Ethics statement

All experiments were conducted in accordance with the guidelines of Waseda University. All protocols were approved by the Committee for the Management of Biological Experiment at Waseda University, and experimental animal care was conducted with permission from the Committee for Animal Experimentation of the School of Science and Engineering, Waseda University (permission # WD18-052; 2018-A104).

Animals

We used sexually immature loaches unless noted. All loaches were purchased from a local supplier, “Okawa Shoten” (Toyohashi, Aichi 441-8481, Japan;  http://dojou-ookawa.com/). The loaches were treated with 50 µg/L malachite green oxalate for one day for disinfection and then kept in breeding tanks under fluorescent light for 12 hr light-dark (LD) cycles (Light:Dark = 12:12, 09:00 Light on, 21:00 Light off). The frozen larva of chironomid “Akamusi” and goldfish pellets were fed at a random time point in the afternoon. The spectrum of the light source (FL25SS EX-D, Panasonic) is shown in  Supplementary Figure S1. (10.2108.zsj.37.177.s1.pdf) The temperature of the circulating water was kept between 25.5°C and 26.5°C.

Behavioral experiment

To determine the influence of light on the daily locomotor activity of Japanese loach, their locomotor activity was measured under LD conditions (Light:Dark = 12:12, 09:00 light on, 21:00 light off) for six days and constant dark (DD) conditions for the subsequent 11 days. The measurement apparatus is illustrated schematically in  Supplementary Figure S2. (10.2108.zsj.37.177.s1.pdf) In each tank, a white LED (LXK 2-PWC4-0180, Lumileds) was used as a light source ( Supplementary Figure S3 (10.2108.zsj.37.177.s1.pdf)), and the light intensity at the bottom face was set to between 134.2 µW/cm² and 272.0 µW/cm².

Rice paddy sand (diameter: about 1 mm; Eiehu Japan Kikaku) was placed on the bottom of the measuring tank for approximately 8 mm ( Supplementary Figure S2 (10.2108.zsj.37.177.s1.pdf)). Loaches were transferred from the breeding tank to the measuring tank at the end of the light period. Measurement was started at the beginning of the dark period. They were fed at zeitgeber time (ZT) 10 on the third day after the start of the measurement with a water temperature kept between 25.5°C and 26.5°C.

In the eye removal experiment, after seven days measurement under LD conditions, the eyes were removed under ice anesthesia at ZT10–ZT11. After 1-day treatment with 50 µg/L malachite green oxalate, their locomotor activity was measured for seven days under LD cycles.

Tissue collection and RNA preparation for expression analysis of clock genes

Loaches kept at around 25.5–26.5°C and Light:Dark = 12:12 (09:00 light on, 21:00 light off) were divided into an LD group and a DD group. The loaches were transferred from the breeding tank to a sampling tank (approximately 8 L) equipped with a white light ( Supplementary Figure S4 (10.2108.zsj.37.177.s1.pdf)) at the end of the light period (ZT12) and kept in the dark for 12 hr at 26°C. The loaches in the LD group were then kept in LD for 24 hr, and their eyeballs were removed under ice anesthesia at ZT1 (10:00, n = 6), ZT6 (15:00, n = 5), ZT11 (20:00, n = 6), ZT13 (22:00, n = 5), ZT18 (03:00, n = 4), and ZT23 (08:00, n = 4). The eyes of the loaches in the DD group were also collected at six time points (10:00, n = 4; 15:00, n = 6; 20:00, n = 6; 22:00, n = 6; 03:00, n = 6; 08:00, n = 7) after incubation under DD conditions for 24 hr. Sampling in the light period was performed under fluorescent light (room light, 282.4 µW/cm²), and sampling in the dark period was performed under dim red light (λ_peak = 664 nm, λ_[5%] = 625–713 nm,  Supplementary Figure S5 (10.2108.zsj.37.177.s1.pdf)). Each eyeball was immersed in RNAlater overnight at 4°C until RNA extraction.

Total RNA was purified according to the recommended protocol of TRIzol^® Reagent (Invitrogen). Electrophoresis was carried out to observe the bands of 18S rRNA and 28S rRNA, thereby confirming that the total RNA was not degraded. ReverTra Ace qPCR Master Mix with gDNA Remover (TOYOBO) was used to digest the contaminating genomic DNA in the total RNA and to synthesize cDNA from 500 ng of total RNA, following the manufacturer's instructions.

Quantitative RT-PCR

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed using recommended conditions. The relative mRNA levels of the target genes were calculated using the Δ Δ Ct method, as described previously (Takeuchi et al., 2018). All the PCR products were subjected to 3% agarose gel electrophoresis to verify the appropriate amplification. Primers were designed using Primer 3 (ver. 0.4.0;  http://bioinfo.ut.ee/primer3-0.4.0/) (Table 1). Primers were designed in regions with relatively low sequence identities among Cry or Per genes, and to encompass an intron in the primer or amplification sequence to avoid genomic DNA-derived amplification. The amplification efficiencies of all the primer sets (Table 1) were tested by standard curve analyses using 3-fold serial dilutions (3⁰ – 3 – ⁶) of mixed loach cDNAs. maEf1α was used for the reference for its relatively constant expression in the eye.

