Ectopic sodium channel expression decreases excitability of Drosophila Kenyon cells

Abstract

Neurons stabilize their physiological properties in part by homeostatic compensation between different ion channel conductances. However, little is known about this process in the central brain in vivo. We studied this problem in Kenyon cells, the third-order olfactory neurons in the fruit fly Drosophila that store olfactory associative memories. We investigated whether Kenyon cells regulate their excitability homeostatically by testing how their activity is affected by ectopic expression of the bacterial voltage-gated sodium channel NaChBac, a manipulation previously reported to increase neuronal excitability in other systems. Surprisingly, NaChBac expression decreases Kenyon cell excitability. Whether expressed constitutively (throughout development) or only in the adult, NaChBac expression in Kenyon cells suppresses Kenyon cell spiking, reduces odour-evoked calcium influx, and prevents olfactory aversive conditioning. However, odour-evoked calcium influx in Kenyon cell axons and dendrites (but not somata) is normal after 4-day adult-only NaChBac expression (but not 2-day adult-only or constitutive expression), suggesting limited homeostatic regulation of calcium influx that is prevented by developmental NaChBac expression. NaChBac expression also decreases expression of endogenous voltage-gated sodium channels (Para) in the spike initiation zone, suggesting homeostatic regulation of sodium influx. Indeed, a compartmental model best fits the data when the exogenous NaChBac conductance is accompanied by a decrease in endogenous sodium conductance. These results suggest that manipulating neuronal activity with ion channels can have unexpected effects depending on compensatory plasticity. KEY POINTS: Neurons stabilize their physiological properties through compensation between key ion channels, but little is known about this process in the central brain in vivo. Here we test homeostatic compensation in vivo in Drosophila Kenyon cells, the neurons that store olfactory associative memories, by ectopically expressing NaChBac, a bacterial voltage-gated sodium channel commonly used to increase neuronal excitability. Surprisingly, NaChBac expression in Kenyon cells decreases their excitability: it reduces expression of endogenous sodium channels, prevents spiking, reduces odour-evoked calcium influx, and impairs learning. The electrophysiological phenotype is reproduced in a compartmental model. However, odour-evoked dendritic/axonal calcium influx (but not spiking or learning) returns to normal if NaChBac is expressed for 4 days only in adults (not during development), suggesting limited homeostatic regulation of voltage-gated calcium influx. These results show that manipulating ion channels can have unexpected effects depending on homeostatic compensation.

Keywords: Kenyon cell; NaChBac; drosophila; homeostatic plasticity; ion channel; mushroom body; sodium channel.

Figure 1. NaChBac expression decreases Kenyon cell excitability and odour responses

A, maximum intensity Z‐projections of confocal stacks of NaChBac‐GFP expression in Kenyon cells driven by R13F02‐GAL4, through the whole mushroom body (upper) or just the calyx and cell bodies (lower). Scale bar: 50 µm. Dashed lines outline the lobes (upper) and calyx (lower). B, current‐clamp traces (baseline‐subtracted) of γ Kenyon cells with 1 s current injection from −40 to +100 pA with 10 pA intervals, with or without NaChBac expression. The example cells shown here and in panel C have the following estimated parameters: membrane capacitance, 17.2 pF (control), 18.3 pF (NaChBac); input resistance, 1.36 GΩ (control), 0.99 GΩ (NaChBac). C, voltage‐clamp traces (baseline‐subtracted, but not leak‐subtracted) of γ Kenyon cells with 1 s voltage steps starting at −100 mV going to −100 to +30 mV with 10 mV intervals, with or without NaChBac expression. The capacitive transients have been truncated. D, spike rate at each level of current injected. Control, n = 7 (black); NaChBac, n = 7 (red). E, peak spike rate for each cell. F, median membrane potential during the current pulse at each level of current injected (excluding the first and last 50 ms). G, input resistance for each cell. H, peak slow inward current density in the first 100 ms of each voltage step. The steady‐state current was subtracted, and the traces were smoothed by a 50 ms moving median to remove the fast inward currents. Control, n = 4; NaChBac, n = 7. I, maximum peak inward current density across voltages from (H). Absolute peak current amplitudes were (mean ± s.d.): control, −7.0 ± 5.0 pA; NaChBac, −124.0 ± 162.6 pA. Membrane capacitances were (mean ± s.d.), control, 14.7 ± 4.4 pF; NaChBac, 16.2 ± 3.6 pF. J, GCaMP6f responses of γ Kenyon cells to 5 s pulses of isoamyl acetate (black bar), quantified at right as the mean ∆F/F during the odour. Top panels: Calyx responses. Control, n given as number of hemispheres (number of brains) = 8(8); NaChBac, n = 8(8). Bottom panels: Lobe responses. Control, n = 9(8); NaChBac, n = 11(8). Traces are means ± SEM. In (B–J), γ Kenyon cells were labelled with GFP or GCaMP6f using the mb247‐GAL4 driver, with expression of lexAop‐GAL80 to restrict expression to the γ lobe driven by R44E04‐LexA. All graphs show mean ± 95% confidence interval. **P < 0.01, ***P < 0.001, Mann–Whitney test. See Table A3 for details of statistical tests.

