Short-hairpin RNA interference (RNAi) strains targeting unc79 (HMC03213) and Nlf-1 (HMS03014) were generated by the Drosophila RNAi Screening Center (Ni et al., 2011). We generated na short hairpin RNAi (termed na139 RNAi) by annealing forward (5′-CTA GCA GTG GTG CAG CTA GAA ACA AAT AGT TAT ATT CAA GCA TAT TTG TTT CTA GCA GCT GCA CCG CG-3′) and reverse (5′-AAT TCG CGG TGC AGC TGC TAG AAA CAA ATA TGC TTG AAT ATA ACT ATT TGT TTC TAG CAG CTG CAC CAC TG-3′) oligonucleotides corresponding to the 5′ UTR of na transcript (Integrated DNA Technologies). The annealed product was cloned into the p-VALIUM20 vector (Ni et al., 2011) and introduced into the attP2 locus using PhiC-31 integration (Bestgene Inc.). Other Drosophila strains have been previously described: tubulinGAL80^ts (McGuire et al., 2003), daughterless-GAL4 (Aldrich et al., 2010), Clock8.0GAL4 (Glossop et al., 2003), elavGAL4 (Lin and Goodman, 1994), UAS-na U4 (Lear et al., 2005), UAS-Nlf-1-V5 (Flourakis et al., 2015), and na^har (Nash et al., 2002).

For RNAi experiments, daughterless-GAL4; tubulin-GAL80^ts females were crossed to males containing either UAS-RNAi or control insertions (attP2 or attP40) (Groth et al., 2004), and crosses were maintained at either 29°C or 19°C. For developmental knockdown experiments, progeny from 29°C crosses were transferred to 19°C within 2 days of eclosion, 2–7 days prior to the start of the 19°C behavioral assay in order to minimize concerns over RNAi persistence (Bosch et al., 2016). Adult-specific knockdown crosses were raised at 19°C and 0–8 day old progeny were shifted to 29°C on the 1st day of the behavioral assay. For ‘non-shifted’ conditions (19°C to 19°C or 29°C to 29°C), 0–8 day old progeny were loaded into behavioral assays. In NA rescue experiments, na^har GAL4 females were crossed to tubGAL80^ts and/or UAS-na males. In some experiments, males also contained UAS-Nlf-1 RNAi or UAS-Nlf-1-V5 transgenes (see Results). Crosses were again raised at either 19°C or 29°C. For development-specific rescue experiments, 0–2 day old 29°C progeny were shifted to 19°C 0–6 days prior to the behavioral assay. For adult-specific behavioral rescue experiments, 0–8 day old 19°C progeny were shifted to 29°C on the day of the behavior assay. Progeny loaded into non-shifted rescue experiments (19°C to 19°C and 29°C to 29°C) were 0–8 days old.

For 19°C behavioral assays, locomotor activity levels of adult male Drosophila were measured for 5 days of 12 h light: 12 h dark conditions followed by 7 days constant darkness (DD) using the Drosophila Activity Monitor system (Trikinetics). For 29°C assays, activity levels were measured for 5–6 days of 14 h light: 10 h dark conditions followed by 7 days DD. For DD rhythmicity analyses, chi-squared periodogram analyses were performed on individual flies over 7 days using ClockLab analysis software (Actimetrics). Flies were considered rhythmic if the chi-squared power was > = 10 above significance, using a 0.01 confidence interval (Lear et al., 2005). To produce light:dark (LD) activity profiles, activity levels of individual flies were normalized and averaged within genotypes over the last 4 days of LD conditions. In one experiment, the four LD days analyzed were non-consecutive due to a ∼12 h disruption in data collection (see Figure legends). LD-DD daily activity profiles include normalized data from the last day of LD conditions followed by the first 3 days of DD conditions. LD and daily activity profiles were generated using the Excel-based program Counting Macro (Pfeiffenberger et al., 2010). Morning activity index (MI) and evening activity index (EI) were calculated from LD data by determining the largest 2–4 h increase in normalized average activity of each genotype over the last 5 h of dark phase (MI) or the last 7 h of light phase (EI) (Lear et al., 2013). For DD MI and EI calculations, normalized activity levels were averaged over three consecutive 30-min time points. For MI, maximum average activity of each genotype was then determined for three consecutive 30-min time points over the 8 h (DD Day 1) or 9 h (DD Day 2) surrounding Circadian Time (CT) 0. For EI, maximum average activity was determined for five consecutive 30-min bins over the 10 h (DD Day 1) or 12 h (DD Day 2) surrounding CT 12. In 29°C assays, the minimum average activity over a 3-h period was determined both before and after the observed maximal morning or evening activity peak. MI and EI were then obtained by subtracting the average of these minimum values from the corresponding maximum activity value (Lear et al., 2009). In 19°C assays, control strains exhibit a less consistent midday decrease in activity in DD conditions than what is observed at 29°C (Majercak et al., 1999). Therefore, we determined the minimum morning activity averaged over a 4-h period preceding morning peak, and the minimum evening activity averaged over a 4-h period following the evening peak.