RNA-sequencing

RNA-sequencing was performed as previously described for the analysis of gold lined spinefoot (Takeuchi et al., 2018) with some modifications: Briefly, brains and eyeballs were excised from three loaches, each at ZT5 and ZT18. The collected tissues were homogenized in TRIzol reagent using an electric homogenizer (Thermo Fisher Scientific). The homogenate was stored at –80°C until further processing. Total RNA was extracted from the homogenate by chloroform extraction, and all the samples were mixed together. Subsequently, genomic DNA was digested with rDNaseI (TaKaRa). Thereafter, total RNA was purified using the miRNeasy Mini kit (Qiagen). The purification method was carried out in accordance with the manufacturer's instructions.

Library construction and sequencing were performed by Eurofins Genomics consignment service. A strand-specific library was prepared, and a paired-end sequence was performed using HiSeq 2500 (Illumina). For the adapter, the sequence of TruSeq LT Kits and TruSeq v1/v2 Kits were used. The quality control and filtering of raw reads (89,158,386 reads) were performed using a personal computer (LM-iH610X, mouse computer). Adapter sequences, short reads less than 50 bases, and reads with a lower sequence quality were removed using a PEAT (Li et al., 2015) and Trimmomatic (ver. 0.36) ( http://www.usadellab.org/cms/?page=trimmomatic) (Bolger et al., 2014). The clean raw reads were assembled using Trinity (version: 2.8.4) ( https://github.com/trinityrnaseq/trinityrnaseq/releases) (Grabherr et al., 2013) Parameters in Trinity were: --seqType fq; --max_memory 200 G; --left read_peat_trim_qf_trim_1.fq; --right read_peat_trim_qf_trim_2.fq; --CPU 24; --min_contig_length 300; --SS_lib_type RF. A BLAST database was constructed from the resultant transcriptome assembly (93,660 contigs).

Table 1.

Primers used in real-time PCR of maCrys and maPers.

Molecular phylogenetic analysis of clock proteins

The molecular phylogenetic trees of four clock proteins (CRY, PER, CLOCK, and BMAL) were constructed. First, multiple alignments, including clock protein sequences, were constructed using MEGA (ver. 6.06), removing the N-terminal and C-terminal if necessary. Using the obtained fasta files, molecular phylogenetic trees were constructed by ClustalW (ver. 2.1) and drawn using NJplot (ver. 2.3).

Identification and molecular phylogenetic analysis of opsin proteins

Using 24 zebrafish Opsin sequences as queries, tBLASTn analysis was performed against the BLAST database constructed from the Japanese loach transcriptome assembly of brain and eyes (see previous section). We estimated the exon–intron junctions in maOpsins from those in zebrafish Opsin genes. Based on these results, we designed RT-PCR primers (Table 2), ensuring that at least one exon/intron junction was included in the amplicon.

Expression analysis of maOpsins

The total RNA was extracted from eyes of five male and five female loaches per time point (ZT0 (n = 5), ZT6 (n = 5), ZT12 (n = 5), ZT18 (n = 5), and ZT0 (n = 5)), and then ocular cDNA was synthesized as described above. The cDNA pools were mixed together and used as a template for RT-PCR. Genomic PCR was performed by using 10.4 ng of template genomic DNA prepared from the loach muscle. Thermal cycling was initiated with an incubation at 95°C for 30 min; 25 steps of PCR were performed, each one consisting of 95°C for 30 s, 60°C for 30 s, and at 68°C for 2 min. Following the final PCR cycle, the samples were incubated at 68°C for 5 min and kept at 4°C. The PCR products were separated with 3% agarose gel electrophoresis and observed with SYBR Green.

Spectrophotometric analysis

Extraction procedures and spectrophotometric analyses were carried out as described previously (Sakata et al., 2015). All the manipulations were performed under dim red light (> 630 nm). The dissected eyeballs were homogenized with a dounce homogenizer (loose: 30 strokes; tight: 30 strokes) at 4°C in buffer P (50 mM HEPES-NaOH [pH 6.6], 140 mM NaCl, 1 mM DTT, 0.1 mM PMSF, 4 mg/L aprotinin, and 4 mg/L leupeptin). The homogenate was centrifuged at 15,000 rpm for 30 min, and the pellet was collected. To solubilize ocular pigments, the pellet was homogenized again (tight: 30 strokes) in 1% CHAPS/buffer P. After centrifugation at 15,000 rpm for 30 min, the clear supernatant was collected and subjected to spectrophotometry using a SHIMADZU UV-2450 spectrophotometer. The sample was kept in an optical cell (light path, 1 cm) at 4°C during the analysis. After measuring the absolute absorbance, hydroxylamine (NH₂OH-HCl, pH 6.6) was added to the sample at a final concentration of 100 mM. The sample was kept for 120 min until the spectral changes disappeared. Then, the photopigments in the sample were bleached by successive irradiations with red and green light ( Supplementary Figures S6 (10.2108.zsj.37.177.s1.pdf) and  S7 (10.2108.zsj.37.177.s1.pdf)).

Immunoblot analysis

SDS-PAGE and immunoblotting analyses were performed as described previously (Sakata et al., 2015; Watari et al., 2012; Toda et al., 2014). A small aliquot (704 ng) of the loach ocular extract for spectrophotometric analyses (see above) was used. Affinity-purified anti-chicken rhodopsin antiserum (cRh-C antibody (Sakata et al., 2015)) or control mouse IgG (I5381-1MG; Sigma) were used as the primary antibody.