Figure 2. Odour‐evoked calcium influx is reduced by constitutive or 2‐d, but not 4‐d, NaChBac expression

A, constitutive expression of NaChBac reduces odour‐evoked calcium influx in Kenyon cells; all Kenyon cells labelled with GCaMP6f driven by R13F02‐GAL4. Left: γ lobe response to 5 s isoamyl acetate (black bar). Right: odour responses for all areas quantified as mean ∆F/F during the odour presentation. Diagrams depict the locations analysed for the calyx and each axonal lobe. Control, n = 16(11), 19(11), 12(7), 19(11), 13(7), 14(7); NaChBac, n = 19(10), 17(10), 21(13), 18(10), 17(13), 20(13). B, schematic of temperature‐sensitive regulation by GAL80^ts. At 18°C, GAL80^ts blocks GAL4 activity, but at 31°C, GAL80^ts is inactivated, releasing GAL4 to drive transcription of the target transgene. C, schematic of the auxin‐inducible gene expression system (AGES). Without auxin present (left), AID‐tagged GAL80 blocks GAL4 activity. With auxin present (right), AID‐tagged GAL80 is targeted for degradation through ubiquitination (Ub), releasing GAL4 to drive transcription of the target transgene. D, calcium responses in all Kenyon cells with GCaMP6f expression driven by R13F02‐GAL4. Cells conditionally expressed NaChBac for either 2 or 4 days using GAL80^ts (top two rows) or auxin‐inducible degradation of AID‐GAL80 (bottom two rows). Left panels: Schematics of the experimental timeline, indicating housing temperature (for GAL80ts) or auxin presence (for AID‐GAL80), and the time of imaging (arrow). Centre panels: γ lobe responses to 5 s isoamyl acetate (black bar) (see Fig. A6 for all lobes). Right panels: odour responses for all areas quantified as mean ∆F/F during the odour presentation. 2 d GAL80^ts, Control n = 13(10), 16(10), 22(13), 16(10), 22(13), 22(13), NaChBac n = 8(5), 8(5), 16(12), 8(5), 16(12), 16(12). 4 d GAL80^ts, Control n = 21(12), 20(12), 22(13), 20(12), 23(13), 23(13), NaChBac N = 16(8), 15(8), 17(9), 16(8), 17(9), 17(9). 2 d AID‐GAL80, Control n = 19(11), 19(11), 19(11), 18(11), 19(11), 19(11), NaChBac n = 16(12), 17(12), 17(12), 17(12), 17(12), 17(12). 4 d AID‐GAL80, Control n = 15(9), 15(9), 15(9), 15(9), 15(9), 15(9), NaChBac n = 16(9), 16(9), 16(9), 16(9), 16(9), 16(9). E, summary of the main effect of genotype (control vs. NaChBac) from mixed‐effects model analysis of panel D and Fig. A6. The graph shows the mean difference in odour response between control and NaChBac flies across all lobes (negative means NaChBac responses were smaller), with 95% confidence interval, in the different timings and methods of NaChBac expression described by the labels at left and depicted in the schematics at right. Graphs show means ± 95% confidence intervals. *P < 0.05, **P < 0.01, ***P < 0.001, mixed‐effects analysis with Geisser‐Greenhouse correction and Šidák's multiple comparison test. Odour evoked calcium traces are means ± SEM. See Table A3 for details of statistical tests.

