To identify circadian patterns of gene expression in skeletal muscle, adult male zebrafish were acclimated for 2 wk to a 12:12-h light-dark photoperiod and then exposed to continuous darkness for 86 h with ad libitum feeding. The increase in gut food content associated with the subjective light period was much diminished by the third cycle, enabling feeding and circadian rhythms to be distinguished. Expression of zebrafish paralogs of mammalian transcriptional activators of the circadian mechanism (bmal1, clock1, and rora) followed a rhythmic pattern with a ∼24-h periodicity. Peak expression of rora paralogs occurred at the beginning of the subjective light period [Zeitgeber time (ZT)07 and ZT02 for roraa and rorab], whereas the highest expression of bmal1 and clock paralogs occurred 12 h later (ZT13–15 and ZT16 for bmal and clock paralogs). Expression of the transcriptional repressors cry1a, per1a/1b, per2, per3, nr1d2a/2b, and nr1d1 also followed a circadian pattern with peak expression at ZT0–02. Expression of the two paralogs of cry2 occurred in phase with clock1a/1b. Duplicated genes had a high correlation of expression except for paralogs of clock1, nr1d2, and per1, with cry1b showing no circadian pattern. The highest expression difference was 9.2-fold for the activator bmal1b and 51.7-fold for the repressor per1a. Out of 32 candidate clock-controlled genes, only myf6, igfbp3, igfbp5b, and hsf2 showed circadian expression patterns. Igfbp3, igfbp5b, and myf6 were expressed in phase with clock1a/1b and had an average of twofold change in expression from peak to trough, whereas hsf2 transcripts were expressed in phase with cry1a and had a 7.2-fold-change in expression. The changes in expression of clock and clock-controlled genes observed during continuous darkness were also observed at similar ZTs in fish exposed to a normal photoperiod in a separate control experiment. The role of circadian clocks in regulating muscle maintenance and growth are discussed.
- circadian rhythms
teleost fish show pronounced circadian rhythms of foraging behavior and locomotor activity that are driven by central oscillators in the brain, synchronized by light cycles (12), and modulated by a variety of environmental cues including temperature (29) and food availability (40, 41). The rhythm and period of circadian biological processes are driven by a complex molecular clock machinery that is highly conserved across the animal kingdom. Knowledge about basic clock mechanisms and functions are largely derived from studies in Drosophila (35, 36) and mice (38), with increasing interest in the zebrafish model (Danio rerio) (reviewed in Ref. 47). The molecular clock involves transcription-translation and posttranslational feedback loops; for example, in mammals, the transcription factor BMAL1 forms a dimer with other PAS domain proteins CLOCK to activate Period (Per1 and Per2) and Cryptochrome (Cry1 and Cry2) genes (38). PER and CRY proteins are translocated into the cell nucleus where they inhibit their own transcription. Secondary feedback loops involving the RORA (transcriptional activator of bmal1) and REV-ERBα genes (NR1D1, transcriptional repressor of bmal1) act to stabilize the clock mechanism, whereas posttranslational modifications of the PER-CRY dimer are required to set the period of the rhythm (43). In mice, light information received by the eyes travels to the suprachiasmatic nucleus that synchronizes the central and peripheral molecular clocks in a hierarchical manner (cf. Ref. 38), i.e., peripheral clocks are entrained and synchronized to the central pacemaker. Studies in the teleosts have demonstrated that the pineal gland is the central photoreceptive organ, which contains intrinsic circadian oscillators that drive the rhythmic secretion of melatonin in response to light input (cf. Ref. 7, 16). The role of the pineal as a hierarchical master clock in fish has been under discussion since observations that dissected zebrafish peripheral tissues (heart and kidney) and cells in cultures can be directly entrained by light and show a robust circadian rhythm (54). There is a growing body of information on zebrafish genes whose expression is inducible by light (44, 46, 53). Of particular interest is the observation that two transcriptional repressors of the circadian rhythm, namely per2 and cry1a, are inducible by light (44, 46). For example, deletion of D- and E-boxes from the promoter of the per2 gene resulted in loss of light responsiveness of this gene (46). Thus, the peripheral clock mechanism in fish have some similarities with that in Drosophila, which is directly responsive to light (36). However, it remains to be determined whether enough light reaches the internal organs of adult zebrafish to render them photoreceptive and photoresponsive and how the putative in vivo peripheral entrainment to light interacts with the neuroendocrine signals from the pineal.
