From darkness to discovery: evolutionary, adaptive, and translational genetic insights from cavefish

Abstract

How genotype determines phenotype is a well-explored question, but genotype-environment interactions and their heritable impact on phenotype over the course of evolution are not as thoroughly investigated. The fish Astyanax mexicanus, consisting of surface and cave ecotypes, is an ideal emerging model to study the genetic basis of adaptation to new environments. This model has permitted quantitative trait locus mapping and whole-genome comparisons to identify the genetic bases of traits such as albinism and insulin resistance and has helped to better understand fundamental evolutionary mechanisms. In this review, we summarize recent advances in A. mexicanus genetics and discuss their broader impact on the fields of adaptation and evolutionary genetics.

1. Genotype-phenotype relationships in environmental adaptation

The question of why an organism looks and behaves the way it does, and not any other way, has long captivated biologists. The genotype of an organism and the environment it inhabits together determine and constrain its phenotype [1]. A deep and nuanced understanding of genotype-phenotype-environment relationships is therefore crucial to any theory of phenotypic evolution. High-throughput sequencing technologies have made it possible to sequence the genomes of diverse organisms. This has led to an explosion of comparative analyses that enable prediction of gene or regulatory network function through homology, and transcriptomic studies that can correlate between gene regulation and specific phenotypes. Yet, how environmental pressures interact with genotypes to shape the phenotypes of organisms over several generations has not been as thoroughly investigated. The study of adaptations and their genetic underpinnings can shed light on the interaction between genotype and the environment. Derived populations that are adapted to extreme environments can show considerable divergence in phenotype from the ancestral population, and therefore provide invaluable models for studying adaptation. Cave-dwelling species offer striking examples of such adaptation. Caves are harsh environments characterized by perpetual darkness and food scarcity (reviewed in [2, 3]). Troglomorphic species have evolved repeatedly in nearly every animal phylum [4]. Remarkably, they have converged on a set of similar phenotypic traits that include the regression of eyes, the loss of melanic pigmentation, and a suite of physiological and behavioral changes. These represent substantial phenotypic deviations from their nearest relatives that inhabit surface environments (Figure 1A). As such, troglomorphic species present excellent models for studying the genetic basis of adaptation.

Figure 1. Troglomorphic species are present across animal phyla and can be leveraged to answer genetic and evolutionary questions.

(A) Examples of cave-dwelling and subterranean species, from left to right: olm or cave salamander Proteus anguinus, mysid shrimp Spelaeomysis quinterensis found in the Dinaric Karst of the western Balkans, Tumbling Creek cavesnail Antrobia culveri found in Missouri, USA, and golden-line fish Sinocyclocheilus anatirostris found in karst areas of southwest China. (B) The Mexican tetra Astyanax mexicanus: a widely used troglomorphic species to study adaptation. This species consists of a sighted river-dwelling surface ecotype (top) and blind cave ecotype (bottom), depicted here alongside their contrasting natural habitats. (C) Map showing the locations of several populations of the cave and surface ecotypes of Astyanax mexicanus distributed in northeastern Mexico. Image sources: (A) olm cropped from [74] figure 1B, licensed under CC BY-NC-ND 4.0; mysid shrimp cropped from [75] figure 1B; cavesnail from Wikimedia Commons public domain; golden-line fish cropped from [76] figure 1B; (B) Astyanax natural habitat photographs courtesy of Riley Kellermeyer; (C) map of Astyanax populations cropped from [9], figure 1B. Images licensed under CC BY 4.0 unless stated otherwise.