Figure 3. Adult‐only NaChBac expression decreases Kenyon cell excitability

A, spike rate at each level of current injected, in control Kenyon cells (black) or Kenyon cells expressing NaChBac (red) for 2 d (upward‐pointing triangles) or 4 d (downward‐pointing triangles). n = 15, 9, 11, 9. B, peak spike rate for each cell. C, median membrane potential during the current pulse at each level of current injected (excluding the first and last 50 ms). n = 15, 9, 11, 9. D, input resistance for each cell. n = 15, 8, 11, 9. E, peak inward current density at each voltage step (subtracted steady‐state current). F, maximum peak inward current density across voltages from (E). Absolute peak current amplitudes were (means ± s.d.): control 2 d, −7.3 ± 6.6 pA; NaChBac 2 d, −294.6 ± 180.0 pA; control 4 d, −7.3 ± 9.0 pA; NaChBac 4 d, −221.0 ± 123.3 pA; membrane capacitances were (means ± s.d.):, control 2 d, 19.3 ± 9.7 pF; NaChBac 2 d, 13.6 ± 5.4 pF; control 4 d, 13.0 ± 5.1 pF; NaChBac 4 d, 13.9 ± 3.4 pF. G, current injection required to induce a NaChBac potential. If the required current is 0, that means that NaChBac potentials occurred at the end of a hyperpolarizing current injection. Cells that did not show NaChBac potentials are not included. n = 9, 9, 5. H, average depolarization for the biggest NaChBac potential for each cell, over the full 1 s current injection (see Methods). Cells that did not show NaChBac potentials are not included. n = 9, 9, 5. I, mean normalized NaChBac current in Kenyon cells expressing NaChBac for 2 d (blue), 4 d (purple), or constitutively (green) (see Methods). Traces are means ± SEM. J, decay time constants for NaChBac currents (fitted to a single exponential decay function). n = 8, 7, 4. Graphs show means ± 95% confidence interval. **P < 0.01, ***P < 0.001, Kruskal–Wallis ANOVA with Dunn's multiple comparisons test. See Table A3 for details of statistical tests.

Figure 4. NaChBac expression increases sparseness of Kenyon cell odour responses

A, schematics show the housing temperatures or auxin exposure applied in each condition. NaChBac and GCaMP6f were expressed under the control of R13F02‐GAL4: (left to right) constitutively, for 2 or 4 d (by GAL80^ts) or for 2 or 4 d (by AGES). The same left‐to‐right order of conditions is used in every panel. B, activity maps of Kenyon cell soma responses to isoamyl acetate, with and without NaChBac expression. Greyscale, baseline fluorescence; false colouring, ∆F/F. Scale bar: 10 µm. C, mean population sparseness of Kenyon cell soma odour responses for the seven odours listed in panel D, with (red) and without (black) NaChBac expression, under the experimental conditions shown in panels A and B. D, matrices show pairwise correlations between Kenyon cell soma odour responses, to: (1) isoamyl acetate, (2) 3‐octanol, (3) butyl acetate, (4) ethyl butyrate, (5) apple cider vinegar, (6) δ‐decalactone, (7) methylcyclohexanol. E, mean inter‐odour correlations from (D), with (red) and without (black) NaChBac expression. F, traces show Kenyon cell soma odour responses averaged across all cells and all odours listed in (D). Black bars indicate odour presentation. Traces are means ± SEM. G, Responses in (F) quantified as the mean ∆F/F during the odour. Graphs show means ± 95% confidence interval. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001, Mann–Whitney's test. From left to right, n = 13(9), 13(10), 19(13), 22(12), 14(9), 14(10), 14(8), 12(8), 10(7), 15(9). See Table A3 for details of statistical tests.

Figure 5. NaChBac expression prevents olfactory learning

A, example traces of individual fly trajectories before and after training. Flies were presented with a choice between air (Ø) or vinegar (1:100; ‘CS+’, orange) on each side of the chamber for 2 min, to assess their naïve preference. Two minutes later, vinegar was paired with electric shock for 1 min; 2 min later, air versus vinegar was presented again for 2 min. Fly position within the chamber (horizontal dimension) is plotted against time (vertical dimension). The dotted line shows the centre of the chamber. B, the change in the percentage of decisions made in favour of the CS+ odour. n, given as number of flies [number of experiments]: R13F02‐GAL4: 124[9], UAS‐NaChBac: 109[7], R13F02‐GAL4, UAS‐NaChBac: 140[10]. C, percentage of decisions in favour of the CS+ (vinegar) before training, with constitutive expression of NaChBac in Kenyon cells driven by R13F02‐GAL4. D, as in (A), but with 2‐d or 4‐d expression of NaChBac. AID‐GAL80: GAL80 tagged with auxin‐inducible degron (AID). E, as in (B), but with 2‐d or 4‐d expression of NaChBac. N = 119[8], 61[5], 120[10], 133[10]. F, as in (C), but with 2‐d or 4‐d expression of NaChBac. **P < 0.01, ***P < 0.001, Kruskal–Wallis ANOVA with Dunn's multiple comparisons test. Innate decision scores do not differ significantly between groups (Kruskal–Wallis test with Dunn's multiple comparison test). # significantly different from 0 (B,E) or 50 (C,F), P < 0.05, Wilcoxon's signed‐rank test with Holm–Bonferroni correction. Box plots show median and interquartile range and whiskers show the min–max range. See Table A3 for details of statistical tests.