Microarray studies have identified several hundred genes with circadian patterns of expression in mouse liver and skeletal muscle (33). Genes that are under control of the clock mechanism are referred to as clock-controlled genes (CCGs), which are responsible for the integration between the clock mechanism and other physiological pathways, and ultimately orchestrate the biological output of the circadian pathway. For example, the transcriptional repressor of the stabilizing loop nr1d1 has been shown to play an important role in the genomic recruitment of histone deacetylase 3 (HDAC3) in mouse liver (17). HDAC3 functions in lipid homeostasis, and absence of nr1d1 caused impaired lipid metabolism with subsequent changes in the phenotype of the liver (17). The clock mechanism is also important for the physiology of other peripheral tissues and has been shown to play a pivotal role in maintaining muscle phenotype in the mouse (2). In this tissue, the clock gene controls the expression of myoD, a member of the myogenic regulatory factor family, which functions in muscle determination and differentiation (2, 32). The absence of a functional clock mechanism in this tissue led to reduced force generation and reduced mitochondrial volume density, mediated by the CCGs myoD and pcg1a/β, respectively (2).
Zebrafish have many advantages as a model system for investigating circadian clocks, including transparent embryos, which facilitate the imaging of fluorescent reporter genes in vivo and the ease of performing large-scale forward genetic screens (47). Danio is a diurnal fish that is mostly active during the subjective light phase of the photoperiod with clear differences in locomotor activity and spawning behavior between the dark and light phases (4, 26). Teleost fish underwent whole genome duplication early in their evolution, and subsequent differential patterns of gene loss have resulted in lineage-specific differences in the paralogs retained (49). For example, the zebrafish and Tiger pufferfish (Takifugu rubripes) have three clock genes, whereas the stickleback and Japanese Medaka fish have two (49). Additional copies of bmal1, cry1, cry2, per1, rora, and rev-erbβ (nr1d2) genes have also been described for zebrafish (19, 28, 48, 50). The expression pattern of some core clock genes of the circadian rhythm in zebrafish has been investigated in the retina, brain, pineal gland, and Z3 cell line (reviewed in Ref. 47).
The main objective of the present study was to provide a detailed description of the expression of 17 clock genes and their paralogs in zebrafish skeletal muscle. Circadian patterns of expression were determined in relation to the subjective light cycle in fish exposed to 3–4 cycles of continuous darkness. Using quantitative PCR (qPCR) and a robust normalization strategy, we also investigated the hypothesis that the expression of myogenic regulatory factors, components of the IGF system, and other selected nutritionally responsive genes in skeletal muscle, is under control of the circadian clock mechanism.
MATERIALS AND METHODS
Fish and water quality.
The F8 generation of a wild-caught population of zebrafish [Danio rerio (Hamilton 1822)] from Mymensingh, Bangladesh was used. All fish were adult males aged 10 mo [total length = 38.1 ± 0.2 mm and body mass = 496 ± 8 mg (means ± SE, n = 130)]. The source colony and experimental animals were kept in a stand-alone freshwater circulating system, which included a UV water sterilizing device and biological, chemical, and particle filters. Nitrite (0 ppm), nitrate (10–20 ppm), ammonia (0 ppm), and pH (7.6 ± 0.2) were tested during acclimation and experimental periods using a Freshwater Master Test Kit (Aquarium Pharmaceuticals, Chalfont, PA). All experiments and animal handling were approved by the Animal Welfare and Ethics Committee, University of St. Andrews, and conformed to United Kingdom Home Office guidelines.
Circadian rhythm experiment.
