Optimization of main components of artificial compound feed (ACF) for Hippocampus kuda based on the synergistic expression of biological clock genes

Abstract

Feed is very important for fish farming. The appropriate composition and proportion of feed ingredients can promote the growth of fish, maintain normal physiology and behavior, and even improve the resistance ability to disease and stress, etc. The core of artificial compound feed (ACF) is the composition and proportion of lipid, protein, and carbohydrate, which are also the main nutritional components required by fish. Appropriate levels and ratios can promote fish growth and save costs, and the improper would affect the biological clock systems of fish, leading to metabolic abnormalities. This study explored the preparation of ACF for H. kuda. The composition and proportion of the three main nutrients in ACF were screened based on the synchronicity between six pairs of clock genes (Clock, Bmal1, Per1, Per2, Per3, Cry1, and Cry2) in the central and peripheral clock systems, as well as the expression of eight lipid-metabolism genes (Hmgcr, Mvk, Mvd, Lss, Fdps, Cetp, Scap, Srebp1, Srebp2) in the liver and their synergy with liver clock genes. The results showed that, based on several parameters such as gene expression cycle, relative expression level, and top phase appearance time, the best synergy between the central and peripheral circadian clock systems was observed in ACF with crude fat content of 8.80%, crude protein content more than 38.4%, and carbohydrate content of 23.5%. Based on the expression relationship between lipid metabolism genes and circadian clock genes in the liver, it was further clarified that the optimal levels of fat, protein, and carbohydrate were determined with 8.80%, 38.4%, and 23.5%, respectively. After 4 weeks of breeding validation, compared with frozen Mysis, the screened ACF fed for H. kuda showed significant advantages in body length specific growth rate (SGR_L), body weight specific growth rate (SGR_W), and feed conversion rate (FCR).

Supplementary Information

The online version contains supplementary material available at 10.1007/s10695-025-01514-x.

Introduction

Feed is very important to fish farming. When the component and its ratio in the artificial compound feed (ACF) is not appropriate, it not only affects the growth of fish, but also leads to abnormal behavior and physiology of fish, reducing their resistance to adverse environments and diseases (Oliva-Teles 2012). Fat, protein, and carbohydrate are the three major nutrients required by fish. Appropriate fat content (6–15% species-dependent) in ACF enhances lipid peroxidation pathways, promoting protein-sparing effects and providing 15–20% energy efficiency for tissue synthesis in carnivorous species like rainbow trout (Oncorhynchus mykiss) (Tocher 2003). Balanced ω−3/ω−6 fatty acid ratios (1:1–1:4) improve membrane fluidity and reduce pro-inflammatory eicosanoid production in Atlantic salmon (Salmo salar) (Sargent et al. 2002). Low fat content (< 3%) can cause a deficiency of fat-soluble vitamins (A, D, E, K) and essential fatty acids (linoleic/α-linolenic acids), while high fat content (> 20%) upregulates Srebp1c and Fasn genes, inducing hepatic steatosis and impairing immunity (Qin et al. 2013).

When the protein content in ACF is insufficient (< 30% for carnivorous species), it limits methionine and lysine availability, reducing muscle synthesis efficiency and feed conversion rate. Carnivorous fish such as groupers (Epinephelus spp.) require ≥ 2.5% methionine in dietary protein to optimize IGF-1-mediated muscle growth and suppress myostatin (MSTN) (Mohanta et al. 2013; Xu et al. 2015; Dong et al. 2013). Conversely, if the protein content is too high (> 50%), excessive protein would be decomposed into energy via glutamate dehydrogenase (GDH), increasing ammonia excretion through the ornithine-urea cycle. Plant-based proteins (> 30% substitution) reduce zinc bioavailability via phytate chelation, impairing scale mineralization in common carp (Cyprinus carpio) (Kumar and De Boeck et al. 2012). This not only increases the cost but also raises un-ionized ammonia (NH3) levels above 0.02 mg/L, causing gill damage and hypoxia stress (Carter et al. 1998; Lam et al. 2008).

Carbohydrate is the cheapest energy source in ACF, and its utilization rate does greatly vary among different fish species. Cold-water species like rainbow trout exhibit prolonged postprandial hyperglycemia (> 200 mg/dL for 24 h) compared to warm-water tilapia (Oreochromis niloticus) (12–18 h), requiring stricter starch restriction (≤ 20%) (Enes et al. 2011). Dietary β-glucan (2–4%) enhances carbohydrate utilization in carp by stimulating AMPK-dependent glycolysis (hexokinase activity ↑35%) via gut microbiota-derived short-chain fatty acids (Do Huu et al. 2016). High-carbohydrate diets (> 40%) in carnivores induce persistent hyperglycemia (> 150 mg/dL for 24 h postprandial) and hepatic insulin resistance (Hemre et al. 1993; Li et al. 2012). Therefore, it is necessary to design ACF with different nutritional ratios for different fish.

Previous researches have shown that changes in ACF may have certain impacts on the biological clock system and nutrient metabolism (Wang et al. 2024). Dietary lipid-protein ratios (e.g., 1:2 vs. 1:4) alter circadian expression of Bmal1 and Per2 in zebrafish (Danio rerio) liver by modulating PPARγ/Rev-erbα crosstalk, thereby disrupting daily rhythms of lipogenesis and gluconeogenesis (Paredes et al. 2021). In fact, there is still relatively little research in this field, mainly focusing on the relationship between lipid metabolism and obesity. For example, a high-fat diet can disrupt the expression of circadian clock genes, thereby affecting the normal metabolism of mice and ultimately leading to metabolic diseases (Honma et al. 2016). In gilthead seabream (Sparus aurata), suboptimal carbohydrate levels (35% vs. 25%) suppress Clock gene expression in the hypothalamus by 40%, correlating with impaired melatonin rhythm and feeding entrainment (López-Olmeda et al. 2009). There have been no reports on the effects of changes in the other main components yet. Therefore, it is great significance that exploring the synergistic effects of ACF with different composition and their ratios on the biological clock systems and nutritional metabolism in farmed organisms.

The yellow seahorse, Hippocampus kuda, is a small marine medicinal and ornamental fish that is mainly fed with natural feed such as Mysis and Artemia (live or frozen) during artificial breeding at present. Preliminary transcriptome data reveal conserved circadian regulators (Bmal1, Per3) in seahorse brain tissues, suggesting feeding schedules may entrain their circadian physiology (Zhang et al. 2022). This feeding not only introduces a large number of unknown pathogens into the breeding system, but also causes water pollution and other problems due to feed spoilage. Therefore, there is an urgent need to develop ACF for the seahorse. This article explored the relationship between the main components of ACF and the biological clock system and conducted preliminary optimization researches on ACF for artificially cultivated seahorses.

