To generate an auxin-inducible system for control of GAL4 activity in Drosophila, we designed a transgene that would ubiquitously express two proteins, TIR1 and AID-tagged GAL80. (Figure 1). We included shorter AID sequences (minimal degron of IAA17) than previously used in Drosophila (Trost et al., 2016). Fusion of this minimal AID to luciferase resulted in a fusion protein with short half-life in plants (~10 min) (Dreher et al., 2006). We also instead used Arabidopsis thaliana TIR1 (AtTIR1), which has two point mutations (improving affinity and auxin sensitivity) that can deplete nuclear and cytoplasmic proteins in C. elegans (Zhang et al., 2015). The P2A sequence between AtTIR1 and AID-tagged GAL80 encodes a self-cleaving peptide that is known to work efficiently in Drosophila cell culture (Daniels et al., 2014). The AtTIR1-T2A-AID-GAL80-AID sequence was codon optimised, synthesised, and cloned downstream of the αTub84B promoter in pattB (Figure 1 and Figure 1—figure supplement 1). The αTub84B (tubulin) promoter drives expression of the bicistronic sequence in all cells, all of the time, and (in the absence of auxin) will inhibit any GAL4 activity. When present, auxin will tether the TIR1 to the AID sequences, triggering the degradation of GAL80 and the release of GAL4 inhibition (Figure 1).

We combined the AGES with c564-GAL4 to see if we could get inducible expression in fat body. A nuclear localised GFP (nls-GFP) was used as a reporter of GAL4 activity. We first examined adult female flies that were placed on food containing different concentrations of 1-naphthaleneacetic acid potassium salt (K-NAA) auxin, which has shown to be more water soluble than NAA (Martinez et al., 2020). A negative control consisting of fly food without auxin (0 mM auxin) was included to investigate whether there was any leaky GAL4 activity in absence of auxin. After 24 hr, GFP fluorescence in the abdomen was assayed (note that there is some naturally occurring autofluorescence in the fly abdomen). At 5 and 10 mM, GFP fluorescence is clearly detectable in the abdomen of live female flies (Figure 2A). Based on this assay, 10 mM for 24 hr results in ~60% of the GFP fluorescence levels seen in the positive control (Figure 2B). At 1 mM there is no detectable GFP above background fluorescence, and the levels are indistinguishable from both the 0 mM and the negative control (lacking both the GAL4 driver and the UAS-nls-GFP). To provide an alternative and more direct assay of transgene induction, quantitative PCR (qPCR) was used to measure the levels of GFP mRNA (Figure 2C). Here, 5 and 10 mM show levels of mRNA equivalent to the positive control, while 0 mM showed no significant difference when compared with the negative control.

Adult males eat less than females (Wong et al., 2009), therefore, this will impact on auxin ingestion and possibly on transgene induction. As with females, we examined both GFP fluorescence levels in the abdomen, and GFP mRNA levels in male flies (Figure 2—figure supplement 1). Although reduced compared to females, there is still a robust induction of GFP expression.

To assess how quickly transgene expression is turned off after removal of auxin from the food, we performed a time course experiment using adult females and 10 mM auxin. After 24 hr, flies were shifted to food without auxin and GFP fluorescence levels in the abdomen measured every 24 hr. At 48 hr (24 hr without auxin), GFP levels were reduced (Figure 2—figure supplement 2A). At 72 and 96 hr, the levels were not significantly different to 0 hr. Furthermore, we tested how stable auxin is in prepared food kept at 4°C. Even after 15 weeks, the auxin food could induce GFP levels to that of newly prepared food (Figure 2—figure supplement 2B).

In addition to the need for temporal control of GAL4 activity in the adult, researchers might also want to induce expression at specific points during larval development. To test if AGES works in larvae, we used the c564-GAL4 fat body driver and tested multiple auxin concentrations (Figure 3A and C). Here, we observed that expression could be induced at lower concentrations of auxin. When using food containing no auxin, there is no detectable GFP, whereas with 1 mM auxin, GFP is expressed, although at lower levels than with higher concentrations of auxin (Figure 3A and C). Five mM auxin is an optimal concentration, as concentrations above that do not increase GFP levels. Five mM auxin can induce low-level GFP expression after 6 hr (Figure 3B and D). However, 18 hr is required to provide maximum levels of expression.

For AGES to work with GAL4 lines expressed in the central nervous system, auxin must be able to pass the selectively-permeable glial membrane, akin to the mammalian blood-brain barrier (Limmer et al., 2014). NAA can cross the blood-brain barrier in Drosophila (Chen et al., 2018). To verify this using K-NAA, we first used elav-GAL4 (expressed in all neurons) and examined the expression of GFP in the adult brain. Similar to the fat body c564-GAL4 driver, we observe robust expression of GFP, when flies are fed auxin food (10 mM) for 24 hr (Figure 4—figure supplement 1). This demonstrates that AGES works in the central nervous system and that K-NAA auxin can cross the blood-brain barrier in Drosophila.

