Differential expression genes in the eye-transcriptomes of the riverine and marine grass puffer (Takifugu niphobles)

Abstract

Many marine fish species migrate to rivers, but little is known about whether these species switch their vision when inhabiting rivers or adapt their vision to the environment of rivers and the sea. Grass puffer (Takifugu niphobles) is a marine fish species frequently migrating to rivers. In this study, we investigated grass puffers from riverine and marine populations and analyzed the gene expression in their eyes. The phylogenetic analysis did not separate riverine and marine grass puffers. Principal component analysis and ADMIXTURE based on genome-wide SNPs showed no population differentiation of these two populations. Gene expression levels by high-throughput RNA sequencing indicated no differences in the expression patterns of vision-related genes in marine and riverine grass puffers. This result indicates that the grass puffers adapt their visual system to both marine and riverine environments rather than switching the expression of vision-related genes. Additionally, riverine grass puffers increase the expression levels of heat shock proteins and related genes, suggesting that they adapt to the environmental differences when they migrate to the river by increasing the expression levels of these genes.

Introduction

The environments in the rivers and the sea differ in many aspects, including nutrients, salinity, light intensity, and wavelength of light. For example, in freshwater with low transparency, water containing micro-particles scatters short-wavelength light, and long wavelengths dominate the light in turbid water^{1, 2–3}. In contrast, long-wavelength light is absorbed by water, while short-wavelength light is predominant in the clear ocean⁴. Since fish species inhabit this heterogeneous light environment, they have adapted their vision to the different light environments^2,3,5,6. Accordingly, the evolution of fish vision is known as a prime example of adaptive evolution.

Vertebrate visual pigments consist of light-absorbing chromophores and the protein component, opsin⁷. Spectral sensitivity is determined by the type of chromophore [with 11-cis retinal (A1-) or 11-cis 3-dehydroretinal (A2-derived retinal)] and by interactions between chromophores and amino acid residues of opsin⁸. The replacement of A1- with A2-derived retinal in the visual pigments shifts the absorption to a longer wavelength^9,10.

The enzyme CYP27c1 is known to be involved in the switch from A1 to A2, and the expression level of this gene correlates with the amount of A2^11,12. Freshwater fish primarily utilize A2 retinal, while marine fish use A1 retinal¹³. A1 and A2 retinal are known to be differentially used at different life cycle stages. Salmon use A1 retinal when living in the sea and switch to A2 when they migrate up the river to spawn¹⁴. In lamprey, A1 and A2 are differentially used by juveniles and upstream-migrating adult lamprey¹¹. Marine fish species migrating to the river use both A1 and A2 retinal¹³. This usage is a characteristic of both riverine and marine environments. However, little is known about whether these species mainly use A2 retinal when they live in rivers or have both A1 and A2 retinal in rivers and the sea. Furthermore, little is known about whether these species express different levels of opsin genes in the rivers and the sea.

Grass puffer (Takifugu niphobles) is a marine fish species that frequently migrates to rivers. Grass puffers are abundant in rivers and the sea during the same season. Fish with marine ancestry that have colonized freshwater have been reported to have adapted to freshwater by duplicating specific genes^15,16. However, the grass pufferfish would not be a colonizing fish because they cannot live in freshwater for long periods¹⁷. Grass puffers possess both A1 and A2 retinal¹³ therefore they seem to migrate to rivers for short periods. Hence, it is unclear how they adapt to different river environments compared to the sea and whether the populations in the rivers and the sea are the same. In this study, we analyzed the phylogenetic and population genetics of riverine and marine populations of the grass puffer. Additionally, we analyzed gene expression levels in the eyes of grass puffers from both riverine and marine populations. The gene expression levels in the eyes of marine and riverine pufferfish species were compared with those of riverine and marine grass puffers to show how grass puffers inhabit the river and marine environments.

