Modulation of RNA processing genes during sleep-dependent memory

Abstract

Memory consolidation in Drosophila can be sleep-dependent or sleep-independent, depending on the availability of food. The anterior posterior (ap) alpha'/beta' (α'/β') neurons of the mushroom body (MB) are required for sleep-dependent memory consolidation in flies fed after training. These neurons are also involved in the increase of sleep after training, suggesting a coupling of sleep and memory. To better understand the mechanisms underlying sleep and memory consolidation initiation, we analyzed the transcriptome of ap α'/β' neurons 1 hr after appetitive memory conditioning. A small number of genes, enriched in RNA processing functions, were differentially expressed in flies fed after training relative to trained and starved flies or untrained flies. Knockdown of each of these differentially expressed genes in the ap α'/β' neurons revealed notable sleep phenotypes for Polr1F and Regnase-1, both of which decrease in expression after conditioning. Knockdown of Polr1F, a regulator of ribosome RNA transcription, in adult flies promotes sleep and increases pre-ribosome RNA expression as well as overall translation, supporting a function for Polr1F downregulation in sleep-dependent memory. Conversely, while constitutive knockdown of Regnase-1, an mRNA decay protein localized to the ribosome, reduces sleep, adult specific knockdown suggests that effects of Regnase-1 on sleep are developmental in nature. We further tested the role of each gene in memory consolidation. Knockdown of Polr1F does not affect memory, which may be expected from its downregulation during memory consolidation. Regnase-1 knockdown in ap α'/β' neurons impairs all memory, including short-term, implicating Regnase-1 in memory, but leaving open the question of why it is downregulated during sleep-dependent memory. Overall, our findings demonstrate that the expression of RNA processing genes is modulated during sleep-dependent memory and, in the case of Polr1F, this modulation likely contributes to increased sleep.

Keywords: D. melanogaster; Drosophila; RNA processing; memory; neuroscience; sleep.

Figure 1.. Differential gene expression after training in mushroom ap α′/β′ neurons.

(A) 5- to 7-day-old mixed sex wCS flies were exposed to one of the following three conditions: Trained-Fed, Trained-Starved and Untrained-Fed. Only Trained-Fed flies are expected to increase sleep after treatment and thus form sleep-dependent memory (Chouhan et al., 2021). Brain dissection, single cell suspension and cell sorting were used to extract ap α′/β′ neurons in each of these three different conditions, and bulk-sequencing of the sorted cells was conducted (created in BioRender.com/k87y777). (B) We sequenced four samples for each condition and subjected them to differential gene analysis of pairwise comparison of the three conditions. We found that 59 genes are significantly different in TrainedStarved vs. TrainedFed and UntrainedFed vs. TrainedFed. (C) Gene ontology (GO) analysis of these 59 genes using FlyEnrichr (https://maayanlab.cloud/FlyEnrichr/) revealed that they encode cellular components of the 90 S preribosome, Cajal body, DNA-directed RNA polymerase complex, nuclear euchromatin, and condensed chromosome.(D) Heatmap of the 59 DEGs including two genes labeled in red, CG13773 (Polr1F) and Regnase-1, which affect sleep and are the focus of this study due to their impacts on sleep. DEGs were identified by DESeq2 with the cutoff of FDR <0.1 and fold change >1.5. Heatmaps were plotted by using TPM values of genes for each sample; data were log-transformed and scaled row-wise for visualization.

Figure 1—figure supplement 1.. PathwayPCA visualization and GO analysis of differentially expressed genes across three conditions.

(A) PathwayPCA plot of all the differentially expressed genes (DEGs) across the three different conditions. The plot shows that genes responsible for PC1, which accounted for 35% of variance, largely encode proteins involved in transcription and biosynthesis, including RNA biosynthesis processes. (B) Top5 significant Gene Ontology (GO) biological process terms of all differentially expressed genes (DEGs) from three pairwise comparisons are shown. The x-axis represents the -log10 of the p-value, and the numbers on the bars indicate the count of genes identified for each GO term.

Figure 2.. The sleep screen of differentially expressed genes identifies Polr1F and Regnase-1 as sleep-regulating genes.

(A–B) Flies carrying the ap α′/β′ neuron driver R35B12-Gal4 were crossed with flies carrying UAS-RNAi constructs targeting DEGs identified from RNA-seq analysis. 5–7 days old female F1 progeny were loaded onto Trikinetics DAM monitors to measure their sleep in a 12 hr light: 12 hr dark (12:12 LD) cycle. Mean total sleep (A) and nighttime sleep (B) were calculated by Pysolo and the difference between experimental flies and Gal4 and RNAi controls was calculated separately for each independent experiment; average values comparing each experimental to its Gal4 control (X-axis) and RNAi control (Y-axis) are shown in the plots. Of all the lines screened, knockdown of Polr1F and Regnase-1 had strongest effects on sleep, producing an increase and decrease in sleep, respectively. (C–F) show the representative sleep traces of R35B12-Gal4>polr1F^RNAi flies and R35B12-Gal4>regnase1 RNAi. N=23–32 per genotype from two independent replicates combined are shown in E and F respectively, and bar graphs show mean + SEM. p values for each comparison were calculated using the Kruskal-Wallis test with Dunn’s multiple comparisons test. **p<0.01, ***p<0.001, ****p<0.0001.

