Mutations in hiw, dsarm, or axed attenuate axon death signaling resulting in morphological preservation of severed axons for approximately 50 days, the average lifespan of Drosophila (Neukomm et al., 2017; Neukomm et al., 2014; Osterloh et al., 2012). In contrast, neuronal expression of prokaryotic NMN-Deamidase (NMNd) to consume NMN results in less than 10% of severed axons being preserved at 7 days post axotomy (dpa) (Hsu et al., 2021). We performed an established wing injury assay to confirm this observation (Paglione et al., 2020). Briefly, a subset of GFP-labeled sensory neurons (e.g. dpr1–Gal4 MARCM clones) expressing NMNd (Hsu et al., 2021) or GFP were subjected to axotomy in 5- to 7-day-old flies with one wing being partially injured and the other serving as an uninjured control (Figure 1—figure supplement 1A). At 7 dpa, we quantified uninjured control axons, axonal debris and severed intact axons, respectively (Figure 1—figure supplement 1B, left), and calculated the percentage of protected severed axons (Figure 1—figure supplement 1B, right). We observed a 40% preservation of severed axons with NMNd (Figure 1—figure supplement 1, genotypes in Source data 1). The modest increase of preservation in our hands is probably due to higher NMNd levels caused by higher Gal4 levels in dpr1 than elaV (Hsu et al., 2021). However, the expression of NMNd fails to attenuate axon death signaling to the extent of axon death mutants, suggesting that NMN does not play an essential role in activating axon death supported by the absence of the NMN-synthesizing enzyme Nampt in flies (Gossmann et al., 2012). Alternatviely, NMNd expression and the resulting NMN consumption is simply not sufficient for potent attenuation of axon death signaling and preservation of severed axons.

Based on the above observations, we decided to generate an N-terminal GFP-tagged NMN-Deamidase (GFP::NMN-D) to increase protein stability (Rücker et al., 2001). Plasmids with GFP-tagged wild-type and enzymatically dead versions of NMN-Deamidase were generated (Di Stefano et al., 2017; Di Stefano et al., 2015), under the control of the UAS regulatory sequence (UAS–GFP::NMN-D, and UAS–GFP::NMN-D^dead, respectively). We found that wild-type and enzymatically dead NMN-D enzymes are equally expressed in S2 cells, as detected by our newly generated anti-PncC/NMNd/NMN-D antibodies (Materials and methods, Figure 1—figure supplement 2). Notably, we observed two immunoreactivities per lane, with the lower band being a potential degradation product.

The similar expression of the NMN-D enzymes prompted us to use the plasmids to generate transgenic flies by targeted insertion (attP40 landing site). To compare in vivo expression levels, NMN-D, NMN-D^dead, and NMNd were expressed with pan-cellular actin–Gal4. We found that NMN-D and NMN-D^dead immunoreactivities were significantly stronger than NMNd (Figure 1—figure supplement 2B,C). In addition, GFP immunoreactivity was also detected in NMN-D and NMN-D^dead, confirming the robust expression of the newly generated tagged proteins (Figure 1A). These results show that our newly generated GFP-tagged NMN-D variants are substantially stronger expressed than NMNd in vivo.

The lower-expressed NMNd resulted in a 40% preservation in our wing injury assay. We repeated the injury assay to assess the preservation of our newly generated and higher-expressed NMN-D variants. While severed axons with GFP or NMN-D^dead degenerated, axons with NMN-D remained fully preserved at 7 dpa (Figure 1B and C, Figure 1—figure supplement 3). This contrasts with the weaker preservation of axons with lower NMNd expression (Figure 1—figure supplement 1). Similarly, strong preservation was seen in cholinergic olfactory receptor neurons (ORNs), where severed axons with NMN-D remained preserved at 7 dpa (Figure 1D). We extended the ORN injury assay and found preservation at 10, 30, and 50 dpa (Figure 1E). While quantifying the precise number of axons is technically not feasible, severed preserved axons were observed in all 10, 30, and 50 dpa brains, albeit fewer at later time points (MacDonald et al., 2006). Thus, high levels of NMN-D confer robust protection of severed axons for multiple neuron types for the entire lifespan of Drosophila.

