelism and variation in the 1061 developing olfactory system of Drosophila species Beyond chemoreception: diverse tasks of soluble olfactory proteins in 1063 insects Shaking B mediates synaptic coupling between auditory sensory 1065 neurons and the giant fiber of Drosophila melanogaster The Q system: A repressible binary system for transgene expression, 1067 lineage tracing, and mosaic analysis Evolution of Acid-Sensing Olfactory Circuits 1069 in Drosophilids Functional integration of 1071 "undead" neurons in the olfactory system Glia actively sculpt sensory neurons by controlled 1073 phagocytosis to tune animal behavior Glial modulation of neural excitability mediated by extracellular pH: A hypothesis. Prog Brain extracellular space: developmental studies in rat optic nerve Gap junctions and hemichannels μPIXE for a μbrain: The vinegar fly's brain, antenna, sensilla hairs and eye ion 1080 concentrations The Q-system: A Versatile Expression System for Drosophila Axon-glia communication evokes calcium signaling in olfactory ensheathing cells of the 1085 developing olfactory bulb The 40-year mystery of insect odorant-binding proteins A Presynaptic Gain Control Mechanism Cell-type-specific roles of Na+/K+ ATPase subunits in Drosophila auditory 1091 mechanosensation Molecular mechanisms of olfactory detection in insects: Beyond receptors: Insect olfactory 1093 detection mechanisms Antagonistic activities of Suppressor of Hairless and Hairless control alternative cell fates in the 1095 Drosophila adult epidermis Suppressor of Hairless, the Drosophila homolog of the mouse recombination signal-binding 1097 protein gene, controls sensory organ cell fates Structure and function of the thecogen cell in contact chemosensitive sensilla of Periplaneta americana L Distinct types of glial cells populate the Drosophila antenna Chemosensory organs as models of neuronal synapses Atlas of olfactory organs of Drosophila melanogaster 1. Types, external organization, 1104 innervation and distribution of olfactory sensilla Atlas of olfactory organs of Drosophila melanogaster 2. Internal organization 1107 and cellular architecture of olfactory sensilla Genetically encoded fluorescent indicators for imaging 1109 intracellular potassium ion concentration Olfaction in invertebrates The Neuroglial Potassium Cycle during Neurotransmission: Role of Kir4 Cryofixation without cryoprotectants. Freeze Substitution and Freeze Etching of an Insect Olfactory Receptor Ions and mucoid substances in sensory organs--microanalytical data from insect sensilla Structure and function of insect olfactory sensilla Volume and surface of receptor and auxiliary cells in hygro-1120 /thermoreceptive sensilla of moths (Bombyx mori, Antheraea pernyi, and A. polyphemus) Caenorhabditis elegans glia modulate neuronal activity and behavior Ionic and volume changes in neuronal microenvironment Epitheilal physiology of insect sensilla Myosin VIIA defects, which underlie the usher 1B syndrome in humans, lead 1128 to deafness in Drosophila Anatomical and molecular design of the Drosophila antenna as a flagellar auditory organ Comparative analysis of binary expression systems for directed gene expression in transgenic 1132 insects Olfaction in Drosophila An ammonium transporter is a non-canonical 1135 olfactory receptor for ammonia Role of astrocytes in the clearance of excess extracellular potassium Effect of OBPs on the response of olfactory receptors A glial DEG/ENaC channel functions with neuronal 1141 channel DEG-1 to mediate specific sensory functions in C. elegans Knockout of glial channel ACD-1 exacerbates sensory deficits in a C. elegans mutant by 1143 regulating calcium levels of sensory neurons Cell Killing by the Drosophila Gene reaper. Science (80-. ) Functional properties of insect olfactory receptors : ionotropic receptors and odorant receptors Ca2+ currents in central insect neurons: Electrophysiological and pharmacological properties A tissue specific cytochrome P450 required for the structure and function of Drosophila sensory 1151 organs Fibroblast growth factor signaling instructs ensheathing glia wrapping of 1153 Proc. Natl. Acad. Sci Robust olfactory responses in the absence of odorant binding proteins The structure and function of auditory chordotonal organs in insects Identification of candidate odorant 1159 degrading gene/enzyme systems in the antennal transcriptome of Drosophila melanogaster Potassium Ion Movement in the Inner Ear: Insights from Genetic Disease and 1162 Asymmetric ephaptic inhibition between 1164 compartmentalized olfactory receptor neurons Chemosensory Proteins: A Versatile Binding Family Active cochlear amplification is dependent on supporting cell gap 1169 junctions All raw and processed datasets and materials used in this study can be found in EDMOND, the data repository 827 of the Max Planck Society, at the following location: https://dx.doi.org/10.17617/3.7m 828 The authors declare that the research was conducted in the absence of any commercial or financial relationships 830 that could be construed as a potential conflict of interest. 831 The figures appear below. Supplementary figures are available in a separate file.