We then tested more restricted GAL4-driver lines in the larval ventral nerve cord (grh-GAL4) and in the adult (Or85a-GAL4). In both cases there was no detectable GAL4 activity in the absence of auxin, while in the presence of auxin (10 mM for adults and 5 mM for larvae) GFP in a pattern consistent with the GAL4 driver is observed (Figure 4A and C) at levels significantly above background fluorescence levels (Figure 4B and D).

For a universally applicable drug-inducible system, the drug should not impact on development or mortality of the flies. We tested whether continuous exposure to different concentrations of auxin (K-NAA) would impact on developmental timing. Time to pupation (from egg-laying) is unaffected for concentrations up to 5 mM, however, 10 mM caused a delay of approximately 1 day (Figure 5A). Ten mM auxin causes some developmental delay, although it has no impact on the survival of the flies through these developmental stages (compared to a no auxin control) (Figure 5B). These data indicate that 10 mM auxin should be avoided for larval induction, however, this is not an issue as >1 mM is sufficient for robust induction of expression (Figure 5A) and 5 mM does not cause any developmental delay. Concentrations of 5 and 10 mM auxin can induce expression in adults (Figure 2 and Figure 2—figure supplement 1). Survival assays on both male and female adults show that concentrations of 5 and 10 mM have no effect on survival and lifespan (Figure 5D and E). Therefore, in summary, up to 5 mM auxin is optimal for use in larvae, whilst up to 10 mM auxin is best for use in adults. We also tested whether auxin affects locomotion of larvae (crawling assay) and adults (climbing assay) and see no significant effects with 5 and 10 mM auxin, respectively (Figure 5—figure supplement 1).

To determine whether AGES is suitable for acute adult behavioural manipulation, we recapitulated a classic GeneSwitch manipulation of clock neuron excitability using AGES (Depetris-Chauvin et al., 2011). A key circuit for control of Drosophila rhythmic behaviour is the ventrolateral neuron (LNv) cluster, of which the small ventrolateral neurons (sLNvs) are both necessary and sufficient for maintenance of free-running locomotor rhythms (Grima et al., 2004; Renn et al., 1999; Stoleru et al., 2004). The four large LNvs and four out of five sLNvs express the neuropeptide pigment dispersing factor (PDF) (Helfrich-Förster, 1995; Renn et al., 1999). PDF is a key clock output neuropeotide required for synchronisation of clock neuron groups, and loss of PDF function results in arrythmicity, desynchronisation of clock oscillators, and altered period length (Lear et al., 2009; Lin et al., 2004; Peng et al., 2003; Sheeba et al., 2008). Adult-specific silencing of PDF+ LNvs by expression of the inwardly rectifying potassium channel Kir2.1 using PDF-GAL4-GeneSwitch is sufficient to nearly ablate circadian locomotor rhythms in constant conditions without resetting the molecular clock (Depetris-Chauvin et al., 2011).

To recapitulate this behavioural experiment, we combined the PDF-GAL4 line with the AGES to allow auxin-inducible control of Kir2.1 expression in PDF+ neurons. We simultaneously replicated GeneSwitch-driven Kir2.1 expression in PDF+ neurons. After eclosion and entrainment to a 12 hr light:12 hr dark (12:12 LD) schedule on standard fly food, flies were individually loaded into activity tubes and placed in the Drosophila activity monitoring (DAM) system (Trikinetics) for 3 days in 12:12 LD, followed by 8 days in constant darkness (DD). Control activity tubes contained standard DAM food (2% agar, 5% sucrose) while experimental food contained RU-486 (200 mg/mL) or NAA (2 or 10 mM). In contrast to the previous experiments, NAA auxin was used here instead of K-NAA auxin. Analysis of the amplitude of the behavioural locomotor rhythm in constant darkness by fast Fourier transform (FFT) (Plautz et al., 1997) showed robust inhibition of locomotor rhythms in both male and female experimental PDF-GAL4; AGES > UAS-Kir2.1 flies on 2 or 10 mM NAA (Figure 6A-D, Figure 6—figure supplement 1A,B). Two mM NAA was sufficient to completely ablate circadian locomotor rhythms in both male and female experimental flies (Figure 6C and D), without a significant effect on average 24 hr locomotor activity of parental controls (Figure 6—figure supplement 2 A-D). Two mM NAA feeding significantly lengthened the behavioural period of male PDF-GAL4/+; AGES/+ parental controls by 19 min, with no effect on females of the same genotype. In agreement with the classic study by Depetris-Chauvin et al., 2011, 200 μg/mL RU-486 feeding also reduced circadian locomotor rhythms in both sexes of flies with PDF-GAL4-GeneSwitch-driven expression of Kir2.1 (Figure 6E and F), though the effect was not as strong as 2 mM NAA feeding in the AGES flies. In addition, RU-486 feeding resulted in a broader distribution of locomotor rhythm strengths and a net increase in rhythmicity in male PDF-GAL4-GeneSwitch parental control flies (Figure 6E), which was accompanied by a 51 min increased period length and increased average 24 hr locomotor activity (Figure 6—figure supplement 2G). RU-486 also increased the period of female PDF-Gal4-GeneSwitch parental controls by 46 min, with no effect on total activity or rhythm strength (Figure 6F, Figure 6—figure supplement 2F,H). In summary, we find that 2 mM NAA feeding is sufficient for AGES-induced expression of Kir2.1 in adult PDF+ clock neurons to ablate circadian locomotor behaviour. Indeed, we observed a stronger ablation of locomotor rhythmicity in the AGES compared to GeneSwitch, with fewer off-target behavioural effects of NAA feeding of parental control flies.