Result

No genetic differentiation between the riverine and marine populations of T. niphobles

For the river population, we collected four grass puffers 1.3 km upstream from the mouth of Tagoe River, Kanagawa, Japan (Fig. S1, Table S1). Based on our observations, the grass puffers are not seen in the rivers during midwinter (mid-January through the end of February). In contrast, we visually observed many grass puffer individuals in the rivers from March through December, even at low tide. For the marine population, we collected four grass puffers on a beach 10 km from the mouth of the Tagoe River. Since this beach is more than 3 km from the nearest river, we expected less direct migration of the river grass puffers. We measured salinity and light environments at the Tagoe River and Tateishi Park Beach to verify the differences between the river and the sea environments (Fig. S2).

As a first step, we investigated the genetic relationships between the river and marine populations of the grass puffer. Total RNAs were extracted from eyes, and 9.3–15.9 Gb sequences were determined from four individuals, each of grass puffers from the riverine and marine populations (Table S2). Single nucleotide polymorphisms (SNPs) were extracted from short reads of all grass puffers from the riverine and marine populations, mapped to coding regions of a reference genome (Takifugu rubripes, PRJDB18283). We calculated Fst values using a total of 172,214 SNPs. The mean Fst value between riverine and marine populations was almost 0 (-0.00026635), indicating no genetic differentiation between these two populations.

For phylogenetic analysis, we mapped the short reads of three marine [(T. rubripes, T. poecilonotus, and C. rivulata (kitaF)] and riverine [T. nigroviridis (tn1)] species to a T. rubripes reference genome and extracted SNPs (Fig. S1, Table S2). The mapping information is listed in Table S2. The SNPs for all individuals were combined with the grass puffer’s SNPs, and a phylogenetic tree was constructed by the ML method using C. rivulata and T. nigroviridis as out-group species with a total of 116,941 sites. Phylogenetic trees revealed that the grass puffer formed a monophyletic group with no population differentiation (Fig. 1). We then conducted a population genetic analysis using 275,155 sites. Using the same data set, a principal component analysis (PCA) showed similar results, with no differentiation between the riverine and marine populations of grass puffer (Fig. 2A). Furthermore, ADMIXTURE analysis using the same data excluding the out-groups showed that the grass puffers formed either a single cluster (K = 2, 3) or multiple clusters (K = 4–6), but the riverine and marine populations did not differentiate (Fig. 2B). These results indicate that riverine and marine grass puffers are from the same population. Grass puffer is known to spawn simultaneously in the surf on the beach in June, and subpopulations are presumed to mix at this time, which may prevent genetic differentiation.

Fig. 1 [Images not available. See PDF.]

Phylogenetic relationships between Tetraodontidae species. The numbers on the branches represent bootstrap values. Maximum likelihood tree based on 116,941 SNPs.

Fig. 2 [Images not available. See PDF.]

Genetic relationships between Takifugu species. Principal Components Analysis (PC1 versus PC2) based on 275,155 SNPs (A). Colored circles correspond to the names of species or populations in the panel. (B) ADMIXTURE results based on the same SNP data with (A) for K = 2–6.

Expression differences in the eyes between the riverine and marine populations of T. niphobles

We performed RNA sequencing to compare gene expression levels in the eyes of grass puffers from the riverine and marine populations. Using the high-throughput sequences mapped to a T. rubripes reference genome sequence, we compared gene expression levels between the riverine and marine populations of grass puffers, correcting for multiple comparisons using False Discovery Rate (FDR). As a result, we found 25 differentially expressed genes with statistical significance (Fig. 3, Table S3). Among the 25 genes, six were highly expressed in the riverine population, and 19 were highly expressed in the marine population (Fig. 3, Table S3). In this study, we focused on six of the 25 genes highly expressed in the riverine population to reveal how marine grass puffer individuals can inhabit a freshwater environment. Among the six genes, three were heat shock protein genes (hsp 9, 47, 90)^18,19. UNC45B has also been reported to be part of a chaperone system that interacts with HSP90²⁰.

Fig. 3 [Images not available. See PDF.]