Materials and methods

ACF preparation

The experiment used fish oil and soybean oil as fat sources; used soy protein concentrate, fish meal, and soybean meal as protein sources; and used dextrin as sugar source to prepare three types of ACF for optimizing the contents of crude lipids, crude proteins, and carbohydrates of the seahorse ACF. The set levels of lipid addition were 4.00%, 8.00%, and 12.00%, with actual measurement values of 4.84%, 8.80%, and 12.59% in prepared ACF. The set levels of protein addition were 34.00%, 38.00%, and 42.00%, with measured values of 34.58%, 38.67%, and 42.34%. The set levels of carbohydrate addition were 18.00%, 22.00%, and 26.00%, with measured values of 19.10%, 23.52%, and 27.02%. The formula and nutritional components of the experimental ACFs are shown in Table 1. The specific preparation followed the conventional method, pressing into a particle size of 1 mm, cutting into 2–3 mm lengths, and curing in a 90 °C oven for 30 min. And then, it was naturally air-dried to a moisture content of about 10%, and stored at − 20 ℃ for use.

Table 1.

The measured nutritional compositions of ACF in treatments (% dry weight)

ACF types		Treatments
Lipid-optimized feed	G1	G2	G3
Dry matter	91.10	90.91	90.67
Crude protein	38.84	39.11	38.54
Crude lipid	4.84	8.80	12.59
Ash content	11.92	12.01	12.02
Protein-optimized feed	G4	G5	G6
Dry matter	90.34	90.56	90.75
Crude protein	34.85	38.42	42.71
Crude lipid	8.16	7.98	8.34
Ash content	10.67	11.91	13.30
Sugar-optimized feed	G7	G8	G9
Dry matter	91.51	90.81	90.12
Crude protein	38.27	38.58	39.12
Crude lipid	8.33	7.92	8.63
Ash content	12.20	11.04	11.01
Sugar-measured value	19.10	23.52	27.02

The bold section indicates the feed components utilized in the experiment

Experimental animal

The juvenile seahorse (H. kuda) used in the experiment was cultured in the breeding base of Ningbo University. Healthy and energetic juveniles were selected for the experiment, with an average weight of 0.6 ± 0.03 g and an average body length of 4.9 ± 0.22 cm. Temporarily cultivated in a recirculating aquaculture system for 2 weeks to avoid interference from other environmental factors in the study. During the temporary culture period, fed frozen Mysis once a day.

Experimental design

The experiments were divided into three groups to investigate the effects of varying lipid, protein, and carbohydrate levels in artificial compound feed (ACF) on the seahorse’s circadian clock system, nutrient metabolism gene expression, and growth, respectively. Each experimental group was designed to specifically evaluate the impact of these nutritional components on the respective biological processes.

Experiment 1#: lipid optimization

A total of 120 seahorse juveniles were randomly selected and divided into three groups and raised in 3 × 1-00L circular plastic containers. Artificial lighting was employed with a photoperiod of L16:D8 and a light intensity of 1000 lux. The seawater was exchanged regularly every day, with a salinity of 25 ± 0.5 and a temperature of 25 ± 0.5 °C. The feeding time set was 2 h after turning on the light, fed once a day, and about 8 g ACF each time. Group 1 (G1) fed ACF with a lipid level of 4.84%; Group 2 (G2) was 8.80%; Group 3 (G3) was 12.59% (Table 1). Continuously raised for 4 weeks and ended the experiment.

Experiment 2#: protein optimization

Group 4 (G4) fed ACF with a crude protein level of 34.58%; Group 5 (G5) was 38.67%; Group 6 (G6) was 42.34% (Table 1). Other conditions were the same as Experiment 1#.

Experiment 3#: carbohydrate optimization

Group 7 (G7) fed ACF with a sugar level of 19.10%; Group 8 (G8) was 23.52%; Group 9 (G9) was 27.02% (Table 1). Others also were the same as Experiment 1#.

Sample collection

The juvenile seahorses were subjected to fasting for 24 h prior to sampling. All treatment groups were sampled at time points of ZT2, ZT6, ZT10, ZT14, ZT18, and ZT22. For each sampling, three juvenile individuals were randomly selected and deeply anesthetized; then, their brain and liver tissues were quickly collected. The tissues were immersed in RNA store, immediately put in liquid nitrogen, and then transferred to a − 80 ℃ freezer for long-term storage.

The efficient and effective extraction of RNA and removal of genomic DNA from biological samples is a critical preliminary step to ensure the purity and integrity of RNA samples for subsequent molecular biological analyses; the reaction mixture was prepared as follows: 2.0 µl of 5 × gDNA Eraser Buffer, 1.0 µul of gDNA eraser, and total RNA were combined in a 20 µl reverse transcription reaction system. For TB Green qPCR, up to 1 µg of Total RNA was used, while for probe-based qPCR analysis, up to 2 µg of Total RNA was utilized. The mixture was prepared in RNase-free ddH2O, adjusted to a final volume of 10 µl, and incubated on ice. The reaction was then carried out at 42 °C for 2 min. To ensure the accuracy of the reaction solution preparation, a Master Mix was prepared in a volume corresponding to the number of reactions plus an additional two reactions. The Master Mix was then aliquoted into individual reaction tubes, and RNA samples were added accordingly. reverse transcription of RNA. The reaction mixture was prepared by combining the following components in a single tube: 10 µl of the pre-prepared solution, 1.0 µl of PrimeScript RT Enzyme Mix I, 4.0 µl of RT Primer Mix, 4.0 µl of 5 × PrimeScript Buffer 2 (for real time), and 1.0 µl of RNase-free ddH2O; the reaction solution is prepared on ice. To ensure accurate preparation and minimize pipetting errors, a Master Mix was prepared in a volume corresponding to the number of reactions plus an additional two reactions. The Master Mix was then aliquoted into individual reaction tubes, with 10 µl dispensed into each tube. The reaction was performed at 37 °C for 15 min, and real-time fluorescence quantitative PCR (q-PCR) operations were referred to Pan et al. (2020).

Verification experiment

According to obtaining optimized levels of fat, protein, and carbohydrate, a new ACF was performed for the seahorse. A validation experiment was conducted compared with frozen Mysis. There were three replicates for each food, and the cultured period was also 4 weeks. Other conditions were the same as above. In this period, the weight and length of all seahorses (36 individuals) in each treatment were measured every week. The growth rate was calculated according to the following formula:

$${\mathit{SGR}}_W=\left[\left(L,n,W_t,-,L,n,W_0\right),/,t\right]\times 100\%.$$

$${\mathit{SGR}}_L=\left[(,{\mathit{LnL}}_t,-,{\mathit{LnL}}_0,),/,t\right]\times 100\%.$$

$$CF=\left(W_t,/,{L_t}^3\right)\times 100\%.$$

$$FCR=\left(W_t,-,W_0\right)/F\times 100\%.$$

where W_t and L_t were the weight (g) and length (cm) at the end of the experiment, W₀ and L₀ were the initial weight (g) and length (cm), and t was time (d). F was the feed consumption during t days (g).

Data analysis

The Acro circadian rhythm analysis was used to analyze the original data to determine whether the expression of the target gene conformed to the cosine function. The peak expression time point (top phase time) of the target gene was fitted by cosine waves (Bo et al. 2022). In addition, statistical differences between different sampling time for each gene were determined through one-way ANOVA, and Tukey’s test was used to determine the significance of differences between sampling time points and the mean values of different sample sets. The Duncan test method was used to compare the significance of differences between treatments, with P < 0.05 being the level of significance. The experimental data was expressed as “mean ± SD.”

Results

The expressions of central clock genes in H. kuda

Seven biological clock genes (Clock, Bmal1, Per1, Per2, Per3, Cry1, and Cry2) were detected and analyzed during the experiments.