Figure 6. NaChBac expression reduces Para expression

A, schematic of FlpTag cassette integration into a target gene. The FlpTag cassette is initially inserted in the reverse orientation relative to the target gene. Following cell type‐specific Flp expression, the cassette is flipped, so that GFP is spliced into the mRNA between two exons. B, diagram of the mushroom body highlighting the spike initiation zone (SIZ), where Para is most highly concentrated. C, example maximum intensity Z‐projections of control and Kenyon cell > NaChBac mushroom bodies. Constitutive NaChBac expression was driven by R13F02‐GAL4; Para‐FlpTag (green) was flipped by mb247‐LexA driving FLP; the direct fusion mb247‐dsRed was included as an anatomical marker (magenta). Dashed white line outlines the mushroom body (excluding the cell bodies). Scale bar: 50 µm. The curly bracket indicates the spike initiation zone (SIZ). D, schematic illustrating the mask and skeleton (blue dashed line) used for analysing the mushroom body. The black dashed line marks the boundary between the calyx and the peduncle. Lower panel: diagram showing how the mushroom body structure was divided into evenly spaced segments for quantifying the localization of Para. Right panel: GFP/dsRed ratio in each segment of the mushroom body, normalized to the average ratio across all segments in controls. The shaded area highlights the regions of the calyx and peduncle where Para is concentrated. Dashed line indicates the cut off between the calyx and peduncle. Control n = 25(17), NaChBac n = 42(28). E, quantification of the ratio of GFP/dsRed signal averaged across the 10 segments in the shaded area of panel (D) (normalized to the average across all segments in the control, mean ± 95% confidence interval). ****P < 0.0001, Mann–Whitney test. F, mean fluorescence (±95% confidence interval) of GFP and dsRed signal within the shaded area of panel (D), normalized to the average across all segments in the control. ***P < 0.001, Kruskal–Wallis test with Dunn's multiple comparison test. G, quantification of the ratio of GFP/dsRed signal averaged across the 10 segments of the peduncle for 2 days and 4 days acute NaChBac expression (normalized to control, mean ± 95% confidence interval). **P < 0.01, Mann–Whitney test. 2 day expression n, control = 9(5), NaChBac = 10(6). 4 day expression n, control = 18(10), NaChBac = 23(14). H, mean normalized fluorescence (±95% confidence interval) of GFP and dsRed signal within the 10 segments of the peduncle for 2 days and 4 days acute NaChBac expression. *P < 0.05, Kruskal–Wallis test with Dunn's multiple comparison test. See Table A3 for details of statistical tests.

Figure 7. Action potentials and fast inward currents are smaller in NaChBac‐expressing cells

