Enhanced memory despite severe sleep loss in Drosophila insomniac mutants

A key function of sleep is hypothesized to be the maintenance of cognitive processes [6], which is supported by evidence from the detrimental consequences of acute and chronic sleep loss, and the beneficial effects of artificially induced sleep in both humans and animal models [11,37,38]. However, the reciprocal relationship between sleep and memory formation is still not well understood on molecular, cellular and circuit levels (Fig 1A). By utilizing the well-established Pavlovian aversive olfactory conditioning [39], we initially screened large populations of a spectrum of Drosophila long and short sleep mutants, reasoning that chronic sleep modulations by these mutations might allow to shed light on the common molecular mechanisms of sleep alterations in regulating cognitive functions.

Fig 1. Olfactory learning and memory screen identified Inc protein as a memory suppressor.

(A) Rationale and approaches for studying the molecular mechanisms for the balance between the amount of sleep and the strength of memory functions. (B–E) Olfactory short-term memory (STM, or learning) tested immediately after training for Shaker mns (B), 1xBRP (C), nf1 (D), and Fbxl4, homer, wake, fmn and inc² (E). n = 18 for mns, 6–9 for 1xBRP, 11–12 for nf1, 12 for Fbxl4, 22 for homer, 16 for wake, and 12–25 for fmn and inc². (F–J) Olfactory middle-term memory (MTM) tested 3 h after training for aus (F), nf1 (G), Fbxl4 (H), homer (I), and wake and inc² (J). n = 6 for aus, 14–22 for nf1, 10–16 for Fbxl4, 21–23 for homer, and 11–19 for wake and inc². (K and L) Anesthesia-resistant memory (ARM) (K) and anesthesia-sensitive memory (ASM) (L) tested 3 h after training for wake and inc². n = 10. (M and N) Odor avoidance of inc² for octanol (M) and 4-methylcyclohexanol (MCH) (N). n = 12. (O) Schematic illustration of both inc¹ and inc² mutants. Note that inc² is a P-element-mediated mutant with a UAS cassette for driving inc expression with the presence of Gal4 [31,35]. (P–U) STM (P), MTM (Q), ARM (R), ASM (S) and octanol (T) and MCH (U) odor avoidance for inc¹. n =  10 for STM, 12 for MTM, 16 for ARM and ASM, and 12 for odor avoidance. (V and W) Sleep profile of inc¹ and inc² flies averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins (V), daily sleep amount, sleep latency at ZT12, P[doze] and P[wake] (W). n =  69–96. Except for mns mutants, which were in an isogenized Canton-S control (Ctrl) background, all other mutants were compared to w¹¹¹⁸ (wt) animals. * p p p S1 Data.

https://doi.org/10.1371/journal.pbio.3003076.g001

Interestingly, most tested mutants, regardless of whether they were long or short sleepers, displayed either normal or reduced performance compared to wild-type (wt)/control flies in short-term memory (STM) measured immediately after training, or middle-term memory (MTM) measured 3 h after training (Fig 1B–1L). Specifically, the short sleepers Shaker minisleep (mns) [40], 1xBRP [11,41], homer [42], wide awake (wake) [43] and argus (aus) [44] mutants were largely normal in olfactory associative learning and memory (Fig 1B–1L). Neurofibromatosis-1 (nf1) and dopamine transporter mutant fmn were previously shown to regulate olfactory memory [45,46], and we could recapitulate these findings (Fig 1D, 1E and 1G). Interestingly, the long sleeping mutant Fbxl4 [47] showed a specific decrease in STM but not MTM (Fig 1E and 1H). Notably, a single severe short sleeping mutant, insomniac² (inc²), exhibited higher STM and MTM (Fig 1E and 1J). MTM consists of consolidated anesthesia-resistant memory (ARM) and unconsolidated anesthesia-sensitive memory (ASM), which can be distinguished by cooling trained flies on ice to induce anesthesia (also see Materials and methods) [11,21,48]. The ARM component was specifically increased in inc² mutants (Fig 1K and 1L). Due to occasional unavailability of certain mutant animals during memory screening, not all memory components were tested for every mutant. Given that the inc² short sleep mutants showed enhanced learning and memory, we decided to focus on dissecting the function of Inc in regulating both sleep and memory.

The inc² allele is a P-element-mediated inc mutation with undetectable Inc protein levels in immunoblots [31,32]. To further demonstrate the important role of Inc in suppressing olfactory learning and memory, we tested the inc¹ null mutants, which harbors a small deletion covering 5′UTR and part of the first exon of inc gene (Fig 1O) [31], for STM, MTM, ARM and ASM. Indeed, the inc¹ mutants also exhibited significantly increased STM, MTM and ARM (Fig 1P–1R), confirming the essential role of Inc in constraining aversive olfactory memories. ASM was also increased in inc¹ mutants (Fig 1S), potentially reflecting different strengths of these alleles. Both inc¹ and inc² mutants did not show any obvious alteration in odor sensation, as shown by their normal odor avoidance, proving a genuine olfactory memory phenotype in these animals (Fig 1M, 1N, 1T and 1U).

In classic Pavlovian aversive olfactory learning and memory assay, flies passively experience a conditioned stimulus (odor-electrical shock pairings), which is less dependent on internal or motivational states (e.g., foraging and mating) compared to other memory assays. Indeed, sleep has been shown to be important for a variety of distinct memory paradigms [10,37,38,41,49]. To further investigate, we examined whether inc² mutants also exhibit enhanced performance in operant conditioning paradigms by studying their courtship behavior. After training naïve wt and inc² male flies with mated wt females, both groups displayed substantial courtship suppression (S1A Fig). Notably, inc² mutants showed significantly higher courtship learning (S1B Fig), consistent with their enhanced performance in Pavlovian aversive olfactory learning and memory (Fig 1).

Both of the two inc mutations were highly wake-promoting (Fig 1V, 1W and S1C–S1J Fig), consistent with previous findings [31,32]. Interestingly, inc mutations provoked sleep loss mainly through an increase of sleep latency and a severe decrease of sleep quality as depicted by the strongly increased probability of awaking from sleep (P[wake]), but leaving sleep pressure unaffected as indicated by the normal probability of falling asleep from wakefulness (P[doze]) [50]. In short, both inc alleles suffer from a substantial decrease of sleep depth and quality (Fig 1W and S1F Fig). The memory hyperfunction and sleep deficits in inc mutants are unlikely to result from enhanced locomotor ability, as evidenced by their decreased locomotion index (S1G–S1J Fig) and normal climbing ability (S1 Video). However, as the total daily locomotor activity was strongly increased in inc mutants (S1G Fig), we cannot completely exclude locomotor differences as a contributing factor to higher memory performance. Overall, these data suggest that inc LOF, characterized by severely decreased sleep, enhances both Pavlovian associative olfactory memory and operant courtship learning.

At first glance, the excessive memory in inc short sleep mutants seems to challenge the conventional concept that sleep loss or sleep deprivation impairs cognitive functions [51]. To better understand this apparent paradox and identify molecular mechanisms in mediating this phenotypic constellation of inc mutants, we undertook a modifier screen by testing autosomal heterozygous mutations for a robust modulation of the inc sleep phenotypes. We here focused on key players in major signaling pathways and biological processes involved in sleep regulation, circadian rhythms, learning and memory, neurotransmission and synaptic plasticity. We reasoned that any heterozygous hits identified through this screen might provide insights into the etiology of the inc sleep and memory phenotypes.