Immunohistochemistry

Sample treatment, processing, and immunohistochemical procedures were performed as described previously (Watari et al., 2012). Loach heads were collected at ZT18 and stored at – 80°C until use. The frozen samples were sectioned transversely at 14 µm. After blocking with 1.5% horse normal serum in 0.05% Triton X-100/PBS (PBS; 10 mM Na-Phosphate [pH 7.4], 140 mM NaCl, 1 mM MgCl₂) for 1 hr at room temperature, the slides were incubated for 16 hr at 4°C with primary antibody (cRh-C or control mouse IgG). The samples were rinsed with 0.05% Triton X-100/ PBS, and then treated with a biotinylated anti-mouse IgG for 1 hr at room temperature. After washing with 0.05% Triton X-100/PBS and PBS, the immunoreaction was enhanced by applying the Avidin-Biotin Complex method using Vectastain Elite ABC kit (vector laboratories) for 30 min. The resulting signals were visualized via treatment with DAB-peroxidase substrate solution (0.05% diaminobenzidine, 0.01% H₂O₂, 50 mM Tris-HCl [pH 7.4], 200 mM NaCl, 1 mM MgCl₂) for 3 min at room temperature.

Table 2.

Primers used in RT-PCR of maOpsins.

Statistical analysis

The locomotor activity data were analyzed by using Actogram J software (Schmid et al., 2011) to obtain actograms and chi-squared periodograms, from which the period lengths were estimated with probabilities of less than 0.01. Amplitude of the estimated period length in the chi-squared periodgram was used to calculate Q value that indicates robustness of rhythmicity. A Wilcoxon's signed-rank test (significant P < 0.05) was used to compare the Q values of the estimated period-lengths in LD with those in DD, or those of intact fish with those of eye-excision fish. Two-way factorial ANOVA was used (significant P < 0.05) to test the variations of locomotor activity and gene expression levels. The Tukey-Kramer method was used for multiple comparisons (significant P < 0.05).

Locomotor activity analysis of Japanese loach under the LD and DD conditions

We investigated the diel locomotor activity of Japanese loach in the laboratory environment (Fig. 1,  Supplementary Figure S8 (10.2108.zsj.37.177.s1.pdf) and  Supplementary Table S1 (10.2108.zsj.37.177.s1.pdf)). Each fish was reared under LD condition for six days, then kept under DD conditions for 11 days. Chi-squared periodograms showed that 13 out of 14 animals had significant 24-hr rhythmicity with periods of 23.6–24.7 hr under LD condition (Fig. 1A–K, M, and N; LD in middle rows), indicative of the synchronization of their locomotor activities to the LD cycles. Their locomotor activity rhythmicity under the DD condition decreased compared to LD condition, except for Fig. 1J and L (Fig. 1, middle rows and  Supplementary Figure S8 (10.2108.zsj.37.177.s1.pdf)). In the analyses of their locomotor activities with 4-hr time windows (Fig. 1, bottom rows), a significant interaction between light conditions and circadian time was observed in five of 14 individuals (Fig. 1A, C–E, and H; s. i. in bottom rows). In these animals, under LD conditions, the locomotor activity during each time point of the light phase was significantly lower than or not higher than those of the dark phase (Fig. 1A, C–E, and H; grey line in bottom rows).

The locomotor activity of the other nine animals did not show significant interactions (Fig. 1B, F, G, and I–N; bottom rows). Among them, six individuals showed significant daily variation (Fig. 1B, G, J, K, M, and N; daggers in bottom rows), and the locomotor activity during each time point of ZT/CT0-ZT/ CT12 was significantly lower or not higher than those of ZT/CT12-ZT/CT24 in five of the six animals (Fig. 1B, G, J, K, and M; dashed line in the bottom rows). The locomotor activity under the LD condition was higher than that under the DD condition in three of the nine animals (Fig. 1F, L, and N; asterisks bottom rows). One individual did not show significant differences in either circadian time or light conditions (Fig. 1I).

Phylogenetic analysis of clock proteins

Japanese loach exhibited nocturnal behavior under the LD condition, however, the rhythmicity was weakened or lost in DD (Fig. 1), implying that locomotor activity is predominantly controlled by light. Commonly, the eye is the primary site for the input pathway of visual and non-visual light information, such as time signals, while the brain and even peripheral tissues are photosensitive in fish species (Whitmore et al., 2000; Moore and Whitmore, 2014). Therefore, we considered the eyes to be a plausible candidate for the photoreceptive site involved in the suppressive photoresponses, although loach eyes are rather small. Therefore, we verified the photic and circadian function of the loach eyes by determining the expression of both clock genes (Figs. 2 and 3) and the photopigments (Figs. 4–6).

Fig. 1.