A volcano plot for the gene expression differences between the riverine and marine grass puffer populations. Each blue dot represents the gene expressed in the eye. The log2 fold changes of average TPM values of the marine grass puffer population compared to those of the riverine grass puffer population and the -log10 p-values resulting from a t-test comparing average TPM values between the riverine and marine grass puffer populations for each gene are plotted on the x- and y-axis, respectively.

Next, we examined whether the expression levels of genes related to vision varied between the riverine and marine populations. The photosensitivity of visual pigments is tuned by the amino acid sequence of opsins and the type of chromophores⁸. Therefore, we first searched for the opsin genes in the genome of T. rubripes and the cyp27c1 gene , which is involved in switching chromophores. The color opsins, blue-sensitive (SWS2), green-sensitive (RH2), and long-wavelength sensitive (LWS) opsin genes were annotated in the genome of T. rubripes. In addition, the scotopic opsin, RH1 gene, and cyp27c1 were present in the T. rubripes genome. Comparisons of gene expression levels of the four opsin genes and cyp27c1 between the river and marine populations of grass puffer showed no difference with statistical significance using a t-test and a one-tailed Bayesian t-test (p > 0.05; BF₁₀ < 1; Fig. 4C, D, F). In addition, the amino acid sequences of the four opsins were identical between the two populations.

Fig. 4 [Images not available. See PDF.]

Expression of opsin and cyp27c1 genes. The relative expression of opsin genes in (A) T. rubripes, (B) T. poecilonotus, (C) riverine population of grass puffer, (D) marine population of grass puffer, and (E) T. nigroviridis. (F) the expression of cyp27c1 gene. We used a t-test and a one-tailed Bayesian test to assess statistical significance. All comparisons yielded P > 0.05 and BF₁₀ < 1.

If the opsin genes we isolated are paralogs, the gene expression analysis may not be accurate. Therefore, we investigated whether the opsin genes are homologous. Based on our previous searches of opsin genes from genomes and transcriptomes, we found that each opsin gene (LWS, SWS2, RH2, RH1) used in this study from Takifugu niphobles, T. rubripes, T. poecilonotus, and Dichotomyctere nigroviridis exists as a single copy. Accordingly, we constructed phylogenetic trees for each opsin gene (LWS, SWS2, RH2, RH1) based on sequences obtained from the four pufferfish species, using the ML method with 1000 bootstrap replicates. By comparing each of these phylogenetic trees (Fig. S3) with the species tree (Fig. 1), we found that the topologies for the SWS2, RH2, and RH1 genes are identical to that of the species tree. This consistency suggests that the SWS2, RH2, and RH1 genes from the examined species are orthologous. In the phylogenetic tree of the LWS gene, the branching within the genus Takifugu lacks strong bootstrap support, likely due to their close genetic relationships²¹. However, since only a single copy of the LWS gene is present in the reference genomes of Takifugu species, it is highly likely that the LWS genes used in our analysis are also orthologs. Therefore, we used appropriate orthologous genes for the gene expression analysis.

Comparisons of gene expression between species in Tetraodontidae

We examined the opsin repertoire for three individuals each of T. rubripes and T. poecilonotus, as well as four individuals of Tetraodon nigroviridis, which is a Tetraodontidae species inhabiting rivers and ponds in Southeast Asia²². We assembled the RNA-seq reads and searched for the SWS2, RH2, LWS, and RH1 genes in the assembled contigs of T. rubripes, T. poecilonotus, and T. nigroviridis. We found the homologs of the SWS2, RH2, LWS, and RH1 genes from these three species. The assembled information is listed in Table S4. Therefore, we infer that the species in the Takifugu and T. nigroviridis, have an opsin repertoire of SWS2, RH2, LWS, and RH1 and use a trichromatic color vision.