2. Adaptation to extreme environments: Astyanax mexicanus as a model

One of the best-studied cave-dwelling species is the Mexican blind cavefish, Astyanax mexicanus (Figure 1B). This teleost fish comprises several populations inhabiting more than 30 caves in northeastern Mexico (Figure 1C) [5, 6]. Commonly studied caves include Pachón, Tinaja and Molino. Like other troglomorphic species, A. mexicanus cavefish are derived from river-dwelling ancestors that have eyes, normal pigmentation, and differ from cavefish in a series of other traits that we will explore in this review. However, A. mexicanus offers several advantages over other cave-dwellers as a model for investigating the mechanisms of adaptive phenotypic evolution. Firstly, descendants of the ancestral river-dwelling A. mexicanus, referred to as surface fish, continue to exist in the river systems neighboring the caves, and are a proxy for the ancestral form. This enables comparisons between cavefish and surface fish at every level of organization, ranging from genomes and physiologies to population structures (Figure 2). Secondly, A. mexicanus has independently colonized caves at least twice [7-9], presenting more than one replicate of the same natural experiment. This unique system permits questions about convergent evolution: were the same mechanisms at play every time A. mexicanus colonized a new cave system, or were there different mechanisms that ultimately led to the same phenotype? Thirdly, and perhaps most importantly, the A. mexicanus cave and surface ecotypes are the same species. Cavefish populations are interfertile amongst themselves and with surface fish [6]. This makes A. mexicanus amenable to quantitative trait locus (QTL, explained in the following section) mapping to identify genomic regions associated with a trait of interest. With numerous genomic and functional studies conducted in recent years, A. mexicanus has proven to be a truly remarkable model to answer general questions concerning convergence, adaptation, and phenotypic evolution. In this review, we discuss the literature that deploys Astyanax mexicanus as a model to understand the genetic basis of adaptation and close with our anticipation of insights yet to be gained from this fascinating system.

Figure 2. Troglomorphic traits in Astyanax mexicanus.

Abbreviation: SF, surface fish; CF, cavefish; ot, optic tectum; ol, olfactory lobe; DASPEI, (2-(4-(Dimethylamino)styryl)-N-ethylpyridinium iodide); microCT, micro-computed tomography ; H&E, hematoxylin and eosin. (A) The head regions of adult fish show the reduced eye size and pigmentation in cavefish. (B) DASPEI staining of the surface fish and cavefish larvae show increased superficial neuromasts in cavefish. (C) microCT images of dorsal cranium show the asymmetry in adult cavefish. (D) Different size and morphology shown by the dissected brains from surface fish and cavefish. (E) Left, the increased amount of visceral adipose tissue in cavefish shown by dissection and H&E staining. Right, H&E staining images of adult fish trunk show the reduced amount of muscle and increased amount of fat in cavefish. Image sources: (B, D) modified with the author’s permission from [6] figure 2F and figure 2C. (C) cropped from [77] figure 4A, 4D and 4E. (E) obtained from [78] figure 1A and [49] figure 1C, with permission from Elsevier. Images licensed under CC BY 4.0 unless stated otherwise.

3. Quantitative trait locus (QTL) mapping and candidate gene approaches

QTL mapping is a genetic tool to identify genomic regions associated with quantitative traits, as illustrated in Figure 3A. In brief, quantitative traits in the F2 generation are generated through crosses between parents exhibiting different phenotypes for the same trait. Molecular markers divide the whole genome into multiple linkage groups. QTL analysis evaluates the associations between traits and the molecular markers. By correlating the quantitative traits with genotypes of molecular markers, specific linkage groups or genomic regions associated with the traits can be identified. In laboratory-based model systems, mutants with segregated phenotypes are typically generated by inbred strains [10, 11]. Astyanax mexicanus, however, provides a unique advantage for QTL mapping, given its existence as a single species comprising two distinct ecotypes: the natural “mutant” cavefish and the wild-type surface fish. This unique feature of A. mexicanus makes QTL mapping a powerful tool in studying the genetic basis of traits in cavefish. So far, several morphological, behavioral, and physiological traits have been mapped to various QTLs [12-21].

Figure 3. Genetic tools in linking genotypes to phenotypes in Astyanax mexicanus.

(A) An illustration of QTL mapping and analysis. Phenotypically and genetically different surface fish (F0, left) and cavefish (F0, right) are crossed to generate F1 hybrids. F1 hybrid fish phenotypically resemble the parental surface fish. The F1 hybrids are intercrossed to generate F2 hybrids, which exhibit a wide range of phenotypes and genotypes. The whole genome is divided into linkage groups by molecular markers. The linkage probabilities (logarithm of the odds score, LOD score) between the quantitative trait and molecular markers are plotted in the linkage groups (bottom) for the F2 generation. Loci with LOD score above the statistical threshold (blue) are considered significantly associated with the trait. Fish illustrations modified from original by Mark Miller. (B) the workflow of candidate gene approaches.

