Introduction

Sleep is essential for animal physiology, and understanding its molecular mechanisms has significant implications for both basic research and clinical applications. Although previous studies have identified key sleep regulators (Du et al., 2021), fundamental questions remain regarding the roles of cellular processes such as protein synthesis and degradation in sleep regulation.

Growing evidence suggests a dynamic interplay between sleep regulation and protein homeostasis. Protein synthesis is particularly active during sleep (Lyons et al., 2023), while Sleep deprivation affects translational initiation (Costa et al., 2019). The proteasome, a critical mediator of ubiquitin-dependent protein degradation, regulates synapse homeostasis by modulating ribosomal components (Costa et al., 2019), thereby influencing synaptic function. Intriguingly, proteasome components exhibit oscillating expression pattern at the transcriptional level in human cells (Desvergne et al., 2016). A recent study in Drosophila showed that mutation in a proteasome component alters sleep patterns (Fernández et al., 2020). Despite these advances, the precise mechanisms by which protein synthesis and degradation influence sleep, and how these processes interact, remain unresolved.

While several factors influencing proteasome assembly and subunit transcription have been identified (Kapetanou et al., 2022), the translational control of proteasome subunits remains poorly understood. Additionally, proteasome inhibition has been shown to alter ribosome function (Galimberti et al., 2016; Costa et sl., 2019; Palanca et al., 2014), raising the question of whether the proteasome itself is regulated by the status of the translation machinery. Exploring this relationship could provide key insights into the interplay between proteasome activity and translation.

In Drosophila, Mettl5 exhibits 18S ribosomal RNA m⁶A methyltransferase activity (Leismann et al., 2020) and interacts with Trmt112 to facilitate this function, a mechanism conserved in human (Leismann et al., 2020; van Tran et al., 2019). Although Mettl5 depletion abolishes m⁶A modification on 18S rRNA without impairing rRNA maturation, it affects fly orientation behavior. In mammalian cells, Mettl5-mediated 18S rRNA N6-methyladenosine (m⁶A) modification regulates stem cell fate determination and neural function (Wang et al., 2020). However, it remains unclear whether Mettl5 modulates global or transcript-specific translation profile in vivo, particularly in tissues rather than cell lines. Moreover, while Mettl5 is linked to intellectual disability (ID) with comorbid sleep disturbances, its mechanistic role in these disorders is unknown. Testing whether Mettl5 loss affects behaviors like sleep in Drosophila could clarify this connection.

We discovered that Mettl5 regulates sleep in Drosophila. To investigate the underlying mechanism, we performed RNA-seq and Ribo-seq on Mettl5 mutants, revealing dysregulation of multiple clock genes and proteasome components. This suggests Mettl5 coordinates protein production and degradation, which are crucial for protein homeostasis. Follow-up experiments confirmed that the protein level of Period was upregulated in Mettl5 mutants, contributing to the sleep phenotype. This study highlights that ribosome defects can perturb proteasome function, uncovering a mechanism that couples protein degradation with synthesis. Additionally, by mapping the genome-wide downstream gene profile of an rRNA methylation modifier, this study offers insights into the gene-specific roles of ribosome function.

Results

Mettl5 regulates sleep in Drosophila

Mettl5 contains a predicted N6 adenine specific nucleic acids methyltransferase domain. In order to study the function of this gene in sleep regulation, we generated CRISPR-Cas9 knockout mutants. We created two alleles. Mettl5^1bp, which deleted 1bp in the coding region results in a truncated version (Figure 1A) and down regulation of Mettl5 at mRNA level (Figure 1B). Another allele Mettl5^9bp, which results in 3 amino acids deleted version of Mettl5 (Figure 1A) does not cause significant change at mRNA level (Figure 1B). We found that heterozygous Mettl5¹ᵇᵖ mutants exhibited significantly reduced nighttime sleep (Figure 1C-G), particularly during early night (ZT12-ZT16; Figure 1C). Quantitative analysis revealed increased wakefulness at ZT14 in mutants (Figure 1H). ’awake %’ was used to indicate the percentage of awake fruit fly population at specific time points (e.g., ZT14). quantitative nighttime sleep latency measurements indicated a delayed sleep start in mutants (Figure S1G). In addition, Mettl5 expression was significantly up regulated during the sleep recovery period after mechanical sleep deprivation (Figure 1I-J). Mettl5 mutant displayed significantly increased sleep rebound in 24 hours after sleep deprivation (Figure 1K-L), indicating its effects on sleep homeostasis. Moreover, results of sleep arousal assay at ZT19 indicated that the percentage of aroused flies are significantly more than the control group (Figure 1M). These results demonstrate that the truncated Mettl5 mutation causes sleep deficits, establishing Mettl5 as a novel sleep regulator in Drosophila.

Figure 1

To confirm the specificity of the sleep phenotype in Mettl5^1bp, we performed genetic rescue experiments. Introduction of a single wild-type Mettl5 copy completely rescued both the reduced sleep amount and increased wakefulness at ZT14 (Figure 1N-R), demonstrating that these phenotypes specifically result from Mettl5 deficiency.

