Discovery and Visualization of Age-Dependent Patterns in the Diurnal Transcriptome of Drosophila

Abstract

Many critical life processes are regulated by input from 24-hour external light/dark cycles, such as metabolism, cellular homeostasis, and detoxification. The circadian clock, which helps coordinate the response to these diurnal light/dark cycles, remains rhythmic across lifespan; however, rhythmic transcript expression is altered during normal aging. To better understand how aging impacts diurnal expression, we present an improved Fourier-based method for detecting and visualizing rhythmicity that is based on the relative power of the 24-hour period compared to other periods (RP24). We apply RP24 to transcript-level expression profiles from the heads of young (5-day) and old (55-day) Drosophila melanogaster, and reveal novel age-dependent rhythmicity changes that may be masked at the gene level. We show that core clock transcripts phase advance during aging, while most rhythmic transcripts phase delay. Transcripts rhythmic only in young flies tend to peak before lights on, while transcripts only rhythmic in old peak after lights on. We show that several pathways, including glutathione metabolism, gain or lose coordinated rhythmic expression with age, providing insight into possible mechanisms of age-onset neurodegeneration. Remarkably, we find that many pathways show very robust coordinated rhythms across lifespan, highlighting their putative roles in promoting neural health. We investigate statistically enriched transcription factor binding site motifs that may be involved in these rhythmicity changes.

Keywords: Fourier spectrum; aging, Drosophila; circadian rhythm; diurnal expression; transcriptomics.

Figure 1

Comparison of RP24 score to F₂₄ score. A. Scatterplot comparing the RP24 score to F₂₄, defined as the fold change of the power of the 24-hour component of the expression profile over random permutations. B. The expression profiles of the two most rhythmic transcripts, CG16798-RA and CG44195-RA in our young (5-day) flies. C. The Fourier power spectrum of the two most rhythmic transcripts in young flies. D. The expression profiles in young flies of the two least rhythmic transcripts based on the RP24 q-value, pain-RA and Nup54-RA, that have a F₂₄ score greater than 3. E. The Fourier power spectrum of pain-RA and Nup54-RA in young flies.

Figure 2

Rhythmicity changes with age. A. The total amount of rhythmicity, as defined by the distribution of RP24 over all transcripts, changes slightly toward greater rhythmicity after aging. B. Euler diagram shows that the specific transcripts with statistically significant RP24 values is substantially different between young and old flies. C. A scatterplot where each transcript is represented by a dot. The x-axis value is the log-transformed RP24 value in young flies, and the y-axis position is the log-transformed RP24 value in old flies. Red dots correspond to transcripts that are significantly rhythmic (FDR ≤ 0.05) in young flies and not in old, blue dots correspond to transcripts that are significantly rhythmic in old flies and not young, and purple dots correspond to transcripts that are rhythmic in both young and old flies. D. Histogram shows the RP24 distribution for ELC compared to RLC transcripts in young flies. E. Histogram shows the RP24 distribution of RLC compared to LLC transcripts in old flies.

Figure 3

Phase changes with age. A. Circular histogram depicts the phase of RLC transcripts in young compared to the distribution of the phase of the same transcripts in old. The light-dark cycles is represented on the histogram as a 24-hour clock, with lights-on (ZT0) at the top and lights-off (ZT12) at the bottom. Phases are binned into increments of 30 minutes. B. Circular histogram similar to panel C compares the phase distribution of ELC transcripts in young to LLC transcripts in old. C. Dot-and-arrow scatterplot shows phase advance for RLC transcripts. Distance from the origin is to the dot is RP24 in young, scaled between 0 and 1 with a logistic function, and the phase is shown as the angle from the positive y-axis (ZT0) to the phase of that gene on the same 24-hour clock as panels C and D. Each transcript is represented moving from a phase/rhythmicity in young (scatter point) to a phase/rhythmicity in old (tip of arrow). Transcripts belonging to the core clock mechanism are shown in orange and labeled. D. Dot-and-arrow scatterplot shows phase delay for RLC transcripts.

Figure 4

Pathway analysis of rhythmic transcript groups. A. Scatterplot representing differentially rhythmic pathways. The x- and y-axes represent the average rhythmicity in young and old, respectively, for DAVID clusters computed from ELC, RLC, and LLC transcripts. The color map represents the DAVID enrichment score, and the size of each dot is proportional to the number of transcripts in that cluster. B. The ELC transcripts with thioredoxin-like fold and glutathione metabolism related function show increased expression on average, but with reduced rhythmicity in young compared to old flies. Phase/rhythmicity dot-and-arrow scatterplots (as in Figure 3E–F) are shown for transcripts from this pathway exhibiting phase advance and phase delay. Color map defines a unique color for each transcript based on phase ordering. C. The RLC transcripts with mitochondrial translation function show consistent rhythmicity in young and old with little change in phase. D. Rhythmicity and phase changes of LLC transcripts with function related to amino acid synthesis.

Figure 5

Vision and rhabdomere related transcripts show diverse changes in rhythmicity. A. The expression profiles of ELC transcripts related to vision comparing the same transcripts in young and old flies. The color of each curve corresponds to phase ordering in young. B. Dot-and-arrow scatterplots show the change in phase and RP24. As with Figure 3E–F, the phase is shown as the angle from the positive y-axis (ZT0) on a 24-hour clock. The color corresponds to the same phase ordering for the same transcripts as panel A. C. The expression profiles for RLC transcripts related to vision, comparing young and old. D. Dot-and-arrow scatterplots show the change in phase and RP24, similarly as panel B, for the same transcripts in panel C. The color map corresponds to phase ordering in young for both panel C and D.

Figure 6

Results of pathway analysis of differentially expressed rhythmic transcript groups. The y-axis of each point corresponds to the log fold change of average expression in old over young for all transcripts in each cluster; the x-axis shows the RP24 averaged over all transcripts in each cluster. Hue denotes DAVID enrichment score, and the size of each point corresponds to the number of transcripts in the cluster. Labeled clusters have at least 5 transcripts, an absolute value log fold change ≥ 0.75, and an enrichment score ≥ 1.3.

Figure 7

Transcription factor binding site motif analysis of promoter regions. A. Heatmap shows enrichment of transcription factor binding site motifs in promoters of transcripts belonging to DAVID clusters. DAVID clusters are on the y-axis, and transcription factors corresponding to binding site motifs are on the x-axis. DAVID clusters shown have at least 10 transcripts, and transcription factors pass an expression threshold of 5 FPKM in young or old flies based on our experimental data. Hue represents q-value on a logarithmic scale. Asterisks mark results with q-value ≤ 0.05. Bar plot shares an x-axis with the heatmap, and shows young (red) and old (blue) expression of each transcription factor. Dimer partners are separated with a forward slash (/); the expression of the dimer partner with the lowest expression (bolded) is represented in the bar plot. Duplicate transcription factor symbols have different motif position weight matrices, and are distinguished by inclusion of sequencing platform information. Diagram of workflow is show in the upper left-hand corner of panel A. B. Boxplot compares RP24 (in old flies) for LLC transcripts involved in calcium ion transmembrane transport, with and without significant So transcription factor binding site motifs. C. Gene expression profile of so in young (red) and old (blue) flies. Lights-on is Zeitgeber Time (ZT) 0, and lights-off is ZT12. D. Binding site motif locations for So in promoters of LLC transcripts involved in calcium ion transmembrane transport. Motif locations are denoted by a downward-pointing (forward strand) or upward-pointing (reverse strand) green triangle. Promoters are ordered by length of first intron (grey bar following rightward pointing arrow) within subgroups containing or lacking motif instances for So.