Lipid optimization

In G1, six (except for Cry2) of all analyzed seven clock genes exhibited their significant circadian rhythms in the brain of H. kuda (Table 2). The relative expression levels of various genes were significant differences, and their expression cycles can be divided two groups, Clock and Per3 were close to 1.0 cycle, and the other four genes were about 1.2 cycles within 24 h. Regarding their expression fluctuations, Clock and Per3 reached their top phases at 3 h after turning on the light (ZT7) and reached the lowest expression level during the alternation of light and dark phases (ZT18–ZT22). Bmal1 and Per2 reached their top phases at ZT9 and also reached their lowest levels during ZT18–ZT22. Per1 reached its peak at ZT8 and Cry1 at ZT10, and they both showed their lowest expression levels at the end of the light phase (ZT18).

Table 2.

Effects of different lipid level diets on the synchronization of circadian clock gene expression in the brain and liver of H. kuda

Group	Gene	Expression cycle (brain/liver)	Relative expression level (brain/liver)	The time of top phase (brain/liver)
G1	Clock	0.97 ± 0.05aA/1.04 ± 0.07aA	131.2 ± 8.68/261.3 ± 24.66	7:26 ± 0:35aA/9:16 ± 0:53bA
	Bmal1	1.18 ± 0.09aA/1.21 ± 0.11aA	355.1 ± 32.37/118.3 ± 17.63	9:10 ± 0:27aA/13:38 ± 1:06bA
	Per1	1.16 ± 0.13aAB/1.39 ± 0.09bA	710.5 ± 30.74/592.2 ± 27.46	8:15 ± 0:34aA/11:43 ± 0:47bA
	Per2	1.23 ± 0.17aA/0.94 ± 0.12bA	335.7 ± 15.28/247.6 ± 11.28	9:45 ± 0:37aA/13:58 ± 1:03bA
	Per3	1.02 ± 0.04aA/1.25 ± 0.12bA	535.7 ± 22.07/502.7 ± 21.69	6:48 ± 1:11aA/9:48 ± 0:46bA
	Cry1	1.17 ± 0.11aA/0.87 ± 0.07bA	345.3 ± 27.37/218.7 ± 10.56	10:05 ± 0:54aA/11:40 ± 0:51aA
G2	Clock	0.83 ± 0.05aB/0.96 ± 0.11aA	142.5 ± 12.76/327.1 ± 10.82	4:05 ± 0:56aB/5:17 ± 0:45aB
	Bmal1	1.08 ± 0.03aA/1.09 ± 0.05aA	143.1 ± 8.74/330.3 ± 17.79	8:59 ± 0:27aA/9:50 ± 0:41aB
	Per1	1.24 ± 0.12aA/1.04 ± 0.10aB	88.2 ± 7.67/241.7 ± 22.69	9:38 ± 0:39aB/8:30 ± 0:46aB
	Per2	0.92 ± 0.09aB/1.06 ± 0.07aA	412.9 ± 21.18/176.3 ± 23.47	8:15 ± 0:54aB/9:48 ± 1:01aB
	Per3	1.28 ± 0.13aB/1.06 ± 0.11aA	115.9 ± 5.56/316.4 ± 13.81	9:22 ± 0:49aB/7:44 ± 1:04bB
	Cry1	1.25 ± 0.09aA/0.95 ± 0.07bA	117.8 ± 7.68/270.1 ± 11.31	9:36 ± 0:51aA/8:24 ± 0:47aB
G3	Clock	0.95 ± 0.07AB/	277.2 ± 13.71/	8:15 ± 0:36A/
	Bmal1	0.86 ± 0.05B/	120.2 ± 6.87/	7:55 ± 0:52B/
	Per1	0.97 ± 0.09B/	145.3 ± 11.12/	8:36 ± 0:49AB/
	Per2	1.26 ± 0.12A/	188.9 ± 12.13/	11:23 ± 1:14C/
	Per3	1.14 ± 0.13AB/	312.3 ± 17.64/	9:07 ± 0:58AB/
	Cry1	1.14 ± 0.11A/	120.5 ± 14.76/	10:45 ± 1:06A/

Different lowercase letters indicated significant differences (p < 0.05) in the relative expression of the same clock gene between different tissues (brain and liver); different capital letters indicated significant differences (p < 0.05) in the relative expression of the same clock gene in the same tissue with different treatments

In G2, these six genes also showed rhythmic oscillations (Table 2), but their expression levels (except Clock and Per2) were lower than those in G1. And their expression cycles also can be roughly divided into two groups. Clock, Bmal1, and Per2 had about 1 cycle, and the other three genes had about 1.2 cycles within 24 h. Compared to G1, there had been changed in the cycles of three genes, included Clock, Per2, and Per3. Furthermore, the expression trend of Clock had a peak at ZT4 (just turning on the light). The other five genes reached their top phase at ZT9, with Bmal1 and Per1 reaching their lowest expression levels during ZT18–ZT22, Per2 showing its lowest level during the early dark phase (ZT20–ZT24), and Per3 and Cry1 having their lowest levels during the end of the light phase (ZT16–ZT18).

In G3, these six genes also exhibited rhythmic oscillations (Table 2). Their expression levels (except Clock) were significantly lower than those in G1, and the oscillation patterns of three genes (Bmal1, Per1, and Per2) markedly differed from those in G2. Clock reached its top phase at ZT8, later 4 h than it in G2. Bmal1 and Per1 also reached their peaks at ZT8, Per3 at ZT9, and Per1 and Cry1 at ZT11. Among them, Clock, Per1, and Per3 had their lowest levels during ZT18–ZT22, Bmal1 and Per2 showed their lowest levels during ZT20–ZT24, and Cry1 reached its lowest level during the dark phase (ZT0–ZT4).

Protein optimization

In G4, six genes of Clock, Bmal1, Per1, Per2, Per3, and Cry1 also exhibited significant circadian rhythms in the brain tissue of H. kuda (Table 3). Except for Per2 and Cry1, which had their expression cycles of 1.4 and 1.3, respectively, the other genes all had about 1.0 cycle within 24 h. Regarding their expression fluctuations, Clock reached its top phase at ZT10, Bmal1 and Per3 at ZT8, Per2 and Cry1 at ZT12, and Per1 at ZT14. Clock and Per3 reached their lowest expression levels during ZT18–ZT22, Bmal1 had at the end of the light phase (ZT16–ZT18), Per1 showed at the late dark phase (ZT2–ZT4), and Per2 and Cry1 had at the dark phase (ZT0–ZT4). In G5, these six genes also had rhythmic oscillations (Table 3). Their expression cycles of all genes were about 1.0 cycle within 24 h. And almost genes reached their top phases at ZT8 or ZT9, but there was a difference in their lowest expression levels, Clock, Per1, and Per2 reached during ZT18–ZT20, Bmal1 and Cry1 had at the end of the dark phase (ZT2–ZT4), and Per3 had at the middle dark phase (ZT22–ZT24). In G6, these six genes exhibited rhythmic oscillations too (Table 3). All genes (except Bmal1) had an expression cycle of 1.0 in 24 h. And all genes (except Per3) reached their top phases at ZT9. Among them, Clock, Per1, Per2, and Per3 had their lowest expression levels during ZT18–ZT20, Bmal1 had at the late light phase (ZT16–ZT18), and Cry1 had during ZT20–ZT22.