The observed expression pattern of Mettl5 further supports its sleep regulatory function. Using Mettl5-Gal4 reporter lines, we found expression in both neurons (colocalizing with ELAV staining; Figure S1A-C) and glial cells (colocalizing with Repo staining; Figure S1D-F). Behavioral analyses revealed additional mutant phenotypes consistent with sleep dysregulation. we tested the climbing ability of Mettl5 mutation which showed an increase (Figure S1H). RNAi knocking down of Mettl5 showed a consistent phenotype of down regulated sleep amount during the night time (Figure S1I-L).

Mettl5 regulates Drosophila sleep through its methyltransferase activity

Previous studies established that Mettl5 interacts with Trmt112 to regulate 18S rRNA m⁶A modification (Leismann et al., 2020; van Tran et al., 2019). Consistent with these findings, our LC-MS/MS analysis revealed significantly reduced m⁶A levels in both total RNA and 18S rRNA from heterozygous Mettl5^1bp (Figure 2A-B). To determine whether Mettl5’s sleep regulatory function depends on its methyltransferase activity, we performed two key experiments. First, we examined the phenotype of Trmt112 knockdown. We found that the Mettl5-Gal4 driven Trmt112 RNAi resulted in a similar phenotype to Mettl5^1bp (Figure 2C-G), suggesting that Mettl5 regulates sleep through its 18S rRNA m⁶A modification activity. The efficiency of the Trmt112 RNAi line was tested in previous study (López-Varea et al., 2021). More important, we performed a rescue experiment with a mutated form of Mettl5 that lacks the NPPF amino acids required for its methyltransferase activity (Figure2A) (Iyer et al., 2016). We observed that this mutant Mettl5 failed to rescue the sleep phenotype caused by Mettl5^1bp (Figure 2H-L). Together, these results demonstrate that Mettl5’s methyltransferase activity is essential for its role in sleep regulation, likely through its function in 18S rRNA m⁶A modification.

Figure 2

RNA-seq and Ribo-seq revealed the downstream gene profile of Mettl5

To better understand the downstream events of Mettl5, we performed RNA-seq and Ribo-seq to assess transcriptomic and translational changes in Mettl5^1bp. Principal Coordinates Analysis (PCoA) of RNA-seq and Ribo-seq datasets revealed clear separation between mutant and control groups (Figure S2A, B, D, E). High reproducibility was found among biological replicates (Figure S2C, F), confirming data quality. Transcriptome analysis identified 1,053 significantly differentially expressed genes (|log₂ (Fold Change) | ≥ 1 & p^adj< 0.05), comprising 217 upregulated and 836 downregulated transcripts (Figure 3A, B). Parallel ribosome profiling revealed 299 translationally regulated genes (|log₂ (Fold Change) | ≥ 0.265 & p^adj< 0.05), with 149 upregulated and 150 downregulated targets compared to the controls (Figure 3C). Heatmaps of the top 100 differentially expressed genes from both RNA-seq and Ribo-seq analyses were generated according to the rlog transformed values (Figure S3A, B).

Figure 3

To explore the biological implication of differentially expressed genes at transcriptional and translational levels, we performed Gene Ontology (GO) enrichment analysis. At the transcriptional level (Figure S4A, C), the most significantly enriched biological processes included cellular response to chemical stimulus, small molecule biosynthetic process, and fatty acid metabolic process (Figure S4A), with supporting gene networks (Figure S4C). Translational-level analysis (Figure S4B, D) revealed prominent enrichment for organic acid and amino acid metabolic processes, lipid catabolism, cellular respiration, and transport mechanisms. Notably, we identified strong associations with circadian regulation, including circadian regulation of gene expression and entrainment of the circadian clock.

The GO results usually contain a long list of enriched terms that have highly redundant information and are difficult to summarize. So, we performed a simplify Enrichment analysis that visualizes the summaries of clusters by word cloud for the GO enrichment result at transcriptional and translational levels. For transcriptionally significant DEGs (Figure S5A), major enriched processes included metabolic and stimulus-response pathways, wing disc and imaginal system development, ion-channel homeostasis and transport, and cytokinesis/cell cycle transitions. Translationally regulated genes additionally showed enrichment for sleep-wake cycles, circadian behavior, and neuronal cell death (Figure S5D). Cellular component analysis highlighted chromosomal structures, membrane-bound complexes, and vesicles (Figure S5B), while molecular functions predominantly involved enzymatic activities, ion/protein binding, receptor activities, and transcription processes (Figure S5C). The strong concordance between RNA-seq and Ribo-seq enrichment profiles (Figure S5E-F) further validated these findings.

