Population Genomics of Premature Termination Codons in Cavefish With Substantial Trait Loss

Abstract

Loss-of-function alleles are a pertinent source of genetic variation with the potential to contribute to adaptation. Cave-adapted organisms exhibit striking loss of ancestral traits such as eyes and pigment, suggesting that loss-of-function alleles may play an outsized role in these systems. Here, we leverage 141 whole genome sequences to evaluate the evolutionary history and adaptive potential of single nucleotide premature termination codons (PTCs) in Mexican tetra. We find that cave populations contain significantly more PTCs at high frequency than surface populations. We also find that PTCs occur more frequently in genes with inherent relaxed evolutionary constraint relative to the rest of the genome. Using SLiM to simulate PTC evolution in a cavefish population, we show that the smaller population size and increased genetic drift is sufficient to account for the observed increase in PTC frequency in cave populations without positive selection. Using CRISPR-Cas9, we show that mutation of one of these genes, pde6c, produces phenotypes in surface Mexican tetra that mimic cave-derived traits. Finally, we identify a small subset of candidate genes that contain high-frequency PTCs in cave populations, occur within selective sweeps, and may contribute to beneficial traits such as reduced energy expenditure, suggesting that a handful of PTCs may be adaptive. Overall, our work provides a rare characterization of PTCs across wild populations and finds that they may have an important role in loss-of-function phenotypes, contributing to a growing body of literature showing genome evolution through relaxed constraint in subterranean organisms.

Keywords: adaptation; cavefish; population genomics; relaxed constraint; stop codons.

Graphical abstract

Fig. 1.

Overview of Mexican tetra populations. Focal populations of surface and cave Mexican tetra in which PTCs were identified. The 12 populations represent two genetically distinct lineages, Lineage 1 and Lineage 2, with cave populations repeatedly derived in independent events. Map indicates location of Mexican tetra cave populations, San Luis Potosí and Tamaulipas, Mexico.

Fig. 2.

Transcript truncation by PTCs. Distribution of the percent of transcript cDNA truncated by PTCs for all transcripts affected by computationally confirmed PTCs. Bars to the left represent less of the cDNA truncated (starting with the bin of 0% to 1% of the gene truncated) and bars to the right represent more of the cDNA truncated (ending with 99% to 100% truncated). Subsequently, bars to the left represent PTCs that occur towards the end of a gene, whereas bars to the right represent PTCs that occur towards the beginning of a gene. PTCs private to cave and surface populations did not have a significantly different distribution from one another.

Fig. 3.

PTC distribution by population. a) Distribution of PTCs in cave and surface Mexican tetra populations. The number of PTCs shown for each population is the total PTCs found in the population divided by the number of individuals sampled. While both cave and surface populations have many PTCs within the population and fewer PTCs which occur at high frequency, cave populations harbor significantly more PTCs at high frequency than are present in surface populations. b) Total PTCs found in each cave broken down by presence in other cave populations (number of PTCs normalized by population sample size). PTCs unique to one cave are found only in the focal cave population. PTCs also present in caves from the same lineage are shared in one or more caves from the same lineage as the focal cave. Finally, PTCs present in caves from both lineages are found in at least one cave from each lineage. As expected, PTCs are shared more frequently among caves from the same evolutionary lineage than between lineages. c) Comparison of total and high-frequency PTCs found on chromosome 12 in surface, cave, and simulated populations. Values for Rio Choy (n = 9), Mante (n = 10), Rascón (n = 14), and Pachón (n = 19) are the observed PTC counts found in the population genomic dataset. Values for SLiM simulated cave populations (population size 1 thousand, 32 thousand, and 250 thousand individuals, respectively) are from 19 random individuals (same sample size as Pachón) output at the end of the simulation (see Materials and Methods).

Fig. 4.

Putative relationships between PTC genes and cave-derived phenotypes. We explored documented functional consequences of 50 annotated candidate PTC genes (those that occurred at high frequency [≥ 0.8] in one or more cave populations, never occurred in a homozygous state in surface fish, and for which no transcripts were predicted to splice out the PTC) in zebrafish, mice, and humans by leveraging the databases Zfin, MGI, IMPC, Orphanet, OMIM, and DECIPHER, and review of current literature (supplementary table S9, Supplementary material online). Each colored bubble represents a suite of phenotypes related to cave-derived traits. The genes listed in each bubble had documented phenotypes related to the phenotype indicated in the bubble. The asterisk denotes candidate genes with evidence of selection in population(s) where the PTC is present, and the triangle denotes candidate genes that overlap a QTL (established by previous studies of Mexican tetra) for a trait related to the phenotype indicated in the bubble. The genes listed here represent the most compelling candidates for PTCs with the potential to contribute to cave adaptation.

Fig. 5.

Functional Impacts of PTCs in pde6c. a) Location of PTCs in pde6c across taxa. PTCs are attributed to the regression of visual protein networks in species that burrow underground or are active strictly at night, including the cape golden mole (C. asiatica) (Emerling and Springer 2014), nine-banded armadillo (Dasypus novemcinctus) (Emerling and Springer 2015), and multiple bat species (Blumer et al. 2022). Black bars represent each exon, and all introns have been scaled to 100 bp. b) Visual comparison of 9 dpf injected pde6c CRISPants with reduced eyes compared with uninjected, wild-type siblings. c) Comparison of relative eye size (eye area/fish length) in 9 dpf injected pde6c CRISPants compared with uninjected, wild-type siblings. Pde6c CRISPant surface fish show significantly reduced eye size; P = 0.01 for both the left and right eyes. d) Comparison of OMR between injected pde6c CRISPants and uninjected, wild-type siblings. OMR index represents the proportion of the total possible distance traveled in the direction of the visual cue. Positive values indicate positive directional response to visual cue, while negative values indicate an inability to discern and orient towards visual cues. Pde6c CRISPants have significantly lower OMR response and often do not orient to the visual cue in the OMR assay and therefore likely have visual defects compared with wild-type siblings (P < 0.0001).