However, QTL mapping often only provides a relatively large genomic region with numerous potential candidate genes in it. Furthermore, the long generation time (~1 year) in A. mexicanus can make QTL mapping time-consuming, although a recent study has shown promising improvements in reducing the generation time to 5 months [22]. Consequently, refining a QTL to a specific quantitative trait gene (QTG) or even the quantitative trait nucleotide (QTN) has remained a major challenge. To address this issue, researchers have incorporated candidate gene approaches (Figure 3B) to identify potential candidates associated with troglomorphic traits in cavefish. In this review, we will discuss a few examples of mutations found using candidate gene approaches, either in addition to or independent of QTL studies (Table 1).

Table 1.

Coding mutations identified in cavefish of Astyanax mexicanus; more can be found in [16, 20, 36-38].

Gene name	Cave populations	Cavefish traits	Methods	Functional validation ?	References
cbsa (cystathionine ß-synthase a)	Pachón	-	QTL mapping and candidate gene approach	No	[46]
cry1a (cryptochrome circadian regulator 1a)	Chica, Pachón, and Tinaja	-	Genome-wide comparisons	No	[40]
dtx2 (deltex E3 ubiquitin ligase 2)	Pachón and Tinaja	Carotenoid accumulation	QTL mapping	No	[12]
hcrtr2 (hypocretin receptor 2)	Pachón, Tinaja, and Molino	-	Genome-wide comparisons	No	[37]
hk2 (hexokinase 2)	Molino	Hyperglycemia	Candidate gene approach	Yes*	[34]
insra (insulin receptora)	Pachón, Tinaja, Yerbaniz, Japonés and Arroyo	Hyperglycemia	Candidate gene approach	Yes**	[33]
mao (monoamine oxidase)	Pachón, Tinaja, Los Sabinos, Curva, Toro and Chica	Behavioral syndrome	Candidate gene approach	Yes	[30, 31]
mc1r (melanocortin 1 receptor)	Pachón, Yerbaniz and Japonés	Brown phenotype	QTL mapping and candidate gene approach	Yes**	[27, 47]
mc4r (melanocortin 4 receptor)	Pachón, Tinaja, Sabinos, Micos, Yerbaniz, Piedras, Molino, Arroyo, and Japonés	Hyperphagia	Candidate gene approach	Yes	[32]
oca2 (oculocutaneou s albinism II)	Pachón, Molino and Micos	Albinism	QTL mapping and candidate gene approach	Yes	[24-26, 48]
per2 (period circadian clock 2)	Pachón, Tinaja, and Molino	Enhanced lipogenesis	Candidate gene approach	No	[49]

^*in HEK293T cells

^**in zebrafish

3.1. Pigment loss, mutations in oca2 and mc1r

Reduced pigmentation in cavefish was found to be associated with two traits – albinism and the brown phenotype [23]. While albinism manifests as a near-absence of pigmentation, the brown mutation results in a brown coloration by affecting both the number and size of melanophores. Protas, et al. [24] found that albinism mapped to the same locus in F2 progeny obtained from both surface/Molino and surface/Pachón crosses. Based on known albinism loci in mice and humans, they identified oculocutaneous albinism II (oca2) as a likely candidate gene in that locus. Indeed, sequencing this gene revealed different loss-of-function mutations in oca2 in both Pachón and Molino populations. This was confirmed by functional assays in cell lines [24] and elegant complementation tests using surface fish oca2-CRISPR mutants [25]. Further studies showed that oca2 can also regulate sleep in cavefish [26], suggesting a pleiotropic function of oca2 in cavefish adaptation.

Similarly, Gross, et al. [27] found a single QTL responsible for the brown phenotype in some cavefish populations. They identified melanocortin 1 receptor (mc1r) to be the most likely candidate gene in this locus, harboring different mutations in several cavefish populations. Functional assays in zebrafish showed that these mc1r mutations are responsible for the brown phenotype. Additionally, its downstream target microphthalamia-associated transcription factor (mitf), which is also involved in melanogenesis [28], has been found to have a large truncation in its coding region in some cave populations, although further validation is needed [29].