Our Gene Set Enrichment Analysis (GSEA) identified pathways enriched in differentially expressed genes at both transcriptional and translational levels. To capture subtle but biologically important changes that might be excluded by stringent statistical thresholds, we analyzed gene sets ranked by log2 fold-change without p-value filtering. The GSEA-GO and GSEA-KEGG results, sorted by Normalized Enrichment Score (NES), revealed several key pathways (Figure S6A-D). Notably, the proteasome pathway showed significant suppression at both transcriptional and translational levels (P.adjust<0.05) (Figure 3E, I). Transcriptional-level analysis highlighted disruptions in arginine and proline metabolism and insect hormone biosynthesis (P.adjust<0.05) (Figure 3D, G). While, while translational changes predominantly affected fatty acid degradation and phosphonate/phosphinate metabolism (P.adjust<0.05) (Figure 3H-K).

To better understand the correlation between transcription and translation levels, we performed correlation analysis of differentially expressed genes based on fold change of DEGs (Figure 3L). 977 genes showed opposing trends between transcription and translation, while only 24 genes exhibited concordant changes. We identified 240 translation-specific and 3,309 transcription-specific DEGs, with each cluster undergoing separate enrichment analysis (Figure S7A-H). Notably, core clock genes (Clk, tim, etc.) displayed significant changes at both levels (Figure 3L).

To isolate translational effects, we calculated translation efficiency (TE) differences between Mettl5^1bp and w¹¹¹⁸controls (Figure 3M). Among 1,204 genes with significantly altered TE, GO enrichment highlighted amino acid metabolism and small molecule biosynthesis (Figure 3N), while KEGG analysis emphasized glycine/serine/threonine metabolism and one-carbon pool by folate pathways (Figure 3O, P).

Ribo-seq revealed the Mettl5^1bp led to changes of some global translation features

Using Ribo-seq, we compared global translation features between the w¹¹¹⁸ and Mettl5^1b groups. The TE (translation efficiency) correlation coefficients among the three biological replicates of each genotype ranged from 0.87 to 0.93, demonstrating high reproducibility within genotypes (Figure S8A-F). We found that the length of ribosome-protected RNA fragments (RPFs) was approximately 28nt (Figure S10A). The RPFs exhibited a significant 3-nt periodicity. Metagene analysis of individual 28nt reads revealed the distribution of RPFs across the gene locus in w¹¹¹⁸ and Mettl5^1bp, respectively (Figure S8, S9). The starting point of translation is 12nt upstream of start-codon and gradually disappears 15nt from stop-codon (Figure S8G-L). RPFs on the metagene plot distribution around the translation start and translation stop site (Figure S9A-F), different coding frames on CDS, 3’-UTR and 5’-UTR, respectively in two groups of samples all showed the periodicity (Figure S9G-L). As expected, all plots show an enrichment of P-sites in the first frame on the coding sequence but not the UTRs, in accord with ribosome protected fragments from protein coding mRNAs (Figure S9G-L).

Analysis of open reading frame (ORF) types revealed alterations in their distribution. Compared to w¹¹¹⁸, Mettl5^1bp exhibited an increased proportion of overlapping downstream ORFs (dORFs) and fewer non-overlapping dORFs (Figure S10B). In both groups, translated upstream ORFs (uORFs) were significantly shorter than untranslated uORFs (P < 0.05) (Figure S10C). Additionally, Mettl5^1bp displayed reduced read counts for both dORFs and uORFs (Figure S10D-E). Based on translation potential, uORFs were classified as either translated or untranslated, and motif analysis was performed separately for each category (Figure S10F-G).

Mutation of Mettl5 altered codon preference

Comparison of codon occupancy (A-site) between the two groups revealed that Mettl51bp preferentially used GAC and GAU, whereas w1118 favored UCC (Figure S10H). This trend was further supported by the cumulative frequency distribution of these codons (Figure S10I-K). Since GAC and GAU both encode aspartate (Asp), we analyzed Asp amino acid occupancy. Intriguingly, Asp was significantly enriched in w1118 during translation (Figure S10L), suggesting that Mettl51bp may exhibit altered translational regulation (see Discussion for details).

Metagene analysis of RPFs revealed distinct translation patterns between Mettl5^1bp and w¹¹¹⁸. The coding sequence (CDS) and flanking regions were segmented into 100 equal bins, and average RPF density was computed for each bin. The resulting plots illustrate differences in ribosome occupancy between Mettl5^1bp and w¹¹¹⁸ along the CDS (Figure S10M), near the translation start site (Figure S10N) and around the translation termination site (Figure S10O). Notably, Mettl5^1bp and w¹¹¹⁸ exhibited divergent ribosome occupancy patterns, particularly along the CDS and near the start codon (Figure S10M, N), suggesting potential differences in translation dynamics and initiation efficiency.

Mettl5 regulates the clock gene regulatory loop

Our findings demonstrate that Mettl51bp disrupts the core clock gene regulatory loop controlling circadian rhythm. We observed significant alterations in both transcriptional and translational levels of multiple clock genes, with Cry and Clock showing upregulation while Tim, per, vri, and pdp1 were downregulated at both levels (Figure 4A-B). Notably, per, vri, and pdp1 exhibited particularly pronounced downregulation in translation efficiency (Figure 4C). These changes occurred without affecting clock neuron morphology at different time points (Figure S11).