A, action potentials in control versus NaChBac‐expressing cells (taken from the current‐clamp traces in Fig. 1B at +30 and +90 pA, respectively; dashed lines show −23 mV and +4 mV, respectively). B, somatic voltage at the time of spike onset (peak second derivative of voltage in the 5 ms before the peak voltage), for the spike with the lowest somatic voltage at spike onset for each cell. The data for 2‐d and 4‐d NaChBac expression are combined (2 d, upward‐pointing triangles; 4 d, downward‐pointing triangles). *P = 0.043, ***P < 0.001, 2‐way ANOVA with Sidak's multiple comparisons test. n (here and panels E–H) = 26, 5, 7, 5. 2‐d and 4‐d data are with AID‐GAL80 from Fig. 3; constitutive data are from Fig. 1. C, spike waveforms for control (black) and NaChBac‐expressing (red) cells. For each cell, the 20 ms around the spike peak were averaged for each current step, and the trace from the current step with the most prominent peak (greatest difference between peak and mean voltage) is shown, to exclude other spikes from the trace window. Horizontal line shows the voltage at time of spike onset. D, as with (C), but with the derivative of the voltage, showing the slope of the spike waveform. Horizontal line shows dV/dt = 0. E, spike height from (C), taken as the difference between the peak voltage and the voltage at the time of spike onset (horizontal line in (C)). **P = 0.00235, 2‐way ANOVA with Sidak's multiple comparisons test. Main effect of genotype, P = 0.00555; no significant interaction between genotype and acute versus constitutive expression (P = 0.149). F, Peak dV/dt from (D). **P = 0.00407, 2‐way ANOVA with Sidak's multiple comparisons test. Main effect of genotype, P = 0.00508; no significant interaction between genotype and acute versus constitutive expression (P = 0.213). G, as with (E), for 2 or 4 days NaChBac expression, but using spike waveforms averaged from spikes whose voltage at spike onset was at different 10 mV bins (e.g. spike height at 0 mV is the height of the waveform averaged across all spikes whose voltage at spike onset was between −5 and +5 mV). H, as with (G), but for peak dV/dt. I, fast inward currents in control versus NaChBac‐expressing cells, taken from Fig. 1C at ‐30 mV and +10 mV, respectively. J, waveforms of fast inward currents for control (black) and NaChBac‐expressing (red) cells. The waveform from the command voltage with the biggest current for each cell is shown. K, amplitudes of fast inward currents from (J) at different command voltages. L, amplitudes of fast inward currents from (J) at the command voltage with the highest current amplitude for each cell. n = 25, 4, 4, 4. *P = 0.0129, Kruskal–Wallis test with Dunn's multiple comparison test. M–O, fast inward current amplitude from (L) is strongly correlated with spike height (M), peak dV/dt (N), and total number of spikes across current steps +10 to +100 pA (O). P, fast inward current amplitude is not correlated with the residual NaChBac current in the last 100 ms of the voltage step. In O and P, cells where we detected no fast inward currents are plotted at 0 on the y‐axis. n = 37 (M,N); 51 (O,P). M–O: Filled symbols, NaChBac current; empty symbols, no NaChBac current. Thick lines show averages across cells; thin lines and transparent markers show individual cells. Black: control; Red: NaChBac. Upward‐pointing triangles, 2 d auxin; downward‐pointing triangles, 4 d auxin; circles, constitutive expression. Graphs show mean ± 95% confidence intervals (where the error bars are very large and extend beyond the axis limits, it is because there are only two cells at that data point). See Table A3 for details of statistical tests.

Figure 8. NaChBac expression does not alter voltage‐gated steady‐state outward current amplitudes

A, example voltage‐clamp traces from control and NaChBac‐expressing cells with AID‐GAL80 and 2 d auxin exposure. The baseline and leak currents were subtracted and the traces were smoothed by a 50 ms moving median filter to remove the fast inward currents shown in Fig. 7. Both the control and NaChBac cell show a steady‐state outward current at command voltages above −30 mV. B, median current in the last 100 ms of the voltage step (shaded area in panel A), with estimated leak and residual NaChBac currents subtracted, at +30 mV. Mean ± 95% confidence interval; P = 0.212, Kruskal–Wallis test. C, as in B, but at all voltage steps. Circles/thick lines show the mean ± 95% confidence interval; thin lines show individual cells. See Table A3 for details of statistical tests.

Figure 9. Compartmental modelling reveals the dual effects of NaChBac on Kenyon cell excitability

A, schematic of the compartmental model of a γ Kenyon cell reconstructed from the Flywire connectome. The model includes compartmentalized sections, soma, dendrites, spike initiation zone (SIZ) and axon. B, current‐clamp recording of wildtype (WT) Kenyon cell to varying current injections. C, simulated current‐clamp of best‐fit model for the cell in (B). D, current‐clamp recording of NaChBac‐expressing Kenyon cell. E, simulation using optimal parameter set recreating NaChBac Kenyon cells response. F, simulation using optimal parameter set from (E) but with no NaChBac. G, simulation using average transient sodium conductance of (WT) models and optimal NaChBac conductances from (E). H–J, NaChBac expression induces a depolarization block. Simulated activation and inactivation gates during 20 pA current injection for the transient sodium channel (NaT), persistent sodium channel (NaP), potassium channel (Ks) and bacterial sodium channel (NaChBac) are shown for (H) the wildtype control model (same parameters as in C), (I) the NaChBac model with reduced Para conductance (same as in (E)) and (J) the NaChBac model without a reduction in Para conductance (same as in (G)). K, comparison of transient sodium (Para) conductance in best‐fit WT and NaChBac‐expressing models. Lines represent means. *P < 0.05, unpaired t‐test. L, simulated number of action potentials in response to varying somatic current injections comparing WT and NaChBac‐expressing cells with and without reduced Para expression. Error bars indicate standard deviation. M, simulated number of action potentials in response to varying current injections with varying conductance of NaChBac within a single WT model (same model parameters as in (B)). Initial resting membrane potential for WT cell and simulation (B–C) is −60 mV and −70 mV for NaChBac‐expressing cell (D) and −69 mV for the corresponding models (D–G).