Table 3.

Effects of different protein level diets on the synchronization of circadian clock gene expression in brain and liver of H. kuda

Group	Gene	Expression cycle (brain/liver)	Relative expression level (brain/liver)	The time of top phase (brain/liver)
G4	Clock	1.11 ± 0.11aA/1.07 ± 0.07aA	118.1 ± 7.16/982.7 ± 41.62	9:51 ± 0:47aA/14:33 ± 1:13bA
	Bmal1	1.08 ± 0.05aA/0.93 ± 0.09aA	157.6 ± 9.34/322.1 ± 15.73	8:01 ± 0:21aA/13:38 ± 0:57bA
	Per1	0.95 ± 0.10aA/0.85 ± 0.08aA	142.6 ± 7.02/418.7 ± 22.74	13:58 ± 0:54aA/14:36 ± 1:09aA
	Per2	1.40 ± 0.14aA/0.75 ± 0.11bA	170.2 ± 7.63/220.3 ± 21.84	11:25 ± 0:38aA/12:12 ± 0:57aA
	Per3	0.97 ± 0.06aA/1.28 ± 0.09bA	123.2 ± 7.31/358.7 ± 23.66	8:01 ± 0:44aA/12:58 ± 1:11bA
	Cry1	1.29 ± 0.13aA/1.06 ± 0.04bA	159.1 ± 7.63/991.7 ± 38.68	12:20 ± 0:47aA/14:38 ± 0:29bA
G5	Clock	1.10 ± 0.04aA/1.13 ± 0.07aA	150.2 ± 17.33/160.4 ± 9.79	9:24 ± 0:33aA/9:04 ± 0:51aB
	Bmal1	1.02 ± 0.09aA/1.13 ± 0.10aB	173.7 ± 8.67/208.7 ± 7.83	7:18 ± 0:42aA/9:13 ± 0:38bB
	Per1	1.17 ± 0.09aB/1.10 ± 0.06aB	330.7 ± 14.37/347.6 ± 13.46	9:45 ± 0:41aB/8:27 ± 0:44aB
	Per2	1.09 ± 0.06aB/1.09 ± 0.08aB	317.8 ± 19.33/174.2 ± 12.87	8:47 ± 0:53aB/7:29 ± 1:02aB
	Per3	1.13 ± 0.12aB/0.83 ± 0.10bB	203.7 ± 10.63/211.7 ± 6.57	10:16 ± 0:28aB/10.22 ± 0:47aB
	Cry1	1.15 ± 0.09aAB/1.05 ± 0.09aA	163.3 ± 12.68/166.8 ± 18.42	8:27 ± 0:33aB/7:49 ± 0:42aB
G6	Clock	0.92 ± 0.08aA/1.04 ± 0.07aA	424.6 ± 19.47/127.6 ± 11.73	9:13 ± 0:39aA/9:48 ± 0:45aB
	Bmal1	1.24 ± 0.14aB/0.95 ± 0.08bA	93.3 ± 8.86/111.7 ± 10.87	9:10 ± 0:24aB/9:16 ± 0:34aB
	Per1	0.97 ± 0.09aA/1.08 ± 0.06aB	156.8 ± 8.67/255.6 ± 17.73	9:07 ± 0:51aB/10:08 ± 0:37aC
	Per2	0.91 ± 0.11aC/1.02 ± 0.09aB	156.7 ± 10.12/171.4 ± 10.52	9:04 ± 0:36aB/10:08 ± 0:34aC
	Per3	1.07 ± 0.07aAB/0.95 ± 0.10aB	139.6 ± 6.78/221.7 ± 13.22	10:19 ± 0:27aB/10:11 ± 0:53aB
	Cry1	1.06 ± 0.10aB/1.12 ± 0.11aA	288.7 ± 11.34/179.3 ± 11.76	9:27 ± 0:36aC/10:42 ± 0:29bC

Different lowercase letters indicated significant differences (p < 0.05) in the relative expression of the same clock gene between different tissues (brain and liver); different capital letters indicated significant differences (p < 0.05) in the relative expression of the same clock gene in the same tissue with different treatments

Carbohydrate optimization

In G7, these six genes also exhibited significant circadian rhythms in the brain tissue of H. kuda (Table 4). Except for Per1, whose expression cycle was 0.65 in 24 h, all genes had a cycle of about 1.1. Regarding their expression trends, Clock reached the top phase at ZT5 and had its lowest expression level at the later stage of the light phase (ZT14–ZT18). Per2 and Cry1 reached their top phases at ZT8, and Bmal1, Per1, and Per3 reached their top phases at ZT9. And, Bmal1 and Per1 had their lowest expressions at the early stage of the dark phase (ZT20–ZT22), Per2 and Per3 reached during the period of light and dark alternation (ZT18–ZT22), and Cry1 reached at the end of the dark phase (ZT2–ZT4).

Table 4.

Effects of different sugar level diets on the synchronization of circadian clock gene expression in brain and liver of H. kuda

Group	Gene	Expression cycle (brain/liver)	Relative expression level (brain/liver)	The time of top phase (brain/liver)
G7	Clock	1.20 ± 0.06aA/0.89 ± 0.05bA	426.3 ± 11.69/156.2 ± 10.74	5:08 ± 0:35aA/5:00 ± 0:28aA
	Bmal1	0.92 ± 0.05aA/0.96 ± 0.07aA	623.7 ± 26.37/82.7 ± 7.66	9:39 ± 0:27aA/7:41 ± 0:53bA
	Per1	0.65 ± 0.11aA/1.22 ± 0.09bA	591.7 ± 20.66/221.6 ± 11.86	9:01 ± 0:37aA/9:56 ± 0:51aA
	Per2	1.10 ± 0.07aA/1.03 ± 0.11aA	672.7 ± 33.83/184.5 ± 16.33	8:30 ± 1:02aA/9:36 ± 0:34aA
	Per3	1.11 ± 0.09aA/1.13 ± 0.07aA	268.7 ± 18.36/711.4 ± 23.36	9:33 ± 0:56aA/9:53 ± 0:39aA
	Cry1	1.10 ± 0.03aA/1.09 ± 0.09aA	417.4 ± 23.68/221.7 ± 13.42	8:07 ± 0:42aA/9:59 ± 0:46bA
G8	Clock	0.94 ± 0.05aB/0.89 ± 0.07aA	203.7 ± 18.64/247.3 ± 23.17	4:45 ± 0:27aA/4:31 ± 0:41aA
	Bmal1	0.98 ± 0.03aA/0.97 ± 0.02aA	218.7 ± 13.63/218.7 ± 17.47	8:18 ± 0:44aB/8:36 ± 0:32aA
	Per1	1.13 ± 0.07aB/1.15 ± 0.04aA	165.2 ± 8.84/138.6 ± 12.48	8:33 ± 0:38aA/8:36 ± 0:26aA
	Per2	1.05 ± 0.06aA/1.10 ± 0.08aA	453.7 ± 23.83/339.2 ± 12.72	8:15 ± 0:37aA/10:11 ± 1:12bA
	Per3	1.09 ± 0.03a/1.07 ± 0.09aA	532.3 ± 18.68/289.2 ± 17.28	7:55 ± 0:45aB/8:56 ± 0:53aA
	Cry1	1.25 ± 0.07aB/1.20 ± 0.05aA	430.4 ± 21.42/174.3 ± 15.17	9:30 ± 0:22aB/9:27 ± 0:34aA
G9	Clock	1.12 ± 0.10aAB/1.11 ± 0.07aB	233.1 ± 13.42/497.6 ± 23.78	10:08 ± 0:55aB/10:42 ± 0:42aB
	Bmal1	1.11 ± 0.09aB/0.99 ± 0.07aA	362.3 ± 11.84/231.7 ± 20.32	8:07 ± 0:33aB/8:27 ± 0:39aA
	Per1	0.99 ± 0.06aC/0.87 ± 0.08aB	287.8 ± 13.66/314.7 ± 11.67	10:05 ± 0:43aB/13:44 ± 1:14bB
	Per2	0.98 ± 0.04aA/0.77 ± 0.07bB	244.1 ± 11.37/321.1 ± 26.11	7:58 ± 1:07aA/13:12 ± 0:57bB
	Per3	1.08 ± 0.08a/1.16 ± 0.08aA	321.2 ± 19.76/356.3 ± 21.79	7:58 ± 0:38aB/8:47 ± 0:52aA
	Cry1	1.10 ± 0.11aA/0.80 ± 0.07bB	311.3 ± 14.66/218.3 ± 11.94	8:29 ± 0:46aA/14:15 ± 1:07bB