Figure 4.

The observed expression patterns revealed an unexpected regulatory relationship: while the canonical clock circuitry positions Per downstream of Clock, our finding that Clock was upregulated while per was downregulated suggests Per may actually function upstream of Clock in Mettl5-mediated regulation. Surprisingly, despite the transcriptional downregulation of per (Figure 4D), we detected increased PER protein levels through both immunostaining and western blot analyses (Figure 4E-H). Detection of the PER protein at different time points indicated that it was increased at both ZT 0 and ZT18 (Figure 4E-F). This apparent contradiction aligns with the observed circadian phenotype, as Mettl5 mutants showed significantly longer period lengths (Table 1), mirroring effects seen when PER stabilization results from reduced kinase activity as previously reported (Philpott et al., 2023). Genetic epistasis experiments further supported this model, with double mutants exhibiting clock gene phenotypes that position both Clock and Per downstream of Mettl5 (Figure 4I-N, Table 1). Together, these results indicate that Mettl5^1bp affects circadian regulation through mechanisms that extend beyond transcriptional control, likely involving post-translational regulation of PER protein stability.

Table 1

To investigate the factors mediating PER protein level changes in Mettl5 mutants, we examined the ubiquitin-proteasome pathway, which plays a well-documented role in Period protein degradation (Grima et al., 2002; Ko et al., 2002; Chiu et al., 2008). Notably, recent evidence indicates that N6-methyladenosine (m6A) regulates the ubiquitin-proteasome system in other biological contexts (Sun et al., 2023). Our integrated analysis of RNA-seq and Ribo-seq data revealed significant downregulation of multiple proteasome pathway components in Mettl5^1bp mutants at both transcriptional and translational levels (Figures 3E, 3I, 4O), suggesting impaired protein degradation capacity. Based on these findings, we propose a model where Mettl5 regulates circadian function through three interconnected mechanisms: first, by directly modulating proteasome components to control PER protein stability post-translationally; second, by transcriptionally and translationally regulating Per and other clock expression. In this model, Mettl5^1bp-induced proteasome downregulation leads to PER accumulation, which is responsible for the phenotypes (Figure 5).

Figure 5

Mettl5^1bp alters axon complexity

Previous studies have established a strong correlation between sleep homeostasis and synaptic complexity (Bushey et al., 2011). Additionally, synaptogenesis has been shown to enhance proteasome activity in axons (Costa et al., 2019). Given our observations of impaired sleep rebound following deprivation and altered expression of proteasome subunits in Mettl5^1bp mutants, we sought to examine potential effects on synaptic complexity.

To assess synaptic complexity, we adapted an established quantification method (Bushey et al., 2011) using Syt-GFP, a marker that colocalizes with endogenous synaptic vesicles, to visualize presynaptic morphology changes. Control experiments with UAS-Fmr1 and Fmr1 mutations successfully replicated the expected decrease and increase in syt-eGFP signal, respectively (Figure 6A-D). Strikingly, Mettl5^1bp mutants exhibited significantly increased syt-eGFP fluorescence in presynaptic terminals (Figure 6E-M), indicating altered synaptic complexity.

Figure 6

Discussion

Our study reveals that Mettl5, a known rRNA methyltransferase, modulates sleep through its RNA methylation activity. Through integrated RNA-seq and Ribo-seq analyses of Mettl5^1bpmutants, we identified Mettl5’s downstream targets at both transcriptional and translational levels. Further investigation demonstrated that Mettl5 influences sleep regulation by affecting two key pathways: the circadian clock gene network and the proteasome system. These findings provide novel mechanistic insights into sleep control, highlighting the coordinated role of protein synthesis and degradation in this process. Notably, our Ribo-seq analysis revealed that Mettl5^1bp alters fundamental translation features, including uORF translation efficiency and codon preference, suggesting rRNA methylation plays a regulatory role in these processes.

This discovery has important clinical implications, as METTL5, the human ortholog of Mettl5, is associated with intellectual disability (ID) when mutated (Richard et al., 2019). Our work expands the understanding of Mettl5’s molecular function and may inform potential therapeutic strategies for ID. The clinical relevance of our findings is underscored by reports of sleep disturbances, particularly reduced sleep duration, in ID patients - a phenotype that parallels our observations in Mettl5^1bp mutants. The mechanistic framework established in this study could explain these clinical sleep abnormalities. However, further validation in vertebrate models is needed to determine whether this regulatory mechanism is evolutionarily conserved and applicable to human sleep disorders.