Before measuring the effect of NMN-D on the NAD⁺ metabolic flux, we measured the activities of the various NAD⁺ biosynthetic enzymes in fly heads (Figure 2A, Figure 2—figure supplement 1A, Amici et al., 2017; Zamporlini et al., 2014). We confirmed that NAD⁺ can be synthesized from nicotinamide (NAM), nicotinamide riboside (NR), and quinolinate. However, the Drosophila Qaprt homolog that catalyzes the conversion from quinolinate to nicotinic acid mononucleotide (NaMN) remains to be identified (Katsyuba et al., 2018). We also confirmed the absence of NAMPT activity (Figure 2—figure supplement 1A, Gossmann et al., 2012). In addition, we confirmed the expression of genes involved in NAD⁺ synthesis and axon death signaling that are involved in NAD⁺ metabolism by measuring respective mRNA abundance in fly heads by qRT-PCR (Figure 2—figure supplement 1B). We note that the expression and activity of NAD⁺ metabolic enzymes can be readily detected fly heads.

Next, we wanted to know whether the sole expression of NMN-D can change the NAD⁺ metabolic flux in vivo (Figure 2A). We compared levels of metabolites in heads by LC-MS/MS among samples that expressed NMN-D and NMN-D^dead (Figure 2B, van der Velpen et al., 2021). Consistent with robust NMN-D activity, NMN levels were sixfold lower and NaMN 12-fold higher. We also found significantly higher NaAD and NaR levels. Importantly, all other metabolites remained unchanged, including NAD⁺ (Figure 2B).

Prompted by such a significant change in the NAD⁺ metabolic flux, we wondered whether the change could alter the expression of genes involved in NAD⁺ metabolism or axon death signaling. However, besides the expected significant increase of the Gal4-mediated expression of NMN-D and NMN-D^dead, we did not observe any notable changes (Figure 2—figure supplement 2). Our observations demonstrate that the expression of NMN-D alone is sufficient to change the NAD⁺ metabolic flux, thereby significantly lowering NMN levels without affecting NAD⁺ in Drosophila heads; they serve as an excellent tissue for metabolic analyses.

Mutations that attenuate axon death signaling robustly suppress the morphological degeneration after axotomy. They also preserve synaptic connectivity. We have previously demonstrated that synaptic connectivity of severed axons with attenuated axon death remains preserved for at least 14 days using an established optogenetic assay (Neukomm et al., 2017; Paglione et al., 2020). Briefly, mechanosensory chordotonal neurons in the Johnston’s organ (JO), whose cell bodies are in the second segment of adult antennae, are required and sufficient for antennal grooming (Hampel et al., 2015; Seeds et al., 2014). The JO-specific expression of CsChrimson combined with a red-light stimulus can specifically and robustly induce antennal grooming.

We used this assay to test individual flies before bilateral antennal ablation (ctl) and at 7 dpa (Figure 3). Flies expressing GFP in JO neurons failed to elicit antennal grooming following red-light exposure at 7 dpa (Figure 3, Video 1). In contrast, flies with JO-specific NMN-D expression continued to elicit antennal grooming at 7 dpa (Figure 3, Video 2). Remarkably, the evoked grooming behavior remained equally robust at 14 dpa (Figure 3—figure supplement 1). Preservation of synaptic connectivity depended on low NMN levels, as flies expressing NMN-D^dead in JO neurons failed to elicit antennal grooming upon red-light exposure at 7 dpa (Figure 3, Video 3). Our findings demonstrate that lowering NMN potently attenuates axon death signaling, which is sufficient to preserve synaptic connectivity of severed axons and synapses for weeks after injury.

Two enzymatic reactions synthesize NMN in mammals: NAMPT-mediated NAM and NRK1/2-mediated NR consumption. Drosophila lacks NAMPT activity, and NMN synthesis relies solely on Nrk-mediated NR consumption (Figure 2—figure supplement 1A). We hypothesized that mouse NAMPT (mNAMPT) expression could increase NMN synthesis and, therefore, lead to faster injury-induced axon degeneration in vivo (Figure 4A). We generated transgenic flies harboring mNAMPT under the control of UAS by targeted insertion (attP40). Western blots revealed proper expression of mNAMPT in fly heads (Figure 4B).