Western blots were performed using protein extracts from adult Drosophila heads obtained during mixed light phase conditions (Zeitgeber Time 0–10 h), using methods described previously (Lear et al., 2013). For adult or developmentally restricted RNAi experiments, 0–2 day old adults were shifted from the permissive (29°C) or restrictive (19°C) temperature to the opposing condition for at least 7 days prior to extraction. Equal amounts of protein were loaded onto each gel (5–6 μg for NA and 10 μg for UNC79 gels), as determined by Bradford assay (Bio-Rad). A minimum of two biological replicates with varying lane order were performed. Protein expression levels were quantified using NIH ImageJ, with the intensity of the bands standardized to the average intensity of samples (Lear et al., 2013). To estimate protein half-life, we used data from four independent experiments in which elavGAL4 na^har; tubGAL80^ts/+; UAS-na/+ flies were exposed to the following conditions in parallel: (1) 29°C development – 29°C adult, (2) 29°C development shifted to 19°C for 5–7 days adulthood, and (3) 19°C development- 19°C adult. For these calculations, the band intensity within each experiment was standardized to the average intensity of samples 1–3 above. Protein half-life was calculated using the formula N_t = N₀ (1/2)^t/t1/2, where N₀ is the average standardized protein level in the 29°C – 29°C condition, N_t is the average protein level at time t at 19°C (t = 6, averaged from 5 to 7 days), and t_1/2 is the half-life estimate.

For DD rhythmicity data, the proportion of rhythmic flies was determined as described above, and genotypes were compared using Fisher’s exact test. LD and DD activity index values (MI/EI) were compared between two genotypes by unpaired Student’s t-test, or among three or more genotypes using ANOVA followed by Dunnett’s multiple comparison test (Graphpad Prism). For protein expression comparisons, significance was determined using unpaired Student’s t-test (Microsoft Excel).