3.2. mao and social behavior

It has been argued that certain behavioral changes in cavefish, such as the loss of aggression, may be associated with elevated levels of monoamine neurotransmitters in the cavefish brain [30]. Candidate gene approaches revealed a genetic modification (P106L) of the upstream enzyme monoamine oxidase (mao) in cavefish, presumably leading to lower MAO activity in cavefish. While the P106L mutation has been implicated in affecting anxiety-like behaviors in cavefish, its effect on other behaviors still remains unclear [31].

3.3. mc4r and appetite regulation

Investigating the metabolic changes in cavefish, researchers specifically noted an increase in appetite within the Tinaja cavefish when provided with unrestricted access to food [32]. Using a candidate gene approach, three coding mutations in melanocortin 4 receptor (mc4r) were identified and shown to be linked to the hyperphagic phenotype. Notably, one of them, G145S, has been linked to human obesity, highlighting the similarities between the mechanisms of appetite regulation in humans and fish.

3.4. insra, hk2 and hyperglycemia

Another metabolic change observed in multiple cave lineages is hyperglycemia. By examining sequences of all known genes in the insulin pathway, Riddle and Aspiras, et al. identified a coding mutation P211L in one of the insulin receptors (insra) in cavefish [33]. This mutation results in decreased insulin binding, which leads to insulin resistance and hyperglycemia in Pachón and Tinaja cavefish. However, Molino cavefish carry the wild-type allele and do not show insulin resistance, yet they have similarly increased glucose levels. By exploring genes involved in glycolysis, a R42H coding mutation in the glycolytic enzyme hexokinase 2 (hk2) was identified in Molino cavefish [34], suggesting a different mechanism for hyperglycemia in Molino cavefish.

Normally, high blood glucose levels would lead to diseases such as diabetes [35]. However, hyperglycemia in cavefish does not result in pathological effects. On the contrary, it may provide a survival advantage in a nutrient-poor environment. Nevertheless, further studies are needed to fully understand the effects of hyperglycemia in cavefish.

4. Genome-wide studies

The first whole-genome sequence of Astyanax mexicanus, obtained from an individual from the Pachón cave was published in 2014 and based on short-read sequencing (AstMex 1.0) [36]. The second genome (AstMex 2.0), which used long-read sequencing to improve sequencing quality, and also provided the first genome of surface fish, was published in 2021 [37]. These genome references enabled genome-wide comparisons between surface fish and cavefish genomic regions, along with the use of chromosomal regions as references for QTL analysis. Consequently, more putative coding changes underlying the cavefish phenotypes have been identified using these new resources [12, 16, 20, 36-38].

With the ongoing development of sequencing technologies, the reference genomes have been continously updated. The Pachón cavefish genome was recently updated [39], and improved genomes of the surface fish and other cavefish populations are in the pipeline. These resources have enabled genome-wide comparisons among different cave populations, thereby leading to the discovery of novel mutations. For example, a genome-wide comparison revealed genetic divergence in seven cave and surface populations, including a coding mutation R263Q in cavefish cryptochrome circadian regulator 1a (cry1a) [40]. Interestingly, the same mutation was also identified in other distantly related cave and subterranean animals such as golden line barbel and naked mole-rat, exemplifying an impressive example of convergent evolution on a molecular level. Cryptochromes, such as CRY1a, function in circadian rhythm feedback loops [41]. Therefore, mutated cry1a might contribute to the altered circadian rhythms observed in cavefish [42-45], although additional research is necessary to confirm this hypothesis.

5. Regulatory changes

So far, we have discussed several examples of coding mutations in cavefish (Table 1). However, phenotypic changes are not solely attributed to coding mutations, but also to an even larger extent to regulatory changes. For instance, a recent study showed that the majority of genomic sites under positive evolutionary selection in cavefish are located within intronic regions [9]. Additionally, one of the earliest studies exploring regulatory changes in cavefish focused on the gene mc1r. As previously discussed, coding mutations in mc1r were found to be associated with pigmentation phenotypes in certain cave populations [27] (Table 1). However, some cave populations, such as Chica, do not carry mutations in the coding region of mc1r, and yet exhibit the brown phenotype. To investigate the potential mechanisms in those cave populations, Stahl and Gross identified a few candidate mutations in the 5’ region of the mc1r gene that could be associated with the brown phenotype [47].