Different lowercase letters indicated significant differences (p < 0.05) in the relative expression of the same clock gene between different tissues (brain and liver); different capital letters indicated significant differences (p < 0.05) in the relative expression of the same clock gene in the same tissue with different treatments

In G8, these six genes also showed rhythmic oscillations (Table 4). Except for Cry1, all genes had a cycle of about 1.0 in 24 h. Almost genes reached their top phases at ZT8, except that Clock reached its peak at ZT5 and Cry1 reached the top at ZT9. In addition, Clock had its lowest expression during the late light phase (ZT14–ZT18); Per1, Per3, and Cry1 showed at the end of the light phase (ZT16–ZT18); Bmal1 and Per2 had during ZT18–ZT22.

In G9, these six genes showed rhythmic oscillations too (Table 4). Only two genes of Per1 and Per2 had an expression cycle of 1.0, while the others had 1.1 cycles in 24 h. Moreover, Bmal1, Per2, Per3, and Cry1 reached the top phase at ZT8, and the others had at ZT10. But, Clock and Per2 reached their lowest expressions during ZT18–ZT22; Bmal1, Per3, and Cry1 had during ZT16–ZT18; and Per1 had during ZT20–ZT22.

The expressions of peripheral clock genes in H. kuda

The expressions of seven biological clock genes (Clock, Bmal1, Per1, Per2, Per3, Cry1, and Cry2) were also detected and analyzed during the experiments.

Lipid optimization

There were six genes exhibited significant circadian rhythms in the liver of H. kuda in G1 treatment (Table 2). Among them, Per2 and Cry1 had their expression cycles of less than 1, and three genes (Bmal1, Per1, and Per3) were more than 1.2 in 24 h. And the peak expression time of all genes was delayed compared to their expressions in the brain. Clock and Per3 reached their top phases at ZT9, Per1 and Cry1 at ZT11, and Bmal1 and Per2 at ZT13. In addition, Clock reached its lowest expression level in the early stage of the dark phase (ZT20–ZT22). Per3, Bmal1, and Per2 had at the late stage of the light phase (ZT16–ZT18). And, Per1 and Cry1 had during the dark phase (ZT0–ZT4). In G2, these six genes also showed rhythmic oscillations (Table 2). The expression cycle of all genes was about 1.0 in 24 h. Only Cry1 had a different cycle between two tissues (p < 0.05). And the top phase time of almost genes (except Per3) was also similar between two tissues (p > 0.05). Clock reached its peak just at turning on the light and its lowest expression level during ZT16–ZT18. Bmal1, Per1, Per2, Per3, and Cry1 reached the top phase at ZT9 or ZT10. And Bmal1, Per1, and Per2 had their lowest levels during the alternation of light and dark phases (ZT20). Per3 showed its lowest level during ZT18–ZT20, while Cry1 reached its lowest level during ZT20–ZT22. In G3, none clock genes showed significant rhythmic oscillations (Table 2).

Compared the synchronicity of all clock genes expression between in the brain and in the liver of H. kuda among these three treatments. In terms of the expression cycle, only two genes in G1, five genes in G2, and none in G3 had similar cycles between the two tissues (p > 0.05). In terms of the timing of top phase, only one gene in G1 has a similar peak expression time (p > 0.05), and there were five genes in G2 with consistent peak expression between the two tissues. Therefore, comparing the expression of various clock genes, there was the best synchronization between central and peripheral clock systems in the G2 treatment with 8.80% crude fat.

Protein optimization

In G4, six genes of Clock, Bmal1, Per1, Per2, Per3, and Cry1 exhibited significant circadian rhythms in the liver (Table 3). Among them, the expression cycles of Per1 and Per2 were less than 1, and only Per3 was more than 1.2 in 24 h. All genes reached the top phase after the mid light phase (ZT12), with Clock, Bmal1, Per1, and Per3 showing the lowest expression levels during the late dark phase (ZT2-ZT4), while Per2 and Cry1 showed the lowest levels during ZT0–ZT2. In G5, these six genes also showed rhythmic oscillations (Table 3). Their expression cycles of all genes (except Per3 = 0.83) were 1.0 or 1.1. All genes reached the top phase at ZT8 or ZT9, with Clock, Bmal1, and Per1 reaching their lowest expression during the alternation of light and dark phases (ZT18–ZT22), Per2 and Cry1 reaching their lowest levels during ZT16–ZT20, and Per3 reaching its lowest level during ZT20–ZT24. In G6, these six genes also had rhythmic oscillations (Table 3). Their expression cycles were about 1.0 and their top phases were at ZT9 (Table 3); they all showed the lowest expression levels during ZT20–ZT24.

Compared their expression synchronicity, there were three clock genes in G4, one gene in G5, and one gene in G6, had a different expression cycle between the two tissues (p < 0.05), respectively. There were four, one, and one genes that were different in each treatment between two tissues (p < 0.05) in terms of the timing of top phase. Therefore, the expression of each gene in the liver of G5 and G6 was closer to that in the brain, which had a better synchronicity with the protein level above 38.4%.

