Repeated evolution of circadian clock dysregulation in cavefish populations

Abstract

Circadian rhythms are nearly ubiquitous throughout nature, suggesting they are critical for survival in diverse environments. Organisms inhabiting largely arrhythmic environments, such as caves, offer a unique opportunity to study the evolution of circadian rhythms in response to changing ecological pressures. Populations of the Mexican tetra, Astyanax mexicanus, have repeatedly invaded caves from surface rivers, where individuals must contend with perpetual darkness, reduced food availability, and limited fluctuations in daily environmental cues. To investigate the molecular basis for evolved changes in circadian rhythms, we investigated rhythmic transcription across multiple independently-evolved cavefish populations. Our findings reveal that evolution in a cave environment has led to the repeated disruption of the endogenous biological clock, and its entrainment by light. The circadian transcriptome shows widespread reductions and losses of rhythmic transcription and changes to the timing of the activation/repression of core-transcriptional clock. In addition to dysregulation of the core clock, we find that rhythmic transcription of the melatonin regulator aanat2 and melatonin rhythms are disrupted in cavefish under darkness. Mutants of aanat2 and core clock gene rorca disrupt diurnal regulation of sleep in A. mexicanus, phenocopying circadian modulation of sleep and activity phenotypes of cave populations. Together, these findings reveal multiple independent mechanisms for loss of circadian rhythms in cavefish populations and provide a platform for studying how evolved changes in the biological clock can contribute to variation in sleep and circadian behavior.

Fig 1

A. Overlap of genes with rhythmic expression between populations. B. Heatmap of genes with rhythmic patterns in surface fish (Río Choy), ordered by gene phase, compared to expression in cave populations. Each column represents gene expression at a single-time-point, sampled every four hours from 0–20 hours. Redder boxes correspond to higher expression. C. Identifying genes with changes in rhythmicity between cave and surface populations. Genes with greater amplitude values have larger differences between their expression peak and trough, where genes with greater periodicity show stronger cyclical oscillation patterns (see Methods). Genes are colored based on their S_DR q-values. Genes with positive values for both show increased rhythmicity in the surface population, where genes with negative values show increased rhythmicity in cave populations.

Fig 2

A. Key circadian genes with changes in rhythmicity in cave populations (see Fig U in S1 Text for all core circadian genes with changes in rhythmicity). Colored lines represent a loess regression of gene expression through time for each population. B. Simplified schematic of the circadian feedback loops based on proposed interactions in zebrafish[9,55]. Grey boxes indicate genes that are either arrhythmic or show significantly reduced rhythmicity between cave and surface. White boxed genes do not show significant differences between cave and surface. Highlighted in yellow is the core loop. Bright yellow circles represent regulating protein complexes. Red lines indicate negative regulation, black lines indicate positive regulation. Notably, cry4 does not repress Clock/Bmal activation in zebrafish and may play a photoreceptor function[9,55]. Genes that are not boxed did not show evidence of rhythmic expression in any cave or surface population. Dotted lines are for visual clarity. Genes without an annotated ortholog in cavefish were not included in the schematic.

Fig 3

A. Distribution of peak expression time of rhythmic transcripts in each population. Bars indicate subjective day and night. B. Cycling genes on average show a delay in phase in cave populations relative to their phase in the surface population.

Fig 4

Temporal expression patterns of (A) per1a (cyan) and arntl1a (magenta) and (B) rorca (cyan) and rorcb (magenta) in brain and liver tissue in Astyanax mexicanus populations. In-situ staining using RNAscope in the midbrain (‘B’, top panels for each timepoint) and liver (‘L’, bottom panels for each timepoint) of surface fish and cavefish (Pachón, Tinaja, Molino) at CT0, CT8, and CT16. Each time point is a single fish sample with probes separated into two channels. Images are representative sections of two fish collected per time point, per population. Scale bar is 25μM.

Fig 5. Quantification of RNA FISH.

A-D. Expression of per1a, arntl1a, rorca and rorcb in brains (A, C) and livers (B, D) at CT0, CT8, and CT16 in surface fish and cavefish populations. RNAscope probe channel intensity was normalized to DAPI channel intensity in identically sized, anatomically matched ROIs to provide an estimate of mRNA expression per cell (see Methods for full details of RNA FISH analysis). Biological replicates are shown as colored points on graph and represent a brain or liver sample collected from a single individual. Bars reflect mean and error bars show SEM of biological replicates. Statistics were calculated for each mRNA probe by comparing each cave population mean to control mean (Surface) within timepoints using ordinary 2-way ANOVA. Dunnett’s test was used to correct for multiple comparisons across populations, timepoints. Adjusted p-values < 0.05 are reported with * using the following scheme: 0.0332 (*), 0.0021 (**), 0.0002 (***), and <0.0001 (****).

Fig 6

A. Melatonin under light-dark conditions. Ten fish larvae were pooled together, homogenized and melatonin was extracted for each datapoint. Melatonin increased at night in Surface fish, Pachón and Molino population (two-way ANOVA analysis). There was no significant change in Tinaja cavefish. *, p = 0.0117; *** p = 0.0002; **** p<0.0001. B. Melatonin under dark-dark conditions. Ten fish larvae were pooled together, homogenized and melatonin was extracted for each data point. Melatonin increased at night in Surface fish population (two-way ANOVA analysis). There was no significant change in Tinaja, Pachón, or Molino cavefish. ***p-value = 0.0004.

Fig 7. Mutant aanat2 and rorca fish reveal a role for these genes in sleep behavior in A. mexicanus.

A. Total sleep is not significantly altered between control and aanat2 crispant surface fish (Mann-Whitney U, p = 0.99). B-C. Day sleep is not significantly altered between WT and crispant aanat2 fish (Mann-Whitney U, p = 0.22). Night sleep is significantly reduced in aanat2 crispants compared to WT controls (Mann-Whitney U, p = 0.015). D. Total sleep is significantly reduced in crispant rorca fish compared to WT controls (Mann-Whitney U, p<0.0001). E-F. Day sleep was not significantly reduced between WT and rorca crispants (unpaired t-test, p = 0.56). Night sleep is significantly reduced in rorca crispants compared to WT controls (Mann-Whitney U, p<0.0001).