The 140 male fish from the same breeding stock were transferred to four separate tanks (n = 35 per tank, 50 liters freshwater) maintained at 27.6 ± 0.3°C range, in a 12:12-h light-dark photoperiod and fed bloodworms (Ocean Nutrition, Essen, Belgium) to satiety twice daily. Fish were acclimated for 2 wk in the experimental tanks under the same environmental conditions as the source colony. Lights were switched-off at the beginning of a light cycle (10 AM) to start the experiment [referred to as cycle 1: Zeitgeber time 0 h (C1:ZT0)] and continuous darkness was maintained for 86 h, corresponding to three complete light-dark cycles of the acclimation period (Fig. 1). Ten fish were randomly sampled from one of the four tanks every 4 h from C2:ZT02 to C4:ZT02, resulting in 13 time points (n = 130 fish), using a very dim torchlight directed to the floor. The fish were not disturbed during sample collection, as no sudden change in activity was observed. After collection at each time point, additional bloodworms were offered to the fish to make sure food was available at all times across the experiment. This experimental design reduced the possibility of stress due to excessive handling, since each tank was minimally disturbed only every 16 h. In addition, no maintenance was necessary for the duration of the acclimation and experimental periods due to the automatic filtration system, which prevented the creation of external cues to which the fish could respond. The fish were killed by an overdose of ethyl 3-aminobenzoate methanesulphonate salt (MS-222; Fluka, St. Louis, MO) and had their total length and body mass measured. The condition factor was calculated as K = [(body mass/100)/(total length/10)^3]. No statistical difference was observed for body mass or the condition factor among the time points over the experiment (not shown). Fast skeletal muscle was dissected from the dorsal epaxial myotomes, flash frozen in liquid nitrogen, and stored at −80°C prior to total RNA extraction. The digestive tract was dissected and fixed in 4% (mass/vol) paraformaldehyde for later quantification of intestine content to the nearest milligram.
A control experiment under normal 12:12-h light-dark photoperiod was performed 8 mo later using the F9 generation of fish, which were 11 mo old and were direct offspring of fish from our F8 laboratory line. Since fish from F9 had the same genetic background and were approximately the same age, we believe that these two separate experiments were comparable. This control experiment was designed to answer a question raised during the review process and had the specific aim of confirming that the oscillation in gene expression observed during the continuous darkness condition would also be observed in a normal 12:12-h light-dark photoperiod. Thirty-five male zebrafish from the F9 generation were transferred to a recirculating system of 50 liters of freshwater and acclimated for 2 wk in the same environmental conditions as previously described for the F8 generation. After the acclimation period, 10 fish were collected at ZT02 (12 AM, 2 h after lights on) and ZT14 (12 PM, 2 h after lights off) (n = 20) and had their body lengths and weight measured; viscera and epaxial skeletal muscle were dissected for gut food content and total RNA extraction, respectively.
Total RNA extraction from skeletal muscle and first-strand cDNA synthesis.
Total RNA was extracted by homogenization in Lysing Matrix D (MP Biomedicals, Qbiogene, Irvine, CA) with TRI reagent (Sigma) in a FastPrep apparatus (Qbiogene). The RNA concentration, 260:280 and 260:230 ratios were measured using a NanoDrop spectrophotometer (ThermoFisher Scientific, Loughborough, Leicestershire, UK) and were between 1.7 and 2.0 and >2.1, respectively. RNA integrity was also checked by agarose gel electrophoresis. A Quantitect reverse transcription kit (Qiagen, Hilden, Germany) was used to produce first-strand cDNA from 1.0 μg of total RNA following the manufacturer's instructions.
Primer design and screening for circadian expression by qPCR.
Primer pairs were designed as previously described (1) for 16 genes described as core-clock genes in other vertebrate models and tissues (bmal1a, bmal1b, clock1a, clock1b, cry1a, cry1b, cry2a, cry2b, per1a, per1b, per2, per3, roraa, rorab, nr1d2a, and nr1d2b) and four myogenic factors genes (myoD, myog, myf5, and myf6) (Table 1). Previously validated primer pairs for 15 genes of the IGF pathway (igf1a, igf2a, igf2b, igf1ra, igf1rb, igf2r, igfbp1a, igfbp1b, igfbp2a, igfbp2b, igfbp3, igfbp5a, igfbp5b, igfbp6a, and igfbp6b), two ubiquitin ligases genes (MAFbx and trim63), and 12 nutritionally responsive genes [odc1, hsp90a.1, fkbp5, sae1, hsp90a.2, foxo1a, klf11b, and nr1d1 (known to be a circadian gene), cited2, bbc3, znf653, and hsf2] were also used (1).
The qPCR reaction mixture contained 7.5 μl 2× Brilliant II SYBR QPCR Low ROX Master Mix (Stratagene, La Jolla, CA), 6 μl 40-fold diluted cDNA, and 0.25 μmol/l of each primer and nuclease-free water (Qiagen) to a final volume of 15 μl in 96-well plates (Stratagene). The reactions were performed in duplicates in a Stratagene MX3005P machine. Negative controls, thermocycling conditions, and data acquisition were as previously described (1) and followed the MIQE guidelines (6). After the qPCR a dissociation curve (from 55 to 95°C) was performed to verify the presence of a single peak. The specificity of each qPCR assay was also validated by directly sequencing the qPCR products in both directions. The efficiency of each primer pair was calculated by the LinReg software (39) (Table 1) and used to calculate arbitrary mRNA copy numbers. Normalization between plates was as described previously (1). Four reference genes [ef1a, bactin2, lman2 (1), and gapdh (Table 1)] were analyzed using Genorm version 3.5 (45) with M set to < 1.5. The two genes with the most stable level of expression across the experiment were ef1a and bactin2 (M = 0.3). The expression of genes of interest was normalized to the geometric average of the two most stable genes, and gene expression was reported as arbitrary units.
