Changes in a Cone Opsin Repertoire Affect Color-Dependent Social Behavior in Medaka but Not Behavioral Photosensitivity

Abstract

Common ancestors of vertebrates had four types of cone opsins: short-wavelength sensitive 1 (SWS1), SWS2, rhodopsin 2 (RH2), and long-wavelength sensitive (LWS) types. Whereas fish and birds retain all the types, mammals have lost two of them (SWS2 and RH2) possibly because of their nocturnal lifestyle during the Mesozoic Era. Considering that the loss of cone opsin types causes so-called color blindness in humans (e.g., protanopia), the ability to discriminate color by trichromatic humans could be lower than that in potentially tetrachromatic birds and fish. Behavioral studies using color-blind (cone opsin-knockout) animals would be helpful to address such questions, but it is only recently that the genome-editing technologies have opened up this pathway. Using medaka as a model, we introduced frameshift mutations in SWS2 (SWS2a and/or SWS2b) after detailed characterization of the loci in silico, which unveiled the existence of a GC-AG intron and non-optic expressed-sequence-tags (ESTs) that include SWS2a in part. Transcripts from the mutated SWS2 loci are commonly reduced, suggesting that the SWS2a/b-double mutants could produce, if any, severely truncated (likely dysfunctional) SWS2s in small amounts. The mutants exhibited weakened body color preferences during mate choice. However, the optomotor response (OMR) test under monochromatic light revealed that the mutants had no defect in spectral sensitivity, even at the absorbance maxima (λ_max) of SWS2s. Evolutionary diversification of cone opsins has often been discussed in relation to adaptation to dominating light in habitats (i.e., changes in the repertoire or λ_max are for increasing sensitivity to the dominating light). However, the present results seem to provide empirical evidence showing that acquiring or losing a type of cone opsin (or changes in λ_max) need not substantially affect photopic or mesopic sensitivity. Other points of view, such as color discrimination of species-specific mates/preys/predators against habitat-specific backgrounds, may be necessary to understand why cone opsin repertories are so various among animals.

Keywords: color discrimination; medaka (Oryzias lapites); reverse genetics; sensory drive; short wavelength sensitive gene; spectral sensitivity.

FIGURE 1

The medaka short-wavelength sensitive (SWS)2a and SWS2b loci. (A) A diagram of the SWS2 loci on the chromosome 5. The paralogs are concatenated tandemly in a head-to-tail manner. Black boxes represent coding regions (from the start codon to the stop codon). Arrowheads are approximate positions of target sequences for CRISPR/Cas9. An asterisk shows a position of the splice donor of the GC–AG intron (the first intron of SWS2b). (B) The expressed-sequence-tag (EST) clones containing a part of SWS2a. A horizontal gray bar with black boxes represents a part of chromosome 5. Black boxes indicate transcribed regions including untranslated regions (UTRs) (those of SWS2s do not include UTRs as A). Horizontal black lines with black boxes are ESTs found in the database at the National Bioresource Project (NBRP) medaka. ras-related protein rab7-like (XM_020703466 and XM_020703467) and host cell factor C1 (XM_011475416) have been registered as predicted genes in GenBank, and corresponding ESTs (olova48a01, olsp57m10, olte5d13, MF01SSA152f07, and MF01SSA193g08) are found presently existing in the database. A dotted line is the region not sequenced in the ESTs. A total of five EST clones (olova10e01, olea36l17, olgi21a18, olki61c09, and olki19a17) contained a part of SWS2a. Three of the five clones (olea36l17, olgi21a18, and olki61c09) additionally contain upstream intergenic regions and a part of the fourth intron. olki19a17 possesses a different intergenic region, and olova10e01 has a part of exons of the upstream ras-related protein rab7-like gene. (C) The GC–AG intron. Electropherograms of the genomic (top) and complementary (bottom) DNA sequences are shown. Uppercase indicates the first and second exons of SWS2b, and lowercase indicates the first intron. The C residue with an asterisk does not follow the GT–AG rule, but the intron is spliced out. (D) mRNA sequences of SWS2a and SWS2b and the sws2 mutations. Translated amino acid sequences are also shown, with the transmembrane domains (predicted by TMHMM; http://www.cbs.dtu.dk/services/TMHMM/) highlighted by green. Arrowheads are positions of introns. Target sequences are shown in magenta, and downstream methionines (potential initiation sites of translation) are highlighted by black circles. Electropherograms obtained for the sws2^{– 4a} and sws2^+14b heterozygotes are shown. Note that peaks are doubled from the position where the frameshift mutation is introduced. The sequences of mutated alleles (bottom) are determined by subtracting the known sequences of the wild type (top) from the doubled electropherogram.