Locomotor activity of Japanese loach under LD and DD conditions. Top rows: Double-plot actograms of locomotor activity of 14 Japanese loaches under LD and DD conditions. The locomotor activity of each fish was measured for six days in LD (12:12) and subsequently for 11 days in DD (shown by a vertical black bar on the left side). Middle rows: Chi-squared periodograms of the locomotor activities. Significance lines (P < 0.01) are colored in grey. The presumed free-running periods under the DD condition are shown. Bottom rows: Analyses of the locomotor activity in a 4-hr window. Circadian time (CT) was estimated by the free-running period. One cycle was divided into six equal periods. Locomotor activity of four daily or circadian cycles (shown by grey bars on the right side of double-plot actograms) were averaged for each period (counts ± SD per min) and plotted by grey lines with squares (LD) or black lines with black circles (DD). The solid and open squares represent dark and light points in LD cycles, respectively. The significance of the interaction between the circadian rhythm and light conditions was examined by two-way factorial ANOVA and was considered a significant interaction (s.i.) if the P-value was less than 0.05. Multiple comparison tests were performed using the Tukey-Kramer method. Asterisks and daggers represent significant variations depending on the light conditions and time periods, respectively. Different letters indicate significant differences between comparisons (P < 0.05). Dashed lines represent the averaged locomotor activities under the LD and DD conditions. They were drawn when a significant interaction was not detected. Significant daily variations were detected using two-way factorial ANOVA.

Continued.

As transcriptome data for the central tissues has yet to be reported, we performed an RNA-sequencing of the central nervous tissues (eyes and brain) prior to the measurement of the expression profiles of clock genes in the loach eye (Long et al., 2013; Luo et al., 2015; Huang et al., 2016). We locally constructed a BLAST database from the transcriptome assembly to identify the transcripts encoding cryptochrome/photolyase and clock (-related) proteins (see below). Phylogenetic analyses located the partial sequences close to their orthologous genes of zebrafish in molecular phylogenetic trees (Fig. 2 and  Supplementary Table S2 (10.2108.zsj.37.177.s1.pdf)). The transcriptome datasets generated and/or analyzed during the current study are available in the DDBJ repository (Accession: DRA007566).

Fig. 2.

Molecular phylogenetic trees (NJ trees) of loach clock proteins. (A) A molecular phylogenetic tree of CRY proteins. (B) A molecular phylogenetic tree of PER proteins. (C) A molecular phylogenetic tree of CLOCK proteins. (D) A molecular phylogenetic tree of BMAL proteins. Molecular phylogenetic trees were constructed using the Neighbor-Joining method (ClustalW ver. 2.1). The values at the branches indicate bootstrap probabilities. Bootstrap values > 95% have been omitted. Multiple alignments were made using MEGA 6.06. Because some of amino acid sequences obtained from the transcriptome database did not cover the entire CDS, the N-terminus and/or C-terminus of the alignment were manually omitted, and then the processed alignments (corresponding to Val²–Ser⁵⁰⁴ in zCRY1aa [panel A], Met¹–Gln⁶⁶⁸ in zPER1a [panel B], Met¹–Gln⁸⁹² in zCLOCK1a [panel C], and Met¹–Leu⁶²⁶ in zBMAL1 [panel D]) were subjected to progressive alignment by using ClustalW ver. 2.1, (DDBJ) to generate molecular phylogenetic trees. Tr, Tetraodon nigroviridis; Sg, Siganus guttatus; Xt, Xenopus tropicalis; ma, Misgurnus anguillicaudatus; d, Drosophila melanogaster; At, Arabidopsis thaliana.

Fig. 3.

mRNA expression levels of maCrys and maPers under the LD and DD conditions. Grey and black lines denote mRNA profiles under the LD and DD conditions, respectively. Solid and open squares represent dark and light points, respectively. The messenger RNA levels of maCrys and maPers were calculated as a value relative to the maEf1α expression levels. Expression levels of each clock gene were compared between all populations using two-way factorial ANOVA followed by Tukey-Kramer multiple comparison test (P < 0.05). Different letters indicate significant differences between comparisons. Asterisks indicate significant differences between LD and DD conditions. Daggers indicate significant differences among time points. Dashed lines represent the average of the mRNA expression levels under the LD and DD conditions. They were drawn when significant interactions were not detected. Significant daily variations were detected by two-way factorial ANOVA. Mean values ± SD were plotted.

Zebrafish has nine Cry genes (zCry1a, zCry1b, zCry2a, zCry2b, zCry3, z(6-4)phr, zCry4, zCry6, and zCry-dash) and four Per genes (zPer1a, zPer1b, zPer2, and zPer3), which may have diverged over the course of three whole genome duplications, which occurred at an early stage of vertebrate (teleost) evolution (Inoue et al., 2015; Sato and Nishida, 2010). The topologies of the phylogenetic trees (Fig. 2) showed no discrepancy based on their taxonomic relationships. As a result of the tBLASTn search of the database, along with the following comparison with zebrafish Cry genes, we identified eight loach paralogs of zCry (Misgurnus anguillicaudatus Cry1a [maCry1a], maCry2a, maCry2b, maCry3, ma(6-4)phr, maCry4, maCry6, and maCry-dash) (Fig. 2A). Similarly, we identified five loach paralogs of zPer (maPer1a, maPer1b, maPer2a, maPer2b, and maPer3) (Fig. 2B), three paralogs of zClock (maClock1a, maClock1b, and maClock2) (Fig. 2C), and three paralogs of zBmal genes (maBmal1, maBmal2, and maBmal3) (Fig. 2D). We did not identify a zCry1b ortholog in the loach transcriptome sequences, possibly due to the lack of Cry1b or the very low expression levels in the central nervous tissues.