Loss of inc suppresses sleep mainly by promoting nighttime sleep latency and decreasing sleep quality indicated by higher P[wake] (Fig 1W). Thus, in our modifier sleep screen, we quantified daily sleep amount, nighttime sleep latency and P[wake], and calculated the absolute differences between inc mutant flies with or without the presence of heterozygous candidate mutations (Fig 2A–2F). We screened ~70 isogenized alleles, finding that inc mutants were sensitive for heterozygous modifiers further exaggerating their short sleep phenotypes (Fig 2A and 2D). As we showed previously [41], removing a single copy of the ELKS-family active zone scaffold protein Bruchpilot (1xBRP) was more profound in promoting wakefulness in inc mutants than in wt/control background (Fig 2G, 2H and S2A, S2B Fig). As mentioned above already, Inc functions as an adaptor of the E3 ligase Cullin-3 [31]. Consistent with the role of Inc in promoting Cullin-3 function to regulate sleep, Cullin-3 heterozygosity further enhanced the inc sleep deficits (Fig 2G, 2H and S2A, S2B Fig). Moreover, genetic modulations of GABAergic signaling dynamically modified the inc sleep phenotypes (Fig 2I, 2J and S2C, S2D Fig).

Fig 2. Genetic modifier sleep screen for inc mutants.

(A–C) Genetic screen in inc² mutant background analyzing the absolute difference in daily sleep (A), sleep latency at ZT12 (B) and P[wake] (C) between inc² and wt or inc² with autosomal heterozygous candidate mutations. n ≥  25. (D–F) Genetic screen in inc¹ mutant background analyzing the absolute difference in daily sleep (D), sleep latency at ZT12 (E) and P[wake] (F) between inc¹ and wt or inc¹ with autosomal heterozygous candidate mutations. Note that some candidates tested in inc² background were also tested in inc¹ background in the screen. n = 12 for inc¹;liprin-α^R60/+ , 13 for inc¹;liprin-α^F3ex15/+ and 15 for inc¹;syd-1^ex3.4/+ , n ≥ 23 for all other genotypes. (G–J) Some examples of candidates shown to tune sleep of inc mutants by their heterozygosity, including Cullin-3^EY11031/+ (Cul3/+) and brp^c04298/+ (1xBRP) (G and H), and GABA-B-R1^attP/+ (B-R1/+) and vGat^attP/+ (vGat/+) (I and J). n = 40–111 for G, 30–51 for H, 40–47 for I, and 30–51 for J. Statistical comparisons were performed either between wt and inc, or between inc with or without heterozygous candidate mutations. *p p p S2 Data.

https://doi.org/10.1371/journal.pbio.3003076.g002

In contrast to these potential hits enhancing the inc short sleep phenotypes, only a few modifiers were able to rescue the sleep phenotypes of inc mutants. The strongest and most robust rescue was observed after establishing heterozygosity for the PKA catalytic subunit Dc0, particularly pronounced for the Dc0^H2 allele in inc¹ background (Fig 2A–2F). Moreover, Neurofibromatosis-1 (Nf1), previously shown to regulate cAMP/PKA signaling in the context of sleep and memory regulation [45,52], also significantly suppressed the short sleep phenotypes of inc mutants by its heterozygosity (Fig 2A–2C). Given the consistent rescue effects after a mild genetic reduction of PKA signaling components, we decided to focus our efforts on the potential relationship between Inc and the cAMP/PKA signaling pathway.

To deepen our understanding towards the exact nature of this rescue scenario, we examined the detailed sleep phenotypes of the two inc mutants in conjunction with heterozygosity of several distinct alleles of the PKA catalytic subunit Dc0 (Fig 3A–3E). Dc0^B3 and Dc0^H2 alleles are LOF mutants harboring point mutations within the Dc0 open reading frame [53]. Heterozygosity of Dc0^B3 and Dc0^H2 was previously shown to trigger moderate but significant reductions in PKA activity [29]. We first found that, while Dc0^H2 heterozygosity provoked more consolidated and deeper sleep than Dc0^B3 heterozygosity in wt/control background indicated by shorter sleep latency and lower P[wake], both alleles exhibited mild alterations in sleep pattern with a slight reduction in daytime sleep and an increase in nighttime sleep, and an overall normal daily sleep amount (Fig 3A and S3A Fig). These data indicate that, different from strong manipulations in cAMP/PKA signaling [28], a mild reduction in PKA signaling triggered by Dc0 heterozygosity has no major sleep phenotypes in wt/control background.

Fig 3. cAMP/PKA signaling mediates the sleep phenotypes of inc mutants.

(A) Sleep profile of wt, Dc0^B3/+ and Dc0^H2/+ flies averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 70–76. (B) Sleep profile of inc¹ with either Dc0^B3/+ or Dc0^H2/+ averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 57–66. (C) Sleep profile of inc² with either Dc0^B3/+ or Dc0^H2/+ averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 73–76. (D) Sleep profile of wt, Dc0^BG/+ and inc¹ with or without Dc0^BG/+ averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 40–71. (E) Sleep profile of wt and inc¹ with either Dc0^DN/+ or Dc0^G4/+ averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 32–48. (F) Sleep profile of wt, nf1/+ and inc² with or without nf1/+ averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 62–80. (G) Sleep profile of wt, PKA.R1^c/+ and PKA.R1^MB/+ flies averaged from measurements over 4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 58–63. (H) Sleep profile of inc¹ with either PKA.R1^c/+ or PKA.R1^MB/+ averaged from measurements over 4 days, including sleep curves plotted in 30-min bins, daily sleep amount, sleep latency at ZT12, P[doze] and P[wake]. n = 61–67. Dashed straight lines indicate mean values of wt/control animals tested simultaneously. *p p p I and J) Absolute differences in daytime (I) and nighttime (J) sleep among different heterozygotes of cAMP/PKA signaling components in wt or inc mutant backgrounds in A–H. *p p p S3 Data.

https://doi.org/10.1371/journal.pbio.3003076.g003

Notably, both daytime and nighttime sleep were substantially restored for both inc¹ and inc² mutants by heterozygosity of either Dc0^B3 or Dc0^H2 compared to inc¹ or inc² mutants, with shortened sleep latency and severely reduced P[wake], indicating higher sleep quality (Fig 3B, 3C and S3B, S3C Fig). As noted earlier (Fig 2D), the sleep restoring effects of Dc0^H2 allele were much stronger in inc¹ than in inc² background, suggesting a higher sleep pressure provoked by this allele (Fig 3B and S3B Fig). Indeed, Dc0^H2 heterozygosity was shown to have lower PKA activity when compared to Dc0^B3 heterozygosity [29]. Thus, the Dc0 allele-specific sleep effects in inc mutant backgrounds might reflect the allelic strength of these PKA alleles (Fig 3I and 3J).