To assess the endogenous temporal requirements for the Drosophila NA channel complex, we used the inducible GAL4 inhibitor tubulin-GAL80^ts (tubGAL80^ts) to control transgenic RNAi expression (McGuire et al., 2003; Ni et al., 2011). We combined the ubiquitous driver daughterlessGAL4 (daGAL4) with tubGAL80^ts and na or unc79 short-hairpin RNA interference (RNAi) constructs, maintaining flies at temperatures that are primarily permissive (19°C) or restrictive (29°C) for GAL80^ts function. We find that ubiquitous expression of na or unc79 RNAi throughout development and adult stages at the restrictive temperature (29°C) causes severe circadian behavioral phenotypes that are comparable to na and unc79 loss of function mutants (Nash et al., 2002; Lear et al., 2013). These phenotypes include decreased morning and evening anticipatory activity during light:dark (LD) conditions (Figures 1A–C, arrows; Supplementary Table S1, p < 0.05 for LD MI, p < 0.01 for LD EI), as well as decreased free-running rhythmicity during constant dark (DD) conditions (Supplementary Figures S1A–C and Table 1, 0–6% rhythmic, p < 0.01 compared to RNAi controls). To determine whether increased channel gene expression in adults can restore rhythmicity, we raised tubGAL80^ts RNAi crosses at 29°C and shifted adult progeny to 19°C. To minimize the effects of RNAi persistence (Bosch et al., 2016), adults were shifted to 19°C at least 2 days prior to the start of the behavioral assay. Despite this, we find that na and unc79 RNAi progeny continue to exhibit multiple circadian phenotypes throughout the 19°C adult behavior assay (5 days LD; 7 days DD). Most prominently, we observe defects in DD rhythmicity (Table 1, < 10% rhythmic, p < 0.01 relative to controls) and LD evening anticipatory activity (Figures 1D–F, gray arrows; Supplementary Table S1, EI < 0.5, p < 0.01) in the 29°C development – 19°C adult shifted condition that are comparable to phenotypes observed when flies are raised and maintained at 29°C (29°C – 29°C). Notably, we find that morning behavior is not significantly disrupted in RNAi progeny in the 29–19°C condition relative to controls (Figures 1D–F, black arrows; Supplementary Table S1, p > 0.28 for MI and EI values). Thus, our data suggest that developmental expression of NA channel subunits can contribute to adult rhythmicity, but the impact on behavioral rhythms may not be uniform.

To verify that GAL80^ts can suppress daGAL4 driven RNAi phenotypes at the presumed permissive temperature, we assessed circadian behavior in flies maintained at 19°C throughout development and adult stages. Under these conditions, unc79 and na RNAi flies exhibit rhythmicity comparable to controls. This includes similar LD anticipatory behavior (Figures 2A–C, arrows; Supplementary Table S1, p > 0.51 for MI and EI except where RNAi value exceeds control) and DD activity peaks (Supplementary Figures S2A–C, arrows; Supplementary Table S1, p > 0.18 for DD MI and EI). While DD rhythmicity is typically weaker at 19°C than at 29°C in our assays, rhythmicity measurements in the 19°C development – 19°C adult condition are similar between RNAi strains and the corresponding controls (Table 1, > = 35% rhythmic, p > 0.07 between RNAi and control strains). To determine whether adult-driven expression of NA channel subunits is required for rhythmic behavior, we raised unc79 and na RNAi flies at 19°C and shifted them to 29°C as adults. Surprisingly, we observe that these RNAi flies retain robust rhythmic behaviors, including LD and DD evening activity peaks (Figures 2D–F, gray arrows; Supplementary Figures S2D–F, gray arrows; Supplementary Table S1, p > 0.89 except when RNAi > control), and free-running DD rhythmicity (Table 1, > 90% rhythmic, p > 0.60). This suggests that sustained adult expression of NA channel subunits is not required for adult behavioral rhythms. While RNAi strains raised at 19°C exhibit significant rhythmicity throughout the 29°C behavioral assay (see Table 1), we observe some differences in DD morning behavior relative to controls. Control strains in the 19–29°C condition retain a substantial morning activity peak on the first 2–3 days of DD (Supplementary Figure S2D, black arrows; Supplementary Table S1), while both na and unc79 RNAi strains exhibit decreased morning behavior beginning on DD Day 2 (Supplementary Figures S2E,F, black arrows; Supplementary Table S1, DD Day 2 MI < = 0.31, p < 0.01 relative to controls).