We then tested the effect of mNAMPT on the NAD⁺ metabolic flux in vivo. Surprisingly, NAM, NMN, and NAD⁺ levels remained unchanged under physiological conditions (Figure 4C). However, we noticed threefold higher NR and a moderate but significant elevation of ADPR and cADPR levels upon mNAMPT overexpression (Figure 4C). We also asked whether mNAMPT impacts on NAD⁺ homeostasis thereby altering the expression of axon death or NAD⁺ synthesis genes. Besides the expected significant increase in the Gal4-mediated expression of mNAMPT, we did not observe any notable changes at the mRNA level (Figure 4—figure supplement 1).

Although mNAMPT expression failed to elevate NMN under physiological conditions, we hypothesized that mNAMPT could boost NMN levels after injury because of the following observations in mammals: NMNAT2 is labile and rapidly degraded in severed axons (Gilley and Coleman, 2010), while NAMPT persists much longer (Di Stefano et al., 2015). We speculated that in flies, in severed axons, dNmnat declines similarly, but not mNAMPT. Consequently, NMN accumulates. Strikingly, in our wing injury assay, while axons with GFP showed signs of degeneration starting from 6 hr post axotomy (hpa), mNAMPT expression resulted in significantly faster axon degeneration with signs of degeneration at 4 hpa (Figure 4D and E). This accelerated degeneration is likely linked to increased NMN production, but other mechanisms cannot be excluded as there is no increase in NMN under physiological conditions.

We next asked whether the faster degeneration of mNAMPT-expressing severed axons requires axon death genes. While mutations in dsarm and hiw completely blocked the degeneration of severed axons expressing mNAMPT, axed showed a partial preservation of 60% at 12 hr after injury (Figure 4F). Importantly, axed mutants, in the absence of mNAMPT expression, showed a similar preservation within the first 12 hr (Figure 4—figure supplement 2). This preservation remained unchanged at 7 dpa, suggesting that mNAMPT expression does not change the preservation provided by axed, dsarm, and hiw (Figure 4—figure supplement 2, Figure 4G). Our observations support that elevated NMN levels require axon death signaling to initiate the degeneration of severed axons.

Overall, our data suggest that NMN accumulation after injury triggers axon degeneration in Drosophila through the axon death pathway. To the best of our knowledge, we provide the first direct in vivo demonstration that an additional source of NMN synthesis–by the expression of mNAMPT–accelerates injury-induced axon degeneration.

We have shown that low NMN attenuates injury-induced axon degeneration, while a more rapid accumulation of NMN, due to expression of mNAMPT, results in faster injury-induced axon degeneration. A cell-permeable form of NMN (CZ-48) binds to and activates SARM1 by changing its conformation (Zhao et al., 2019). Crystal structures of the ARM domain of dSarm (dSarm^ARM), as well as the full-length human SARM1 (hSARM1), support the observation that NMN acts as a ligand for dSarm^ARM (Figley et al., 2021; Gu et al., 2021). NMN binding to the ARM domain requires a critical residue, lysine 193 (K193), to induce a conformational change in the ARM domain of dSarm/SARM1. Consistent with this, a mutation of the lysine residue (e.g. K193R) results in a dominant-negative injury-induced axon degeneration phenotype in murine cell cultures (Bratkowski et al., 2020; Figley et al., 2021; Geisler et al., 2019; Loreto et al., 2021; Zhao et al., 2019).

To confirm whether NMN activates dSarm in vitro and in vivo, we generated dsarm constructs encoding wild-type and the human K193R-equivalent K450R mutation. Crucially, an isoform we used previously, dSarm(D), fails to fully rescue the dsarm⁸⁹⁶ defective axon death phenotype (Figure 5—figure supplement 1A, B, Neukomm et al., 2017). Therefore, among the eight distinct dsarm transcripts, which all contain the ARM, SAM, and TIR domains, we chose the shortest coding isoform, dsarm(E) (Figure 6—figure supplement 1A). We generated untagged and C-terminal FLAG-tagged dSarm(E), with and without K450R, under the control of UAS and confirmed the FLAG-tagged proteins are expressed at similar levels in S2 cells (Figure 5A). We also directly tested the immunopurified FLAG-tagged proteins for constitutive and NMN-inducible NADase activity (Figure 5B). While wild-type and K450R dSarm(E) had similar constitutive activities in the presence of 25 µM NAD⁺ alone, we found that only wild-type NADase activity was induced further with the addition of 50 µM NMN. At the same time, K450R remained essentially unchanged (Figure 5C). This confirmed the critical role of K450 in NMN-dependent activation of the dSarm NADase, equivalent to the role of K193 in hSARM1 (Figley et al., 2021; Loreto et al., 2021).