Expression analyses of loach clock genes under LD and DD conditions

To determine whether the expression of clock genes is controlled by environmental light and/or the internal circadian oscillator, we measured the transcript levels in the eye by qRT-PCR (Fig. 3 and Table 1). We measured the mRNA levels of four Cry (orthologs of animal Cry1/2) and four Per genes in the eyes, which were collected at six time points under LD or DD conditions (Fig. 3 and  Supplementary Table S3 (10.2108.zsj.37.177.s1.pdf)).

In maCry1a (Fig. 3A), daily fluctuations were observed, with higher expression levels in LD compared to DD. The expression levels of maCry2a (Fig. 3B) showed daily changes, with a peak in the evening (ZT11–ZT13), and the mRNA levels at these points were significantly higher in LD than DD. Significant variations were observed even under the DD condition (Fig. 3B, black circles). These results suggest that the ocular expression of maCry1a and maCry2a are likely regulated by both the circadian clock and external light signals. MaCry2b mRNA levels (Fig. 3C) increased during the light period under LD condition, similar to maCry2a, but did not show any significant fluctuations in DD, implying photic regulation.

No significant interactions between daily temporal variations and light conditions were detected in maCry3, maPer1b, and maPer3 (Fig. 3D, F and H), all of which in turn showed clear daily changes independent of light conditions, indicative of circadian regulation. MaPer1a showed significant interactions between daily temporal variations and light conditions; however, neither photic nor circadian variations were observed (Fig. 3E). The mRNA levels of maPer2a were significantly higher in the light and just after dark (ZT1–ZT13 in LD, Fig. 3G) than the other points in the dark (ZT1–ZT23 in DD; ZT18 and ZT23 in LD), indicating light-dependent upregulation.

Identification and molecular phylogenetic analysis of maOpsins

We determined the types of Opsin genes that were expressed in the eye by RT-PCR. Firstly, we performed tBLASTn analysis to identify Opsin transcripts using our transcriptome assembly, and found 21 contigs which could possibly encode Opsins ( Supplementary Table S4 (10.2108.zsj.37.177.s1.pdf)). We next designed primers for RT-PCR to examine their presence in the loach genome and expression in the eyes (Table 2).

Fig. 4.

PCR analysis of loach Opsin genes. PCR was performed in the presence of eye cDNA template (top row), genomic DNA template (bottom row), or in the absence of eye cDNA template (middle row). PCR products were separated with 3% agarose gel electrophoresis and observed with SYBR Green.

For 16 of 21 primer sets, we detected single amplified products in genomic PCR (Fig. 4, Genome). Then, the contigs corresponding to the 16 primer sets were assumed to exist in the loach genome and divided into several groups by molecular phylogenetic analysis (Fig. 5). We identified seven visual Opsin genes (five cone opsins and two rhodopsins) and nine non-visual Opsins. Except for three primer sets (maPeropsin, maOpn4m1, and maOpn4m2; lanes 13, 16, and 18, respectively), the band sizes were larger than those estimated from the transcriptome, probably because all of the primer sets were designed to amplify the genomic DNA encompassing introns. As for the maPeropsin, maOpn4m1, and maOpn4m2 genes, these may be single-exon genes, as in zebrafish Opn4m2 (Davies et al., 2011), or alternatively, the amplified regions contain very short introns.

Next, the expression levels of the 16 maOpsin genes in the eyes were investigated by RT-PCR using ocular cDNA as a template (Fig. 4, Eyes, RT+). Single bands were detected for visual Opsins (Fig. 4, lanes 2–5 and 7–9), among which the intensity of the bands for maRh1, maRh1-2, and maRed-opsin (Fig. 4, lanes 4, 5, and 9) were relatively higher than those for the other genes. This indicated the relative abundance of these mRNAs, although the present RT-PCR analysis is not highly quantitative. Among non-visual Opsin genes, single bands were detected for maPeropsin (lane 13) and maOpn4m3, (lane 17) and weak signals were observed for maParapinopsin (maPara, lane 11) and maOpn4m2 (lane 18). Although the RT-PCR bands of maPeropsin and maOpn4m2 had almost the same mobility as those obtained using genomic PCR, we considered only the signals originating from cDNA due to the lack of their bands in control experiments, excluding the RT reaction (Eyes, RT-). We did not detect any expression of other non-visual Opsin genes, including the pineal-specific rhodopsin Exorh (lane 6).

Immunohistochemistry of retina and spectrophotometry of eyeball extract

The detection of opsin protein(s) in loach eyes was performed by western blot, immunohistochemistry, and spectroscopy. MaRod-opsin protein in the loach eyeball extract was analyzed by western blot analysis using an affinity-purified anti-chicken rhodopsin antiserum (cRh-C antibody (Sakata et al., 2015)). A rhodopsin-like immunoreactive band was observed at a mobility of 39 kDa (Fig. 6A), which is likely to correspond to the monomer of rhodopsin or porphyropsin, a photopigment found in the retinal rods of freshwater fish, using A₂-retinal (dehydroretinal) as the chromophore instead of A₁-retinal. Additional bands were found at a mobility of 86 kDa, as well as the higher positions in both the loach eyeball extract and purified chicken rhodopsin, which were considered to be dimers and multimers of those opsins. In the immunohistochemistry analysis, rhodopsin-like immunoreactivities were found in many rods in the retina (Fig. 6B–E).