Genes were screened for circadian expression using pools containing equal amounts of cDNA from 10 fish per time point. Transcript levels were analyzed using the ARSER algorithm (56) with period window of 20–28 h and a false discovery rate set to 0.05 (3). Individual reactions were carried out for all genes that passed these screening criteria for rhythmic expression plus cry1b and nrld2a (two genes paralogous to core-clock genes). The screening step was robust since: 1) the results of mRNA levels calculated by the screening and individual reactions were highly correlated (Spearman's correlation test, n = 500, R = 0.85, P < 0.001); and 2) periodicity parameters calculated by the ARSER algorithm using the results from the individual reactions resulted in very similar values to those calculated for the screening reactions.
Gene expression was determined in the control samples to confirm that the oscillation observed during continuous darkness also occurred during normal light-dark conditions. This was possible after normalizing the level of expression of the control samples to four random samples from the first experiment, which were assayed in the same plate. For logistical reasons, we selected only 15 of the 37 genes assayed in the first experiment, which represented all possible transcriptional responses to photoperiod observed in the first experiment (3 reference genes: ef1a, bactin2, and gapdh; 3 transcriptional activators: clock1a, bmal1b, and rorab; 3 repressors: cry2a, per1b, and per3; 3 candidate CCGs: myf6, hsf2, and igfbp3; and 3 noncircadian and non-CCGs: murf1, myog, and hsp90a.2).
Data analysis and statistics.
All data was analyzed for normal distribution and equality of variance. Normally distributed data was analyzed using ANOVA followed by Tukey post hoc tests using PASW Statistics 18 (SPSS, Chicago, IL). Kruskal-Wallis nonparametric tests followed by Conover post hoc tests in BrightStat software (42) were used for the data that was not normally distributed. Hierarchical clustering of gene expression and heat maps were produced using PermutMatrix (http://www.lirmm.fr/∼caraux/PermutMatrix/EN/index.html). Correlation of mRNA levels between genes was analyzed by Spearman's correlation test in SPSS.
Fish continued to eat during the experiment, but showed a marked change in feeding behavior between C2 and C3, resulting in a decoupling of feeding activity from the light-dark cycle of the acclimation period. Gut food content (% body mass) was ∼1.5 at C2:ZT02, increased to ∼3.7 at C2:ZT10, and then declined to 0.8 at C2:ZT22. The food content of the gut only showed a modest increase to ∼1.4 during what would have been the light period of C3, declining to ∼0.3 at C4:ZT02, indicating a marked reduction of foraging behavior (Fig. 2). In the control experiment, a very similar pattern was observed with an increase of approximately twofold in gut food content from ZT02 (1.1%) to ZT14 (2.1%) under a 12:12-h light-dark photoperiod (Fig. 8A).
Noncircadian gene expression in skeletal muscle.
Twenty-five out of 32 of the non-clock genes screened showed no evidence for circadian patterns of expression (sae1, odc1, hsp90a.2, klf11b, foxo1a, fkbp5, cited2, bbc3, znf653 MAFbx, myf5, myoD, myogenin, igf1a, igf2a, igf2b, igf1ra, igf1rb, igf2r, igfbp1a, igfbp1b, igfbp2a, igfbp2b, igfbp5a, and igfbp6b) (Fig. 3). In addition, trim63, hsp90a.1, and igfbp6a passed the screening criteria, but failed to show strong evidence for circadian expression based on the individual reactions (adjusted r2 < 0.3 and false discovery rate > 0.07). Cry1b was the only zebrafish paralog of a core clock gene that had a noncircadian pattern of expression (Fig. 3).
The transcription levels of some genes were significantly correlated with the food content in the gut. Transcripts of igf1rb, MAFbx, bbc3, igf1ra, igfbp5a, and igf2b were negatively correlated (Spearman's correlation less than −0.5, P < 0.05), whereas odc1, igf2a, igfbp2b, and sae1 were positively correlated with gut food content (Spearman's correlation > 0.5, P < 0.05).