To further explore the role of PKA signaling in regulating the sleep phenotypes of inc mutants and to particularly address whether these effects were Dc0 allele-specific, we also utilized two P-element-mediated mutations (Dc0^BG and Dc0^G4 lines) and one additional dominant negative mutant (Dc0^DN line). These three additional lines are all adult homozygous lethal, but only Dc0^BG and Dc0^G4 in heterozygosity showed similar effects on restoring sleep to inc mutants when compared to Dc0^B3 (Fig 3D, 3E and S3D, S3E Fig), while Dc0^DN as a dominant negative allele distinct from the Dc0 null mutants [54] had no effects on the sleep of inc mutants (Fig 3E and S3E Fig). Our data thus show that heterozygosity of Dc0 null or strong hypomorphic alleles consistently suppresses the sleep phenotypes of inc mutants, while having mild to no effects in wt/control animals (Fig 3I, 3J and S3I–S3L Fig). Thus, it seems that inc mutants are sensitized towards PKA signaling modulation, and the level of PKA signaling seems to be directly associated with the severity of their sleep deficits (Fig 3I and 3J).

In its inactive state, PKA forms a tetramer consisting of two catalytic and two regulatory subunits [55]. cAMP activates PKA by binding to the regulatory subunits and subsequently releasing the catalytic subunits so as to phosphorylate target proteins [55]. Thus, the regulatory subunit directly antagonizes the kinase function of the catalytic subunit, and the synthesis of cAMP by adenylate cyclases such as Rutabaga (Rut) is essential for activating/disinhibiting PKA signaling. Along this line, in addition to Dc0, modulating other regulatory components of cAMP/PKA signaling could potentially also tune the sleep phenotypes of inc mutants.

Through our genetic modifier sleep screen for inc mutants (Fig 2), we identified Nf1 as a significant modifier (Fig 2A–2C). nf1 heterozygosity restored a great extent of sleep in inc mutant background by increasing P[doze] and simultaneously reducing P[wake], whereas it had only very mild effects in wt/control background (Fig 3F, 3I, 3J and S3F, S3M Fig). In Drosophila, Nf1 has been studied in the regulation of circadian rhythms [56], sleep [52,57,58] and memory (Fig 1D and 1G) [45]. Importantly, Nf1 was shown to be proteasome-degraded through Cullin-3 E3 ligase-mediated ubiquitination in vitro [59]. Furthermore, Nf1 promotes cAMP/PKA signaling by increasing Rut adenylate cyclase activity and subsequently the synthesis of cAMP and activation of PKA signaling [45]. Thus, it seems possible that a hierarchical signaling cascade, normally elicited by Inc/Cullin-3 complex-mediated Nf1 degradation and subsequent PKA activity suppression, feeds forward to fine-tune the levels of sleep (see Discussion and S6D Fig).

To potentially tune PKA signaling bidirectionally, we further tried to increase PKA signaling by establishing heterozygous scenarios for PKA regulatory subunit type 1 (PKA.R1) both in wt/control and inc backgrounds. In contrast to Dc0 heterozygosity (Fig 3A–3E and S3A–S3E Fig), the heterozygosity of two alleles of PKA.R1 enhanced the short sleep phenotypes of inc mutants, while having very mild or no obvious effect in wt/control background (Fig 3G–3J and S3G, S3H, S3N, S3O Fig). These data again suggest that inc mutants are sensitized towards bidirectional changes of PKA signaling, which seems to directly mediate the sleep phenotypes of inc mutants.

The PKA kinase targets a spectrum of downstream effectors and has widespread effects on various biological processes [60, 61]. We thus asked whether the effects of Dc0^H2 heterozygosity on sleep are specific for inc mutants (Figs 3A–3C, 4A and 4B), or generalize to other genetic sleep deficit scenarios as well. For this purpose, we chose two different approaches: (1) distinct short and long sleep mutants (Fig 4C–4H, 4K and 4L) and (2) activating previously described wake-promoting sleep circuits (Fig 4I and 4J). For most of these scenarios, we did not observe any interference by Dc0^H2 heterozygosity on daily sleep, including the short-sleeping voltage-gated potassium channel Shaker mns mutants [40] (Fig 4C and S4C Fig), sleepless (sss) mutants [62] (Fig 4D and S4D Fig), 1xBRP [11,41] (Fig 4E and S4E Fig), wide awake (wake) mutants [43] (Fig 4G and S4G Fig), molting defective (mld) mutants [63] (Fig 4H and S4H Fig), and long-sleeping mGluRA mutants [64] (Fig 4K and S4K Fig) and T-type like voltage-gated calcium channel (Ca-α1T) mutants [65] (Fig 4L and S4L Fig). Interestingly, the short sleep phenotype of cacophony^H18 (cac^H18) [66] was further enhanced (Fig 4F and S4F Fig). Furthermore, the Dc0^H2 heterozygous scenario did not interfere with the reduced daily sleep caused by the activation of dopaminergic neurons [67] or helicon cells [68] by expressing low-threshold voltage-gated sodium channel NaChBac [69] (Fig 4I, 4J and S4I, S4J Fig). Taken together, our data show that inc mutants are specifically sensitized towards a reduction of PKA signaling.

Fig 4. The effects of Dc0 heterozygosity on promoting sleep are specific to inc mutants.

(A–L) Sleep profile of Dc0^H2/+ in control (A), inc² (B), mns (C), sss (D), 1xBRP (E), cac^H18 (F), wake (G), mld (H), TH>NaChBac (I), R24B11 > NaChBac (J), mGluRA (K) and Ca-α1T (L) backgrounds, averaged from measurements over 2–4 days, including sleep curves plotted in 30-min bins, daily sleep amount, and P[wake]. n = 48–52 for A, 32–38 for B, 48–56 for C, 47–72 for D, 48–51 for E, 48 for F, 54–56 for G, 29–48 for H, 32–48 for I and J, 39–47 for K, and 48 for L. Dashed lines indicate mean values of wt/control animals tested simultaneously. (M and N) Absolute differences in daytime (M) and nighttime (N) sleep among different genetic sleep manipulations in A–L and the diverse effects of Dc0^H2/+ in these backgrounds of manipulations. *p p p S4 Data.

https://doi.org/10.1371/journal.pbio.3003076.g004

As mentioned earlier, it is worth noting that Dc0^H2 heterozygosity drove a slight re-organization of daytime and nighttime sleep, indicated by a slight decrease of sleep during daytime but a mild increase of sleep during nighttime, resulting in an indistinguishable total daily sleep amount compared to wt (Figs 3A and 4A). Consistently, cAMP/PKA signaling was previously shown to regulate circadian rhythms [70], and Dc0 expression exhibits daily oscillations [52], suggesting different functions for the regulation of daytime versus nighttime sleep. We next systematically compared the effects of Dc0^H2 heterozygosity in different backgrounds of genetic sleep manipulations, taking daytime and nighttime into separate considerations (Fig 4M and 4N). Indeed, Dc0^H2 heterozygosity was wake-promoting during daytime in most cases, except for inc mutants, in which it became sleep-promoting (Fig 4M). During nighttime, Dc0^H2 heterozygosity generally showed rather minor sleep-promoting effects, which became much more prominent in inc mutant background (Fig 4N).

