Supplemental Methods and Material, Tables A-K, Figs A-W.

Table A. Numbers of rhythmic genes in each population. Table B. Number of genes with loss in rhythmicity (P > 0.5) in cave populations compared to rhythmic expression in surface (FDR < 0.1 and FDR < 0.05). Table C. Known circadian regulators that are arrhythmic in one or more cave populations. Table D. Timing of peak expression of core clock genes (primary and accessory loops) compared between zebrafish and A. mexicanus populations. Table E. Phase shifts between surface and cave populations. Table F. The number of significant circadian binding motifs identified in promoter proximal regions of genes with evidence for transcription (p<0.05) in the surface population. Table G. Arrhythmic genes in cave populations where motif sequences are also lost. Table H. Average timing difference in peak expression of circadian feedback loop targets. Table I. Number of genes with a significant differential rhythmicity score. Table J. Genes associated with GO term DNA-repair are upregulated in cave populations more than expected by chance. P-values based on Fisher’s exact tests. Table K. A. mexicanus orthologs of genes that are light induced in zebrafish are not more often upregulated in cavefish compared to surface fish. P-values for each surface-cave comparison produced with a hypergeometric test. Fig A. Raw reads per sample for Molino. Fig B. Raw reads per sample for Pachón. Fig C. Raw reads per sample for surface fish. Fig D. Raw reads per sample for Tinaja. Fig E. A. PC1 and PC2 (explaining 19.1% and 18% of variation, respectively) show that the primary axes of differentiation among samples is ecotype. B. PC3 (explaining 7.1% of variation) separates Molino from other populations. Fig F. Cave populations show shifts in phase at per1a/b and cry1a, with gene expression peaking later in cave populations compared to the surface population. Expression is represented as normalized read counts. Fig G. A-B. In the circles are the circadian phase distributions of predicted targets of RRE and E-BOX for surface fish. Grey bars represent the proportion of each motif seen in each phase. Highlighted in light grey are intervals of phase-specific enrichment for surface fish for each motif. Genes with the RRE motif (C) and EBOX (D) motifs show shifts in the timing of peak expression in cavefish populations. Fig H. In the circles are the circadian phase distributions of predicted targets of the D-Box (NFIL3) for surface fish. Grey bars represent the proportion of motifs seen in each phase. Highlighted in light grey is the interval with most significant phase-specific enrichment for surface fish. Fig I. Region of interest used in all brain images (white box). Regions include the optic tectum (TeO) and periglomerular grey zone (PGZ) shown in relation to coronal section stained with DAPI. Scale bar is 50μM. Fig J. DAPI staining in brain (‘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. A. DAPI channel for sections included in Fig 4A. B. DAPI channels for sections included 4B. Fig K. Expression patterns of per1a and arntl1a in Tinaja and Molino brain images adjusted to correct for oversaturation. Fig L. Temporal expression patterns of (A) per1a and (B) arntl1a in midbrain and liver tissue in Astyanax mexicanus populations. In-situ staining of rorca (A) and rorcb (B) using RNAscope in 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. Images are representative sections of two fish collected per time point, per population. Scale bar is 25μM. Fig M. Temporal expression patterns of (A) rorca and (B) rorcb in midbrain and liver tissue in Astyanax mexicanus populations. In-situ staining of rorca (A) and rorcb (B) using RNAscope in 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. Images are representative sections of two fish collected per time point, per population. Scale bar is 25μM. Fig N. Per1a expression at each time point measured with RNAseq in whole fry and in the brain and liver with RNA FISH. Grey dotted lines represent a loess regression for visualization purposes. Fig O. Arntl1a expression at each time point measured with RNAseq in whole fry and in the brain and liver with RNA FISH. Grey dotted lines represent a loess regression for visualization purposes. Fig P. Rorca expression at each time point measured with RNAseq in whole fry and in the brain and liver with RNA FISH. Grey dotted lines represent a loess regression for visualization purposes. Fig Q. Rorcb expression at each time point measured with RNAseq in whole fry and in the brain and liver with RNA FISH. Grey dotted lines represent a loess regression for visualization purposes. Fig R. Analysis of mutagenesis in aanat2 crispant F₀ fish. A. Genotyping gel of uninjected control and injected embryos. A portion of aanat2 genomic region was amplified by PCR from DNA extracted from individual embryos. Labeled D is half of the PCR product that was digested with BbvI. Unlabeled is undigested PCR product. Indels can disrupt the restriction enzyme site, leading to undigested PCR product in injected embryos. B. Diagram of aanat2 gene based on the surface fish reference genome (Ensembl v98). Boxes indicate exon and lines indicate introns. The empty boxes are 5’ and 3’ UTR and the closed boxes are coding sequence. A gRNA was designed targeting exon 1. The gRNA target site is in blue and the PAM sequence is in red. The underlined sequence is the BbvI restriction enzyme recognition sequence used for genotyping. The arrow indicates the predicted Cas9 cut site. Gene structure was generated using http://wormweb.org/exonintron and then modified. C. Sequence of wildtype surface fish and sequence of six clones from the restriction enzyme resistant band from aanat2 injected individuals. The total number of base pairs less than the wildtype sequence is indicated to the right of each clone. Fig S. Analysis of mutagenesis in rorca crispant F₀ fish. A. Genotyping gel of uninjected control and injected embryos. A portion of rorca genomic region was amplified by PCR from DNA extracted from individual embryos. Labeled D is half of the PCR product that was digested with Cac8I. B. Diagram of rorca gene based on the Pachón Ensembl v93 genome. Boxes indicate exon and lines indicate introns. The empty boxes are UTR and the closed boxes are coding sequence. A gRNA was designed targeting exon 6. The gRNA target site is in blue and the PAM site is in red. The underlined sequence is the Cac8I restriction enzyme recognition sequence used for genotyping. The arrow indicates the predicted Cas9 cut site. Gene structure was generated using http://wormweb.org/exonintron and then modified. C. Sequence of wildtype surface fish and sequence of 3 clones from the restriction enzyme resistant band from rorca injected individuals. The total number of base pairs more or less than the wildtype sequence is indicated to the right of each clone. Fig T. A. Expression of per2, a light-activated clock gene, over the course of the day in surface and cave populations (JTK_cycle p-values: surface, p = 0.09, q = 0.7; Molino, p = 0.07, q = 1; Pachón, p = 0.003; q = 0.16; Tinaja, p = 0.016; q = 0.22). Rhythmicity was not found to be different between populations (all comparisons, S_DR p>0.68, q = 1). B. Base level expression of per2 between populations. Per2 has lower base level expression in the surface than Pachón (log₂-fold change = 0.82) and Tinaja (log₂-fold change = 0.56), but higher expression than Molino (log₂-fold change = 1.3). Fig U. Core circadian genes and melatonin regulator aanat2 show differentiated rhythmicity from surface fish in at least one cave population. Fig V. Exo-rhodopsin has robust rhythmic expression in the surface population (p = 9.26 x 10⁻⁵, q = 0.006), but is not strongly in rhythmic in cave populations (Pachón, p = 0.04, q = 0.32; Tinaja, p = 1, q = 1.0; Molino, p = 0.37, q = 1.0). Fig W. Phylogenetic tree of species used for RELAX analysis for changes in selection intensity. Black branches indicate “reference” branches, where blue branches (cavefish lineages of A. mexicanus) have been used as foreground branches to test for changes in selection intensity.

(PDF)