As shown in Table 1, the Mettl5^1bp/+ mutant exhibits a robust long-period phenotype, with circadian rhythms significantly extended to 28.3 ± 0.4 hours compared to the wild-type’s 23.9 ± 0.05 hours. This prolonged period perfectly aligns with the observed behavioral phenotypes, including delayed nighttime sleep onset, later daytime waking, and the overall shift in sleep profile. This is indeed quite similar to previous report on Period3 variant (Zhang et al., 2016). We think that the prolonged circadian period contributes to the observed sleep phenotype. However, since total sleep time was significantly reduced in the mutant, we cannot attribute the phenotype solely to period lengthening. Furthermore, our 24-hour PER expression analysis in Mettl5 mutants revealed elevated PER protein levels at ZT1 and ZT18, while ZT6 and ZT12 showed no significant changes, with no apparent phase shift. These findings collectively suggest that the phenotype primarily results from PER protein stabilization and accumulation.

We found that Mettl5 heterozygotes showed significant reductions in total RNA and 18S rRNA methylation levels, contrasting with mouse studies where heterozygous knockouts maintained normal 18S rRNA m6A methylation (Sepich-Poore et al., 2022). This discrepancy may stem from either fundamental differences in rRNA methylation regulation between Drosophila and mice, or distinct biological consequences of knockdown versus knockout approaches, as complete gene elimination often triggers compensatory mechanisms (Teng et al., 2013; Rossi et al., 2015; Vu et al., 2015; Ma et al., 2019; El-Brolosy et al., 2019).

Our study uncovers a previously unrecognized connection between circadian clock genes and proteasome function. While previous work demonstrated that the circadian clock rhythmically regulates proteasome components in Drosophila fat bodies under dietary restriction (Hwangbo et al., 2023), we now show that Mettl5 modulates clock protein synthesis and degradation in clock neurons by influencing proteasome activity. This regulation likely occurs through Mettl5-mediated ribosomal methylation in clock neurons, which impacts the proteasome degradation pathway (Costa et al., 2019), ultimately altering clock protein dynamics.

Mettl5 represents a novel integrator of rRNA methylation and proteasome function, providing a mechanism to balance protein synthesis and degradation. Interestingly, we observed specific effects on both transcriptional and translational outputs, with particular proteasome subunits showing differential regulation. This specificity may arise from either selective translational control by Mettl5 or additional layers of regulation through protein-protein interactions. These findings suggest an intricate regulatory network coordinating these processes, though further studies are needed to elucidate the underlying mechanisms.

Our results reveal complex relationships between clock genes and sleep regulation. While cyc loss-of-function alleles show enhanced sleep rebound in females (Shaw et al., 2002), Mettl5^1bpmutants with elevated PER protein exhibit reduced rebound. Notably, Mettl5’s regulatory effects display tissue specificity - although absent in canonical clock neurons, Mettl5-Gal4 is expressed in distinct neurons and glia that appear crucial for sleep regulation. This expression pattern, combined with Clock’s broader distribution (Patop et al., 2023), suggests specialized circuits for sleep homeostasis that warrant further investigation.

Evidence indicates that sleep functions in development, metabolism and neuronal plasticity (Anafi et al., 2019). Sleep exerts effects on neuronal plasticity by modifying synapses. The synaptic homeostasis theory proposes that sleep has a role in downscale synaptic strength (Tononi and Cirelli, 2006). Indeed, synapse markers progressively decrease during sleep in both mammals and Drosophila brains (López-Varea et al., 2021; Vyazovskiy et al., 2008; Gilestro et al., 2009; Liu et al., 2010) More importantly, evidence indicates that sleep need and protein levels in synapse are tightly linked. Studies of the presynaptic active zone have shown that synaptic plasticity regulates sleep homeostasis (Huang et al., 2020). Genome-wide proteomic studies of synapse in mouse brain indicate that the synaptic proteins peak around dusk and dawn (Noya et al., 2019; Brüning et al., 2019). Sleep deprivation experiments have demonstrated that the sleep drive is significantly dependent on the cycling of proteins and phosphoproteins in synapses, in contrast to mRNAs (Noya et al., 2019; Brüning et al., 2019). These results suggest that the gene expression regulation at the protein level is crucial for sleep. Consistently, we detected changes in the quantity of axons in Mettl5^1bp, suggesting a possible regulation of neural circuits by Mettl5.

The mechanism by which METTL5 regulates translation warrants further investigation. Previous studies have demonstrated that METTL5 influences translation (Rong et al., 2020; Peng et al., 2022), but whether the mechanisms identified here are conserved across other systems remains an intriguing question. In our analysis, we observed increased usage of aspartate (Asp) codons in Mettl5 mutants. Notably, prior work has linked codon usage to PER protein function—specifically, a codon-optimized version of PER failed to rescue circadian rhythmicity in per mutant flies, unlike the wild-type version (Fu et al., 2016). Further analysis revealed that PER protein levels were elevated in these mutants, suggesting that codon optimization enhances PER expression (Figure 2B in Fu et al., 2016). Strikingly, when we examined the codon-optimized region from Fu et al. (2016), we found that GAC (Asp) was highly enriched, raising the possibility that Mettl5 mutation affects PER protein accumulation by altering GAC codon usage. Additional experiments will be needed to validate this hypothesis. Furthermore, we detected changes in upstream open reading frames (uORFs) in Mettl5 mutants, but their relationship to translational regulation requires further exploration.

