The endoribonuclease Arlr is required to maintain lipid homeostasis by downregulating lipolytic genes during aging

Abstract

While disorders in lipid metabolism have been associated with aging and age-related diseases, how lipid metabolism is regulated during aging is poorly understood. Here, we characterize the Drosophila endoribonuclease CG2145, an ortholog of mammalian EndoU that we named Age-related lipid regulator (Arlr), as a regulator of lipid homeostasis during aging. In adult adipose tissues, Arlr is necessary for maintenance of lipid storage in lipid droplets (LDs) as flies age, a phenotype that can be rescued by either high-fat or high-glucose diet. Interestingly, RNA-seq of arlr mutant adipose tissues and RIP-seq suggest that Arlr affects lipid metabolism through the degradation of the mRNAs of lipolysis genes - a model further supported by the observation that knockdown of Lsd-1, regucalcin, yip2 or CG5162, which encode genes involved in lipolysis, rescue the LD defects of arlr mutants. In addition, we characterize DendoU as a functional paralog of Arlr and show that human ENDOU can rescue arlr mutants. Altogether, our study reveals a role of ENDOU-like endonucleases as negative regulator of lipolysis.

Fig. 1. arlr mutants have reduced lipid storage.

All panels in this Figure are from females. a Relative expression levels of arlr at various developmental times. n = 5 or 6 biologically independent experiments. ns, no significant difference with the control, P = 0.7; ***P < 0.001 from 1 week to 5 weeks. b Lipid staining illustrating the increase of small LDs in aged arlr mutant flies. LDs (Nile Red, red) shown are adjacent to the nucleus (DAPI, blue) at the tissue mid-plane unless mentioned elsewhere. 3-week-old arlr mutants (arlr²⁶² and arlr³⁶⁴) have smaller LDs, a phenotype that is more severe in 5-week-old flies. c, d Reduction in average LD area in arlr mutants. The percentage of large LDs (>14 μm²) was reduced, while the small LDs (<4.0 μm²) was significantly increased (d’). For (d’), we sorted all LDs by size for each genotype and counted their numbers which were then divided to the total LD number. n = 6 biologically independent animals (3 samples in each animal). *P = 0.02 for arlr²⁶² and *P = 0.04 for arlr³⁶⁴ in (c); ***P < 0.001 in (d). e Relative reduction in TAG amount in arlr mutant flies. **P = 0.007 and *P = 0.01 at 3 weeks; ***P < 0.001 at 5 weeks. f, g Glucose and trehalose amounts of the whole body were normal. n = 3 biologically independent experiments in (e–g). ns P = 0.0535 for arlr²⁶² and P = 0.2254 for arlr³⁶⁴ in (f); ns P > 0.99 in (g). h Lifespan of female flies. Dashed lines indicate the median lifespan. n = 1 biologically independent experiment. ***P < 0.001. Statistical data were analyzed by one-way ANOVA with Tukey’s multiple comparison test in (a) and (c–g), and by Log-rank (Mantel-Cox) test in (h). Error bars represent SEM. Scale bars in (b) are 20 μm.

Fig. 2. The EndoU-like domain is required for the rescue of lipid defects.

All panels in this Figure are from females. a, b LDs in control flies (ppl > UAS-GFP and control). c 5-week-old arlr²⁶² mutant flies with small LDs. d Ectopic expression of full-length arlr (arlr-HA) in the fat body showed normal LDs. e Expression of full-length arlr restored LD defects. f Ectopic expression of arlr that lack the signal peptide (arlr^ΔSP-HA) in the fat body showed larger LDs. g Expression of arlr^ΔSP-HA restored LD defects in arlr mutants. h, i arlr lacking the EndoU-like domain (arlr^N-HA) failed to restore the lipid defects. j Quantification of the rescue phenotype of LD size following expression of arlr in arlr mutants. n = 6 biologically independent animals. Data were analyzed by Mixed-effects analysis Dunnett’s multiple comparison test (two-tailed). ns, no significant difference with the ppl > UAS-GFP control, P = 0.68 for Control, P = 0.98 for ppl > UAS-arlr-HA, P = 0.25 for arlr²⁶²;ppl > UAS-arlr-HA, and P = 0.3 for ppl > UAS-arlr^N-HA; ***P < 0.001. k Quantification of the TAG rescue following expression of arlr in arlr mutants. n = 3 biologically independent experiments. ns P > 0.99; **P = 0.003 for ppl > UAS-arlr^ΔSP-HA, **P = 0.003 for arlr²⁶²;ppl > UAS-arlr^ΔSP-HA; ***P < 0.001. Data were analyzed by one-way ANOVA with Tukey’s multiple comparison test. Error bars represent SEM. Scale bars in (a–h) are 20 μm.

Fig. 3. arlr mutants show accelerated lipid consumption.

All panels in this Figure are from females. a Lipid staining under standard food and starvation conditions. The amount of lipids in arlr mutants fed on standard food was comparable to the control at 1 week. After starvation for 24 h, the mutants showed fewer lipids. b High sugar diet (HSD) and high-fat diet (HFD) restored lipid levels in aged mutants. 5-week-old arlr²⁶² mutant flies fed with HSD or HFD for 24 h showed similar fat storage as control. c Quantification of the size of LDs. n = 6 biologically independent animals. ns P = 0.26 for Feed, P = 0.14 for HSD, P = 0.08 for HFD; ***P < 0.001. d Quantification of the relative TAG amounts. n = 3 biologically independent experiments. ns P = 0.31 for Feed, P = 0.12 for HSD, and P = 0.59 for HFD; ***P < 0.001. e The lifespan of arlr mutants was restored by high nutrition diet in females. Dashed lines indicate the median lifespan. n = 1 biologically independent experiment. ***P < 0.001; *P = 0.01. Statistical data were analyzed by independent two-sample t tests (two-tailed) in (c) and (d), and by Log-rank (Mantel-Cox) test in (e). Error bars represent SEM. Scale bars in (a) and (b) are 20 μm.