FIGURE 2

Expression of the cone opsin genes in the short-wavelength sensitive (sws)2 mutants. (A) RT-PCR of all the cone opsin genes in medaka. Gene names and λ_max of the proteins are shown on the left. Long-wavelength sensitive (LWS)a and LWSb could not be analyzed separately because of their high sequence identity. Genomic backgrounds [actin beta (Actb)-somatolactin alpha (SLα):green fluorescent protein (GFP) or color interfere (ci)] and genotypes for the SWS2 loci [ + (homozygotes of the wild-type allele) or – (homozygotes of the mutated allele)] are shown on the top. From the left lane in each genomic background, results of the sws2^{– 4a+2b}, sws2^{– 4a– 2b}, sws2^+29a, sws2^+8a, sws2^{– 8b}, sws2^+14b, and two wild-type fish are shown. Note that expression is largely the same in the 16 individuals in terms of SWS1, rhodopsin (RH)2a, RH2b, and LWSa/b (that of RH2a may fluctuate slightly). Apparent differences are found in the expression of SWS2b, SWS2a, and RH2c. In RH2c, the expression is commonly higher in Actb-SLα:GFP than the expression in ci. In SWS2b and SWS2a, the expression is also higher in Actb-SLα:GFP than it is in ci (see results in the wild type). Additionally, the SWS2 expression differs depending on the genotype, i.e., when mutated, the expression is decreased on both genomic backgrounds. The number of PCR cycles was 30 for SWS2s, 26 for SWS1, RH2s, and LWSs, and 20 for Actb. (B) Examples of the stepwise RT-PCR for determining the appropriate number of PCR cycles (i.e., before plateau) for (A). Here, results at every two cycles between 24 and 34 cycles of two wild type and two SWS2a/b-double mutants (sws2^{– 4a+2b} and sws2^{– 4a– 2b}) with the ci background are shown for three genes (SWS2a, SWS2b, and LWSa/b). Note delayed amplifications in the sws2 mutants for SWS2a and SWS2b, but not LWSa/b. We used forward primers different from those listed in Table 6 (namely, f: 5′-AACAAGAAGCTTCGATCCCA for SWS2a and f: 5′-TTGTTGCTTCTACGGGTTCC for SWS2b), which is why products shorter than those in (A) were amplified. (C) RT-PCR of the SWS2 genes in three wild-type (i.e., with no mutations on any cone opsin genes) strains that express SLα differently. The Actb-SLα:GFP, ci, and HNI strains excessively, never, and ordinarily express SLα, respectively (Fukamachi et al., 2004, 2009b). The expression of SWS2a and SWS2b is stronger in Actb-SLα:GFP than in ci as in (A). Expression in HNI seems to be more similar to that in ci than to that in Actb-SLα:GFP, suggesting that the expression of SWS2 is enhanced in Actb-SLα:GFP (rather than suppressed in ci; see section “Discussion”).