We focused on four opsin genes, cyp27c1, and genes with higher expression levels in the riverine population than in the marine population of the grass puffer. We compared the expression levels of these genes in the populations of the grass puffer with those of T. rubripes, T. poecilonotus, and T. nigroviridis. The relative expression of RH2 in marine T. rubripes and T. poecilonotus exceeded 30% of total opsin expressions, and the expression of SWS2 and LWS was also relatively higher than that of the grass puffer (Fig. 4A, B). Therefore, these two marine species primarily express color opsin genes. In contrast, the relative expression of RH1 in riverine T. nigroviridis exceeded 95% of total opsin expressions, indicating that a scotopic opsin gene was predominantly expressed in this species (Fig. 4E). In the grass puffer, RH1 was predominantly expressed, but less than in T. nigroviridis, with 67% and 84% in the riverine and marine populations, respectively (Fig. 4C–E). SWS2, RH2, and LWS were highly expressed in grass puffer than in T. nigroviridis. cyp27c1 was expressed at lower levels in the marine species T. rubripes and T. poecilonotus than in the grass puffer and at higher levels in T. nigroviridis compared to the grass puffer (Fig. 4F). Among the six genes with higher expression levels in the riverine population compared to the marine population of the grass puffer, we focused on four highly expressed genes a transcript per million (TPM) value greater than 5. These four genes expressed higher levels in the river population of the grass puffer than in the other species (Fig. 5).

Fig. 5 [Images not available. See PDF.]

Expression of heat shock protein and related genes. The comparisons of the expression of (A) Hsp90, (B) Hsp47, (C) Hsp9, and (D) Unc45b in T. rubripes, T. poecilonotus, riverine and marine populations of grass puffer, and T.nigroviridis. P values from the t-test are shown on the line.

Discussion

When marine fish migrate to a river, they encounter an environment that is different from the sea. Therefore, we hypothesized that marine fish migrating to rivers might respond to these distinct environments. This study investigated mRNA expression in the eyes of the grass puffer, a species that inhabits both river and marine environments. As for vision, the repertoire of opsin genes and their expression levels were comparable between the riverine and marine populations of the grass puffer. The expression levels of cyp27c1, representing the amount of A2 retinal, were also comparable. Thus, grass puffers are presumed to have the same vision usage in marine and riverine environments.

In contrast, RH1 was predominantly expressed in the river species T. nigroviridis, while cone opsins were predominantly expressed in the marine species T. rubripes and T. poecilonotus. In the expression of the opsin genes of all species used in this study, the grass puffer showed an expression pattern intermediate between that of marine and river species. The expression of cyp27c1 was also intermediate between marine and riverine species. According to these results, we inferred that the usage of the visual system in the grass puffers is the same in both the river and the sea. In other words, the grass puffers adapt their visual system to two environments, the river and the sea, rather than switching their visual system by changing the expression of vision-related genes in response to the ambient environments. While no changes were observed in the expression levels of vision-related genes, it remains plausible that changes occurred at the protein level, such as post-translational modifications, alterations in the downstream signaling activity of opsins, or differences in opsin turnover rates. Future studies should investigate these potential regulatory mechanisms in the puffer fish. Grass puffers are relatively active swimmers and tend to inhabit shallow waters where turbidity has less effect than in the bottom environment. Therefore, the difference in the light environment between riverine and marine habitats may be slight. This ecological condition may have contributed to the absence of changes in the opsin gene expression patterns. In contrast, for benthic fish species that inhabit deeper environments where light filtration by turbid water affects them more than shallow water, environmental differences in light conditions may affect the expression of opsin genes in fish eyes. In future studies, we aim to conduct comparative analyses of habitat light environments, swimming abilities, and opsin gene expression levels across multiple fish species.