Figure A1. Correlations between electrophysiological parameters

A, maximum peak slow inward current plotted against input resistance, including cells excluded from the main figures due to low input resistance (marked by ‘x’s). Black, control; red, NaChBac. Circles, constitutive expression; upward‐pointing triangles, 2‐d expression, downward‐pointing triangles, 4‐d expression. Empty symbols: No NaChBac current; filled symbols, NaChBac current; lightly filled symbols, no voltage‐clamp recording. B, maximum peak slow inward current density (from Fig. 1I, F) plotted against input resistance. C, maximum spike rate plotted against input resistance. Note that control cells spike more than NaChBac cells at every input resistance level. D, maximum peak slow inward current plotted against uncompensated series resistance estimated from the capacitive transient. Cells with high series resistance tend to have small NaChBac currents. E, command voltage giving the largest slow inward current, plotted against uncompensated series resistance. Cells without NaChBac currents are distributed randomly along the y‐axis, but cells with NaChBac currents consistently have their peak slow inward current at command voltages around −70 to −30 mV, regardless of the uncompensated series resistance. F, command voltage giving the largest slow inward current, plotted against the predicted voltage error from series resistance, calculated according to V _error = I _NaChBac*R_s. G, input resistance plotted against uncompensated series resistance to test whether high series resistance leads to overestimating the input resistance. H, maximum spike rate plotted against series resistance. NaChBac recordings have few spikes regardless of the series resistance. I–J, spike height (I) and peak dV/dt (J) plotted against series resistance. Spikes in NaChBac cells have reduced amplitude and upstroke slope even when considering only cells with low series resistance. r values are Pearson's correlation coefficients for the NaChBac cells only.

Figure A2. Heterogeneity in NaChBac‐GFP fluorescence across Kenyon cells

A, example maximum intensity Z‐projections of R13F02‐GAL4 driving NaChBac‐GFP (green) and mCherry (magenta) expression. Scale bar: 50 µm. B, fluorescence intensities of NaChBac‐GFP and mCherry were quantified in a random selection of Kenyon cells (n = 12 cells from eight brains per group). Graph shows mean ± 95% confidence interval. C, NaChBac‐GFP fluorescence intensity plotted against mCherry in the same cells to assess relative expression levels. Each point represents a single Kenyon cell, NaChBac‐GFP (n = 12[8]) and mCherry (n = 12[8]); each colour represents a different brain. D, normalized average fluorescence per fly from panel (B) for NaChBac‐GFP and mCherry. Graph shows mean ± standard deviation. E, quantification of the coefficient of variation within flies from panel (B) for NaChBac‐GFP and mCherry fluorescence intensities. Graph shows mean ± standard deviation.

Figure A3. Excluded data points from Figs 1 and 3

Cells excluded from Figs 1 and 3 due to input resistance less than 0.8 GΩ are shown here marked by ‘x’s, to illustrate that our results are unaffected by including the excluded cells. Panel letters follow those of Figs 1 and 3; see main figure legends. See Table A3 for details of statistical tests and Dataset S1 for raw data.

Figure A4. R13F02‐GAL4 labels approximately all Kenyon cells

A, example images showing maximum intensity Z‐projections of OK107 > Stinger‐GFP and R13F02‐GAL4 > Stinger‐GFP. Scale bar = 15 µm. B, Kenyon cell count for (A), comparing OK107 (N = 6) versus R13F02‐GAL4 (N = 6) (means ± 95% confidence intervals).