Carbohydrate optimization

In G7, Clock, Bmal1, Per1, Per2, Per3, and Cry1 also exhibited significant circadian rhythms in the liver of H. kuda (Table 4). Among them, almost genes had an expression cycle of about 1 in 24 h, except that Per1 was 1.22. And the peak expression time of almost genes was at ZT9–ZT10, except that Clock and Bmal1 were at ZT5 and ZT7, respectively. In addition, Clock reached its lowest expression at the end of the light phase (ZT16–ZT18); Bmal1, Per1, and Per3 reached the lowest levels during the alternation between light and dark phases (ZT18–ZT22), while Per2 had at the early dark phase (ZT20–ZT22). In G8, these six genes also showed rhythmic oscillations (Table 4). Their expression cycles were similar to their respective cycles in the brain. The top phase of all genes was also similar, except that Per2 was delayed to ZT10. In addition, Clock and Per1 had the lowest expression level at ZT16–ZT18. Bmal1, Per3, and Cry1 reached their lowest levels during ZT18–ZT20, and Per2 had its lowest level at ZT20–ZT22. In G9, these six genes also showed rhythmic oscillations (Table 4). Except for Per2 and Cry1, whose expression cycles had been reduced, all other genes were similar in the brain. Moreover, the top phases of Per1, Per2, and Cry1 were delayed to the afternoon. Clock reached the top phase at ZT10–ZT11 and showed its lowest expression at ZT20–ZT24. Bmal1 reached the lowest at ZT18–ZT22. Per1, Per2, and Cry1 reached at the end of the dark phase (ZT2–ZT4), and Per3 had at ZT16–ZT18.

Compared the expression synchronicity, no gene was different in G8 between the two tissues in terms of the expression cycle. And there were two, one, and three genes that were different with the top phase time in G7, G8, and G9, respectively. Therefore, it was found that the expression of various genes in G8 treatment was well coordinated between two tissues, indicating good synchronicity with the carbohydrate level of 23.5%.

The expressions of lipid metabolism genes in H. kuda

The expression of nine genes related to lipid metabolism (Hmgcr, Mvk, Mvd, Lss, Fdps, Cetp, Scap, Srebp1, Srebp2) was also detected and analyzed in the liver during the experiments.

Lipid optimization

In G1, eight of all detected genes showed significant circadian rhythms (Table 5). According to the expression cycle, it can be roughly divided into three groups: Mvd, Lss, and Cetp were less than 1 cycle; Scap and Srebp1 were about 1; and Hmgcr, Mvk, and Fdps were bigger than 1 in 24 h. About the expression trends of all genes, Srebp1 and Scap reached their peaks at ZT9 and reached their lowest expression levels during ZT18–ZT22. Hmgcr, Mvk, Mvd, Lss, Cetp, and Fdps reached their peaks in the mid light phase (ZT12–ZT13), with Hmgcr, Mvd, Lss, and Cetp showing the lowest expressions at the end of light phase (ZT2–ZT4), while Mvk and Fdps were expressed at lower levels throughout the whole dark phase.

Table 5.

Effects of different lipid level diets on the expression rhythms of lipid metabolism-related genes in the liver tissue of H. kuda

Group	Gene	Expression cycle	Relative expression level	The time of top phase
G1	Hmgcr	1.26 ± 0.11A	287.3 ± 13.44A	13:00 ± 0:49A
	Mvk	1.32 ± 0.08A	281.7 ± 13.67A	12:17 ± 0:36A
	Mvd	0.85 ± 0.03A	315.2 ± 14.17A	14:18 ± 1:07A
	Lss	0.84 ± 0.06A	309.5 ± 23.84A	14:15 ± 0:54A
	Fdps	1.25 ± 0.09A	377.1 ± 20.67A	11:40 ± 0:38A
	Cetp	0.81 ± 0.05A	655.7 ± 47.42A	14:38 ± 1:11A
	Scap	1.10 ± 0.10A	360.4 ± 35.84A	9:10 ± 0:33A
	Srebp1	1.05 ± 0.07A	107.6 ± 8.68A	9:01 ± 0:28A
G2	Hmgcr	1.00 ± 0.04B	551.3 ± 27.66B	8:33 ± 0:31B
	Mvk	1.03 ± 0.06B	682.7 ± 33.82B	9:42 ± 0:42B
	Mvd	1.13 ± 0.12B	189.3 ± 13.34B	9:45 ± 0:39B
	Lss	1.06 ± 0.07B	316.7 ± 13.82A	9:27 ± 0:47B
	Fdps	1.22 ± 0.09A	227.8 ± 13.68B	12:18 ± 0:57A
	Cetp	0.94 ± 0.05B	324.7 ± 23.87B	13:09 ± 0:58A
	Scap	1.14 ± 0.03A	378.3 ± 19.37A	9:36 ± 0:44A
	Srebp1	1.17 ± 0.10A	124.1 ± 13.82A	9:45 ± 0:37A

Different capital letters indica ted significant differences in the expression of the same gene in different treatments (p < 0.05)

In G2, these eight genes also showed significant circadian rhythms (Table 5). The expression cycles of four genes (Mvd, Fdps, Scap, and Srebp1) were above 1.1; the other four genes were about 1 in 24 h. Among them, the oscillation amplitudes of Hmgcr and Mvk were significantly lower than those in G1 (p < 0.05), while Mvd and Lss were bigger (p < 0.05). Their cycles were closer to 1.0 in G2. Furthermore, Srebp1, Hmgcr, Scap, Mvk, Mvd, and Lss reached their peaks at ZT9, the peak time of the middle four genes was significantly earlier than in G1, and these six genes had their lowest expressions during ZT18–ZT22. Cetp and Fdps reached their peaks at ZT12–ZT13 and had their lowest expressions in the early stage of the dark phase.

In G3, these lipid-metabolism genes did not show significant oscillatory rhythmicity (Table 5). In addition, compared the relative expression levels of each gene in G1 and G2, Hmgcr and Mvk were significantly increased (p < 0.05), while Mvd, Fdps, and Cetp were significantly decreased (p < 0.05).

From Table 5, it can be seen that there were significant differences in the expression of each gene between G1 and G2 treatments, including the expression cycle, relative expression level, and appearance time of top phase. It can be clearly seen that the expression cycle of each gene in G2 was closer to 1, and the top phase time was earlier and closer to the feeding time.

Protein optimization

In G4, among these nine lipid-metabolism genes, only Srebp1 and Scap showed significant rhythmic expression (Table 6); their rhythm cycle was 1.15. And both genes reached their peaks in the mid light phase (ZT13). Srebp1 had its lowest expression during the dark phase (ZT0–ZT4), while Scap showed its lowest expression level in the late of the dark phase (ZT2–ZT4).

Table 6.