Figure 2—figure supplement 1.. Effect of Polr1F and Regnase-1 knockdown on activity and sleep architecture.

(A–B) Total activity of flies is not altered by knockdown of Polr1F and Regnase-1. One-way ANOVA was used to calculate the p-values for each comparison. (C–D) Knockdown of Polr1F significantly increases the nighttime average sleep episode (SE) length in R35B12-Gal4>polr1 F RNAi flies, while knockdown of Regnase-1 reduces it in R35B12-Gal4>regnase1 RNAi flies. The p-values for each comparison were calculated using the Kruskal–Wallis test with Dunn′s multiple comparisons test. (E–F) Knockdown of Polr1F significantly reduces the nighttime average sleep episode (SE) number in R35B12-Gal4>polr1FRNAi flies, but it is not significant compared to control groups in R35B12-Gal4>regnase1 RNAi flies. The p-values for each comparison were calculated using the Kruskal–Wallis test with Dunn′s multiple comparisons test. ns = not significant, p>0.05, *p<0.05, **p<0.01, ***p <0.001, ****p < 0.0001.

Figure 3.. Acute and chronic effects of pan-neuronal knockdown of Polr1F on sleep in adult flies.

(A) Schematic representation of transient and chronic sleep measurements in nSyb-GS>polr1f^RNAi flies (created in BioRender.com/t22a408). (B) Representative sleep traces and transient activity plot of flies expressing Polr1F RNAi under the control of an inducible pan-neuronal driver (nSyb-GS>polr1f RNAi) with and without RU treatment. (C) Quantification of sleep during the first 3 hours (ZT 9–12) after F1 progeny flies were loaded into RU- or RU+ DAM tubes at ZT8-T9. Sleep was measured starting at ZT9. N=39–40 individual flies per replicate with data from three independent replicates combined. The Mann-Whitney test was used to compare RU+ group and RU- groups. (D) Representative average sleep traces of nSyb-GS>polr1f RNAi in the RU- and RU+ DAM tubes for 3 consecutive days. Chronic sleep effects of pan-neuronal knockdown Polr1F were measured based on sleep data from day 3 to day 5. (E) Quantification of average total sleep of nSyb-GS>polr1f^RNAi and controls in the DAM tubes from (D). Unpaired t-test was used to compare between RU- and RU+ groups. ****p<0.0001.

Figure 3—figure supplement 1.. Adult specific knockdown of Polr1F promotes sleep.

(A) Representative sleep traces of adult flies expressing Polr1F RNAi under the control of R35B12-Gal4/Tubulin-gal80ts. At permissive temperature, Gal80ts inhibits Gal4 activation of UAS; however, upon switching to restrictive temperature, Gal80ts is deactivated, leading to downstream activation of the UAS-regulated gene. (B) Sleep patterns were quantified over 4 consecutive days, with flies under permissive temperature on the first and last days and restrictive temperature on the second and third days. (C) Sleep quantification during the initial 3 hr period (ZT 0–3) after light onset for each day is shown. N=46–48 individual flies per replicate with data from three independent replicates combined. The Kruskal-Wallis test with Dunn′s multiple comparisons test was used to compare three groups. ns = not significant, p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 3—figure supplement 2.. Acute and chronic effects of pan-neuronal knockdown of Regnase-1 on sleep in adult flies.

(A) Schematic representation of the protocol for transient and chronic sleep measurements in nSyb-GS>regnase-1 RNAi flies (created in BioRender.com/t22a408). (B) Representative sleep traces plot of flies expressing Rengase-1 RNAi under the control of an inducible pan-neuronal driver (nSyb-GS>regnase-1 RNAi) with and without RU treatment (C) Quantification of sleep during the first 3 hr (ZT 9–12) after F1 progeny flies were loaded into RU- or RU + DAM tubes at ZT8-T9. Sleep was measured starting at ZT9. N=22–24 individual flies per replicate with data from two independent replicates combined. The Mann-Whitney test was used to compare RU + group and RU- groups. (D) Representative average sleep traces of nSyb-GS>regnase-1 RNAi in the RU- and RU + DAM tubes for three consecutive days. Chronic sleep effects of knockdown of Regnase-1 were measured based on sleep data from day 3 to day 5. (E) Quantification of average total sleep of nSyb-GS>regnase-1 RNAi and controls in the DAM tubes from (D). Unpaired t-test was used to compare between RU- and RU + group. ns = not significant.