Nlf-1 (aka mid1) functions as a positive regulator of NA/NALCN/NCA channel expression and/or localization. In Drosophila, it has been proposed that circadian regulation of NA channel function in adult clock neurons may be mediated through rhythmic expression of Nlf-1 (Flourakis et al., 2015). To further evaluate the requirements for Nlf-1, we again employed the tubGAL80^ts inducible RNAi strategy. As with na and unc79, Nlf-1 RNAi flies raised and assayed at the permissive temperature (19°C – 19°C) exhibit rhythmic behavior comparable to controls, including LD and DD activity peaks (Figures 3A,B, arrows; Supplementary Figures S3A,B, arrows; Supplementary Table S1, p > 0.9 or RNAi MI/EI values greater than control) and DD rhythmicity (Table 1, p = 0.49). In contrast, Nlf-1 RNAi flies raised and assessed at the restrictive temperature (29°C – 29°C) display defective rhythms (Figures 3C,D, arrows; Supplementary Figures S3C,D and Table S1, p < 0.05 except DD MI values; Table 1, p < 0.01). Moreover, like na and unc79, the adult behavioral phenotypes associated with Nlf-1 RNAi correlate mainly with the developmental temperature. Nlf-1 RNAi flies raised at 19°C and shifted to 29°C exhibit largely rhythmic behaviors, including intact evening activity peaks in LD and DD (Figures 3E,F, gray arrows; Supplementary Figures S3E,F, arrows; Supplementary Table S1), as well as robust DD rhythmicity (Table 1, 85% rhythmic). RNAi flies raised at 29°C and shifted to 19°C as adults exhibit significant defects in evening behavior (Figures 3G,H, gray arrows; Supplementary Table S1, p < 0.01 for LD EI) and DD rhythmicity (Table 1, 0% rhythmic, p < 0.01; Supplementary Figures S3G,H). Similar to unc79 and na, Nlf-1 RNAi strains shifted from 19 to 29°C as adults display a decrease in morning behavior relative to controls (Figures 3E,F, black arrows; Supplementary Figures S3E,F, black arrows; Supplementary Table S1, p < 0.05 for LD MI, p < 0.01 for DD day 2 MI). Despite this, rhythmicity over 7 days DD in Nlf-1 RNAi flies in the 19 to 29°C condition is not significantly lower than controls (Table 1; p = 0.17).

These behavioral data suggest a predominant developmental requirement for na and its key regulators. To determine how this relates to the expression of NA channel complex proteins, we performed Western blot analyses on daGAL4 tubGAL80^ts RNAi flies raised at either permissive (19°C) or restrictive (29°C) temperatures. We assessed protein levels for both NA and the auxiliary channel subunit UNC79, as antibodies to these two channel subunits have yielded consistent Western blot data in previous assays (Lear et al., 2013). We find that na, unc79, or Nlf-1 RNAi flies maintained at 19°C throughout development and adult stages retain substantial expression of both NA and UNC79, with average protein levels ≥ 78% of controls (Figure 4A). The minor decreases in protein levels observed in some RNAi samples (Figure 4A, p < 0.01 for NA levels in unc79 RNAi, others p > 0.06) may reflect residual daGAL4/ RNAi activity at 19°C and/or variability in the Western blot assay. In contrast, RNAi strains raised and maintained at 29°C exhibit major decreases in channel complex expression (Figure 4C), with minimal NA expression in all three RNAi genotypes (<4% of controls; p < 0.01), and UNC79 expression ranging from < 1% (unc79 RNAi) to 27% (Nlf-1 RNAi) of controls (p < 0.05). These results are consistent with previous findings in loss-of-function mutants and constitutive pan-neuronal RNAi (Lear et al., 2013; Flourakis et al., 2015). We find that RNAi strains raised at 19°C and shifted to 29°C for seven or more days during adulthood express NA and UNC79 at similar levels to the 19°C – 19°C group (Figure 4B; p ≥ 0.13). This indicates that substantial expression of NA channel complex proteins is retained in adults even when na, unc79, or Nlf-1 gene expression is inhibited. For RNAi strains raised at 29°C and shifted to 19°C for a minimum of 7 days, we observe defects in protein expression comparable to the constitutive 29°C group (Figure 4D). Protein levels in the 29–19°C condition are significantly lower than controls (NA < 5% of controls, UNC79 < 20% of controls, p < 0.01) and in most cases are also significantly lower than the corresponding 19°C – 19°C samples (p < 0.05 except UNC79 in Nlf-1 RNAi, p = 0.06). Taken together, these data suggest that much of the NA channel complex that functions in Drosophila adults is produced during development.