Subsequent studies further revealed the importance of regulatory changes in generating phenotypic diversity [46, 50, 51]. Combining RNA-seq and whole-genome bisulfite sequencing on RNA and DNA, Gore, et al. showed increased DNA methylation upstream of eye genes including opn1lw1, gnb3a, and crx [50], potentially resulting in reduced eye size. Furthermore, Krishnan, et al. performed genome-wide epigenetic profiling in the liver of surface fish, Pachón and Tinaja cavefish to uncover cis-regulatory elements that regulate lipid metabolism [51]. They used RNA-seq, ATAC-seq (assay for transposase-accessible chromatin sequencing), and ChIP-seq (chromatin immunoprecipitation sequencing) to systematically annotate and correlate gene expression, accessible chromosomal regions, and histone modification profiles. They identified one mutated cis-regulatory element in Pachón cavefish leading to increased expression of the gene 4-hydroxyphenylpyruvate dioxygenase b (hpdb). Hpdb is involved in the tyrosine metabolism pathway, and tyrosine serves as a substrate for both the tricarboxylic acid (TCA) cycle and melanin production, indicating a possible energy trade-off between pigmentation and metabolism in cavefish.

6. Astyanax mexicanus provides insights into evolutionary mechanisms

We have discussed how the advent of sequencing technologies resulted in the capacity to identify differences between surface fish and cavefish at single-nucleotide resolution. Sequencing and genotyping, however, have also permitted comparisons of genomes to infer answers about questions concerning evolutionary mechanisms and origins.

6.1. Genetic mechanisms underlying variation and adaptation

Several factors influence the amount of variation in a population on which natural selection can act upon adapation to a novel environment. In this section, we discuss how genetic and genomic analyses of Astyanax mexicanus shed light on the mechanisms through which this variation is produced and maintained, and how it can fuel evolutionary changes.

Genetic drift.

Genetic drift is the random change in allele frequencies in a population from one generation to the next [52]. In the case of A. mexicanus, assessing the relative contributions of selection versus drift to traits such as eye loss and pigmentation reduction has long been mired in controversy (reviewed in [53]). There are two major lines of evidence supporting the role of drift in cave adaptation. The first line of evidence comprises measurements of effective population sizes (N_e). Intuitively, N_e is a measure of the genetic variability of a population, with smaller N_e values reflecting reduced variability (see [54] for a rigorous definition and treatment of N_e). In a population with reduced genetic variability, therefore, a certain allele is more likely to get driven to fixation or extinction. In other words, genetic drift is likely to exert a greater influence on populations of small effective sizes. Several studies have found that cavefish populations have consistently reduced N_e values and genetic variability compared to surface populations [9, 55-57]. The second line of evidence comes from QTL mapping studies which uncover QTLs with opposing polarities influencing a certain trait of interest. For instance, Protas, et al. found that melanophore QTLs had mixed polarities [58], suggesting that melanic pigmentation was under relaxed selection, allowing for a larger contribution of drift in shaping pigmentation loss in Astyanax mexicanus.

Selection.

When a population exhibits a range of trait values for a given phenotype of interest, some trait values may be more beneficial than others under the prevailing environment. Individuals possessing these beneficial trait values are more likely to survive and reproduce than those lacking the beneficial traits. As a result, the beneficial traits spread in the population and are said to be selected. De novo mutations are a well-appreciated source of genetic variation in a population. However, selection can also occur on standing genetic variation. Standing genetic variation refers to the presence of more than one allele at a given genetic locus in a population [59]. In other words, it is the amount of genetic diversity that exists in a population at a given point in time. Understanding the relative contributions of selection acting on these two kinds of raw material – de novo mutation versus standing genetic variation - can elucidate patterns of allele reuse and genomic regions where mutations tend to arise frequently.