Effects of different protein level diets on the expression rhythms of lipid metabolism-related genes in the liver tissue of H. kuda

Group	Gene	Expression cycle	Relative expression level	The time of top phase
G4	Scap	1.15 ± 0.11A	305.2 ± 17.74A	12:37 ± 0:56A
	Srebp1	1.15 ± 0.07A	253.7 ± 13.49A	13:44 ± 0:46A
G5	Hmgcr	1.04 ± 0.05A	89.4 ± 6.55A	7:58 ± 0:32A
	Mvk	1.04 ± 0.09A	161.7 ± 10.74A	10:13 ± 0:52A
	Mvd	0.94 ± 0.04A	156.7 ± 9.68A	10:16 ± 0:28A
	Lss	1.21 ± 0.06A	142.5 ± 11.37A	8:53 ± 0:34A
	Fdps	1.08 ± 0.09A	187.9 ± 13.77A	12:46 ± 0:54A
	Cetp	0.63 ± 0.03A	194.2 ± 12.73A	13:29 ± 0:51A
	Scap	0.84 ± 0.10B	161.3 ± 17.34B	5:55 ± 1:04B
	Srebp1	0.86 ± 0.07B	177.1 ± 13.94B	5:29 ± 0:38B
G6	Hmgcr	0.98 ± 0.12A	145.2 ± 11.74B	9:24 ± 0:42B
	Mvk	0.83 ± 0.05B	231.4 ± 13.68B	8:07 ± 0:33B
	Mvd	1.09 ± 0.08B	229.4 ± 13.17B	10:13 ± 0:35A
	Lss	0.97 ± 0.10B	208.7 ± 10.69B	8:41 ± 0:37A
	Fdps	1.16 ± 0.07A	106.3 ± 8.77B	10:31 ± 0:39B
	Cetp	1.15 ± 0.11B	159.4 ± 11.32B	10:36 ± 0:44B
	Scap	1.01 ± 0.05A	244.7 ± 12.37C	9:04 ± 0:43C
	Srebp1	1.08 ± 0.06A	217.6 ± 19.31A	8:50 ± 0:55C

Different capital letters indicated significant differences in the expression of the same gene in different treatments (p < 0.05)

In G5, a total of eight genes showed significant circadian rhythm (Table 6). Among them, the expression cycle of Lss was bigger than 1.2; there were four genes with expression cycles close to 1 and three genes below 0.9 in 24 h. The relative expression levels of all genes were not high, and there was a significant difference in the timing of their peak phase appearance. Srebp1 and Scap reached the peak at ZT5, Hmgcr and Lss did at ZT8, Mvk and Mvd did at ZT10, and the rest of genes (Fdps and Cetp) did at ZT13. Furthermore, Srebp1 and Lss reached its lowest expression at the end of the light phase (ZT16–ZT18), Scap and Hmgcr had at the transition between bright and dark phases (ZT18–ZT22), Mvk and Mvd got at the beginning of the dark phase (ZT20–ZT24), while Cetp and Fdps showed their lowest expressions at the end of the dark phase (ZT2–ZT4).

In G6, these eight genes also showed significant circadian rhythms (Table 6), all expression cycles were close to 1.0 in 24 h, their expression levels were all high, and the timing of top phase was relatively concentrated at ZT9–ZT10. In addition, Srebp1 reached its lowest expression during ZT18–ZT20, while the other genes showed their lowest expression levels in the early stage of the dark phase (ZT20–ZT22).

According to significant differences in the expression of each gene among three treatments, it can be clearly seen that the results in G6 was the best.

Carbohydrate optimization

In G7, the above eight genes also showed significant circadian rhythm (Table 7). Among them, there were three genes with an expression cycle of around 0.8 and another three genes with an expression cycle of 1.2 in 24 h. Additionally, there were two genes (Fdps and Scap) with relatively low expression levels. About the top phase time, Srebp1 appeared at ZT6, Mvk and Mvd at ZT8, and the rest genes occurred after ZT11. Furthermore, Srebp1, Scap, Hmgcr, and Lss showed their lowest expression in the early stage of the dark phase (ZT20–ZT24), Mvk reached at ZT18–ZT22, and Mvd had during the late light phase (ZT16–ZT18). Fdps and Scap reached the lowest expression during the late dark phase (ZT2–ZT4).

Table 7.

Effects of different carbohydrate level diets on the expression rhythms of lipid metabolism-related genes in the liver tissue of H. kuda

Group	Gene	Expression cycle	Relative expression level	The time of top phase
G7	Hmgcr	0.86 ± 0.04A	302.1 ± 20.67A	10:45 ± 0:44A
	Mvk	1.08 ± 0.08A	450.1 ± 24.79A	8:18 ± 0:28A
	Mvd	1.10 ± 0.07A	533.7 ± 25.83A	8:04 ± 0:35A
	Lss	1.24 ± 0.11A	509.4 ± 21.71A	11:20 ± 1:11A
	Fdps	1.20 ± 0.09A	111.2 ± 9.37A	13:41 ± 1:02A
	Cetp	1.21 ± 0.13A	262.3 ± 16.76A	12:46 ± 0:57A
	Scap	0.89 ± 0.06A	84.6 ± 7.42A	10:05 ± 0:44A
	Srebp1	0.81 ± 0.03A	193.2 ± 13.85A	6:15 ± 0:33A
G8	Hmgcr	0.98 ± 0.08B	548.7 ± 26.77B	8:18 ± 0:34B
	Mvk	0.87 ± 0.05B	234.5 ± 20.17B	8:01 ± 0:37A
	Mvd	1.02 ± 0.09A	288.3 ± 21.76B	8:47 ± 0:45A
	Lss	1.13 ± 0.10A	617.4 ± 26.79B	8:59 ± 0:42B
	Fdps	1.16 ± 0.07A	243.7 ± 20.14B	10:36 ± 0:53B
	Cetp	1.03 ± 0.07A	274.5 ± 19.27A	8:47 ± 0:36B
	Scap	1.02 ± 0.06B	247.2 ± 13.82B	9:07 ± 0:55AB
	Srebp1	1.14 ± 0.05B	298.7 ± 12.33B	9:36 ± 0:29B
G9	Hmgcr	1.22 ± 0.13C	284.3 ± 11.94A	13:03 ± 0:56C
	Mvk	1.20 ± 0.09A	329.6 ± 17.75C	13:21 ± 0:52B
	Mvd	1.15 ± 0.07A	227.4 ± 12.12C	14:12 ± 1:07B
	Scap	1.09 ± 0.10B	297.6 ± 17.79B	8:47 ± 0:35B
	Srebp1	1.15 ± 0.09B	259.4 ± 11.88B	9:53 ± 0:40B

Different capital letters indicated significant differences in the expression of the same gene in different treatments (p < 0.05)

In G8, the same eight genes also showed obvious circadian rhythms (Table 7). The expression cycle of all genes (except Mvk) was close to 1 in 24 h and with relatively high expression levels. Their peak expression appeared in the morning, centered at around ZT9. Besides, Srebp1, Scap, Lss, and Cetp reached their lowest expression during ZT18–ZT22, while Hmgcr, Mvk, Mvd, and Fdps showed at ZT20–ZT24.

In G9, only five genes (Srebp1, Hmgcr, Scap, Mvk, and Mvd) showed significant rhythmic oscillations (Table 7). All of their expression cycles exceeded 1.1 in 24 h. Srebp1 and Scap reached their peaks at ZT9 and showed their lowest expression during ZT18–ZT22. Hmgcr, Mvk, and Mvd reached their peaks in the mid light phase (around ZT13), with Hmgcr showing the lowest expression in the early dark phase (ZT20–ZT24), while Mvk and Mvd showed the lowest expression during ZT18–ZT22.

According to significant differences in the expression of each gene among three treatments, it can be clearly seen that the results in G8 was the best.

Experimental verification results

After 4 weeks of feeding, there were significant differences in the body length and body weight of H. kuda between the two groups (Table 8). The ACF group showed significant advantages in body length specific growth rate (SGR_L), body weight specific growth rate (SGR_W), and feed conversion rate (FCR) compared to the frozen Mysis group. There was no difference in condition factor (CF) between the two groups. According to the analysis of variance, the SGR_L, SGR_W, and FCR of H. kuda were significantly affected by the feed (Table 8), while CF was not related to the feed.