FIGURE 3

Mate choice of the short-wavelength sensitive (sws)2 mutants. Male fish of wild type, sws2^+1a+14b, sws2^{– 4a– 2b}, sws2^+29a, and sws2^+14b with the color interfere (ci) or actin beta (Actb)-somatolactin alpha (SLα):green fluorescent protein (GFP) background (shown by gray or orange, respectively) were given a choice between ci and Actb-SLα:GFP female fish. The cone opsin genotypes of the choice females were identical to those of the test males, except that we presented ci females with the sws2^+1a+14b mutation and Actb-SLα:GFP females with the sws2^+29a mutation to the wild-type males. Note that the wild-type males still exhibited strong preferences toward females of the same strain, indicating that body colors of the wild type and the cone opsin mutants are indistinguishable for not only humans but also medaka. Each dotted vertical line represents a male fish, and each circle on it is a result in a mate-choice trial (any trial with less than 10 courtships was ignored and is not shown). A box with a horizontal line shows the mean and 95% confidence interval of the ratio of courtship of male fish to female fish of the other strain (i.e., a relative sexual preference of the male fish). A one-way ANOVA followed by a Dunnett post hoc test revealed significant increases (shown by asterisks) in all the sws2 mutants by comparison with the wild type (P < 0.05) on both genomic backgrounds.

FIGURE 4

Behavioral photosensitivity of the short-wavelength sensitive (sws)2 mutants. The optomotor response (OMR) of adult fish was tested individually under monochromatic light at λ = 400 nm for the wild type, SWS2a/b-double, and SWS2b-single mutants (left) and λ = 440 nm for the wild type and SWS2a-single mutants (right). Each circle represents the result for one fish (n = 5 for each strain). Closed circle, wild type; open circle, sws2 mutants. Colors represent genomic backgrounds of the strains [gray, color interfere (ci); orange, actin beta (Actb)-somatolactin alpha (SLα):green fluorescent protein (GFP)]. The OMR was quantified by three parameters: (A) delay: the time elapsed until the fish started the OMR after switching the direction of stripe rotation, (B) duration: the proportion of time the fish was exhibiting the OMR, and (C) distance: the distance the fish swam in the direction of stripe rotation during the test (Matsuo et al., 2018). Mean and standard error of the mean is shown as a box with a horizontal line. No significant difference was detected between the wild type and the sws2 mutants in any of the six comparisons (i.e., the delay, duration, or distance at λ = 400 or 440 nm) (P > 0.05; one-way ANOVA followed by a Dunnett post hoc test using the wild type as a control). (D) The OMR tests under mesopic conditions. The results at 53 μmol/m²/s (photopic condition) are those in (C). Closed circle, wild type; open circle, sws2^+1a+14b (n = 5 each). See panel (C) for other symbols. The results at 0.089 μmol/m²/s (i.e., approximately –4 to + 4 rounds per individual and –2 to + 2 rounds in average) seem to indicate OMR-negative. From this standpoint, the OMR at 0.34 μmol/m²/s might be negative in the mutants, but positive in the wild type, although the difference is not significant (P > 0.05, Student’s t-test). (E) Further examination of the OMR under mesopic conditions. As the wild type (closed brown circles), we used the HNI strain (n = 9–10; one fish died during the tests). As the sws2 mutants (open gray circles), we used the sws2^+1a+14b with the ci background (n = 10). The differences are not significant at any photon flux density (P > 0.05, Student t test). The results in panels D,E may be somewhat inconsistent (e.g., the response at 0.60 μmol/m²/s seems to be higher in D than it is in E), which would reflect, for example, experiments on different days using different fish, experiments at different times of the day (i.e., diurnal fluctuation of cone opsin expression), potential differences in light angles or rotation speeds, or deterioration of the xenon arc lamp [we adjusted the photon flux densities using neutral density filters and the slits of the Okazaki Large Spectrograph (OLS)].

FIGURE 5

Potential effects of Renilla green fluorescent protein (GFP) on the vision of actin beta (Actb)-somatolactin alpha (SLα):GFP. (A) Excitation (blue) and emission (green) spectra of Renilla GFP (hrGFPII; Agilent Technologies) (Ward and Cormier, 1979). Excitation and emission peaks are 500 and 506 nm, respectively. The spectra are largely symmetrical using 503 nm as a border. (B) Net spectral absorbance (minus values) and fluorescence (plus values) by Renilla GFP based on the data in (A). This graph indicates that a part of the blue light (λ < 503 nm) impinging on the eyes of Actb-SLα:GFP (through the cornea and lens that weakly express Renilla GFP; Fukamachi et al., 2009b) would be absorbed, converted to green light (λ > 503 nm), and scattered into all directions before reaching the retina. The λ_max of the medaka cone opsins are shown by colored arrows.