Figure 3—figure supplement 3.. Adult specific knockdown of Regnase-1 has no effect on sleep.

(A) Representative sleep traces of adult flies expressing Regnase-1 RNAi under the control of R35B12-Gal4/Tubulin-gal80ts. (B) Sleep was quantified over four consecutive days, with flies under permissive temperature on the first and last days and restrictive temperature on the second and third days. N=46 individual flies per replicate with data from three independent replicates combined. The Kruskal-Wallis test with Dunn′s multiple comparisons test was used to compare three groups. ns = not significant, p>0.05, *p<0.05, **p<0.01.

Figure 3—figure supplement 4.. Adult specific knockdown of Polr1F or Regnase-1 has no effect on peak activity.

(A) Maximal activity per minute for R35B12-Gal4; TubGal80ts > polr1 F RNAi was quantified over four consecutive days, with flies under permissive temperature on the first and last days and restrictive temperature on the second and third days. (B) Maximal activity per minute for R35B12-Gal4; TubGal80ts > Regnase-1 RNAi was quantified. N=40–48 individual flies per replicate with data from three independent replicates combined. The Kruskal-Wallis test with Dunn′s multiple comparisons test was used to compare three groups. ns = not significant, p>0.05, *****p<0.0001.

Figure 4.. Regnase-1 expression is essential for sleep-dependent and sleep-independent memory.

(A) Schematic representation of the memory test protocol (created in BioRender.com/z46i047). (B, C) Sleep-dependent and sleep-independent memory tests were conducted under fed and starved conditions, respectively. Knockdown of Regnase-1 significantly reduces long-term memory performance in both fed and starved flies. However, knockdown of Polr1F in ap α′/β′ neurons does not affect long-term memory performance. N≥6 biological replicates, each replicate containing 100–150 flies. (D, E) Fed UAS-regnase-1-RNAi/+and R35B12/+flies exhibit a significant increase in sleep after training, while R35B12-Gal4>regnase-1 RNAi flies fail to show a comparable increase in post-training sleep. The total sleep in the ZT8-ZT12 interval is shown in (E). Polr1F knockdown in ap α′/β′ neurons does not affect the post-training increase in sleep. N≥32. (F) Compared to R35B12-Gal4/+and + /UAS-Regnase-1 RNAi flies, R35B12-Gal4>reganse-1 RNAi flies show a significant decrease in the performance index in short-term memory. N≥6 biological replicates, each containing 100–150 flies. ns = not significant, p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 4—figure supplement 1.. Regnase-1 overexpression in the ap α′/β′ neurons does not affect sleep.

(A) The efficacy of the RNAi used in the experiments was verified by real-time quantitative PCR. The different RNAi lines were crossed with nSyb-GS, and adult progeny were placed in either RU- or RU + tubes overnight before being dissected for qPCR. Welch’s test was used for statistical analysis across all three RNAi, with *p<0.05 indicating significance. (B) Representative sleep traces and quantification of adult flies expressing Regnase-1 under the control of R35B12-Gal4. Sleep data were averaged over three consecutive days, with N=31–32 individual flies per replicate, and combined data from two independent replicates were analyzed.

Figure 5.. Knockdown of Polr1F results in high translation.

(A) The nSyb-GS >polr1 f RNAi flies exhibit a significant increase in pre-rRNA levels. (B) The ex vivo puromycin immunostaining assay was used to measure translation in dissected whole brains. The results show that knockdown of Polr1F using the pan-neuronal nSyb-GeneSwitch (GS) system increases translation relative to control flies that were not treated with RU. The normalized mean grayscales from the RU- and RU +groups are compared using an unpaired t-test. The analysis includes data from 8 to 11 flies per group, with results from two independent replicates combined. Scale bar: 100 µm. (C) The schematic model (created in BioRender.com/h27z926) illustrates the major transcriptome features of ap α′/β′ neurons under trained and fed conditions, revealed in this manuscript. Two RNA processing genes Polr1F and Regnase-1 are prominently downregulated during memory consolidation in trained and fed flies, respectively. Polr1F is involved in regulating ribosomal RNA synthesis, and its decrease in levels in trained and fed flies promotes sleep and translation. In contrast, Regnase-1 is localized in the ribosome and involved in mRNA decay, and its downregulation causes deficits in sleep and memory.

Figure 5—figure supplement 1.. rRNA inhibitor (CX-5461) feeding does not affect sleep.

Feeding of the ribosome RNA inhibitor CX-5461 at a concentration of 0.2 mM does not affect sleep in fed or starved flies. Statistical analysis shows that there is no significant difference between CX-5461-fed and control flies. The sample size is N=32 per group from two independent replicates. ns, not significant.