Figure A5. GAL4 efficacy with temporally limited suppression by GAL80^ts or AID‐GAL80

Fluorescence levels in the mushroom body were compared in R13F02‐GAL4 > UAS‐GCaMP6f flies under the different conditions shown by the open and closed circles. N = 8, 9, 12, 8, 12, 13, 8, 10. Graph shows means ± 95% confidence intervals. *P < 0.05, **P < 0.01 ***P < 0.001, ****P < 0.0001, Kruskal–Wallis, Dunn's multiple comparisons test.

Figure A6. Odour‐evoked calcium influx reduced by constitutive or 2‐d, but not 4‐d, NaChBac expression: additional traces and graphs supporting Fig. 2

Odour response traces to isoamyl acetate were recorded from the calyx and each lobe of the mushroom body under various experimental conditions, all using the R13F02‐GAL4 driver. Left panels indicate experimental protocol, where the shaded area shows NaChBac expression: constitutively at 25°C; 18°C during development then 2 days at 31°C with GAL80^ts; 18°C then 4 days at 31°C with GAL80^ts; 2 days auxin with AID‐GAL80; 4 days auxin with AID‐GAL80; 18°C then 8 days at 31°C GAL80^ts, control n = 8(5) 8(5) 6(5) 6(5) 6(5) 6(5) NaChBac n = 13(8) 16(8) 12(6) 12(6) 12(6) 12(6); 18°C then 4 days at 31°C without GAL80^ts, control n = 4(3) 7(4) 8(4) 7(4) 10(5) 8(4), NaChBac n = 5(3) 4(2) 6(3) 5(3) 6(3) 6(3); 18°C until 2 days after eclosion and then 2 days at 31°C with GAL80^ts, control n = 8(6) 9(7) 12(7) 9(5) 14(7) 13(7), NaChBac n = 10(5) 6(5) 12(7) 6(6) 12(6) 11(7). Odour stimulus duration is indicated by the black bar (5 s). Graphs show means ± 95% confidence intervals. *P < 0.05, **P < 0.01 and ***P < 0.001, Mixed‐effects analysis, Geisser–Greenhouse correction and Sidak's multiple comparison test. Traces are means ± SEM.

Figure A7. Multi‐parameter comparison of 2 versus 4 day NaChBac expression

A–D, rationale for multi‐parameter analysis: if data points exist in a parameter space where some parameters are highly correlated, two groups may not appear different when examining each parameter individually, but may be clearly different when plotting parameters against each other. A, in this theoretical example, group A has parameter x sampled from the uniform distribution U(1,11), and parameter y is the same as x plus Gaussian noise: y = x + N(μ = 0, σ = 0.2). Group B has x = U(1,11), y = 0.9*x + N(μ = 0, σ = 0.2). Groups A and B are obviously different when both parameters are plotted together. PC1 and PC2 are the two principal components, PC1 = (1,1), PC2 = (−1,1). B, However, Groups A and B are very overlapping when considering parameters x and y on their own. C, the difference becomes clear and statistically significant when plotting y/x, which captures the fact that Group A has y = x and Group B has y = 0.9*x. D, the difference is also clear when projecting the data onto the second principal component (P‐values for B–D, Mann–Whitney test). E, cells with higher input resistance tended to have stronger NaChBac current density (r = 0.34). Magenta upward‐pointing triangles, 2 day expression; blue downward‐pointing triangles, 4 d expression. F, the few cells that spiked with NaChBac expression tended to have smaller NaChBac currents (correlation, spike rate vs. NaChBac current r = −0.21). G, applying the approach from panel (C): normalizing NaChBac current density to input resistance. P = 0.69, Mann–Whitney test. H, scores for each principal component (PC), for the parameters from (D,E). Principal components analysis was performed after normalizing the data to have mean 0 and standard deviation 1 for each parameter. The PC weights were, in the order (input resistance, NaChBac current density, spike rate): PC1 (−0.70, 0.71, 0.02); PC2, (0.36, 0.33, 0.87); PC3, (0.61, 0.62, −0.49). P = 0.0721, 0.463, 0.281, Mann–Whitney test. Graphs show means ± 95% confidence intervals. See Table A3 for details of statistical tests.

Figure A8. Systematic expression of NaChBac alone does not recreate physiological response of NaChBac‐expressing Kenyon cells

A, current‐clamp simulation, recording somatic responses to varying current injections using optimal parameter set of WT model from Fig. 9B. B–L, current‐clamp simulations as in (A) but with increasing expression of NaChBac throughout all compartments of the model, without decreasing Para. Initial resting membrane potential for all models is –60 mV.