Inc was reported to be broadly expressed in the nervous system [31,32], including the mushroom body within which Inc was shown to contribute to sleep-promoting effects [35]. To directly visualize the expression pattern of Dc0 by driving GFP reporter expression, we utilized the Dc0^BG allele, which is a Gal4 gene trap of the Dc0 locus and whose heterozygosity (Dc0^BG/+) rescued the sleep of inc mutants (Fig 3D). We observed that Dc0 was also broadly expressed in adult central brain and ventral nerve cord, with particularly prominent expression in the mushroom body (Fig 5A), consistent with previous reports [29,71]. Given that a mild reduction in PKA signaling in Dc0/+ rescued the sleep phenotypes of inc mutants, we asked whether the Dc0 protein level was directly changed in inc mutants. Due to the lack of a functional antibody targeting Dc0, we expressed and visualized FLAG-tagged Dc0 in the mushroom body Kenyon cells driven by the mushroom body-specific VT30559-Gal4 (Fig 5B). Importantly, Gal4-driven expressed Dc0-FLAG localized to the mushroom body properly in wt background (Fig 5C), similar to Dc0 antibody staining [29]. The overall Dc0-FLAG staining intensity did not differ between wt and inc mutants (Fig 5C and 5D), suggesting that Dc0 protein abundance is unlikely to be regulated by Inc. However, the distribution of Dc0-FLAG was markedly altered in inc mutants, characterized by circular structures with highly concentrated Dc0-FLAG staining (Fig 5C and 5E). This pattern may indicate altered PKA kinase function and signaling. Additionally, inc mutants displayed abnormal mushroom body morphology, including overgrown structures and missing lobes (Fig 5C and 5F), consistent with previous findings [35,71].

Fig 5. Elevated PKA kinase activity in the mushroom body drives sleep deficits of inc mutants.

(A) Whole-mount brain expressing mCD8-GFP driven by the Dc0^BG Gal4 gene trap line demonstrating the expression pattern of Dc0 in brain and ventral nerve cord (VNC). Dc0 has broad expression but highly enriched in the mushroom body (MB) lobes and calyx. Native GFP fluorescence is shown. Scale bar: 100 μm. (B) Whole-mount brain expressing mCD8-GFP driven the transgenic VT30559-Gal4 line demonstrating its highly restricted MB-specific expression pattern. The brains were stained with Nc82 antibody, and native GFP fluorescence is shown. Scale bar: 50 μm. (C–F) Whole-mount brain immunostaining against FLAG for wt/control and inc¹ flies expressing Dc0-FLAG in the mushroom body driven by VT30559-Gal4, including representative images (C), average intensity (D) and FLAG puncta number (E). inc mutants exhibited abnormal mushroom body lobe morphology (F). n = 14. Scale bar: 20 μm. (G–J) Whole-mount brain immunostaining against GFP for wt/control and inc¹ flies expressing PKA kinase activity sensor PKA.SPARK in the mushroom body driven by VT30559-Gal4, including representative images (G), average intensity (H) and SPARK puncta number (I). inc mutants again exhibited abnormal mushroom body lobe morphology (J). n = 15−16. Scale bar: 20 μm. (K) Simplified schematic illustration for the regulation of PKA kinase activity by cAMP and PKA regulatory subunit in wt background or under the expression of either a non-functional regulatory subunit (PKA.R1^GG), which does no longer bind to the catalytic subunit, or a constitutive active regulatory subunit (PKA.R1^BDK), which constitutively binds to the catalytic subunit and is insensitive to cAMP. (L and M) Sleep profile from measurements over 4 days of flies expressing non-functional (GG) or active (BDK) PKA regulatory subunit in the mushroom body driven by VT30559-Gal4 in wt/control background, including sleep curves plotted in 30-min bins (L), daily sleep amount, sleep latency at ZT12, P[doze] and P[wake] (M). n= 76. (N and O) Sleep profile from measurements over 4 days of flies expressing non-functional (GG) or active (BDK) PKA regulatory subunit PKA.R1 in the mushroom body driven by VT30559-Gal4 in inc¹ mutant background, including sleep curves plotted in 30-min bins (N), daily sleep amount, sleep latency at ZT12, P[doze] and P[wake] (O). n = 83−85. (P and Q) Sleep profile from measurements over 2−4 days of flies expressing non-functional (GG) or active (BDK) PKA regulatory subunit PKA.R1 in the mushroom body driven by VT30559-Gal4 in 1xBRP background, including sleep curves plotted in 30-min bins (P), daily sleep amount, sleep latency at ZT12, P[doze] and P[wake] (Q). n = 27−48. (R) Absolute difference in daily sleep between flies expressing non-functional (GG) and active (BDK) PKA regulatory subunit in the mushroom body in wt/control, inc¹ or 1xBRP background. n = 48−83. (S) Simplified model suggesting that the Inc/Cullin-3 complex regulates mushroom body PKA signaling to promote sleep. **p p   0.001; ns, not significant. Error bars: mean ±  SEM. Underlying data can be found in S5 Data.

https://doi.org/10.1371/journal.pbio.3003076.g005

To directly assess the PKA kinase activity level, we expressed the PKA activity reporter PKA.SPARK [72] in the mushroom body Kenyon cells driven by VT30559-Gal4. Upon PKA activation, SPARK signal forms puncta from otherwise diffuse signal in vitro and in vivo [72,73]. Consistent with this, we observed an increase in the number of PKA SPARK puncta in inc mutants, while the average level of SPARK intensity did not change (Fig 5G–5I). Additionally, structural abnormalities in the mushroom body were evident in inc mutants (Fig 5G and 5J). These data suggest elevated PKA kinase activity in inc mutant Kenyon cells, which might be directly causal for the severe sleep defects of inc mutants.

As mentioned above, PKA forms a tetramer consisting of two catalytic and two regulatory subunits [55] (Fig 5K). In its inactive state, PKA catalytic subunit kinase activity is suppressed by its regulatory subunit [55] (Fig 5K). cAMP then activates PKA by binding to the regulatory subunits and subsequently releases and disinhibits the catalytic subunits so as to allow for phosphorylating target proteins [55] (Fig 5K). As shown earlier, reducing the level of PKA regulatory subunit could further reduce the sleep of inc mutants (Fig 3H). To further test whether PKA kinase activity within the mushroom body is causal for the sleep deficits of inc mutants, we expressed an active form of the regulatory subunit (PKA.R1^BDK), which is insensitive to cAMP and constitutively inhibits the catalytic subunit (Fig 5K) [74], to suppress PKA kinase activity in the mushroom body. As control, a non-functional form (PKA.R1^GG), which has no binding activity to the catalytic subunit, was expressed in the mushroom body. We first show that expressing PKA.R1^GG in the mushroom body driven by VT30559-Gal4 did not significantly affect sleep in both wt and inc mutant backgrounds (S5A and S5B Fig). Interestingly, we found that expression of PKA.R1^BDK triggered a moderate increase of sleep when compared to PKA.R1^GG expression in wt/control background (Fig 5L, 5M and S5C Fig). The sleep-promoting effects of mushroom body PKA.R1^BDK expression were more pronounced in inc mutant background, but not in 1xBRP background (Fig 5N–5R and S5D, S5E Fig), similar to the effects of Dc0^H2 heterozygosity (Figs 3B, 3C and 4B). Collectively, these data suggest that a higher PKA kinase activity within the mushroom body contributes to the inc sleep phenotypes.