Because we did not examine the immunohistochemical analysis of cones due to the lack of antibody to loach cone photoreceptors, we instead performed a spectrophotometric analysis of the photoreceptors in the loach eyes to examine the photoreceptive function of the visual pigments. Characteristic peaks were observed around 410 nm, 540 nm, and 580 nm in the absorption spectrum of the CHAPS-solubilized ocular extract (Fig. 6F, curve 1, “Eyeball extract”), which are indicative of heme proteins, which are widely observed in a variety of animal tissue extracts. We added NH2OH to the extract for the rapid conversion of photointermediate meta-porphyropsin or metarhodopsin, which are formed by light irradiation, into retinal oximes and opsin apoproteins. After judging the termination of the NH₂OH-induced dark reaction from no spectral change (Fig. 6F, curve 2, “100 mM NH₂OH 120 min”), the extract was irradiated with red and green light successively until the spectrum showed no significant changes (total irradiation of 32 min and 8 min, respectively). A green-sensitive pigment, possibly porphyropsin or rhodopsin, was bleached by these irradiations. The difference in spectra was calculated by subtracting each irradiation spectrum (Fig. 6G, curve 3, “Red light 32 min”; curve 4, “Green light 8 min”) from that before the irradiation (Fig. 6F, curve 2, “100 mM NH₂OH 120 min”), which provided an absorption maximum at 520 nm. Importantly, they showed an isosbestic point around 440 nm, suggesting the bleaching of a single component.

Fig. 5.

Molecular phylogenetic tree of loach opsins. The molecular phylogenetic tree was constructed using the Neighbor-Joining method (ClustalW ver. 2.1). The values at the branches indicate bootstrap probabilities. Bootstrap values > 95% are omitted. Multiple alignments were made using MEGA 6.06. Because some of amino acid sequences obtained from the transcriptome database did not cover the entire CDS, the N-terminus and/or C-terminus of the alignment were manually omitted, and then the processed alignment (corresponding to Glu³–Pro³⁵⁶ in zRed opsin1) was subjected to progressive alignment by using ClustalW ver. 2.1, (DDBJ) to generate the molecular phylogenetic tree. Tr, Tetraodon nigroviridis; Xt, Xenopus tropicalis; ma, Misgurnus anguillicaudatus; At, Arabidopsis thaliana.

Fig. 6.

Immunohistochemical localization and spectrophotometry of Japanese loach rod pigment in the retina. (A) Immunoblot analysis of Japanese loach ocular proteins. Arrowhead indicates a presumed monomeric form of loach porphyropsin. Eyeball extracts of Japanese loach containing 704 ng of porphyropsin or purified 4 ng chicken rhodopsin were loaded. (B) A bright-field image of Japanese loach retinal section immunostained by anti-chicken rhodopsin cRh-C. (D) A differential interference contrast (DIC) image of the same section as in (B). (C, E) Control sections using horse normal serum. RPE, retinal pigment epithelium; VPC, visual pigment cells; ONL, outer nuclear layer; INL, inner nuclear layer. (F) Absorption spectra of Japanese loach eyeball extract. To the extract (“eyeball extract”), NH2OH-HCl was added to a final concentration of 100 mM (“100 mM NH2OH”). Then, the sample was kept in dark for 120 min until the spectral changes dissipated. Subsequently, the sample was successively irradiated with red light (727.0 µW/cm², λ_max = 660 nm) for 32 min and green light (574.2 µW/ cm², λ_max = 520 nm) for 8 min (“Red light 32 min” and “Green light 8 min”). (G) Difference absorbance spectra before the initial irradiation (“100 mM NH2OH 120 min” in (F)) and after each subsequent irradiation (“Red light 32 min” and “Green light 8 min” in panel (F)). The inset shows the comparison between the visual pigment templates of λ_max = 520 nm and normalized difference absorption spectrum for the spectral changes during red light irradiation for 16 min after a total of 16 min red light irradiation (curve 5). Curve 3 and curve 4 indicate the template spectra of A1- and A2-based rhodopsins, respectively. Templates were calculated using the method described in (Okano et al., 1989). (H) The sample was prepared separately from Fig. 6F and G. Difference absorbance spectra before the first irradiation and after red light irradiation (202.8 µW/cm², λ_max = 670 nm: Curve 1) or green light irradiation (158.0 µW/cm², λ_max = 533 nm: Curve 2).

We did not observe any spectral changes resulting from the red opsins in the spectroscopic analysis, despite the strong signal of the maRed-opsin in the RT-PCR analysis of the eye. Thus, we carried out a further spectrophotometric analysis after regenerating the photopigments by the addition of 11-cis-retinal in the dark, since the red-opsin may have been degenerated by the dim-red light used during the sampling (Fig. 6H). However, no spectral changes were detected in the long wavelength region compared with the experiment without the addition of 11-cis-retinal (Fig. 6G and H), implying that the Japanese loach has few red cones in its eyes.

The spectral shapes of retinal photopigments change depending on whether the chromophore is A1- or A2-retinal. Therefore, we compared the difference spectra of the green-sensitive component to the templates for A1- and A2-retinal-based pigments (Govardovskii et al., 2000). As shown in the inset of Fig. 6G, the difference spectra were precisely matched with the template spectrum calculated for A2-based rod opsin (porphyropsin), better than that for A1-based rhodopsin (Fig. 6G, inset). This result coincides with the results of loach retinal analysis by Toyama et al. (Toyama et al., 2008). These findings confirm that the Japanese loach has porphyropsin rather than rhodopsin in their eyes.