Expression of core clock genes in skeletal muscle.
The expression of zebrafish paralogs of the transcriptional activators of the circadian mechanism (bmal1 and clock1) and the transcription activator of bmal1 (rora) followed a circadian pattern in skeletal muscle. The two paralogs of bmal1 and clock1 were expressed in phase with each other showing peak expression at ZT14 (Fig. 4, A–D). The bmal1a and bmal1b showed an 8.5-fold change in expression between maximum and minimum values (Fig. 4, A and B). In contrast, clock1a was more responsive to the light-dark cycle than clock1b showing a 6.0- and 2.2-fold change in expression, respectively (Fig. 4, C–D). The highest expression of the two paralogs of the rora gene (roraa and rorab), known to activate bmal1 in mammals, occurred in a different phase from bmal1 and clock1, with an ∼4.5-fold change in expression of both paralogs between maximum (at ZT02) and minimum (at ZT18) (Fig. 4, E and F).
With the exception of cry1b, the expression of the transcriptional repressors determined in this study (cry1a, cry2a, cry2b, per1a, per1b, per2, per3, nr1d1, nr1d2a, and nr1d2b) followed a circadian pattern. Expression of the cry1a gene peaked at ZT02 (∼4.0-fold upregulation) (Fig. 5A). In contrast with the expression of cry1a, the expression of the two paralogs of the cry2 gene (cry2a and cry2b) occurred in phase with bmal1 and clock1, with both paralogs showing similar amplitude of expression in relation to the photoperiod (Fig. 5, B and C).
The transcript levels of all four per genes assayed (per1a, per1b, per2, and per3) occurred in phase with the transcriptional repressor cry1a. A small shift in the phase of expression was observed among the per genes; per1b and per3 mRNA levels were at their highest 4 h later than the peak expression of per1a and per2 genes (Fig. 5, D–G). The fold-change in transcription level of the per1a (∼51.7-fold) and per3 (∼23-fold) were among the highest of all circadian genes studied (Fig. 5, D–G).
Expression of the nuclear receptors genes nr1d1, nr1d2a, and nr1d2b, which belong to the negative loop of transcriptional regulation of the circadian rhythm in mammals, also occurred in phase with expression of the cry1a gene (Fig. 6, A–C). Peak expression of nr1d1 was ∼47-fold higher than its lowest expression (Fig. 6A). A relatively weak, but significant, negative correlation was found between the expression of this gene and food gut content (R = −0.56, P = 0.040).The maximum change in expression of nr1d2a (∼7.3-fold) was significantly higher than for nr1d2b (∼2.3-fold) (Fig. 6, B and C).
Putative clock-controlled genes.
Expression of two IGF-binding proteins (igfbp3 and igfbp5b) and one myogenic regulatory factor (myf6) occurred in phase with the paralogs of the transcriptional activators bmal1 and clock1, with an average of ∼2.0-fold regulation (Fig. 7, A–C). Expression of the heat shock transcription factor 2 gene (hsf2) occurred in phase with cry1a and other genes that belong to the negative arm of the transcriptional regulation network of the circadian rhythm (Fig. 7D). Transcripts of this gene were ∼7.0-fold higher at ZT02 than ZT14 (Fig. 7D).
Expression of circadian genes and CCGs under 12:12-h light-dark photoperiod.
The pattern of gene expression in skeletal muscle under a normal photoperiod was similar to the one observed under a continuous darkness condition (Spearman's correlation coefficient = 0.699, P < 0.001). However, the amplitude of expression between the light and dark periods was higher in skeletal muscle from fish exposed to a normal photoperiod compared with continuous darkness (Fig. 8).
Gene clustering and correlation analysis.
The 20 genes found to have a circadian rhythm of expression could be grouped in two major clusters (Fig. 9). Cluster I comprises genes with peak expression around the middle of the subjective dark photoperiod and included bmal1a, bmal1b, cry2a, clock1b, myf6, clock1a, cry2b, igfbp3, and igfbp5 (Fig. 9). Transcript level of paralog genes in this cluster were highly positively correlated, the two paralogs of the bmal1 gene had the highest correlation coefficient (R = 0.92, P < 0.001), followed by the paralogs for the clock gene (R = 0.84, P < 0.001), and the lowest correlation was found for the two paralogs of the cry2 gene (R = 0.76, P = 0.002) (Table 2). Genes with peak expression during the last time point of the subjective dark photoperiod and the two time points from the subjective light photoperiod were grouped in cluster II (cry1a, hsf2, per1b, per3, nr1d2b, roraa, rorab, nr1d1, per1a, per2, and nr1d2a) (Fig. 9). In this cluster, only the paralogs of the rora gene had highly significant positive correlation in mRNA levels (R = 0.79, P = 0.001) (Table 2). Weak, but statistically significant, negative correlations were found between gut food content and transcription level of per1a (R = −0.57, P = 0.040) and nr1d1 (R = −0.56, P = 0.040). No significant positive correlation was found between the expression of circadian genes and gut food content.