To complement the inducible RNAi approach, we also performed tissue-specific rescue of na^har mutants. For behavioral experiments, we used the broad circadian neuron driver Clock8.0GAL4 (Clk8.0GAL4) in combination with UAS-na and tubGAL80^ts. We have previously demonstrated that constitutive Clk8.0GAL4 driven expression of UAS-na promotes robust rescue of na behavioral phenotypes in LD and DD conditions (Lear et al., 2005). In the inducible scheme, we find that tubGAL80^ts is quite effective at blocking Clk8.0GAL4 UAS-na rescue of na^har mutant phenotypes at 19°C, as flies raised and assayed at 19°C are poorly rhythmic (Table 2, < 20% rhythmic, p = 0.11; Supplementary Figure S4, arrows; Supplementary Table S2; p > 0.1 except DD day 1 EI). In contrast, na^har; tubGAL80^ts/+; Clk8.0GAL4/ UAS-na flies maintained at 29°C throughout development and adulthood exhibit strong rhythmicity (Table 2, p < 0.01; Supplementary Figure S4, black arrows; Supplementary Table S2, p < 0.01 for LD MI and DD Day 1 MI/EI). When we restrict UAS-na expression to adulthood (19°C development – 29°C adult), LD behavior initially appears comparable to mutant controls (Figures 5A,C; black arrows; Supplementary Table S1, p > 0.43 for LD MI and EI). However, by the 1st day of constant darkness, locomotor activity patterns of adult rescue flies are substantially more rhythmic than mutant controls (Figures 5B,D, arrows; Supplementary Table S2, p < 0.01 for DD day 1 MI/EI), and rhythmicity over 7 days of DD is similar to constitutively expressed rescue (Table 2, p = 1.0). These data indicate that adult-driven transgenic na expression can promote behavioral rhythmicity. Notably, mRNA of the NA regulators Nlf-1, unc79, and unc80 is endogenously expressed in the adult Drosophila brain (Robinson et al., 2013). To determine whether adult Nlf-1 expression is required for adult-specific na rescue, we co-expressed UAS-Nlf-1 RNAi with UAS-na in the 19°development – 29°adult condition. We find that RNAi knock down of Nlf-1 in adults blocks robust adult-specific na behavioral rescue, with prominent decreases in DD morning behavior (Figures 5D,H, black arrows; Supplementary Table S2, p < 0.01 for DD day 1 MI) and DD rhythmicity (Table 2, p < 0.01). Moreover, co-expression of UAS-Nlf-1 with UAS-na in the adult rescue scheme also alters the behavioral profile, enhancing LD morning behavior (Figures 5C,I and Supplementary Table S2, p < 0.01) while decreasing DD rhythmic power (Table 2, p < 0.01). These data suggest that adult Nlf-1 expression is important for adult-specific production of the functional NA channel complex. Conversely, we performed development-specific rescue of na by raising crosses at 29°C and shifting to 19°C in adulthood. Here, developmental expression of transgenic na is sufficient to promote significant restoration of adult rhythmicity (Supplementary Figure S5 and Table 2, p < 0.01). However, in this condition, morning behavior is not strongly rescued (Supplementary Figures S5C,D, black arrows; Supplementary Table S2, p > 0.3 for MI comparisons), and DD rhythmicity is decreased relative to constitutive expression (Table 2; p < 0.01). Overall, our rescue data support a role for developmentally produced NA channel complex in the adult circadian pacemaker. These data also indicate that adult-driven channel expression may contribute in some contexts, including within the clock neurons that promote morning behavior.

