Circadian- and Light-driven Metabolic Rhythms in Drosophila melanogaster

Abstract

Complex interactions of environmental cues and transcriptional clocks drive rhythmicity in organismal physiology. Light directly affects the circadian clock; however, little is known about its relative role in controlling metabolic variations in vivo. Here we used high time-resolution sampling in Drosophila at every 2 h to measure metabolite outputs using a liquid-chromatography tandem mass spectrometry (LC-MS/MS) approach. Over 14% of detected metabolites oscillated with circadian periodicity under light-dark (LD) cycles. Many metabolites peaked shortly after lights-on, suggesting responsiveness to feeding and/or activity rather than the preactivity anticipation, as observed in previous transcriptomics analyses. Roughly 9% of measured metabolites uniquely oscillated under constant darkness (DD), suggesting that metabolite rhythms are associated with the transcriptional clock machinery. Strikingly, metabolome differences between LD and constant darkness were observed only during the light phase, highlighting the importance of photic input. Clock mutant flies exhibited strong 12-h ultradian rhythms, including 4 carbohydrate species with circadian periods in wild-type flies, but lacked 24-h circadian metabolic oscillations. A meta-analysis of these results with previous circadian metabolomics experiments uncovered the possibility of conserved rhythms in amino acids, keto-acids, and sugars across flies, mice, and humans and provides a basis for exploring the chrono-metabolic connection with powerful genetic tools in Drosophila.

Keywords: Drosophila; LC-MS metabolomics; circadian clock; conserved oscillators; photic metabolism; ultradian rhythms.

Figure 1.

Identification of metabolites expressed rhythmically in fly bodies. (A) Metabolites that significantly cycle with 20- to 28-h periods in WT-LD (p < 0.05, q < 0.3, ARSER algorithm). (B) Metabolites that significantly cycle with 20- to 28-h periods in WT-DD (p < 0.05, q < 0.3, ARSER algorithm). (C) Density of circadian periods by group, whereby ARSER testing occurred at 8-h, 12-h, or 20- to 28-h searches.

Figure 2.

Effects of light and the clock on the pattern of cycling metabolites. (A) Time-course concentrations of the 4 carbohydrate species that oscillate with 20- to 28-h periods in WT and a 12-h period in per, with larger amplitudes noted in WT-LD compared with WT-DD and Per-LD by way of a paired t test (p = 0.004 and p = 0.013, respectively). (B) Overlapping cycling metabolites across all 3 groups with unique periods.

Figure 3.

Radial plot indicating phases for significant (A) 20- to 28-h cyclers in WT-LD and WT-DD and (B) 12-h cyclers in all 3 groups, as calculated by ARSER. (C) Phases of the 7 circadian overlaps in WT-LD and WT-DD. Note that the dark grey areas in (A) are indicative of overlap between WTLD and WTDD counts.

Figure 4.

Principal components scores plot for all samples using all detected metabolites. Variances explained for the first 2 components were 27.7% and 10.5%, respectively.

Figure 5.

OPLS-DA scores for discriminant analysis of all 3 groups in each of the 4 time windows. Significant models were fit to (B) ZT6-12 samples (R²X = 0.585, R²Y = 0.874, Q² = 0.654, p = 0.003) and (C) ZT12-18 samples (R²X = 0.458, R²Y = 0.488, Q² = 0.265, p = 0.02) but not to (A) ZT0-6 (p = 0.12) or (D) ZT18-24 (p = 0.24), noting the convergence of metabolite profiles in WT-LD and WT-DD at the end of the dark phase in ZT18-24.