Our study demonstrates that ribosomal components can exert gene-specific regulatory functions, building upon previous work showing context-dependent ribosome specialization (Simsek et al., 2017). In Mettl5 mutants, we observed distinct alterations in the translational efficiency of specific genes. These effects could potentially arise through two non-exclusive mechanisms: (1) Mettl5-mediated rRNA modifications may modulate ribosomal binding affinity for particular mRNA sequences, or (2) these modifications might contribute to the formation of specialized ribosome populations that preferentially translate specific subsets of mRNAs. Future studies will be required to distinguish between these possibilities and fully elucidate the underlying molecular mechanisms.

Material and methods

Drosophila Strains

Fly stocks used in this study were maintained under standard culture conditions. We used the w¹¹¹⁸ as the control strain. The following flies were obtained from Bloomington Stock Center: w¹¹¹⁸ (Bl: 5905), Mettl5-Gal4 (Bl: 19514), Repo-Gal4 (Bl: 7415), nSyb-Gal4 (Bl: 51941), pdf-Gal4 (Bl: 41286), Clk^JRK (Bl: 80927), per⁰¹(Bl: 80917), Fmr1^Δ50M(Bl: 6930), UAS-Trmt112RNAi (VDRC: 101515), UAS-DenMark, UAS-syt.eGFP (Bl: 33064), UAS-GFPstinger (from Yi Rao’s lab), UAS-Fmr1 (Bl: 6931), UAS-Mettl5 (FlyORF: F000760) are from Fly ORF collection. UAS-Mettl5m-3HA (generated in this study).

Mettl5^1bp and Mettl5^9bp mutants were generated by the CRISPR/Cas system as described previously (Cheng et al., 2020). The target gRNA was designed with an online tool: http://tools.flycrispr.molbio.wisc.edu/targetFinder/. A target sequence was chosen that has the sequence of 5’-GAGTGGGGTACTGTTCCAACAGG with the PAM sequence in bold. Mutations were verified by genomic PCR and sequencing.

Sleep behavior assays

All sleep assays were conducted in a controlled environment incubator maintained at 25 ± 1°C with 60% ± 5% relative humidity. We maintained a 12:12 light-dark cycle with lights on at ZT0 (06:30) and off at ZT12 (18:30). Fly activity was monitored using the Drosophila Activity Monitoring System (Trikinetics, Waltham, MA). Following a 2-day acclimation period, we recorded locomotor activity at 1-minute intervals for 3 consecutive days. Data analysis was performed using pysolo (Gilestro and Cirelli, 2009), with sleep defined as ≥5 minutes of continuous inactivity.

Mechanical sleep deprivation was performed using the SNAP method to keep flies awake for 12 h overnight (Shaw et al., 2002). Sleep deprivation is applied by mechanical stimuli using a timer-controlled rotating shaker. The intensity of stimuli was 1500 rpm once every minute (2 s/1 min) (Shimizu et al., 2008). Baseline sleep was established from 24-hour recordings prior to deprivation. Sleep loss and recovery were quantified according to previous publication (Cirelli C et al., 2005).

Regarding the ’awake %’ metric, it indicates that at specific time points (e.g., ZT14), the percentage of awake fruit fly population at that moment. At ZT19, we evaluated arousal thresholds by administering a standardized gentle mechanical stimulus. Responsiveness was determined by monitoring activity for 1 minutes post-stimulation. Flies showing no activity during this window were scored as non-responsive. We calculated arousal percentages from the proportion of flies that awakened in response to stimulation.

Circadian rhythm of individual male flies was measured using the Drosophila Activity Monitoring (DAM) System (Trikinetics). Male flies were loaded individually into glass tubes with a length of 65 mm and an inner diameter of 5 mm. The tubes contained standard cornmeal fly food at one end and were sealed with a cotton stopper at the other end. The flies were entrained to a 12 h light/12 h dark cycle for 3 days and then released to constant darkness for at least six days to measure their rhythmicity.Data analysis is done on a Macintosh computer running the FaasX (Fly activity analysis suite) software.

Statistical analysis

All statistical analyses were performed using GraphPad Prism 5 software. For sleep parameter comparisons, we used nonparametric tests including the two-tailed Mann-Whitney test for pairwise comparisons and one-way ANOVA with Tukey’s post hoc test for multiple comparisons. qPCR data were analyzed using unpaired Student’s t-tests, while lifespan data were evaluated with Log-rank (Mantel-Cox) tests. The specific statistical test used for each experiment is indicated in the corresponding figure. In all analyses, a p-value of less than 0.05 was considered statistically significant.