The AtTIR1-T2A-AID-GAL80-AID sequence was synthesised by Twist Biosciences (twistbioscience.com) (Figure 1—figure supplement 1). The GAL80 sequence (from Saccharomyces cerevisiae) and the AtTIR1 sequence (Zhang et al., 2015) were codon optimised for Drosophila for more effective translation. The αTub84B promoter was amplified from a tubulin-eGFP plasmid (gift from M Dionne) using the following primers: tub_FWD: GATATCAAGCTTGCACAGGTCC and tub-RV: GTACCTTCACGCTGTGGATGAGG. The αTub84B and AtTIR1-T2A-AID-GAL80-AID sequences were cloned into pattB (Bischof et al., 2007) using Gibson assembly (Gibson et al., 2009). Successful clones were sequence verified. Annotated sequence is in Figure 1—figure supplement 1—source data 1. The plasmid is publically available at DGRC: the Drosophila Genomics Resource Center (DGRC: ): https://dgrc.bio.indiana.edu/product/View?product=1568 Microinjection was performed by Cambridge Fly Facility using the VK00040 line, which has an attB site at location att3B (87B10) on chromosomal arm 3R. Injected adult males were collected and mated to w¹¹¹⁸ virgin females to identify transgenics (orange eyes). tub-TIR1-T2A-AID-GAL80-AID is available at the BDSC (stock #92470) or the VDRC Stock Center (stock #311020).

In this study, c564-GAL4, UAS-nls-GFP/CyO-actin-GAL4-GFP flies were used to perform fluorescent reporter experiments in the adult and larval fat body, and for quantification of GFP mRNA expression. Moreover, these flies were used in adult survival assays. CantonS flies were used in the developmental survival and developmental timeline experiments. For fluorescent reporter experiments in the adult brain, we used elav-GAL4, UAS-nls-GFP. For behavioural experiments, Iso31 flies were used as a background strain (Ryder et al., 2004), and we used PDF-Gal4 (Park et al., 2000) and PDF-GeneSwitch-GAL4 (BDSC 81116) to drive expression of UAS-Kir2.1 (BDSC 6597). Flies were kept at 25°C on and standard Drosophila food (recipe in supplementary material) was supplemented (just before being aliquoted into vials/bottles/plates) with auxin (K-NAA available from Phytotech [#N610] or Glentham Life Sciences [GK2088]) at varying concentrations. For behavioural experiments, flies were raised and entrained on standard food, and transferred to activity monitoring tubes containing 2% agar, 5% sucrose, supplemented with indicated concentrations of RU-486 in ethanol (Mifepristone, Sigma-Aldrich) or NAA (Phytotech, #N600).

Adult and larval brains were dissected in 1× PBS and fixed for 25 min at room temperature in 4% formaldehyde (methanol free) in 0.3% Triton X-100 PBS (PBST). They were washed four times for 1 hr with 0.3% PBST. Normal goat serum (2% in PBST) was used for tissue blocking (RT, 15 min to 1 hr) and subsequent overnight primary antibody incubation. All tissue washes were done in PBST.

Primary antibodies used were chicken anti-GFP (Abcam #13970, 1:2000), guinea pig anti-Dpn 1:10,000 (Caygill and Brand, 2017) and rat anti-Elav 1:500 (Developmental Studies Hybridoma Bank [DSHB]). Secondary antibodies used include Alexa Fluor 488, 545, and 633 at a concentration of 1:200 (Life Technologies) and tissue was incubated for 1.5 hr at room temperature. Tissue was mounted on standard glass slides in Vectashield Mounting medium (Vector Laboratories). Brains were imaged using a Zeiss LSM 510 microscope. Analysis of acquired images was done using Fiji (Schindelin et al., 2012).