Unlike the vision-related genes, the expression levels of the six genes were significantly upregulated in the riverine grass puffer population compared to the marine population. In the four highly expressed genes among these six genes, the expression levels were upregulated in the riverine grass puffer population compared to the marine T. rubripes and T. poecilonotus. Interestingly, the expression levels of these four genes were higher in the riverine grass puffer population than in the riverine T. nigroviridis. These four genes were heat shock proteins (HSP), HSP90, HSP47, HSP9, and UNC45B. HSP functions in protein folding, maintenance of structural integrity, and proper regulation for a wide range of proteins¹⁸. HSP90 has been reported to play a role in lens formation in the eye²³. In addition, UNC45B is known to interact with HSP90 and is involved in lens formation²⁰. HSP47 is a molecular chaperone specific to collagen and plays an essential role during ocular development, especially in corneal morphogenesis²⁴. Considering that these three chaperone-related genes function in lens and cornea formation, these genes may be involved in normal eye formation and maintenance in freshwater environments with lower salt concentrations than seawater in the grass puffer. T. nigroviridis also inhabits the river, but the expression levels of these three genes were not upregulated. This species adapts to the river; thus, the riverine environment is probably not a major stressor for them. HSP9 has been proposed to be involved in the immune response against pathogenic bacteria¹⁹. The grass puffer individuals that have migrated to the river likely face a different type of pathogenic bacteria than those in the sea, and have upregulated expression of hsp9.

Taken together, we hypothesize that the grass puffer individuals adapt to the environmental differences (low salinity and pathogenic bacteria) that they face when they migrate to the river by increasing the expression levels of heat shock proteins and related genes.

Conclusion

In this study, we investigated how the grass puffer inhabits two different environments, the river, and the sea, based on gene expression in its eyes. Opsin and cyp27c1 genes were expected to be adapted to riverine and marine environments by making their expression patterns intermediate between those of riverine and marine species. On the other hand, the high expression of hsp genes in the riverine grass puffer population suggests that they adapt to the river environment by expressing heat shock proteins. Future studies of other marine fish migrating to rivers will reveal whether other fish species share these two adaptive strategies in different environments.

Methods

Measurement of salinity and light environments

We measured salinity at the Tagoe River and Tateishi Park (Fig. S1). In the Tagoe River, we collected water from a depth of 50 cm every hour from low to high tide at the point where we collected the grass puffers and measured salinity using a hydrometer (SpectrumBrands Japan, Kanagawa Japan) (Fig. S2A). In Tateishi Park, we measured salinity only once since salinity does not change with tides. We measured the wavelength of light in the horizontal direction at a depth of 50 cm with an Ocean Optics spectrophotometer (Ocean Photonics, Tokyo, Japan) at the sites where we collected the grass puffers in the Tagoe River and Tateishi Park (Figs. S1 and S2 B).

Sampling

Grass puffers were collected from Tagoe River and Tateishi Park (Fig. S1) by fishing. All T. niphobles individuals were sampled in early summer during the mid-afternoon.

Individuals of T. rubripes and T. poecilonotus were obtained from a commercial fishing boat, and individuals of each species were collected by fishing in Tokyo Bay (Figure S1). C. rivulata was collected from Tateishi park (Fig. S1) by fishing. T. nigroviridis individuals were obtained from tropical fish dealers. The information of each species used in this study is listed in Table S1. Two pufferfish species (grass puffer and C. rivulata) we collected from the wild do not require collecting permission because they are out of the protected species. The guidelines for experimental animal management of SOKENDAI were followed throughout the study. The Institutional Animal Care and Use Committee of SOKENDAI approved the animal protocols and procedures (permission number: SKD2021AR001). This study was designed under ARRIVE guidelines (https://arriveguidelines.org).