Quantitative PCR

Total RNA was extracted from cells and tissues using the TRNzol Universal Reagent (Tiangen #DP4-02). For cDNA synthesis, we employed the PrimeScript RT reagent Kit with gDNA Eraser (TAKARA #RR047A) following the manufacturer’s protocol. Quantitative PCR was carried out using SuperReal PreMix Plus (SYBR Green) (Tiangen #DP4-02). RP49 served as the endogenous reference gene for normalization across samples. All experiments included three independent biological replicates to ensure reproducibility. The primers used in this experiment are RP49-F: CGGTTACGGATCGAACAAGC; RP49-R: CTTGCGCTTCTTGGAGGAGA; Mettl5-F: CGGTTTCTGGAGGTGGC; Mettl5-R: GCTGGCGTCGATGTTGTAC..

Imaging and analysis of axon volume

We quantified small LNvs axon volume using ImageJ’s Object Counter 3D plugin to measure pixel counts from raw images. Image processing involved applying a standardized threshold that clearly visualized intact axons in control samples. For consistent measurements, we specifically analyzed the axon span between the first axonal bifurcation and the terminal tip, as indicated by the white rectangular markers in Figure 8.

Sample collection and library construction for Ribo-seq and RNAseq

For sample collection, w¹¹¹⁸ and Mettl5^1bp Drosophila were harvested at ZT15, immediately transferred to centrifuge tubes, and flash-frozen in liquid nitrogen. Fly heads were subsequently separated and collected while frozen, with three biological replicates prepared for each genotype. Each sample was equally divided for parallel RNA-seq and Ribo-seq analyses. For Ribo-seq library preparation, samples were lysed in buffer containing 50 mg/mL cycloheximide (Novogen, China) to preserve ribosome positioning, followed by RNase I digestion to generate ribosome-protected fragments (RPFs). Monosomes were isolated using MicroSpin S-400 HR size-exclusion chromatography, followed by rRNA depletion and PAGE purification to select 20-38 nt RPFs. Purified fragments underwent end repair, adapter ligation, reverse transcription, and PCR amplification before Illumina PE150 sequencing. For RNA-seq, total RNA was extracted using TRIzol reagent, with cDNA libraries prepared and sequenced using Illumina PE150 by Novogen. Raw sequencing data in FASTQ format were processed to remove adapter sequences, reads containing N bases, and low-quality reads, while simultaneously calculating Q20/Q30 scores and GC content to generate clean reads for downstream analysis.

RNA-seq analysis

We performed genome alignment and transcriptome analysis using the following pipeline: First, we built a Hisat2 index (v2.0.5) for the Drosophila melanogaster reference genome (dm6 assembly). Clean paired-end reads were then aligned to this reference using Hisat2 (v2.0.5) (Kim et al., 2019). The resulting alignments were processed using StringTie (v1.3.3b) (Pertea et al., 2015) for reference-based transcript assembly. For gene-level quantification, we used FeatureCounts (v1.5.0) (Liao et al., 2014) to count reads mapped to each annotated gene. Differential expression analysis between w¹¹¹⁸ and Mettl5^1bp was conducted with DESeq2 (v1.20.0) (Pertea et al., 2015), with genes meeting both criteria (adjusted p-value < 0.05 and absolute log2 fold change ≥ 1) considered statistically significant.

Ribo-seq analysis

We implemented a comprehensive Ribo-seq analysis pipeline beginning with quality filtering using Bowtie (Liao et al., 2014) to remove reads aligning to non-coding RNAs (rRNA, tRNA, snoRNA, snRNA from FlyBase Release 6.13) with a 2-mismatch allowance (-v 2). The remaining ribosome-protected fragments (RPFs) were mapped to the Drosophila melanogaster genome (FlyBase Release 6.13) using STAR (v2.7.3a) (Dobin et al., 2013), followed by transcript-level alignment to protein-coding sequences using Bowtie (v1.2.2) (Langmead et al., 2009) with parameters "-a -v 2". CDS-aligned RPFs were quantified using featureCounts (Subread v1.6.3) and normalized as RPKM. Differential expression analysis was performed with DESeq2 (v1.14.1) (Love et al., 2014) using thresholds of |log2FC| ≥ 0.265 and padj < 0.05, while translation efficiency differences were assessed using RiboDiff. Data quality was verified through riboWaltz (v1.1.0) (Lauria et al., 2018) for 3-nt periodicity and reading frame analysis, with Ribocode (Xiao et al., 2018) employed for P-site positioning and uORF motif analysis. Functional enrichment analyses (GO, KEGG, GSEA) were conducted using clusterProfiler (v4.5.2.002) (Wu et al., 2021), with results visualized through ggplot2-generated plots and GO term simplification performed using SimplifyEnrichment (Gu and Hübschmann, 2022).