Fig. 4. Genetic interactions of arlr and lipolytic genes.

a Volcano plot of the lipid metabolism-associated genes. b Genetic interactions of arlr²⁶² and lipolytic genes in males at 5 weeks. c Quantification of LD size. n = 6 biologically independent animals. ***P < 0.001 for arlr²⁶² vs. arlr²⁶²;ppl>Lsd1–RNAi, for arlr²⁶² vs. arlr²⁶²;ppl>yip2–RNAi and for ppl > UAS-GFP vs. ppl>Lsd1–RNAi; **P = 0.001; *P = 0.02 for arlr²⁶² vs. arlr²⁶²;ppl > CG5162–RNAi, *P = 0.05 for ppl > UAS-GFP vs. ppl>Lsd-1-mCherry, *P = 0.05 for ppl > UAS-GFP vs. arlr²⁶²;ppl>Lsd-1-mCherry; ns P > 0.99 for ppl > UAS-GFP vs. ppl>regucalcin–RNAi, P = 0.99 for ppl > UAS-GFP vs. ppl>yip2–RNAi, and P = 0.63 for ppl > UAS-GFP vs. ppl > CG5162–RNAi. Data were analyzed by Mixed-effects analysis Dunnett’s multiple comparison test (two-tailed). d Relative TAG amounts. n = 3 biologically independent experiments. ***P = 0.001 for arlr²⁶²;ppl>regucalcin–RNAi and ***P < 0.001 for other three genotypes; *P = 0.03; ns P = 0.03 for ppl>regucalcin–RNAi, P = 0.97 for ppl>yip2–RNAi, and P = 0.69 for ppl>Lsd-1-mCherry. Data were analyzed by one-way ANOVA with Tukey’s multiple comparison test. e Adults expressing Lsd-1 showed subtle LD defects. The statistic of LD size is in (c). Panels (b–e) are from males. Error bars represent SEM. Scale bars are 20 μm.

Fig. 5. Arlr binds and regulates the mRNA level of lipolysis-associated genes.

a Luciferase reporter assay. In the presence of Arlr, the target genes were downregulated compared to the blank control. n = 3 biologically independent experiments. ns P = 0.54; *P = 0.03; **P = 0.006 for pmirGLO-regucalcin and **P = 0.002 for pmirGLO-CG5162. b Genomic loci analysis of the peaks using the IGV software. Black peaks show the binding ability of Arlr-GFP with selected mRNAs. c qRT-PCR results revealed higher relative enrichment of selected mRNAs in Arlr-GFP-immunoprecipitated RNA samples. PGRP-LE was a negative control and iab-7 was a positive control. n = 3 biologically independent experiments. ns P = 0.14 for PGRP-LE and P = 0.33 for iab-7; **P = 0.002; *P = 0.02 for regucalcin, *P = 0.04 for CG5162, and *P = 0.01 for yip2. Data in (a) and (c) were analyzed by independent two-sample t tests (two-tailed). Error bars represent SEM.

Fig. 6. DendoU and Arlr have overlapping function.

All panels in this Figure are from females at 5 weeks. a, b LDs in control and arlr²⁶² mutant flies. c Ectopic expression of dendoU in the fat body showed larger LDs. d Expression of dendoU restored LD defects in arlr mutants. e Knockdown of dendoU showed fewer LDs. f Expression of arlr restored the LD defects in dendoU-knockdown flies. g Reduction of both arlr and dendoU showed more severe LD defects than either mutants. Note that the experiment with dendoU RNAi was performed with a single RNAi line so the conclusion may need to be confirmed with additional lines in the future. h Quantification of the LD size in the above flies. n = 6 biologically independent animals. ***P < 0.001; ns P = 0.83 for ppl>dendoU-RNAi and P > 0.99 for other three genotypes. i Quantification of the LD number per sample. n = 3 biologically independent animals. *P = 0.04, 0.03, and 0.03, respectively. j Relative TAG amounts. n = 3 biologically independent experiments. ***P < 0.001; ns P > 0.99; *P = 0.01. Data in (i) were analyzed by Mixed-effects analysis Dunnett’s multiple comparison test (two-tailed) and others by one-way ANOVA with Tukey’s multiple comparison test. Error bars represent SEM. Scale bars in (a–g) are 20 μm.

Fig. 7. Human ENDOU rescue arlr mutants.

Panels a–e are from females at 5 weeks. a, b LDs in arlr²⁶² mutant flies and the ppl > UAS-GFP control. c Ectopic expression of human ENDOU in the fat body showed normal LDs. d Expression of human ENDOU restored the LD defects in arlr mutants. e Quantification of LD size. n = 6 biologically independent animals. ***P < 0.001. f Relative TAG amounts were restored by human ENDOU. n = 3 biologically independent experiments. ***P < 0.001. g CG5162 and yip2 were reduced by expressing human ENDOU. n = 3 biologically independent experiments. **P = 0.007; *P = 0.01, 0.05, and 0.03, respectively. Data were analyzed by one-way ANOVA with Tukey’s multiple comparison test in (d) and (e), and by independent two-sample t tests (two-tailed) in (g). Error bars represent SEM. Scale bars in (a–c) are 20 μm. h Model: In wild-type flies, EndoU binds to the mRNAs of lipolysis genes and degrades them (pathway 1) to keep the homeostasis of LDs (pathway 2). In arlr mutant flies, these mRNAs fail to be degraded, leading to accelerated lipid consumption.