There is evidence for the role of selection on de novo mutation as well as on standing genetic variation in cavefish adaptation. Independently derived cavefish populations often show different mutations in the same pathway, such as glucose metabolism [33, 34] and circadian rhythms [49] suggesting that these mutations evolved de novo in different lineages. For selection acting on standing genetic variation, one example is the gene oca2 (Table 1). O’Gorman, et al. found that oca2 experienced soft selective sweeps (see Box 1) in independently derived populations [26]. Another study examining the genetic diversity of over 500 single-nucleotide polymorphisms (SNPs) distributed throughout the cavefish genome found that many of these SNPs were shared across independently derived cavefish populations, suggesting that selection has occurred repeatedly on standing genetic variation to favour the same SNPs [60]. Another interesting example is the discovery of a “cave” allele of the mc4r gene at a low frequency in wild surface fish (Table 1) [32]. Therefore, it is possible that the mc4r cave allele was already present in surface populations as standing genetic variation, and became fixed by selection in cave populations. Finally, there is evidence for cryptic genetic variation in Astyanax mexicanus surface fish that can be revealed by the cave environment. Cryptic genetic variation is standing genetic variation that does not currently affect the phenotype (observable traits) of the organism, but can potentially influence phenotype under certain environmental conditions or genetic backgrounds. One study demonstrated that A. mexicanus surface populations possess cryptic genetic variants regulating eye size, which can be revealed through raising surface fish in low-conductivity water characteristic of cave environments [61]. This study illustrates how an environmental change can expose selectable variation that was hidden in the ancestral environment.

Box 1. Hard and soft selective sweeps.

When a beneficial allele or mutation is selected for and increases in frequency in a population, linked alleles near this sequence also get selected by a hitchhiking effect. This leads to a drop in the genetic variability in the vicinity of this beneficial allele (Figure I). This process is known as a selective sweep [73]. Searching for signatures of selective sweeps in the genome can be informative about past selection events.

Hard sweeps occur when a rare but beneficial mutation is selected for and spreads in the population, leading to a sizeable drop in the genetic variation of that region of the genome. Hard sweeps are therefore interpreted as signatures of selection acting on de novo mutation.

In contrast to the hard sweeps discussed above, a soft sweep occurs when a pre-existing allele becomes beneficial in a new environment and is selected for. Since this allele could be present in multiple genetic backgrounds (see Figure I), the reduction in genetic variation around this locus in the population after the sweep is not as drastic. Soft sweeps are interpreted as signatures of selection acting on standing genetic variation.

Figure I.

To what extent selection on de novo mutation versus standing genetic variation are driving cavefish adaptation, however, has been difficult to address. Bridging this knowledge gap, a recent study employing whole-genome resequencing of several cave and surface populations identified genes that had evidence of selective sweeps (see Box 1) shared across cave populations but neutrally evolving in surface populations and found signatures of hard sweeps (e.g. per1b, associated with circadian rhythms) and soft sweeps (e.g. otomp, involved in otolith development) in the genome [9]. These “cave-adaptive” genes were associated with both regressive traits historically attributed to drift, such as eye morphogenesis (e.g. pitx2) and pigment development (e.g. oca2), and gain-of-function traits such as DNA repair (e.g. ogg1) and lateral line nerve development (e.g. tmc2). Strikingly, this study also estimated nearly equal contributions of selection on de novo mutation and pre-existing variation to cave adaptation. This is a surprising finding, because rapid adaptation (see section 6.2) is thought to occur majorly through selection on pre-existing variation and gene flow, whereas selection on de novo mutations is expected to take longer. The authors of this study raise the intriguing possibility that cavefish may exhibit higher mutation rates than surface fish, although further work is needed. Taken together, these studies support a role for selection in driving cave-associated traits, including traits considered regressive and attributed to drift or disuse (see Section 7.1). This is of great significance not only to the Astyanax mexicanus field, but also in changing how we think about trait loss and adaptation.

Pleiotropy.

Pleiotropy refers to a gene’s property of governing more than one function. When a certain gene gets passed on to the next generation, either through selection or through drift, the associated phenotypic trait values also get passed on. Thus, a trait could end up getting indirectly favoured even if it is of no direct adaptive significance. Examples of pleiotropy in Astyanax mexicanus include the gene oca2, which is under positive selection in cavefish populations. oca2 was determined to control melanic pigmentation and to also associate with sleep loss in cavefish [26]. Thus, selection on sleep loss could drive pigment loss, or vice versa. Notably, it has been suggested that eye loss is attributable to the pleiotropic effects of selection on other traits [36, 62, 63].