Next, we generated transgenic flies expressing tagged and untagged wild-type and mutant dSarm(E) variants–by targeted insertion of the UAS–dsarm(E) plasmids (attP40)–and confirmed pan-cellular expression of the FLAG-tagged variants by immunoblotting (Figure 5D). We used all variants for dsarm⁸⁹⁶ axon death defective rescue experiments in our wing injury assay. We found that the expression of wild-type dSarm(E) (both tagged and untagged) almost entirely rescued dsarm⁸⁹⁶ mutants, whereas dSarm(E^K450R) proteins completely failed to rescue the phenotype at 7 dpe (Figure 5E). We demonstrate that a non-inducible NADase variant, dSarm(E^K450R), in the absence of wild-type dSarm, fails to execute injury-induced axon degeneration in vivo.

We have now established that NMN activates dSarm to trigger the degeneration of severed axons in Drosophila. While NMN induces a conformational change in a pocket of the ARM domain, NAD⁺ prevents this activation by competing for the same pocket (Bratkowski et al., 2020; Figley et al., 2021; Jiang et al., 2020; Zhao et al., 2019). We therefore wanted to test whether the preservation provided by lower NMN is reverted by a simultaneous reduction of NAD⁺ synthesis. We generated NMN-D-expressing neurons containing RNAi-mediated knockdown of Nadsyn (nadsyn^RNAi). At 7 dpa, the 100% preservation provided by NMN-D was partially reduced to 60% by nadsyn^RNAi in vivo (Figure 5—figure supplement 2). This observation supports the degenerative NMN and the protective NAD⁺ function by activating and inhibiting dSarm in injury-induced axon degeneration in Drosophila.

Lowering levels of NMN confers very robust protection against axon degeneration in Drosophila, similar to that achieved by targeting other mediators of axon degeneration, such as hiw, dsarm, axed, and the over-expression of dnmnat (dnmnat^OE) (Fang et al., 2012; Neukomm et al., 2017; Neukomm et al., 2014; Osterloh et al., 2012; Paglione et al., 2020; Xiong et al., 2012). We therefore assessed the genetic interaction among these regulators in vivo.

The current model, supported by our data, predicts that NMN accumulation occurs upstream of dSarm activation. Consistent with this, the induced expression of constitutively active dSarm lacking its inhibitory ARM domain (dSarm^∆ARM) is sufficient to pathologically deplete NAD⁺, triggering axon- and neurodegeneration in the absence of injury (Essuman et al., 2017; Neukomm et al., 2017). We asked whether lowering levels of NMN can delay or prevent neurodegeneration induced by dSarm^∆ARM-mediated NAD⁺ depletion. dsarm^∆ARM clones with forced NAD⁺ depletion rapidly degenerated within 5 days after adults were born (days post eclosion, dpe) (Figure 6A). As expected, lowering NMN levels by NMN-D in dsarm^∆ARM clones did not alter the kinetics of neurodegeneration (Figure 6A). These observations further support that NMN accumulation occurs upstream of dSarm activation, and that once neuronal NAD⁺ is low, neurodegeneration cannot be halted by low NMN.

We also asked whether lowering NMN interferes with loss-of-function mutations of axed, dsarm, double mutants, hiw, and dnmnat^OE. As expected, the attenuated axon death phenotype of low NMN did not change in these mutant backgrounds at 7 dpe (Figure 6B).

Loss-of-function mutations in dnmnat also activate axon death signaling, leading to neurodegeneration in the absence of injury (Neukomm et al., 2017). dNmnat is the sole enzyme with NAD⁺ biosynthetic activity in Drosophila. dnmnat¹ mutant clones lack NMN consumption and NAD⁺ synthesis (Zhai et al., 2006), and they degenerate with similar kinetics as dsarm^∆ARM clones (Neukomm et al., 2017). We asked whether low NMN levels can delay or prevent dnmnat¹-induced neurodegeneration. Surprisingly, although not expected to restore NAD⁺ synthesis, lowering NMN levels significantly delayed neurodegeneration (Figure 6C). Between 1–5 dpe, while clones with dnmnat¹ fully degenerated, NMN-D expressing dnmnat¹ clones remained morphologically intact, similar to controls. After 5 days, these clones gradually started to deteriorate (Figure 6C). This protective delay of neurodegeneration depends on lowering NMN levels, as the expression of NMN-D^dead completely failed to protect dnmnat¹ neurons (Figure 6C).