RNA sequencing and comparison of expression levels

Pufferfish individuals were placed in icy water and killed immediately, and their eyes were dissected. Total RNAs were extracted from eyes of C. rivulata, T. rubripes, T. poecilonotus, T. nigroviridis, and grass puffer individuals using TRIzol RNA Isolation Reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. RNA libraries were constructed using the NEBNext Poly(A) mRNA Magnetic Isolation Module and the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Bio Labs, Ipswich, MA) following the manufacturer’s instructions. Short cDNA sequences (paired-end 150 bp) were determined from the libraries using the Illumina HiseqX platform (RNA-seq). RNA-seq reads from each of three or four individuals of C. rivulata, T. rubripes, T. poecilonotus, T. nigroviridis, and the grass puffers (6.43–19.55 Gbp) were trimmed to remove adaptor sequences and mapped to the reference genome of T. rubripes (PRJDB18283). For Takifugu niphobles, we obtained 9.34–15.9 GB of RNA-seq data; for T. rubripes, 17.77–19.55 GB; and for T. poecilonotus, 6.48–19.43 GB, all of which were used for the analysis. Reads showing similarity (90%) with 90% read lengths were mapped to the reference genome and expression levels were calculated using CLC Genomics Workbench ver. 21 (https://www.qiagenbioinformatics.com/). TPM (Transcript Per Million) was used for normalized expression values. The average TPM of the four individuals of the grass puffer were compared. Differential expression was assessed using the TMM (trimmed mean of M values) method²⁵ implemented in the CLC genomics workbench using default parameters. False Discovery Rate (FDR) correction was applied to the calculated p-values and differentially expressed genes with corrected p-value < 0.05 were considered significant.

To examine the opsin gene repertoire, the RNA-seq reads of T. rubripes (n = 3), T. poecilonotus (n = 3), and Tetraodon nigroviridis (n = 4) were assembled (see Table S4) by CLC genomics workbench with “Auto” parameters.

As for T. nigroviridis, the mapping rate of the short reads (15.57%) to the reference genome of T. rubripes were lower than the individuals in a genus Takifugu (> 99.9%). Therefore, we mapped the short reads of T. nigroviridis to the assembled contigs of T. nigroviridis (tn1) for TPM comparison. The mapping information on T. nigroviridis individuals is listed in Table S5. TPM was used for normalized expression values. The TPM values of T. nigroviridis were normalized with that of the grass puffer using TPM values of GAPDH gene. We compared the average TPM values of hsp90, hsp47, hsp9, and unc45b between species or populations using a one-tailed t-test.

Phylogenetic, principal component, and ADMIXTURE analyses

The mapping data of eight grass puffers, two Takifugu species, and two outgroup species were exported in bam file format and sorted and indexed using samtools²⁶. The duplicated reads in bam files were marked by the MarkDuplicates algorithm in GATK v4.2 (https://gatk.broadinstitute.org/hc/en-us). We performed haplotype calling using the HaplotypeCaller algorithm in GATK v4.2. Haplotypes of all individuals were output as gvcf format (-ERC GVCF option). All gvcf files were combined into a single gvcf format file by the CombineGVCFs algorithm in GATK v4.2 and filtered by the VariantFiltration algorithm in GATK v4.2 with default parameters. The combined file was genotyped by the GenotypeGVCFs algorithm in GATK v4.2. We removed all indels, singleton, and doubleton sites to eliminate PCR and sequencing errors, and extracted bi-allelic sites with coverage equal to or more than five in all individuals and with GQ values equal to or more than twenty in all individuals.

The variants in a vcf file were converted to PHYLIP format. 10 kb sequences from the 5′ end of the PHYLIP format file were extracted and a model for Maximum Likelihood method was selected using MEGA ver. X²⁷. A phylogenetic tree was constructed using the Maximum Likelihood (ML) method using PhyML ver. 3.2²⁸ with a model selection option “-m GTR” and with 100 bootstrap replications (Fig. 1: 116,941 sites).

We selected the Takifugu species from a gvcf file and filtered with the same condition described above, and performed a principal component analysis (PCA) using PLINK ver. 1.9²⁹ with an option “–indep-pairwise 50 10 0.1” to explore the genetic relationship among all individuals (Fig. 2A: 275,155 sites).

ADMIXTURE ver. 1.3³⁰ was run on the same dataset (Fig. 2B: 275,155 sites) assuming 2 to 6 clusters (K = 2–6).

Acknowledgements

This work was supported by JST SPRING, Grant Number JPMJSP2104.

Author contributions

MO: research concept, all experiments, data analysis, and manuscript preparation. YT: research concept, research planning, data analysis, and manuscript preparation.