Danios show an endogenous circadian nocturnal feeding behavior when under a 12:12-h light-dark photoperiod (12). In the present study, however, continuous darkness led to an inhibition of the feeding response by the third subjective light cycle, as evidenced by direct observation of the food present in the gut, enabling the transcriptional responses due to feeding and circadian rhythmicity to be distinguished. Feeding activity in teleosts initiates well-characterized transcriptional responses (1, 5, 37). Transcripts for the ubiquitin ligase MAFbx, IGF-1 receptors (igf1ra, igf1rb), and the mitochondrial proapoptotic BCL2 binding protein component 3 (bbc3) were inversely correlated with gut food content (Fig. 3) as previously reported (1). In contrast, transcripts for ornithine decarboxylase (odc1) and sumo-activating enzyme (sae1), previously shown to be positively correlated with gut food content (1), had peak expression during the subjective light phase of the second diurnal cycle and reduced expression throughout the whole of the third diurnal cycle (Fig. 3).
Expression of core-clock genes in zebrafish skeletal muscle.
BMAL1 and CLOCK are considered the central transcriptional activators of the circadian mechanism due to their ability to bind to E-box elements and activate the transcription of most of the core-clock genes (Fig. 10). In zebrafish skeletal muscle the expression of the respective paralogs of the two transcriptional activators (bmal1a, bmal1b, clock1a, and clock1b) clustered together (Fig. 9) and were highly correlated (Table 2). Our results are very similar to the expression pattern described in organs that are considered central pacemakers of the zebrafish clock mechanism (8, 54).
Period proteins (Per 1, 2, and 3) together with the Cryptochrome proteins 1 and 2 are responsible for the negative loop of the circadian mechanism (47). In addition, period proteins have been shown to be important for maintaining the pace of the clock machinery (21, 46). Among the period genes assayed in cultured zebrafish cells only per2 expression was inducible by light (44) (Fig. 10). In skeletal muscle of the zebrafish per1a and per2 showed peak expression at the end of the subjective dark period (ZT22), whereas Per1b and Per3 showed highest expression at the start of the light period (ZT02) (Fig. 5, D–G). Interestingly, expression of per3 in the zebrafish skeletal muscle was similar to that found in Z3 zebrafish cells (34) and in the retina and optic tectum of the nocturnal flatfish Solea senegalensis (30), with highest expression at around ZT02 (Fig. 5G).The observation that diurnal and nocturnal fish have the same pattern of expression in central and peripheral tissues might be valuable in investigating its function.
The paralogs of both cry1 and cry2 have been demonstrated to inhibit bmal: clock-directed transcriptional activation (47). Cry1a is considered to act on the core clock machinery and was the only transcript from the cryptochrome genes whose expression was induced by light in zebrafish cell cultures (44) (Fig. 10). The expression of cry1a, cry2a, and cry2b in skeletal muscle (Fig. 5, A–C) were similar to that described in the eye and brain of the zebrafish (28), considered central organs of the circadian mechanism. The difference in phase of expression of cry1a and paralogs of the cry2 in the muscle were as previously described in the eye and brain (28), with peak expression of cry1a occurring at ZT02 and cry2 paralogs at ZT14. Expression of cry1b in the muscle (Fig. 3), however, did not follow the circadian pattern described for the central organs (28). Furthermore, cry2 genes in zebrafish (Fig. 5, B and C) and mouse muscle are expressed in anti-phase (32). In the current model of the circadian clock in the zebrafish, CRY1 and CRY2 proteins form heterodimers that translocate to the nucleus where they inhibit the BMAL:CLOCK-dependent transcriptional activation (47). In the skeletal muscle, however, the CRY1B might not be part of the pool of CRY proteins available to form dimers with PER proteins (Fig. 10).