In order to estimate the total amount of porphyropsin in a loach eye, we further repeated the extraction of porphyropsin from Japanese loach eyes. The content was estimated to be 0.0073 OD per eyeball. From the sizes of the eyes used for the pigment quantitation, the average surface area of the retina and the density of porphyropsin were calculated to be 0.098 cm² and 0.074 OD/cm², respectively (Table 3).

Locomotor activity analysis of blinded Japanese loach

In the analyses of the locomotor activity with a 4-hr time window ( Supplementary Table S5 (10.2108.zsj.37.177.s1.pdf) and Fig. 7, bottom rows), the locomotor activity was found to decrease in five of eight individuals after removal of the eyes, compared to before the removal (Fig. 7A, B, D, G, and H). However, all individuals seemed to show nocturnal activity, even after enucleation, according to double-plotted actograms (Fig. 7A, B, D, and H; top rows). In fact, chi-squared periodograms showed that all animals had significant 24-hr rhythmicity with periods of 22.5–25.6 hr. These periods are similar to those found in intact animals in LD (Fig. 1, LD and Fig. 7, Intact) rather than intact animals in DD (Fig. 1, DD), A significant difference was not detected between the loaches before and after the enucleation in terms of the amplitudes or period-lengths ( Supplementary Figure S9 (10.2108.zsj.37.177.s1.pdf)). In addition, the enucleated animals still showed acute light responses just after the lights were turned off ( Supplementary Figure S10 (10.2108.zsj.37.177.s1.pdf)). These data strongly suggest that Japanese loaches can synchronize their locomotor activity rhythm with the LD cycle without the input of a light signal from the eyes.

Regulation of the locomotor activity by light and circadian clock

The present observation of nocturnal locomotor activity of the loach is in part consistent with that reported by a previous study (Naruse and Oishi, 1994), in which Japanese loaches showed nocturnal activity when fed at midnight, with strong activity in the morning (ZT0–ZT6) when fed at noon (ZT6). Furthermore, the feeding affected the activity patterns in the following DD or LD without feeding (Naruse and Oishi, 1994); the dark-active pattern under the scheduled feeding at midnight was maintained in the condition without food, whereas the morning-active pattern under the scheduled feeding at noon did not continue in the subsequent LD condition without food. In the present study, the loaches were fed before measurement and fed only once (day 3, ZT10) during the measurement at late daytime, to minimize the possible effects of feeding time on locomotor activity. Under the present condition, eight of 14 fish showed nocturnal locomotor activity without the anticipatory activity around the end of daytime (Fig. 1A–H) and no individual fish showed clear diurnal activity. Therefore, these results suggest that environmental light is a stronger determinant of the behavioral pattern than feeding time, although we did not determine the precise contribution of feeding schedule to the behavioral pattern.

Under the LD condition, some animals did not show any significant increase of locomotor activity in the dark phase (Fig. 1I–N, top rows), while others showed clear nocturnal behavior (Fig. 1A–H). These non-nocturnal individuals showed lower activity throughout the day (Fig. 1I–L and N; bottom rows), which may be the reason for the technical difficulty in the detection of a significant daily variation. Although it is difficult to measure the behavioral difference between the active and inactive individuals, genetic diversity may play a role in regulating activity levels (Kjaer, 2017). However, no individuals showed any diurnal locomotor activity in the present study, and, therefore, our results suggest that loach locomotor activity is likely to be suppressed by light and weakly regulated by the circadian clock.

Table 3.

Features of human, mouse, and loach eyes.

Fig. 7.

Locomotor activity of blinded Japanese loach. The locomotor activities of intact fish were measured for seven days in LD (12:12). On day 7, the eyes were enucleated under ice anesthesia at ZT10–ZT11 (black arrow heads) and treated for one day with 50 µg/L malachite green oxalate (grey horizontal lines). Their locomotor activity was subsequently measured for seven days under LD cycles. Top rows: Double-plot actograms of the locomotor activities of eight Japanese loaches. Middle rows: Chi-squared periodograms of the locomotor activities. Significance lines (P < 0.01) are colored in grey. The presumed free-running periods are shown. Eye-X; eye excision. Bottom rows: Analyses of the locomotor activities in a 4-hr window. One cycle was divided into six equal periods. Locomotor activities of five daily cycles (shown by grey bars on the right side of double-plot actograms) were averaged for every period (counts ± SD per min). The averaged values of intact fish for dark and light points were plotted by solid and open grey squares, respectively. The averaged values of Eye-X fish for dark and light points were plotted by solid and open black circles, respectively. The significance of the interaction between the circadian rhythm and light conditions was examined using two-way factorial ANOVA and shown as significant interaction (s.i.) if P < 0.05. Multiple comparison tests were performed using the Tukey-Kramer method. Asterisks and daggers represent significant variations depending on the eye experiment conditions and time periods, respectively. Different letters indicate significant differences between comparisons (P < 0.05). Dashed lines represent averaged locomotor activities with intact and Eye-X condition, which were drawn when a significant interaction was not detected. Significant daily variations were detected by two-way factorial ANOVA.

Visual pigment of Japanese loach

The spectral shape of the major photopigment in the loach eye correlated well with that of the template for porphyropsin (Fig. 6G, inset), and the difference absorption maximum of the pigment (520 nm) was very close to that of porphyropsins in the eyes of other freshwater fish species (520–524 nm; Wald, 1939). Based on these results, together with a previous report (Toyama et al., 2008), the main pigment, as observed in the spectroscopy, was porphyropsin.