Similar to other short-sleep mutants, inc animals were previously shown to be short-lived [31] and hypersensitive to oxidative stress [13]. Given the strong rescue effects of Dc0 heterozygosity toward the inc mutant sleep deficits, we hypothesized that it might also rescue their survival under oxidative stress and longevity phenotypes. For this purpose, we first measured the survival of the Dc0^H2 heterozygous animals treated with 2% hydrogen peroxide (H₂O₂) and their life span upon normal aging. Interestingly, Dc0^H2 heterozygous animals were more vulnerable than wt/control in response to 2% H₂O₂ treatment (Fig 6A). In wt/control background, the life span of Dc0^H2 heterozygous animals was significantly increased compared to wt/control (Fig 6B), consistent with a previously reported tendency of increased life span in Dc0^H2 heterozygous flies [29]. We then tested Dc0^H2 heterozygosity in both inc¹ and inc² mutants. Similar to a previous report [13], we confirmed that inc mutants were hypersensitive to H₂O₂ treatment, while Dc0^H2 heterozygosity did not have any additive effect here (Fig 6C and 6E). However, in terms of longevity, inc¹ mutants were clearly improved and inc² mutants were fully rescued to the extent of Dc0^H2 heterozygous animals (Fig 6D and 6F). These improvements in longevity, accompanied by a significant rescue in sleep (Fig 3B and 3D), suggest a healthier physiological state being established for inc mutants by a mild down-regulation of PKA signaling. Again, to test the specificity of this rescue of longevity in the absence of a rescue in the survival upon H₂O₂ treatment, we wondered if Dc0^H2 heterozygosity was also sufficient to improve the survival of other short-sleeping scenarios including Shaker mns, sss and 1xBRP upon H₂O₂ treatment and in longevity. Notably, however, for these short sleep mutants, no rescue was observed (Fig 6G–6L). Thus, our data suggest the scenario that a moderate downregulation of PKA signaling is sufficient to specifically promote longevity as a trait unique to inc among the tested short sleep mutants.

Fig 6. cAMP/PKA signaling mediates the longevity phenotypes of inc mutants.

(A and B) Survival curves under 2% H₂O₂ treatment (A) or normal aging (B) for wt and Dc0^H2/+ flies. For H₂O₂ treatment, n = 133 for wt and 124 for Dc0^H2/+. For longevity, n = 299 for wt and 280 for Dc0^H2/+. (C and D) Survival curves under 2% H₂O₂ treatment (C) or normal aging (D) for wt, Dc0^H2/+ and inc¹ with or without Dc0^H2/+. (E and F) Survival curves under 2% H₂O₂ treatment (E) or normal aging (F) for wt, Dc0^H2/+ and inc² with or without Dc0^H2/+. Note that C and E share the same wt and Dc0^H2/+, as well as in D and F. For H₂O₂ treatment, n = 228 for wt, 177 for Dc0^H2/+, 143 for inc¹, 163 for inc², 117 for inc¹;Dc0^H2/+ and 170 for inc²;Dc0^H2/+. For longevity, n = 234 for wt, 150 for Dc0^H2/+, 182 for inc¹, 232 for inc², 149 for inc¹;Dc0^H2/+ and 225 for inc²;Dc0^H2/+. (G and H) Survival curves under 2% H₂O₂ treatment (G) or normal aging (H) for wt, Dc0^H2/+ and mns with or without Dc0^H2/+. For H₂O₂ treatment, n = 264 for wt, 87 for Dc0^H2/+, 71 for mns and 62 for mns;Dc0^H2/+. For longevity, n = 156 for wt, 156 for mns and 154 for mns;Dc0^H2/+. (I and J) Survival curves under 2% H₂O₂ treatment (I) or normal aging (J) for wt, Dc0^H2/+ and sss with or without Dc0^H2/+. For H₂O₂ treatment, n = 98 for wt, 73 for Dc0^H2/+, 98 for sss (n = 98, p sss,Dc0^H2/+. For longevity, n = 112 for wt, 110 for sss and 109 for sss,Dc0^H2/+. (K and L) Survival curves under 2% H₂O₂ treatment (K) or normal aging (L) for wt, Dc0^H2/+ and 1xBRP with or without Dc0^H2/+. For H₂O₂ treatment, n = 97 for wt, 66 for Dc0^H2/+, 91 for 1xBRP and 81 for 1xBRP,Dc0^H2/+. For longevity, n = 147 wt, 150 for 1xBRP and 139 for 1xBRP,Dc0^H2/+. (M and N) Survival curves under 2% H₂O₂ treatment (M) or normal aging (N) for flies expressing non-functional (GG) or active (BDK) PKA regulatory subunit PKA.R1 in the mushroom body driven by VT30559-Gal4 in wt and inc¹ mutant backgrounds. For H₂O₂ treatment, n = 210 for GG, 238 for BDK, 181 for inc¹;GG and 210 for inc¹;BDK. For longevity, n = 165 for GG, 189 for BDK, 110 for inc¹;GG and 168 for inc¹;BDK. (O and P) Survival curves under 2% H₂O₂ treatment (O) or normal aging (P) for wt, nf1/+ and inc¹ with or without nf1/+. (Q and R) Survival curves under 2% H₂O₂ treatment (Q) or normal aging (R) for wt, nf1/+ and inc² with or without nf1/+. Note that O and Q share the same wt and nf1/+, as well as in P and R. For H₂O₂ treatment, n = 188 for wt, 191 for nf1/+, 136 for inc¹, 180 for inc², 141 for inc¹;nf1/+, and 177 for inc²;nf1/+. For longevity, n = 227 for wt, 148 for nf1/+, 140 for inc¹, 232 for inc², 143 for inc¹;nf1/+ (n = 143) and 148 for inc²;nf1/+ (n = 148). *p p p S6 Data.

https://doi.org/10.1371/journal.pbio.3003076.g006

Given that suppressing PKA signaling by PKA.R1^BDK expression specifically in the mushroom body was sufficient to restore sleep in inc mutants (Fig 5L–5R), we next asked if longevity and survival upon H₂O₂ treatment could also be rescued. Consistent with the effects of Dc0 heterozygosity on longevity (Fig 6B, 6D and 6F), mushroom body-specific PKA.R1^BDK expression showed a clear tendency towards extended life span in wt background and significantly increased the life span of inc mutants (Fig 6N). Interestingly, these flies were also longer-lived upon H₂O₂ treatment when compared to control flies expressing PKA.R1^GG in both wt and inc mutant backgrounds (Fig 6M), different from the effects of Dc0 heterozygosity (Fig 6A, 6C and 6E). These data suggest that mushroom body-specific PKA suppression promotes survival during both normal aging and acute oxidative stress.

To test if the rescue effects on longevity of inc mutants was specific to Dc0 heterozygosity or would extend to other PKA signaling components, we further measured the longevity as well as survival upon H₂O₂ treatment of nf1 heterozygosity in wt/control and inc mutant backgrounds, which was shown above to restore sleep to inc mutants (Fig 3F). While nf1 heterozygosity did not show any obvious difference in both survival paradigms compared to wt/control (Fig 6O–6R), it substantially improved the longevity of inc mutants (Fig 6P and 6R), leaving their survival phenotype upon H₂O₂ treatment unaffected (Fig 6O and 6Q). Thus, it is likely that a mild reduction in PKA signaling, provoked by the heterozygosity of PKA regulatory component nf1, triggered effects on the longevity of inc mutants comparable to those observed with Dc0 heterozygosity. Taken together, Nf1 might be directly involved in mediating molecular signal transduction from Inc to PKA signaling.