To confirm that NMN-D delays dnmnat¹-mediated neurodegeneration, we wanted to generate tissue-specific CRISPR/Cas9 dnmnat knockouts. CRISPR/Cas9 tools significantly facilitate genetics in Drosophila. Instead of using mutants in a specific genomic locus, mutations are generated by co-expressing Cas9 and sgRNAs.

First, we generated transgenic flies harboring tRNA-flanked sgRNAs to target four distinct loci in dnmnat under the control of UAS (Port and Bullock, 2016). We made similar transgenic flies to target all other axon death genes (Figure 6—figure supplement 1A). We then tested our novel tools for their ability to attenuate axon death by assessing preserved axonal morphology (Figure 6—figure supplement 1B) and synaptic connectivity after injury (Figure 6—figure supplement 1C). We found that the preservation depended on the combination of the tissue (e.g. Gal4 driver) and the Cas9 source (e.g. UAS–cas9 vs. actin–cas9). Our observations highlight that the combination of Cas9 and sgRNAs must be carefully determined in each tissue targeted by CRISPR/Cas9.

We then asked whether dnmnat^sgRNAs can trigger neurodegeneration by analyzing neuronal survival (Figure 6—figure supplement 1D) and synaptic connectivity over time (Figure 6—figure supplement 1E). Notably, we observed synthetic lethality in UAS–dnmnat^sgRNAs actin–Cas9 flies; we thus used UAS–cas9. Neurons with CRISPR/Cas9-targeted dnmnat degenerated as fast as dnmnat¹ mutants (Figure 6—figure supplement 1D). In line with these findings, we observed reduced synaptic connectivity in 7- and 14-day-old flies (Figure 6—figure supplement 1E). We also found the expression of NMN-D in dnmnat^sgRNAs clones resulted in similar neuroprotection as observed with dnmnat¹ mutants. They remained morphologically intact during the first 5 days and then gradually degenerated (Figure 6—figure supplement 1F). Therefore, NMN-D can also delay neurodegeneration in CRISPR/Cas9-targeted dnmnat clones by preventing NMN accumulation.

Taken together, our in vivo results suggest that in the absence of dnmnat, NMN-D prevents NMN accumulation and therefore delays neurodegeneration. However, neurons subsequently degenerate because NAD⁺ synthesis halts, and NAD⁺ gradually decays below the threshold of survival. Similarly, NMN-D fails to delay neurodegeneration when dSarm^∆ARM forcefully depletes NAD⁺. These results support the role of NMN as an activator rather than an executioner in axon death signaling.

Flies (Drosophila melanogaster) were kept on Nutri-Fly Bloomington Formulation (see resources table) with dry yeast at 20 °C unless stated otherwise. The following genders were scored as progeny from MARCM crosses: females (X chromosome); and males & females (autosomes, chromosomes 2 L, 2 R, 3 L, and 3 R). We did not observe any gender-specific differences in clone numbers or axon death phenotype. Gender and genotypes are listed in Source data 1.

NA was converted to NaAD by the consecutive actions of the ancillary enzymes PncB (bacterial NaPRT) and NadD. The reaction mixture consisted of ethanol buffer, 40 mM HEPES/KOH pH 7.5, 10 mM KF, 10 mM MgCl₂, 2.5 mM ATP, 2 mM PRPP, 0.3 mM NAM, 6 U/ml ADH, 0.067 mg/ml BSA, 0.5 U/ml PncB and 1 U/ml NadD. A control mixture was prepared in the absence of NAM. The generated NaAD was converted to NAD which was quantified as described above.