Live imaging of GFP in adult flies and larvae was performed using a Nikon SMZ 1500 microscope. In larvae, to identify earliest auxin induction effects, animals were then placed on 1 mM auxin-supplemented food and imaged at 0, 1, 5, 10, and 25 hr intervals. Adults were placed on food containing varying auxin concentrations for 24 hr prior to imaging.

Live GFP levels were quantified in six animals per condition (including negative and positive controls). Binary images were created by using a threshold (Outsu’s thresholding) and the Wand Tool in Fiji was used to trace the GFP-positive area as region of interest (ROI). The ROI corresponded to abdomens in adults or whole larvae. Mean pixel intensity per ROI was calculated using the Measure plugin in Fiji for a total of six biological replicates. Statistical significance analysed using one-way ANOVA (normally distributed) or Kruskal-Wallis test with Dunn pair-wise comparison (non-normally distributed). Parametric t-test or non-parametric Wilcoxon tests were used to compare individual conditions with respective controls. Statistical tests were completed in R (v3.6.3) and plots were performed using the software GraphPad Prism version 9 for Windows.

All animals were kept at 25°C. To determine the effects of auxin on larval development, 10 mated Canton-S flies were allowed to lay eggs on fly food containing 0, 1, 5, or 10 mM of auxin for a maximum of 24 hr, or until ~50 eggs were counted. Five replicates were obtained for each concentration, except for 0 mM where four replicates were used. Daily emergence of L3 larvae and presence of pupae were recorded to determine the time required for egg-to-L3, and egg-to-pupae development for each concentration.

Adult survival assays were performed on c564-GAL4, UAS-nls-GFP/AGES flies on three replicates, with 20 flies per replicate. Male and female animals were separated 5 days post-eclosion. Each replicate was continuously exposed to 0, 5, and 10 mM auxin food. Death was scored daily until all flies were deceased. Statistical tests and plots were performed using the software GraphPad Prism version 9 for Windows.

In order to quantify whether levels of GFP mRNA expression changed in the presence of auxin, the c564-GAL4, UAS-nls-GFP/AGES flies were placed on 0, 5, and 10 mM auxin-supplemented food for 24 hr. The c564-GAL4, UAS-nls-GFP/+ flies were used as a positive control whilst the AGES flies were used as a negative control (lacking both the GAL4 driver and the UAS-nls-GFP). The 0 mM was included to investigate whether there was any leaky GAL4 activity in absence of auxin.

Total RNA was extracted from whole adult flies, three per replicate per condition, using a standard TRIzol extraction protocol (Scientific, 2016). DNA was degraded using RNAse-free DNAse (Thermo Scientific) and cDNA synthesis was performed using the iScript cDNA synthesis kit (Bio-Rad), following manufacturer’s instructions. Real-time qPCR was performed with iTaq Universal SYBR Green Supermix (Bio-Rad) using the StepOnePlus Real-Time PCR System (Applied Biosystems).

The gene RpL4 (Ribosomal Protein L4) was used as reference gene. The GFP mRNA expression levels were calculated using the ΔΔCt method (ΔCt = Ct [mean Ct of reference gene] – Ct [target]; ΔΔCt = ΔCt [target] − mean ΔCt [control]) (Livak and Schmittgen, 2001). Target refers to the auxin concentrations (0, 5, 10 mM) and control refers to the AGES flies. Primers used: RpL4_FW - 5’-TCCACCTTGAAGAAGGGCTA-3’, RpL4_RV – 5’-TTGCGGATCTCCTCAGACTT-3’, GFP_FW – 5’-GAGCTGTACAAGAGCAGGCA-3’, GFP_RV – 5’-GTTGACGGCGTTTTCGTTCA-3’. Statistical significance analysed Kruskal-Wallis test with Dunn pair-wise comparison in R (v3.6.3).

We evaluated circadian locomotor rhythms using the DAM system (Trikinetics). Male and female flies of the indicated genotype were entrained in a 12:12 LD cycle prior to loading individual 5- to 7-day-old flies into glass DAM tubes containing control or experimental diets. For AGES flies, the control diet consisted of standard DAM food. For GeneSwitch flies, the control diet consisted of 1% ethanol vehicle in standard DAM food. Flies were monitored via DAM assay for 3 days in 12:12 LD at 25°C, followed by 8 days in constant darkness (DD). Circadian locomotor parameters were analysed using ClockLab software (Actimetrics) for data from days 2 to 8 of DD. Period length was determined by χ² periodogram analysis, and relative rhythm power at 24 hr was determined using FFT. FFT power and average daily activity counts for each sex were compared by two-way ANOVA by genotype and food substrate, with Tukey’s post hoc comparisons. Because period length could not be estimated for arrhythmic flies, period length was compared by Student’s t-test for flies of the same genotype on different food substrates.