Our findings that the intrinsic elevation of PKA signaling underlines the sleep and longevity phenotypes of inc mutants suggest that their excessive memory might also be rescued by Dc0 heterozygosity. To directly test this hypothesis, we first confirmed the Dc0^H2 heterozygosity rescue effects on sleep in the same fly cohort prepared for olfactory memory (S6A Fig). While Dc0^H2 heterozygosity did not exhibit any significant effects in wt/control background, it further slightly increased STM (Fig 7A and S6B Fig) and significantly further promoted the higher 3 h MTM of inc² mutants (Fig 7B and S6C Fig). The ARM component was also further increased by establishing Dc0^H2 heterozygosity in inc² mutants (Fig 7C). Notably, Dc0^H2 heterozygous animals displayed elevated ARM and diminished ASM (Fig 7C and 7D), consistent with a previous study [75].

Fig 7. Excessive memory functions in inc mutants are not restored via PKA manipulations.

(A–D) Olfactory memory tested immediately or 3 h after training for wt, Dc0^H2/+ and inc² with or without Dc0^H2/+, including STM (A), MTM (B), ARM (C) and ASM (D). n = 9–15 for STM, 12–14 for MTM, and 16 for ARM and ASM. (E) A simplified model for roles of PKA signaling in coordinating memory, sleep and longevity. inc loss-of-function (LOF) is memory-promoting potentially through unknown signaling pathways, which is limited simultaneously by elevated PKA signaling. Thus, it is plausible that memory hyperfunction in inc mutants provokes adaptive higher PKA activity, which in turn suppresses memory functions. As a consequence, excessive PKA kinase activity provokes severe sleep deficits and compromised life span in inc mutants. The rescue effects of Dc0^H2/+ on life span and its effects on further increasing the excessive memory of inc mutants might be either a direct role of PKA, or a consequence of the restored sleep. (F and G) Representative images of whole-mount brain expressing mCD8-GFP in subtypes of mushroom body α/β_c (F) or α′/β′ (G) Kenyon cells in wt/control, Dc0^H2/+ and inc¹ with or without Dc0^H2/+ backgrounds. α/β_c neurons form discrete clusters in morphology, but not for α′/β′ neurons. Scale bar: 50 μm. (H–L) Statistics of the cluster number of α/β_c neurons (H), the number of lobes (I and K) and the volume of cell body and dendrites (J and L). n = 30–42 for H and J, 15–21 for I, and 32–36 for L. (M) Hypothesized model for the performance in sleep and memory controlled by PKA activity. Our data and previous studies suggest a bidirectional sleep regulation by PKA signaling. In contrast, memory is bidirectionally regulated likely only when PKA signaling is tuned within a physiological range. Severe and pathological PKA signaling reduction or increase is detrimental for memory performance. inc mutants seem to have higher setpoints for both memory and PKA activity. Restoring sleep to inc mutants by mildly titrating PKA signaling to half dose further promotes memory performance. *p p p S7 Data.

https://doi.org/10.1371/journal.pbio.3003076.g007

How might Inc interact with PKA signaling in the regulation of memory? Both Inc and PKA are widely expressed in the fly brain but enriched in the mushroom body [31,32,71]. As shown above, inc mutants exhibit gross mushroom body structural defects (Fig 5C, 5F, 5G and 5J). Furthermore, severe circuit defects in subtypes of mushroom body neurons in inc mutants, especially for the α/β core (α/β_c) and α′/β′ Kenyon cells, have been previously reported [35]. In this regard, a simple hypothesis might be that the further increased memory performance of inc mutants by Dc0 heterozygosity (Fig 7A–7E) could be due to a rescue of the mushroom body structural defects. To address this question, we expressed a GFP reporter in either α/β_c or α′/β′ Kenyon cells and evaluated the effects of Inc/PKA signaling in regulating the morphological properties of mushroom body subtype neurons. Consistent with previous findings [35], inc mutants showed an increased number of clusters of cell bodies, severely affected lobe structure, and an increased volume of cell bodies and dendrites of α/β_c neurons (Fig 7F and 7H–7J). While the cluster number phenotype remained unaffected (Fig 7H), the severe lobe structural phenotype was even more frequently triggered by Dc0 heterozygosity (Fig 7I), while the cell body and dendritic volume phenotypes were slightly suppressed (Fig 7J). In addition to α/β_c neurons, inc mutants did almost never show any phenotype in the number of lobes of α′/β′ neurons, whose cell body and dendritic volume was also increased by Dc0 heterozygosity (Fig 7G, 7K and 7L). Both lobe number and volume of cell bodies and dendrites of inc mutants were obviously further increased by Dc0 heterozygosity (Fig 7G, 7K and 7L). These data suggest that the further increased memory performance of inc mutants with Dc0 heterozygosity is unlikely to be achieved by restored mushroom body circuitry. Instead, the higher olfactory memory performance of inc mutants might be at least partially caused by overgrowth and abnormal arborization of these circuits. To a certain extent, the elevated PKA signaling might suppress mushroom body Kenyon cell overgrowth to constrain memory performance, and at the same time drive sleep loss in inc mutants. Regarding the functions of sleep, the substantially restored sleep in inc mutants by Dc0 heterozygosity likely contributes to the memory benefits associated with the physiological functions of sleep.

Suppressing PKA signaling in the mushroom body by PKA.R1^BDK expression restored sleep, survival upon H₂O₂ treatment, and longevity in inc mutants (Figs 5L–5R, 6M and 6N). Comparing to Dc0 heterozygosity (Fig 7A–7E), we wondered whether mushroom body-specific PKA.R1^BDK expression would regulate memory in a similar manner in wt and inc mutants. In addition, it is unclear how mushroom body-specific PKA signaling reduction affects the mushroom body structural phenotypes of inc mutants. To address these questions, we systematically examined memory (S6E–S6H Fig) and mushroom body structural phenotypes (S6I-S6K Fig) of animals expressing PKA.R1^GG (control) or PKA.R1^BDK driven by VT30559-Gal4 in either wt or inc mutant background. We found that STM was not affected by mushroom body-specific PKA.R1^BDK expression (S6E Fig). However, while MTM was not affected in wt background, it was surprisingly decreased in inc mutant background with either PKA.R1^GG or PKA.R1^BDK expression (S6F Fig). This decrease in MTM is due to a clear reduction in ASM component (S6H Fig). Importantly, the ARM component was consistently increased in inc mutants and in PKA.R1^BDK-expressing animals (S6G Fig). PKA.R1^BDK expression did not further increase the ARM of inc mutants (S6G Fig), however. Similar to the interaction between inc and Dc0 heterozygosity, PKA.R1^BDK expression did not rescue the mushroom body structural phenotypes of inc mutants (S6I-S6K Fig).

These data suggest that mushroom body-specific PKA.R1^BDK expression partially mimicked Dc0 heterozygosity in regulating mushroom body structure, sleep, life span and ARM component. The distinctions between these two genetic scenarios might derive from (1) additional circuits and tissues besides the mushroom body Kenyon cells being involved in dictating the effects of Dc0 heterozygosity in memory and survival under H₂O₂ treatment; and (2) the different strength of PKA signaling reduction between these two different genetic PKA manipulations accounting for these behavioral differences. cAMP/PKA signaling has been shown to regulate sleep bidirectionally [28], while both severe increase and decrease of cAMP/PKA signaling are detrimental for memory (Fig 7M). Thus, it seems possible that Dc0 heterozygosity triggers a level of PKA signaling optimal for restoring life expectancy but further promoting memory functions of inc mutants. In contrast, PKA.R1^BDK expression may result in much stronger PKA signaling reduction that is detrimental for memory of inc mutants (Fig 7M).