Migration, hybridization, and gene flow.

Gene flow between populations can promote adaptation to a novel environment by expanding the substrate of genetic variation that natural selection acts on. There is a growing body of literature providing evidence for hybridization and gene flow between A. mexicanus populations, especially from surface populations to cave populations, and between cave populations. The earliest studies positing the existence of gene flow between A. mexicanus populations observed morphologically hybrid fish in certain cave populations (reviewed in [5]). Over time, morphological comparisons were augmented by the use of genetic markers such as random amplified polymorphic DNA (RAPD) markers, microsatellite loci, and mitochondrial DNA (reviewed in [64]). The first A. mexicanus genome assembly [36] provided opportunities for large-scale genome-wide comparisons between individuals and populations. Herman, et al. resequenced 45 whole genomes from two surface populations (Rascón and Río Choy) and three cave populations (Pachón, Tinaja and Molino), and found evidence for historical and contemporary gene flow between all three cave populations, and between cave and surface populations [8]. Recently, whole-genome resequencing and admixture analyses provided strong evidence that the Chica cave comprises a hybrid population, addressing a long-standing question in the evolutionary history of this cave [40]. Other similar studies also using whole-genome resequencing found evidence for extensive gene flow into the Río Subterráneo and Pachón caves from their nearby surface localities [9, 55]. Taken together, these studies are consistent with the role of gene flow in shaping the evolutionary history of A. mexicanus and suggest that gene flow may have catalyzed the adaptation of ancestral A. mexicanus to cave environments.

6.2. Estimating timescales of phenotypic change

Traditional Darwinian theory posits that evolution is a slow, incremental process requiring millions of years. However, multiple findings suggest that evolution can occur over much shorter timescales (reviewed in [65]). Estimating the time of divergence of a derived population from its ancestor can shed light on the pace of evolutionary adaptations. Astyanax mexicanus offers the unique advantage of extant ancestral populations that can be directly compared to the derived cave populations, circumventing the need to reconstruct a picture of the ancestral populations. Early research efforts, utilizing variations in mitochondrial DNA, have suggested that the cave colonization events occurred between 1.8 and 4.5 million years ago [66]. However, more recent studies employing whole-genome data have drastically revised these timelines to somewhere between 30,000 [57] and 160,000-190,000 generations ago [8, 9]. These studies also provide evidence to suggest that cave-adapted traits arose almost simultaneously with cave colonization events, supporting the idea that cavefish adapted rapidly to their novel environment to develop troglomorphic phenotypes.

6.3. Genetic mechanisms underlying parallel phenotypic evolution

One intriguing characteristic of Astyanax mexicanus is the existence of multiple cave populations (Figure 1C). Comparing independently derived populations allows for exploring whether the same genetic pathways are repurposed each time, or if distinct genetic mechanisms can lead to the same phenotypic outcome. The earliest hints that the cave ecotype may have multiple origins came from complementation studies. Crossing cavefish from different caves leads to the rescue of a troglomorphic trait in certain cases, indicating that different genomic loci must be involved in producing the same trait in different caves [67, 68]. There is also strong phylogenetic evidence to support multiple origins of cavefish. Phylogenies constructed using mitochondrial DNA comparisons, microsatellite loci, and whole-genome sequencing have counted between two to five independent origins, but they all agree that Astyanax mexicanus cavefish are not descended from a single cave-dwelling ancestor (Figure 4). Rather, they are derived from at least two ancestral stocks of surface fish ([8, 9, 55, 56, 69], see [64] for a review of older studies). This provides at least two independent replicates of the same natural adaptation experiment to investigate how often the same or different mechanisms underlie parallel or convergent phenotypes.

Figure 4. Astyanax mexicanus colonized caves at least twice independently.