If much of the NA channel complex that functions in the adult brain is produced during development, this implies that the channel complex is very stable. To address this further, we performed inducible na rescue experiments using the pan-neuronal driver elavGAL4. Notably, constitutive elavGAL4 driven expression of UAS-na restores both behavioral rhythmicity and NA channel complex expression to na^har mutants (Lear et al., 2005, 2013). We find that elavGAL4 na^har; tubGAL80^ts/+; UAS-na/+ flies maintained at 19°C throughout development and adult stages express very little NA protein compared to a wild-type (na+) control (Figure 6A, top panel, lanes 1–2; quantified in Figure 6B, <4% of wild-type, p < 0.01). Moreover, UNC79 levels in this genotype (Figure 6A, bottom panel, lanes 1–2; Figure 6B, <10% of wild-type, p < 0.05) are comparable to levels previously observed in na loss-of-function mutants (Lear et al., 2013). Thus, GAL80ts appears highly effective at blocking elavGAL4 at 19°C, although we cannot exclude low level transgene expression. However, when this strain is raised and maintained at 29°C, we observe strong expression of both NA and UNC79 proteins (Figure 6A, lane 3). Moreover, when elavGAL4 na^har; tubGAL80^ts/+; UAS-na/+ flies are raised at 29°C and then shifted to 19°C as adults, we continue to observe high levels of channel complex proteins for at least 5–6 days after the temperature shift (Figure 6A, lanes 4–6; Figure 6B, 53–85% of 29°C – 29°C protein levels). Thus, the NA channel complex produced during developmental stages is indeed quite stable.

To compare the capacity to produce the NA channel complex during development vs. adulthood, we restricted transgenic UAS-na rescue to either condition and compared protein expression levels. Similar to above, we observe that developmentally restricted UAS-na expression yields high levels of NA and UNC79 protein in adults assayed 6–7 days later (Figure 7A, lanes 1–2; Figure 7B, red vs. dark blue bars). Co-expression of Nlf-1 during development does not significantly alter NA or UNC79 levels in adults (Figure 7A, lanes 2–3; Figure 7B, dark blue vs. dark teal bars, p > 0.16). In contrast to developmental expression results, transgenic rescue of na in adults (elavGAL4 na^har; tubGAL80^ts/+; UAS-na/+ 19°C shifted to 29°C for 7–8 days) produces only modest levels of NA and UNC79 protein (Figure 7A, lane 4; Figure 7B, pink bars, ∼27–30% of 29°C – 29°C levels, p < 0.01 for NA). Thus, transgenic expression of UAS-na in the adult nervous system can promote production of the channel complex, but likely at much lower levels than developmentally restricted or constitutive expression of the transgene. Based on our behavioral results (Figures 5I,J), we also considered whether Nlf-1 could be a limiting factor for the production of new channel complex in adults. However, we find that co-expression of Nlf-1 with na in adult neurons does not produce a discernable increase in NA or UNC79 levels in head extracts (Figure 7A, lanes 4–5; Figure 7B, pink vs. light teal bars). This suggests that other factors independent of Nlf-1 expression levels limit the production of the NA channel complex in the Drosophila adult nervous system.

We also pooled data from developmentally restricted transgenic rescue experiments in order to estimate the half-life of the NA channel complex. We compared protein expression within elavGAL4 na^har; tubGAL80^ts/+; UAS-na/+ flies raised and maintained at 29°C (i.e., 29°C – 29°C) to those shifted from 29°C development to 19°C for 5–7 days of adulthood (e.g., Figure 6A, lane 6; Figure 7A, lane 2). From these pooled data, we find a small but significant decrease in NA expression levels in 29°C to 19°C head extracts as compared to the constitutive 29°C to 29°C samples (Supplementary Table S3, 86.8% of 29–29°C levels; p < 0.05). For UNC79, we observe greater variability in expression, with a larger but non-significant decrease in expression in the 29–19°C extracts (Supplementary Table S3, 73.7%; p = 0.2). We used the more consistent difference in NA expression levels to estimate protein half-life. Assuming an average time difference for the 29–19°C condition of 6 days (range 5–7 days), our calculation yields a half-life of ∼29 days for NA protein (see Materials and Methods). This half-life estimation for NA combined with the non-significant change in UNC79 levels confirms that the NA ion channel complex is likely very stable in the Drosophila nervous system.