Enzymatic activities were determined by directly measuring the newly synthesized NAD as follows:

Drosophila Schneider cells (S2) are sold and authenticated by Thermo Fisher (R69007). The Master Seed Bank has been tested for contamination of bacteria, yeast, mycoplasma, and virus and has been characterized by isozyme and karyotype analysis. S2 cells were maintained at 26 °C in Drosophila Schneider’s medium (Thermo Fisher) supplemented with 10% Fetal Bovine Serum (Thermo Fisher) and 1xPenicillin-Streptomycin (Thermo Fisher). 8x10⁵ cells were plated out in 10 mm plates 24 hr prior to transfection. Cells were co-transfected with either UAS–dsarm(E)::flag, UAS–dsarm(E^K193R)::flag, UAS–GFP::NMN-Deamidase, or UAS–GFP::NMN-Deamidase^dead constructs and pAc–GAL4 (Addgene) to a final concentration of 10 µg DNA/well using Mirus TransIT-Insect (Mirus Bio). Forty-eight hr post-transfection, cells were harvested with the original medium in tubes on ice. Cells were centrifuged at 5000 g for 30 s, the supernatant discarded, the cells resuspended in 5 ml of cold PBS and centrifuged again at 5000 g for 30 s. After discarding the supernatant, cells were resuspended in 300 µl/plate of cold KHM lysis buffer (110 mM CH₃CO₂K, 20 mM HEPES pH 7.4, 2 mM MgCl₂, 0.1 mM digitonin, Complete inhibitor EDTA free (Roche)) and incubated for 10 min at 4 °C while briefly vortexing for 5 s and up-and-down pipetting five times every minute. Samples were then centrifuged at 3000 rpm for 5 min to pellet cell debris. Protein concentration of the supernatant was determined using the BCA protein quantification assay (ThermoFischer).

Rabbit anti-PncC antibodies were generated by Lubioscience under a proprietary protocol. The immunogen used was purified from Escherichia coli, strain K12, corresponding to the full protein sequence of NMN-D. The amino acid sequence is the following:

MTDSELMQLSEQVGQALKARGATVTTAESCTGGWVAKVITDIAGSSAWFERGFVTYSNEAKAQMIGVREETLAQHGAVSEPVVVEMAIGALKAARADYAVSISGIAGPDGGSEEKPVGVWFAFATARGEGITRRECFSGDRDAVRRQATAYALQTLWQQFLQNT.

Four to 12% surePAGE gels (genescript) were used with MOPS running buffer (for higher molecular weight proteins) or MES running buffer (for lower molecular weight proteins). Gels were subjected to 200 V. A molecular weight marker Precision Plus Protein Kaleidoscope Prestained Protein Ladder was used (Biorad). Proteins were transferred to PVDF membranes with the eBlot L1 system using eBlot L1 Transfer Stack supports (Genscript) and the resulting membranes were washed three times with TBS-T (Tris-buffered saline containing 0.1% Tween 20 (Merk)). Membranes were blocked with 5% milk (Carl-Roth) in TBS-T at room temperature (RT) for 1 hr. Membranes were incubated at 4 °C with corresponding primary antibodies overnight (O/N). Membranes were washed three times with TBS-T for 10 min and incubated with secondary antibodies in 5% milk in TBS-T at RT during 1 hr. Membranes were washed three times with TBS-T for 10 min.

Fluorescent signals were acquired using Odissey DLx (LI-COR). Images were quantified by densitometric analysis using ImageJ (NIH).

Wings of aged flies (0–10 days post eclosion (dpe)) were observed and imaged with a spinning disk microscope to assess for intact or degenerated neurons and axons.

The plasmids listed below were generated and used for PhiC31 integrase-mediated targeted transgenesis (Bestgene) (5xUAS, w⁺ marker). attP40 target site: UAS–GFP::NMN-Deamidase, UAS–GFP::NMN-Deamidase^dead, UAS–dsarm(E), UAS–dsarm(E)::flag, UAS–dsarm(E^K450R), UAS–dsarm(E^K450R)::flag, UAS–mNAMPT. VK37 target site: UAS–4 x(tRNA::axed^sgRNAs), UAS–4 x(tRNA::hiw^sgRNAs), UAS–4 x(tRNA::dsarm^sgRNAs), UAS–4 x(tRNA::dnmnat^sgRNAs). All plasmids are available as *.gb files on Addgene.