The nr1d1, nr1d2, and ror genes code for nuclear receptors involved in the stabilizing loop of the circadian clock mechanism (15, 47) (Fig. 10). The rev-erbα and -β receptors, coded by the zebrafish nr1d1 and nr1d2, respectively, are considered constitutive transcriptional repressors of bmal1, whereas ror genes are transcriptional activators of bmal1 (20) (Fig. 10). In addition, nr1d1 has been recently suggested to act as a transcriptional repressor for the bmal1 partner clock (11), regulating the transcription of both main transcriptional activators of the circadian mechanism (Fig. 10). In zebrafish skeletal muscle, the expression of nr1d2 and rora clustered together with cry1a (Fig. 9). However, nr1d1, nr2d2a, and nr2d2b transcripts levels peaked at the end of the subjective dark period (ZT22), while transcripts of roraa and rorab were at their highest levels at the beginning of the subjective light period (ZT02–04) (Fig. 9). This small difference in phase of expression of the two components of the stabilizing loop might reflect their tight control of regulation over the circadian mechanism, which is reflected in the activation/repression of bmal1 and clock1 expression. In addition to their role in regulation of circadian rhythm, these genes are known transcription factors for genes involved in lipid metabolism (14). In the present study, a negative correlation was found between expression of nr1d1 in skeletal muscle and gut food content in accordance with previous findings (1). It is plausible that nrld1 may play a role in integrating circadian and metabolic rhythms in skeletal muscle.
Expression of putative clock-controlled genes in zebrafish skeletal muscle.
In the zebrafish, the insulin-like growth factor pathway is comprised of four ligands (IGF1A, IGF1B, IGF2A, and IGF2B), their respective receptors (IGF1AR, IGF1BR, and IGF2R) and nine IGF-binding proteins (IGFBP1A, IGFBP1B, IGFBP2A, IGFBP2B, IGFBP3, IGFBP5A, IGFBP5B, IGFBP6A, and IGFBP6B). Interaction between the ligands and IGF1-receptors ultimately leads to tissue growth, with the binding proteins playing important roles in regulating the concentration of the ligands in the plasma and their release in target tissues. We have previously shown that igf1a and igf2b are upregulated during feeding, while igf1ra, igf1rb, igfbp1a, and igfbp1b are upregulated during fasting (1). In the present study, expression of other two IGF-binding proteins genes (igfbp3 and igfbp5b) was rhythmic and peaked at the onset of the dark phase (ZT14), in phase with the transcriptional activators bmal1 and clock1 (Fig. 9). It is possible that changes in transcript levels are reflected in protein levels with effects on biological functions, although this is not always the case due to the complexity of regulatory mechanisms affecting mRNA stability and degradation, translational control, etc. Overexpression, knockdown and knockout systems have been previously employed to study the biological importance of the IGF-binding proteins in skeletal muscle (reviewed in Ref. 13). Most circulating IGFs in the plasma are found to be conjugated to IGFBP3, and this binding protein serves as a modulator of the IGF action in target tissues by prolonging hormone half-life (18, 55). IGFBP3 also have IGF-independent actions in inhibiting cell proliferation in cancer lines (55). IGFBP5 is known to play a crucial role in muscle growth and differentiation (reviewed in Ref. 13) and circadian expression of this growth-related gene has been previously reported in the skeletal muscle of mouse (33). Given the importance of IGFBP3 and IGFBP5 in the growth axis and the involvement of the clock pathway in the cell cycle, a plausible hypothesis is that the cyclic expression of these two IGF-binding proteins is related to the local regulation of cell-cycle and growth.
Myogenic regulatory factors (MRFs) (myoD, myf5, myf6, and myogenin) are a class of helix-loop-helix transcription factors that play a pivotal role in myogenesis (9, 10, 23). Myf6 (also known as MRF4) was shown to function in myogenic determination and differentiation in myf5/myoD double knockout mouse (27). Myf6 was found to play an important role in muscle fiber alignment in zebrafish embryos, using the morpholino technique to knockdown two splice-variants of myf6 transcripts (51). In the present study, the expression of myf6 was not correlated with food intake in C3, but it did exhibit a circadian expression pattern peaking in phase with bmal1 and clock1 at the beginning of the subjective dark period (Fig. 7C). Similar circadian patterns of myf6 expression were reported previously in skeletal muscle of the horse, a mammalian species with higher physical activity during daylight hours (31). In the mouse, myoD is a direct target of CLOCK and BMAL, which bind in a rhythmic fashion to the core enhancer in the myoD promoter (2, 32). ClockΔ19 and Bmal1−/− mutants showed similar phenotypes to myoD−/− mutants with reduced force generating capacity relative to wild-types due to a disruption of myofilament organization (2). In contrast, we found no evidence for circadian expression of myoD in zebrafish (Fig. 3). It is plausible that the well-known redundancy of MRFs may have resulted in lineage-specific differences in their regulation by clock genes. A plausible hypothesis would be that the rhythmic expression of myf6 in zebrafish muscle parallels that described for myoD in mouse muscle, with potential effects on the maintenance of myofibrillar structure.