It seems paradoxical that defective mushroom body morphology is coupled with excessive memory in inc mutants (Fig 7). Inc has been shown to be active during a specific time window of Drosophila brain development [35]. The overproduction of the mushroom body Kenyon cells with structural defects in inc mutants suggests that the mushroom body is the crucial neuron population from where Inc functions during mushroom body development to regulate both sleep and memory. Notably, we found that Dc0^H2 heterozygosity further enhances the mushroom body structural phenotypes of inc mutants (Fig 7F–7L), but at the same time increases their excessive memory performance (Fig 7A–7E) and restores their sleep level (Fig 4B and S6A Fig). Thus, this overgrowth of mushroom body Kenyon cells might be causally related to the memory hyperfunction of inc mutants.

As discussed above, inc LOF exhibited enhanced memory, and downregulating PKA signaling in inc background further exacerbated this phenotype, suggesting that an elevated PKA signaling limits the memory hyperfunction of inc mutants. However, it is currently unknown how inc LOF triggers excessive olfactory memory in the first place. It appears likely that inc LOF promotes memory performance through unknown signaling pathways (Fig 7E).

While PKA signaling acting like a “brake” for higher memory performance in inc mutants (Fig 7A–7D, 7M and S6D Fig), what might be the unknown signaling pathways directly mediating the memory-promoting effects of inc mutants? As an adaptor protein for Cullin-3 ubiquitin E3 ligase, Inc might promote the degradation of a direct target of Cullin-3 ligase, whose accumulation due to impaired ubiquitination and degradation might be responsible for the memory-promoting effects of inc mutants. In this regard, a potential target of Inc/Cullin-3 complex might be the dopaminergic signaling involved in both sleep and memory regulation, as it was previously shown to be important in mediating the sleep phenotypes of inc mutants [32]. Furthermore, given that inc mutants have profound mushroom body circuit phenotypes, it will be important in the future to screen for the interactions between Inc and other factors involved in regulating mushroom body development. Indeed, mushroom body-specific expression of a gain-of-function Rho kinase (ROCK or Rok) promotes memory formation [79], and triggers mushroom body structural defects [80], similar to inc mutants. While potential effects of ROCK gain-of-function on sleep have not been addressed, inc LOF might promote excessive memory and mushroom body structural defects by modulating Ras/Raf/ROCK signaling.

Our data suggest that the elevated PKA signaling seems to constrain the excessive memory in inc mutants. What might be the exact action of Inc in regulating PKA signaling? The evolutionarily conserved nature of Inc as an adaptor of Cullin-3 E3 ligase for ubiquitination of target proteins suggests that PKA/Dc0 might be a direct target of Inc/Cullin-3. Indeed, in vitro studies suggest that PKA catalytic subunit is regulated through CHIP E3 ligase mediated ubiquitination and proteolysis [81]. However, there is no evidence for a role of Cullin-3 in PKA ubiquitination so far and Dc0 protein level does not seem to differ in inc mutants (Fig 5C and 5D). Given the elevated PKA kinase activity in inc mutants (Fig 5G–5I), and that their sleep phenotypes can be substantially rescued by suppressing PKA kinase activity with the expression of an active form of its regulatory subunit (Fig 5K–5S), it is more plausible that Inc indirectly regulates PKA activity through cAMP/PKA signaling regulators, for example Nf1 (Fig 3F), adenylate cyclase Rut and/or PKA regulatory subunits (Fig 3G and 3H). Thus, it will be interesting in the future to explore the direct Inc/Cullin-3 ubiquitination targets, which may contribute to the sleep deficits and memory hyperfunction of inc mutants.

cAMP/PKA signaling was among the first identified pathways controlling learning and memory [82]. Furthermore, cAMP/PKA signaling levels bidirectionally and negatively regulate sleep (Fig 7M) [28], consistent with our finding that the elevated PKA kinase activity of inc mutant directly drives their sleep deficits (Fig 5G–5S).

Notably, genetically and pharmacologically inducing sleep in the cAMP adenylate cyclase mutant rutabaga (which should suffer from reduced PKA activity) rescues their memory phenotypes [38]. Thus, lower cAMP/PKA signaling in rutabaga mutants might require excessive sleep for achieving a certain level of memory performance. In this regard, Dc0 overexpression in the mushroom body suppresses memory at young age [29,83]. Conversely, a mild reduction in the protein level of Dc0 in Dc0 heterozygotes prominently suppresses age-associated memory decline, while leaving memory unaffected at young age [29]. We here show that such genetic manipulation does not seem to provoke a major sleep phenotype in daily sleep amount in young wt/control animals (Figs 3A and 4A), while in inc background it specifically rescued the sleep defects (Figs 2A–2F and 3A–3E) and further increased memory performance (Fig 7A–7E). Different from rutabaga mutants whose changes of PKA activity might rather be outside the physiological range, Dc0 heterozygosity might mimic physiologically relevant scenarios (Fig 7M), allowing for maintenance of longevity and memory (Fig 7E and S6D Fig).

Intriguingly, Dc0 heterozygous animals are compromised in survival upon acute oxidative stress (Fig 6A), despite being ultimately longer-lived compared to wt/control animals upon normal aging (Fig 6B). Why would a reduction in PKA signaling promote longevity but fail to protect against oxidative stress? This may be explained by the differential regulation of downstream pathways by PKA in responding to acute stress versus chronic aging. Alternatively, PKA signaling might be differentially required for survival under acute oxidative stress and the chronic stress associated with aging. Indeed, acute H₂O₂ exposure directly activates PKA, followed by subsequent phosphorylation of substrate proteins in vivo [84], which might be critical for oxidative stress resistance. Interestingly, cAMP level, Dc0 protein level and PKA kinase activity in fly heads remain unchanged during early aging at age 20-day [29], whereas a significant reduction in Dc0 level is observed in 30-day-old fly wing neurons [85]. Thus, mushroom body cAMP/PKA signaling might become adverse for longevity and memory during aging.

PKA phosphorylates a wide range of targets, including critical synaptic proteins involved in neurotransmission and synaptic plasticity [86–88]. In mice, the AMP-activated protein kinase SIK3-mediated protein phosphorylation promotes sleep [89], while PKA suppresses sleep potentially by phosphorylating and inactivating SIK3 activity [90]. In SIK3 Sleepy mice, while their overall synaptic protein phosphorylation likely represents a signature of sleep need [89], its contribution to cognitive functions remains uncertain. In inc mutants, elevated PKA activity may induce a phosphorylation state driving wakefulness. Furthermore, Dc0^H2 heterozygosity likely renormalizes the phosphorylation state and subsequently the level of sleep of inc mutants, which further supports even higher memory functions (Fig 7A–7E). Future studies will utilize quantitative proteomic and phosphoproteomic analyses in combination with sleep and memory screens to elucidate these phosphorylation states, providing insights into mechanisms underlying age-associated memory decline.