Data availability

The nucleotide sequences were deposited in the DDBJ Sequenced Read Archive under a Bioproject PRJDB18283.

Declarations

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Maan, ME; Hofker, KD; van Alphen, JJ; Seehausen, O. Sensory drive in cichlid speciation. Am. Nat.; 2006; 167, pp. 947-954. [DOI: https://dx.doi.org/10.1086/503532] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16615032]

2. Seehausen, O et al. Speciation through sensory drive in cichlid fish. Nature; 2008; 455, pp. 620-626.2008Natur.455.620S [DOI: https://dx.doi.org/10.1038/nature07285] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18833272]

3. Terai, Y et al. Visual adaptation in Lake Victoria cichlid fishes: Depth-related variation of color and scotopic opsins in species from sand/mud bottoms. BMC Evol. Biol.; 2017; 17, 200. [DOI: https://dx.doi.org/10.1186/s12862-017-1040-x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28830359][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5568302]

4. Kirk, JT. Light and Photosynthesis in Aquatic Ecosystems; 1994; Cambridge University Press: [DOI: https://dx.doi.org/10.1017/CBO9780511623370]

5. Terai, Y et al. Divergent selection on opsins drives incipient speciation in Lake Victoria cichlids. PLoS Biol.; 2006; 4, e433. [DOI: https://dx.doi.org/10.1371/journal.pbio.0040433] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17147472][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1750929]

6. Nagai, H et al. Reverse evolution in RH1 for adaptation of cichlids to water depth in Lake Tanganyika. Mol. Biol. Evol.; 2011; 28, pp. 1769-1776. [DOI: https://dx.doi.org/10.1093/molbev/msq344] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21172834]

7. Shichida, Y. In The Retinal Basis of Vision (ed J. Toyoda) (Elsevier Science, 1999).

8. Yokoyama, S. Molecular evolution of vertebrate visual pigments. Prog. Retin. Eye Res.; 2000; 19, pp. 385-419. [DOI: https://dx.doi.org/10.1016/S1350-9462(00)00002-1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10785616]

9. Harosi, FI. An analysis of two spectral properties of vertebrate visual pigments. Vision. Res.; 1994; 34, pp. 1359-1367. [DOI: https://dx.doi.org/10.1016/0042-6989(94)90134-1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8023444]

10. Parry, JW; Bowmaker, JK. Visual pigment reconstitution in intact goldfish retina using synthetic retinaldehyde isomers. Vision Res.; 2000; 40, pp. 2241-2247. [DOI: https://dx.doi.org/10.1016/S0042-6989(00)00101-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10927111]

11. Morshedian, A et al. Cambrian origin of the CYP27C1-mediated vitamin A1-to-A2 switch, a key mechanism of vertebrate sensory plasticity. Royal Soc. Open Sci.; 2017; 4, 170362.2017RSOS..470362M [DOI: https://dx.doi.org/10.1098/rsos.170362]

12. Enright, JM et al. Cyp27c1 red-shifts the spectral sensitivity of photoreceptors by converting vitamin A1 into A2. Curr. Biol.; 2015; 25, pp. 3048-3057. [DOI: https://dx.doi.org/10.1016/j.cub.2015.10.018] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26549260][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4910640]

13. Toyama, M et al. Presence of rhodopsin and porphyropsin in the eyes of 164 fishes, representing marine, diadromous, coastal and freshwater species—A qualitative and comparative study. Photochem. Photobiol.; 2008; 84, pp. 996-1002. [DOI: https://dx.doi.org/10.1111/j.1751-1097.2008.00344.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18422881]

14. Beatty, DD. A study of the succession of visual pigments in Pacific salmon (Oncorhynchus). Can. J. Zool.; 1966; 44, pp. 429-455.1966CaJPh.44.399B [DOI: https://dx.doi.org/10.1139/z66-045] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/5932749]

15. Ishikawa, A et al. A key metabolic gene for recurrent freshwater colonization and radiation in fishes. Science; 2019; 364, pp. 886-889.2019Sci..364.886I [DOI: https://dx.doi.org/10.1126/science.aau5656] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31147520]