Crosses were performed on standard cornmeal agar containing 200 µM all-trans retinoic acid in aluminum-wrapped vials to keep the progeny in the dark (Paglione et al., 2020). Adult progenies were aged at 20 °C for 7-14 days before starting the experiment. CsChrimson experiments were performed in the dark, and flies were visualized for recording using an 850 nm infrared light source at 2 mW/cm² intensity (Mightex, Toronto, CA). For CsChrimson activation, 656 nm red light at 6 mW/cm² intensity (Mightex) was used. Red light stimulus parameters were delivered using a NIDAQ board controlled through Bonsai (https://open-ephys.org/). Exclusion criteria: to avoid spontaneous grooming behavior, during the recording, flies that groomed within the first 30 s were excluded from the analysis. Red-light stimulation (10 Hz for 10 s) was followed by a 30 s interstimulus recovery (3 repetitions in total). Flies were recorded, and videos were manually analyzed using VLC player (http://www.videolan.org/). Grooming activity (ethogram) was plotted as bins (1 bin, grooming event(s) per second). Ethograms were visualized using R (https://cran.r-project.org/). The ablation of 2nd antennal segments did neither damage the head nor lead to fly mortality. Flies that died during the analysis window (7–15 dpa) were excluded.

Total RNA from forty fly heads for each genotype was isolated with TRIzol LS Reagent (Invitrogen, 10296010). The isolated RNA was treated with TURBO DNase (Invitrogen, AM2238) at 37 °C for 20 min and purified using RNA Clean & Concentrator-5 (Zymo Research, R1015). First strand cDNA was synthesized using random hexamers (Invitrogen, N8080127) and SuperScript IV first-strand synthesis system (Invitrogen, 18091050). Quantitative PCR was performed for each sample using PowerUp SYBR green master mix (Applied Biosystems, A25741) with technical triplicates for both +RT and −RT for each genotype in MicroAmp optical 96-well reaction plates (Applied Biosystems, 4306737) and QuantStudio 1 real-time PCR system (Applied Biosystems). Relative transcript abundance was calculated using the ΔΔCt method. α-tubulin, an mRNA that remains unchanged in mutant and transgenic flies based on qPCR analysis, was used for normalization. Statistical significance was calculated using the one-way ANOVA test. The post hoc Tukey test was performed for statistically different groups determined by the ANOVA test (p-value <0.05). The resulting p values from the Tukey test are reported.

Primers used (forward and reverse, respectively, 100–200 bp amplicons, all isoforms included):

For all experiments, at least 3 biological replicates were performed for each genotype and/or condition. No inclusion/exclusion criteria was applied except for optogenetics, which is stated in the subsection ‘optogenetics’ of Materials and methods.

Image-J and photoshop was used to process wing and ORN pictures. Software for optogenetics is included in the optogenetics section. Graphpad prism 9 was used to perform all the statistical analysis. For tests applying a false discovery rate (FDR) correction, the adjusted p value we report is the q value.

The following R code was used to generate ethograms from excel flies.

library(readxl) nmn_7 <- read_excel("nmn 7 days.xlsx") head(nmn_7) # custom function using image to emulate an ethograph ethogram <- function(zeroOneMatrix, color=’skyblue',xlab='behaviour',ylab='animals'){ m <- as.matrix(zeroOneMatrix) m[m==0] <- NA nAnimals <- nrow(m) nTimeSlots <- ncol(m) image(x=1:nTimeSlots,  y=1:nAnimals,   z=t(m[nAnimals:1,]),   col = c(color),   xlab = xlab,   ylab = ylab,   yaxt = 'n') } # let’s plot ethogram(nmn_7, color='lightskyblue1') data_t=t(nmn_7) head(data_t) colnames(data_t)=rev(c(1:ncol(data_t))) rownames(data_t)=c(1:nrow(data_t)) head(data_t) data_long <- reshape2::melt(data_t) colnames(data_long)=c("Time", "Animal", "Val") data_long$Val <- factor(data_long$Val) head(data_long) data_zeros =data.frame(data_t) data_zeros[is.na(data_zeros)] <- 0 rs = rowSums(data_zeros) rs data_sum = data.frame(Time = c(1:length(rs)), Count = rs) head(data_sum) library(ggplot2) e2=ggplot()+  geom_tile(data_long, mapping = aes(x=data_long$Time, y=data_long$Animal, fill = data_long$Val), color="white", size = 0.5)+  labs(x="behaviour", y="animals", title="OPTO")+  theme_bw()+theme(axis.text.x=element_text(size = 9, angle = 0, vjust = 0.3),    axis.text.y=element_text(size = 9),    plot.title=element_text(size = 11))+  theme(panel.border=element_blank())+  scale_fill_manual(values = c("skyblue"))+  geom_line(data_sum, mapping = aes(x=data_sum$Time, y=data_sum$Count), color="black", size = 0.7) e2

We followed the ARRIVE guidelines for reporting work involving fly research.