The rhythmic expression of the chaperone transcriptional regulator hsf2 was previously described in the pineal tissue of chicken (22) and zebrafish larvae (53). The expression of two chaperone genes (hsp90a.1 and hsp90a.2) in zebrafish skeletal muscle showed no discernible pattern of periodicity with respect to photoperiod, while expression of hsf2 was rhythmic and peaked at onset of lights on, in phase with the transcriptional repressor cry1a (Fig. 7D). The role of hsf2 as a clock-controlled gene in the circadian output is not known, but evidence from experiments with chicken point to the activation of specific stress-response factors in response to light (22). In zebrafish larvae, expression of hsf2 was concomitant with expression of genes involved in the response of oxidative stress and chaperone genes (53). Exposure to light is known to cause oxidative stress through production of hydrogen peroxide in zebrafish cells (24, 25) with subsequent activation of stress-responsive genes, including the (6-4) pyrimidine-pyrimidone dimer DNA photolyase involved in DNA repair (25). The production of hydrogen peroxide and subsequent activation of stress-responsive genes and the MAPK signaling pathway has recently been considered one of the potential mechanisms that render peripheral tissues to be photoreceptive and photoresponsive, since these events regulate transcription of cry1a in the zebrafish with noticeable effects on the circadian mechanism (24, 25, 47).
In the present study, we characterized the expression of the main clock of the circadian mechanism in the skeletal muscle of the zebrafish (Fig. 10). Most of these genes had a similar expression pattern to that described for the central organs (retina and brain) of the circadian mechanism (reviewed in Ref. 47). In addition, we provide evidence for differences in the responsiveness of clock1 and nr1d2 paralogs to circadian stimuli and the loss of a circadian rhythm for cry1b in skeletal muscle. Finally, gene expression of two IGF-binding proteins (igfbp3 and igfbp5b) and a myogenic regulatory factor (myf6) involved in igf-mediated growth and terminal muscle differentiation in fish, respectively, were identified as clock-controlled gene in zebrafish skeletal muscle. This finding points to an important physiological role of the clock mechanism in regulating muscle mass homeostasis through integration with the IGF pathway and MRFs.
Perspectives and significance
We have described the transcriptional regulation of the main circadian genes and identified CCGs in skeletal muscle of the zebrafish. These studies provide a foundation for investigating the integration of the clock system with physiological processes in teleosts. We found that myf6 was a clock-driven MRF rather than MyoD as reported in mice where it regulates aspects of phenotype maintenance including force generation and myofibrillar structure [Andrews et al. (2). PNAS 107, 19090–19095]. In addition, the finding that the circadian expression of many genes in this peripheral tissue is similar to those described for organs considered central photoreceptive and pacemakers of the circadian mechanism is valuable for future investigations on the hierarchy of the systemic clock, i.e., the integration between the neuroendocrine signals from the pineal, the central pacemaker organ, and peripheral clocks in fish.
This research was funded by the European Community's Seventh Framework Programme FP7/2007–2013 under Grant Agreement 222719-LIFECYCLE. I. P. G. Amaral was supported by research studentships from the Programme Alβan, the European Union Programme of High Level Scholarships for Latin America Scholarship E07D402823BR, from the Sir Harold Mitchell Fund, and from the Capes Foundation, Ministry of Education of Brazil Scholarship BEX 0449-10-5.
No conflicts of interest, financial or otherwise, are declared by the author(s).
I.P.G.A. and I.A.J. conception and design of research; I.P.G.A. performed experiments; I.P.G.A. analyzed data; I.P.G.A. and I.A.J. interpreted results of experiments; I.P.G.A. prepared figures; I.P.G.A. and I.A.J. drafted manuscript; I.P.G.A. and I.A.J. edited and revised manuscript; I.P.G.A. and I.A.J. approved final version of manuscript.
- Copyright © 2012 the American Physiological Society