Astyanax multi-species-coalescent phylogenetic tree cropped from [9], figure 1C, licensed under CC BY 4.0. Numbers at nodes indicate bootstrap node support values. Lineage 1 and Lineage 2 (new and old lineage, respectively, in older publications) are cave lineages from different ancestral surface fish stocks. Population names are color-coded according to lineage. As shown in the simplified map graphic on the left, lineage is correlated with geographical proximity of populations. Since surface and cave populations are interspersed in the tree, cave populations do not form a monophyletic clade and have multiple origins. The Subterráneo population (Micos region) is thought to have arisen from a Lineage 2 surface population but now extensively hybridized with Lineage 1 surface fish in the Micos region [9].

7. Concluding Remarks and Future Directions

In this review, we have made a case for the power of Astyanax mexicanus as a model to understand the genetic basis of evolution and adaptation. This is possible chiefly due to the existence of interfertile surface and cave populations, and to the multiple independent origins of the cave ecotype. We conclude with our perspectives on the genetic and biological insights this model system may yield in the future (see also the Outstanding Questions).

Outstanding questions.

• What are the specific genetic, epigenetic, and genomic changes driving transcriptomic and phenotypic variations in cavefish, and how do they work together to achieve the unique adaptations to the cave environment?

• How can gene-editing tools, such as transgenesis and CRISPR, be effectively utilized to study the genetic basis of adaptive traits in cavefish?

• How is the study of genetics and evolution in cavefish contributing to the understanding of general evolutionary principles and mechanisms in cavefish and other species?

• How do cavefish survive with traits that are typically considered diseases in the human context?

• How can the study of non-pathological traits in cavefish provide valuable insights into the development of therapies and treatments for human diseases?

7.1. Regressive and constructive traits

Eye and pigmentation loss in Astyanax mexicanus have classically been considered “regressive” traits, whereas gain-of-function traits, such as expanded nonvisual sensory systems, have been classic examples for “constructive” traits. Historically, traits that regressed were believed to do so under relaxed selection, while constructive traits arose under selection and are adaptive. However, this dichotomy is of little relevance, since it does not correlate with the adaptive value of classified traits. For instance, there is evidence supporting the selection of eye loss through direct and indirect means [63, 70, 71], which was “attribute(d) solely to disuse” by Darwin. We call for abandoning this dichotomy altogether and classifying traits on the basis of their adaptive value in the prevailing environment.

7.2. Genome editing in Astyanax mexicanus

The use of classical genetic tools and sequencing-based technologies have provided an unprecedented understanding of the biology of Astyanax mexicanus. The development of genome manipulation tools, however, still remains challenging with emerging model systems, and Astyanax mexicanus is no exception. The major hurdle in the case of Astyanax mexicanus has been its long generation time (~1 year). Despite this, Astyanax mexicanus researchers have successfully edited the surface fish genome using TALENs [72] and CRISPR/Cas9 mutagenesis [25] to confirm the role of oca2 in pigmentation. A recent publication showed that by altered husbandry practices, Astyanax mexicanus generation time could be reduced to 5 months [22]. We anticipate that future research will focus on functional genetic analyses of several more cavefish alleles, now rendered possible under practically feasible timeframes.

7.3. Implications for human diseases

Cavefish share similar traits and conserved mutations [27, 32, 33] (Table 1) with several human diseases such as albinism and obesity. However, despite these shared characteristics, cavefish do not exhibit pathological effects. Investigating how cavefish cope with these traits that lead to pathologies in humans would provide valuable insights into therapeutic strategies for human diseases. Additionally, studying cavefish offers the opportunity to identify novel mutations associated with specific phenotypes, expanding our understanding of disease pathology in humans. We argue that utilizing natural model systems, such as Astyanax mexicanus, will yield unparalleled insights not only into the genetic variation underlying adaptation to new environments but also reveal potential avenues for the development of novel therapeutic interventions and treatment strategies for various human diseases.

Highlights.

• Distinct genetic mechanisms underly similar phenotypic outcomes in independently derived cavefish populations

• Genomic and transcriptomic analyses highlight hitherto underappreciated roles for the regulatory genome and epigenome in shaping cave adaptation

• Revised estimates of cave colonization times indicate that phenotypic evolution may be much faster than previously assessed

• Trait loss can be driven by active selection, and is not necessarily caused by the removal of a selective pressure