Immunofluorescence experiments

We performed immunofluorescence staining on 7-15 days old adult flies (unless otherwise specified). Flies were anesthetized with CO₂ and dissected in ice-cold 0.03% PBST (1× PBS with 0.03% Triton X-100; Sigma, T9284). Samples were fixed in 2% paraformaldehyde (PFA) for 55 minutes at room temperature (RT), followed by four 15-minute washes in 0.03% PBST at RT. After blocking overnight at 4°C in 10% Normal Goat Serum (NGS; in 1× PBS with 2% Triton), samples were incubated with primary antibodies for 24 hours at 4°C. Primary antibodies included rat anti-Elav (DSHB, 9F8A9; 1:200) and mouse anti-Repo (DSHB, 8D12; 1:200), diluted in antibody buffer (1.25% PBST, 1% NGS). Following four 15-minute washes in 1× PBS with 1% Triton at RT, samples were incubated overnight at 4°C with secondary antibodies: Alexa Fluor™ 568 (Thermo Fisher, A11004; 1:200) and Alexa Fluor™ 647 (Thermo Fisher, A21247; 1:200). After four additional 15-minute washes in 1× PBS with 1% Triton at RT, samples were mounted using DAPI-containing antifade mounting medium (Solarbio, S2110).

Images were acquired using a Leica SP8 confocal microscope with LAS X software, applying auto Z-brightness correction when needed for signal uniformity. Images were processed in Adobe Photoshop CS6 and figures assembled in Adobe Illustrator 2020. PDF immunofluorescence followed the same protocol. All experiments included ≥3 biological replicates, each containing ≥10 flies. Fluorescence intensity was quantified using ImageJ.

Western Blotting

Protein samples were prepared by homogenizing approximately 30 fly heads in RIPA lysis buffer (150 mM NaCl, 1.0% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 8.0) supplemented with protease (CW2200S) and phosphatase (CW2383S) inhibitor cocktails according to manufacturer specifications. Lysates were mixed with 2× SDS loading buffer, boiled at 100°C for 5 min, and immediately cooled on ice.

For immunoblotting, membranes were probed with rabbit anti-PER primary antibody (1:5000 dilution; kindly provided by Dr. Jeffrey Price’s laboratory, University of Missouri-Kansas City) overnight at 4°C, followed by incubation with HRP-conjugated goat anti-rabbit IgG secondary antibody (1:1500; ABclonal, AS014) for 4 hr at room temperature. Protein signals were detected using ECL substrate (ABclonal, RM00021P) and imaged with an Amersham ImageQuant 800 system (GE Healthcare). Band intensities were quantified using ImageJ software, with three biological replicates performed for statistical reliability.

LC-MS/MS analysis of m⁶A levels

Total RNA was isolated from cells and tissues using TRNzol Universal Reagent (Tiangen, #DP4-02). For 18S rRNA purification, we separated total RNA by polyacrylamide gel electrophoresis followed by gel extraction. Our analysis utilized ribonucleoside standards including adenosine (rA) and N6-methyladenosine (N6 mA), with a mobile phase consisting of methanol:ddH2O (v/v).

For each biological replicate, 1 μg of total RNA or purified 18S rRNA was digested to single nucleosides using Nucleoside Digestion Mix (NEB, #M0649). Proteins were precipitated by adding a 4:1 methanol:digest ratio and incubating at -20°C for 2 hours. Quantitative analysis was performed using multiple reaction monitoring (MRM) with the following transitions: 268.10275→136.0621 (rA), 282.11835→150.0774 (N6 mA).

Negative geotaxis RING assay

Climbing assays were performed following a 12-hour recovery period after CO2 anesthesia, using 200 flies per genotype distributed across 10 vials (20 flies/vial) marked at 90 mm height. Each assay consisted of three trials separated by 15-minute intervals, initiated by sharply tapping vials three times to induce negative geotaxis, with the number of flies reaching the 90 mm mark within 10 seconds recorded per trial. Five groups were tested per genotype (15 trials total). All assays were video recorded under standardized conditions with the camera positioned 20 cm from vials and uniform backlighting provided by a white open-faced box to ensure consistent imaging quality.

Supplementary figures

Figure S1

Figure S2

Figure S3

Figure S4

Figure S5

Figure S6

Figure S7

Figure S8

Figure S9

Figure S10

Figure S11

Data availability

The RNA-seq and Ribo-seq data from this publication have been deposited to the NCBI bioproject database https://www.ncbi.nlm.nih.gov/bioproject/ and assigned the identifier PRJNA994860.

Acknowledgements

This work was supported by National Natural Science Foundation of China (Grant No. 32070492 and 32122017) to Juan Du. We would also like to acknowledge the 2115 Talent Development Program of China Agricultural University.

Additional information

Author contributions

XY, XW and TF designed and performed the experiments, analyzed the data and wrote the manuscript. XY performed the bioinformatics analysis, analyzed the data and edited the manuscript. YR analyzed data and edited the manuscript. JD designed experiments, analyzed data, wrote and edited the manuscript. All authors have read and approved the final manuscript.

Funding

MOST | National Natural Science Foundation of China (NSFC) (32070492)

• Juan Du

MOST | National Natural Science Foundation of China (NSFC) (32122017)

• Juan Du