Collectively, by intensively studying the seemingly paradoxical inc short sleep mutants, we propose a cellular and behavioral hyperfunction scenario likely originating from the developmental overgrowth of mushroom body Kenyon cells, an essential brain center for sensory integration and sleep regulation in Drosophila. Our work identifies a molecular signaling cascade elicited by Inc in controlling PKA signaling, which tunes the balance between memory functions and the amount of sleep. While behavioral hyperfunction, coupled with sleep deficits and cognitive imbalances, mirrors hallmark traits of neurodevelopmental disorders such as autism [91, 92], the potential etiology to these disorders remains obscure. As Inc functions as an adaptor for Cullin-3 E3 ligase-mediated ubiquitination [31,36], and Cullin-3 mutations have been associated with autism spectrum disorder [93,94], our findings provide a potential mechanistic connection between neurodevelopmental hyperfunction and the etiology of autism.

Pavlovian aversive olfactory conditioning was performed as previous reported [11,41]. Briefly, two aversive odors, 3-Octanol (OCT) and 4-methylcyclohexanol (MCH), served as behavioral cues (odors were diluted in mineral oil at a 1:100 ratio), and 120 V AC current electrical shocks were used as a behavioral reinforcer. Briefly, during one training session, around 100 naive flies were placed in a T-maze and exposed to the first odor (conditioned stimulus, CS⁺, MCH or OCT) paired with electric shocks (unconditioned stimulus, US) for 60 s. Afterwards, these flies were allowed to rest for 60 s followed by exposure to the second odor (CS⁻, OCT or MCH) without electric shocks for 60 s. During testing session, these flies were simultaneously exposed to CS ⁺ and CS⁻ at the two sides of the T-maze and they had 60 s to choose between the two odors. A reciprocal experiment, in which the second odor was paired with electric shocks as CS⁺, was performed. A performance index was calculated as the number of flies choosing CS⁻ minus the number of flies choosing CS⁺, divided by the total number of flies, and multiplied by 100 to get a percentage. The final performance index was the mean of the two performance indices from the two reciprocal experiments.

Courtship conditioning was performed similarly as previously reported [38]. Briefly, virgin wt and inc² males were collected at eclosion (Day1) and individually housed until age 5-day before training (Day5). In parallel, virgin wt females and additional wt males were separated, collected and housed in groups until Day4. At Day4, in order to freshly pre-mate virgin wt females within 12 h prior training, 4-day-old wt virgin females were crossed to the additionally collected wt males. Pre-mated wt females were used at Day5 for training individually housed wt and inc² males.

Sleep experiments were performed exactly as previously reported [41]. Briefly, sleep of single male flies was measured by Trikinetics Drosophila Activity Monitors (DAM2) from Trikinetics (Waltham, MA) in 12/12 h light/dark cycle with 65% humidity at 25 °C. 3 ~ 4-day-old single male flies were loaded into Trikinetics glass tubes (5 mm inner diameter and 65 mm length) which have 5% sucrose and 2% agar in one side of the tube. Flies were entrained for at least 24 h and their locomotor activity was measured at 1 min interval. Due to entrainment to new environment, data from the first ~ 24 h were excluded. A period of immobility without locomotor activity counts lasting for at least 5 min was determined as sleep [22]. Sleep from multiple days of recordings was analyzed and averaged using the Sleep and Circadian Analysis MATLAB Program (SCAMP) [96].

To carry out genetic modifier sleep screen for the X-linked inc mutants, we crossed female virgin inc mutants to either wt as control or isogenized autosomal candidate mutants to introduce heterozygosity of candidate mutations to inc male hemizygous mutants. For each measurement or biological replicate, differences between inc mutant with or without heterozygous mutations were calculated and then different measurements or replicates were pooled to generate Fig 2. At least two biological replicates were performed for each candidate mutant. As inc mutants, especially inc² mutants, show more profound sleep phenotypes during nighttime than during daytime, night P[wake] was specifically analyzed for the genetic modifier screen in Fig 2 so as to show major effects of different candidate heterozygous mutants on night sleep quality. For other P[wake] analysis, especially that Dc0 heterozygosity has similar effects between day and night for inc mutants, daytime and nighttime were averaged for simplicity in main figures. Detailed daytime and nighttime sleep parameters were separately shown in supplementary figures (S1–S6 Figs).

Climbing assay was carried out as previous described [41]. Briefly, flies were sorted into groups of 10 flies after eclosion and aged to 5-days old. During testing, flies were gently transferred to the testing tubes (2.1 cm diameter) and allowed to rest for at least 5 min. Afterward, both wt and inc mutant flies were simultaneously and quickly tapped down to the bottom. Flies reacted to the tapping and started to climb, and this process was visualized and videotaped.

Whole-mount brain immunostaining was performed similar to previous report [11,41]. Brains of 5-day old male flies were used for all staining experiments. Flies were kept on ice for 1–2 min and then dissected in ice-cold Ringer’s solution (130 mM NaCl, 5 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, 5 mM HEPES, 36 mM Sucrose, pH =  7.3). Dissected brains were fixed immediately in 4% Paraformaldehyde (w/v) in PBS for 30 min on an orbital shaker at room temperature. Samples were washed in 0.7% Triton-X (v/v) in PBS (0.7% PBST) for 20 min ×  3 times, followed by blocking with 10% normal goat serum (v/v) in 0.7% PBST for at least 2 h at room temperature on shaker. After overnight antibody (rabbit anti-GFP conjugated with Alexa 488, 1:500; Nc82, 1:50; anti-FLAG, 1:200; anti-FasII, 1:200) incubation at 4 °C in darkness, brains were washed in 0.7% PBST for 30 min ×  6 times at room temperature. Afterwards, for GFP Alexa 488 staining, brains were directly mounted on microscope slides in Vectashield and kept in a dark place at 4 °C before being scanned. For Nc82, FLAG and FasII staining, overnight secondary antibody (Alexa 488 or Cy5, 1:300) incubation at 4 °C in darkness followed by 6 times washing in 0.7% PBST was performed before mounting. Additional antibody information can be found within Table 1.

Whole-mount adult brain samples were imaged on a Leica TCS SP8 confocal microscope from Leica Microsystems, and images were obtained using the Leica Application Suite X. For analyzing PKA SPARK intensity and puncta, the mushroom body lobes were imaged using a 63× 1.40 NA oil objective with a voxel size of 0.2405 µm ×  0.2405 µm ×  1 µm at a speed of 400 Hz. For morphology analysis, mushroom body lobes and calyx were imaged using a 40 × 1.30 NA oil objective with a voxel size of 0.3788 µm ×  0.3788 µm ×  1 µm. All the parameters were kept constant for the same set of experiments. Image stacks were processed and analyzed with Fiji (https://fiji.sc/).

Longevity experiments were carried out exactly as previously reported [11]. Briefly, male flies were sorted into a population of approximately 20–25 flies per vial/replicate at the age of 2 days. To reduce the variability in life span, a few different cohorts of flies were used for each experiment. The longevity flies were regularly transferred onto fresh food every second day or over the weekend and the number of dead flies in each vial was recorded at each time of transfer until the death of the last fly of a replicate. Comparison of longevity curves was performed with GraphPad Prism using Log-rank analysis.

Statistics was performed as previously described [41]. GraphPad Prism 6 was used to perform most statistics and figures were generated using both Adobe Illustrator and Prism. Student t test was used for comparison between two groups and one-way ANOVA with Tukey’s post hoc tests were used for multiple comparisons between multiple groups (≥3). For mushroom body lobe analysis (Fig 7I, 7K and S6J, S6K Fig), Kruskal-Wallis test with Dunn’s multiple comparisons was used. Additional software information can be found within Table 1.