16. Ishikawa, A; Yamanouchi, S; Iwasaki, W; Kitano, J. Convergent copy number increase of genes associated with freshwater colonization in fishes. Philos. Trans. R. Soc. B; 2022; 377, 20200509. [DOI: https://dx.doi.org/10.1098/rstb.2020.0509]

17. Kato, A; Maeno, Y; Hirose, S. Brief migration of the grass puffer, Takifugu niphobles, to fresh water from salt water. Ichthyol. Res.; 2010; 57, pp. 298-304. [DOI: https://dx.doi.org/10.1007/s10228-009-0150-x]

18. Sørensen, JG; Kristensen, TN; Loeschcke, V. The evolutionary and ecological role of heat shock proteins. Ecol. Lett.; 2003; 6, pp. 1025-1037. [DOI: https://dx.doi.org/10.1046/j.1461-0248.2003.00528.x]

19. Abumourad, I. Tilapia heat shock protein: Molecular cloning and characterization. Science; 2009; 7, pp. 46-57.

20. Hansen, L et al. The myosin chaperone UNC45B is involved in lens development and autosomal dominant juvenile cataract. Europ. J. Hum. Genet.; 2014; 22, pp. 1290-1297. [DOI: https://dx.doi.org/10.1038/ejhg.2014.21] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24549050][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4200433]

21. Park, YJ et al. Classification of Takifugu rubripes, T. chinensis and T. pseudommus by genotyping-by-sequencing. PLoS ONE; 2020; 15, e0236483. [DOI: https://dx.doi.org/10.1371/journal.pone.0236483] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32853203][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7451653]

22. Rainboth, WJ. Fishes of the Cambodian Mekong; 1996; Food & Agriculture Org:

23. Hooven, TA; Yamamoto, Y; Jeffery, WR. Blind cavefish and heat shock protein chaperones: A novel role for hsp90alpha in lens apoptosis. Int. J. Dev. Biol.; 2004; 48, pp. 731-738. [DOI: https://dx.doi.org/10.1387/ijdb.041874th] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15558465]

24. Tanaka, Y et al. Expression of 47 kDa heat shock protein (HSP47) during development of mouse cornea. Exp. Eye Res.; 1996; 63, pp. 383-393. [DOI: https://dx.doi.org/10.1006/exer.1996.0128] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8944545]

25. Robinson, MD; Oshlack, A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol.; 2010; 11, pp. 1-9. [DOI: https://dx.doi.org/10.1186/gb-2010-11-3-r25]

26. Li, H et al. The sequence alignment/map format and SAMtools. Bioinformatics; 2009; 25, pp. 2078-2079. [DOI: https://dx.doi.org/10.1093/bioinformatics/btp352] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19505943][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2723002]

27. Kumar, S; Stecher, G; Li, M; Knyaz, C; Tamura, K. MEGA X: Molecular evolutionary genetics analysis across computing platforms. Mol. Biol. Evol.; 2018; 35, 1547. [DOI: https://dx.doi.org/10.1093/molbev/msy096] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29722887][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5967553]

28. Guindon, S et al. New algorithms and methods to estimate maximum-likelihood phylogenies: Assessing the performance of PhyML 3.0. Syst. Biol.; 2010; 59, pp. 307-321. [DOI: https://dx.doi.org/10.1093/sysbio/syq010] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20525638]

29. Purcell, S et al. PLINK: a toolset for whole-genome association and population-based linkage analyses. Am. J. Hum. Genet.; 2007; 81, pp. 559-575. [DOI: https://dx.doi.org/10.1086/519795] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17701901][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1950838]

30. Alexander, DH; Lange, K. Enhancements to the ADMIXTURE algorithm for individual ancestry estimation. BMC Bioinf.; 2011; 12, pp. 1-6. [DOI: https://dx.doi.org/10.1186/1471-2105-12-246]