Mice, a well-known nocturnal animal species, have a rod-dominant retina and poor eyesight (Hensch, 2005; Prusky and Douglas, 2004). Lyubarsky et al (2004) reported that the axial density of rhodopsin and the length of the rod outer segment are 0.019 OD/µm and 24 µm, respectively, whereby, the density of rhodopsin can be calculated to be 0.46 OD/cm² (Table 3). Compared with these values, the density of Japanese loach porphyropsin is estimated to be approximately 16% of that of mouse rhodopsin. In addition, Japanese loaches have relatively small eyes (relative to body mass) in contrast to known nocturnal reef fishes, such as the haemulid fish H. sciurus and the apogonid fish P. kauderni, whose eyes are larger than those of diurnal species due to adaptations to scotopic image formation (Schmitz and Wainwright, 2011). As such, Japanese loaches may have no or reduced vision with specialized eyes for detecting environmental light conditions. In general, nocturnal animals adapt to dark environments using different strategies. Some nocturnal animals have rod-enriched retinas to achieve a high sensitivity of scotopic vision. In other cases, they have small eyes or become blind by developing alternative non-visual sensory systems, such as chemosensory or touch sensory systems. Japanese loaches adopt the latter strategy, because they live in muddy water, where they use well-developed barbels for sensory systems instead of vision to find and capture prey.

Non-visual opsin expressed in loach eyes

In addition to visual opsins, we detected the expression of Opn4m3 mRNA in the eyes (Fig. 4, lane 17). Opn4s (melanopsins) are known as circadian photoreceptors that transmit light signals to the circadian clock system (Lucas et al., 2007; Lucas et al., 2001). ZOpn4m3 has been reported to be photo-responsive in zebrafish (Davies et al., 2011). Therefore, among the five types of maOpn4 genes identified in the present study (Fig. 5), maOpn4m3 is a plausible candidate gene encoding a circadian photoreceptor in loach eyes.

Light-input pathway to regulate the locomotor activity

Despite being small, loach eyes retain both their circadian and photoresponsive functions, and their removal did not have any effect on nocturnal locomotor activity (Fig. 7). However, this is not wholly indicative of the lack of or small contribution of the eyes to behavioral regulation, since extraocular photosensitive tissues, such as the pineal gland or deep brain photosensitive tissues (Moore and Whitmore, 2014; Okano et al., 1994; Vivid and Bentley, 2018; Cassone and Menaker, 1984), may also contribute to the regulation of locomotor activity in parallel with the light-input pathway in the retina. As such, extraocular photoreceptive site(s) should be investigated in future experiments, using local excision or illumination of the brain in living fish.

Under the LD condition, the loaches showed locomotor activity of onsets and offsets which clearly coincided with lights being turned on and off, respectively (Fig. 1). This suggests that environmental light signals are likely transmitted to the central nervous system to control the locomotor activity in a circadian-clock-independent manner, and that the circadian clock weakly affects the locomotor activity, along with the light-dependent pathway(s). Alternatively, the neuronal center regulating the locomotor activity may detect the environmental light directly.

Ecological aspects of light-dependent nocturnal locomotor activity in Japanese loach

The physiological and ecological importance of the strong photic and weak circadian regulation of the locomotor activity in the Japanese loach may be relevant to the desynchronization between the locomotor activity and the internal circadian clock. Such a desynchronization has been reported in a cavefish, A. mexicanus (Beale et al., 2013), wherein the circadian clock and light-dependent action may be important factors for the maintenance of internal homeostasis and for the adaptation of the fish to the local light environment in the cave, respectively. Although the Japanese loach does not inhabit a cave environment, its ecological environment may have several aspects in common to cave environments. Japanese loaches prefer to hide in tunnel-shaped tubes in artificial breeding environments, as well as to bury themselves in the sand. This type of behavior, in which they shield themselves from light, may provide them with a strategy for avoiding predation by birds. Indeed, the predators of loaches, such as Japanese crested ibis and Japanese white stork, seem to show diurnal behavior, like many birds (Cassone et al., 2014). Loaches are presumed to hide from predator birds during the daytime and become active, i.e., out of the sand, for eating during the nighttime.

Future perspectives

In this study, we identified clock genes in the Japanese loach and investigated their role in the photic and circadian regulation of the eyes. The analysis of Opsin genes and porphyropsin in the eyes suggested their visual and circadian photoreceptive function. However, eye excision experiments demonstrated that the eyes do not play an essential role in the regulation of functional behavior. Rather, taking into consideration the muddy water as the natural habitat of the Japanese loach, it may be interesting to investigate whether its eyes may also be specialized for magnetoreception. In migratory birds, the eyes operate to detect geomagnetic fields, possibly using cryptochromes as magnetoreceptors (Ritz et al., 2000). Geomagnetosensing is also found in the non-migratory animals, such as chicken (Wiltschko et al., 2007; Freire et al., 2005) and zebrafish (Takebe et al., 2012). This study may be of relevance to industrial fisheries. That is, considering that the Japanese loach is a potential food resource that is extraordinarily rich in proteins and vitamins, further ethological and chronobiological experiments would help to establish Japanese loach as an edible model fish.