Table 8.

Comparison of growth of H. kuda fed with ACF and iced Mysis

	ACF	Iced Mysis	F-value	P-value
Initial length/cm	6.33 ± 0.36	6.32 ± 0.34
Initial weight/g	1.21 ± 0.08	1.20 ± 0.06
Terminal length/cm	8.31 ± 0.34	7.72 ± 0.21
Terminal weight/g	3.14 ± 0.27	2.26 ± 0.31
SGRL/%	0.67 ± 0.06	0.33 ± 0.12	19.778	0.000
SGRW/%	1.65 ± 0.17	0.92 ± 0.22	19.986	0.000
FCR/%	1.04 ± 0.17	0.44 ± 0.14	31.434	0.000
CF/%	0.36 ± 0.06	0.37 ± 0.04	0.006	0.939

Discussions

Numerous studies have shown that the biological clock system played a crucial regulatory role in the nutritional metabolism of organisms (Honma et al. 2016). Among various environmental factors, food was a crucial factor affecting the peripheral circadian clock system (Wang et al. 2024). Therefore, through temporal meal alignment, the homeostasis of circadian clock systems can be maintained, enabling organisms to achieve optimal metabolic states (Wang et al. 2024). In addition to feeding time (Wang et al. 2024), the nutritional composition and their proportion in food also played an important role in affecting the peripheral circadian clock system (Honma et al. 2016). For example, a high-fat diet can cause changes in the clock system of mammals, thereby disrupting normal metabolism and ultimately leading to various metabolic diseases (Honma et al. 2016). Feed is crucial for aquaculture production and is also a major component of aquaculture costs. Therefore, researches on ACF were an important component of the industry. This study analyzed the effects of lipids, proteins, and carbohydrates on H. kuda through the expression of biological clocks and metabolic genes.

Different lipid contents had a significant impact on the expression of circadian clock genes and lipid metabolism genes in H. kuda. Among them, the impact on central clock genes was mainly reflected in the expression level and the timing of top phase, except for one gene (Clock) whose expression cycle was prolonged. The expression level was relative to the reference gene and varied for each treatment, making it difficult to compare among different groups. The following mainly discussed the changes in the top phase time and expression cycle about each gene. Except for the Clock gene, its expression peaks appeared earlier in the mid- and low-fat feed treatments, and the top phase time of all other clock genes appeared relatively later, indicating that different lipid contents did not have a significant impact on the expression of central clock genes, and further proving that light is the main factor affecting the central clock system (Wang et al. 2024). However, lipid content had a significant impact on the expression of peripheral clock genes. Low-fat feed not only delayed the expression time of Clock and Per3 genes to a later morning, but also caused Bmal1, Per1, Per2, and Cry1 genes to reach their maximum expression levels in a later afternoon. That was, compared to the central clock genes in the low-fat treatment, there was a significant delay in the expression time of peripheral clock genes. As for genes related to lipid metabolism, their expression trends were similar to those of peripheral clock genes. This indicated that there may be a close relationship between the expression of peripheral clock genes and metabolic genes and also suggested that when low-fat feed was fed, the peripheral clock system will detach from the central clock system, losing its original synchronization. The reason may be due to the low intake of lipid substances. In order to maintain normal functioning, the body’s lipid synthesis pathway will be relatively inhibited, and even promotes the breakdown of lipid substances in the liver to meet the energy needs of H. kuda. Therefore, the expression of key enzymes in the lipid synthesis pathway was affected, indirectly affecting the expression of clock genes closely related to it (Pan et al. 2020; Wang et al. 2024), which was called the entrainment effect of environmental factors (Wang et al. 2024). When fed with mid-fat feed, the expression of peripheral clock genes and central clock genes in H. kuda showed superior synchronization, and most lipid metabolism genes also reached their peaks quickly after feeding, indicating that the lipid levels in this feed were more suitable for the growth of the seahorse. However, when fed with high-fat diet, the circadian rhythm of peripheral clock genes in H. kuda was completely lost, and the lipid metabolism genes in the liver were also disrupted, indicating that high-fat feed disrupted the normal rhythmicity of lipid metabolism genes in the liver, leading to the loss of circadian rhythm of peripheral clock genes. This phenomenon had also been observed in mice. When mice continued to eat a high-fat diet, it could cause insulin secretion disorders and further reduce the circadian rhythm of clock genes (Honma et al. 2016). With the extension of feeding time, the peak expression of fat-metabolism related genes and their transcription factors will also decrease, and the rhythm of clock genes will be completely disrupted (Honma et al. 2016).

Different protein levels had also different effects on the circadian clock genes and metabolic genes in H. kuda. Under the feeding of low-protein feed, the clock genes of Clock, Bmal1, and Per3 in the seahorse reached their highest expression quickly after turning on the lights in the morning, while the other clock genes reached their peaks in the afternoon. When fed with mid- and high- protein feed, all central clock genes quickly reached the top phase after turning on the lights. This result suggested that the low-protein feed may affect the expression of circadian clock genes, and only when the protein content reached a certain level or above will it not affect the expression of clock genes. Proteins are composed of amino acids and are essential nutrients for the growth of all animals. When the protein intake of organisms was too low, the synthesis of clock genes and proteins closely related to the clock system may be affected to some extent (Glencross 2006). For example, in the liver tissue of low-protein treatment in this study, although the expression of peripheral clock genes still exhibited rhythmicity, there was a loss of synchronization with central clock genes (inconsistent expression cycles, changes in top phase time). In addition, the expression rhythmicity of some key enzyme genes in the lipid metabolism pathway was also disrupted. This indicated that low protein feed can affect the expression of lipid metabolism genes and peripheral clock genes. However, it can be seen that high protein feed did not have as significant an impact on the circadian clock system and metabolic genes as high-fat feed from this study, although there were reports that excessive protein intake by organisms can produce excessive peptides after digestion, which can block opioid receptors in the intestine, inhibit appetite, and disrupt the normal functioning of the circadian clock system and metabolism (Hemre et al. 2002b, a; He et al. 2011). However, seahorses are carnivorous animals with high activity of protein hydrolytic enzymes in their bodies, and they had a high demand for protein in feed (Huneau et al. 2019; Chuang 1990). Therefore, a slightly higher protein content in feed did not affect their metabolism and growth.

Different levels of sugar also have different effects on the circadian clock genes and metabolic genes in the seahorse. Feeding high-sugar content feed, the central clock genes of H. kuda reached their peak expression quickly after turning on the lights in the morning. However, only Clock, Bmal1, and Per3 in the peripheral clock genes reached their highest expression at the same time, while the other genes delayed reaching their peak until the afternoon. Only Srebp1 and Scap in lipid metabolism genes reached their peak at the same time, there were three genes of Hmgcr, Mvk and Mvd reached their peak in the afternoon, while Lss, Fdps, and Cetp lost their significant rhythmicity. This indicated that high-sugar feed had a certain impact on the expression of peripheral clock genes and lipid metabolism genes in the seahorse. When fed with low- or medium-sugar content feeds, the expression of biological clock genes and lipid metabolism genes showed significant rhythmicity, and there was good coupling and synergy between the central clock system and the peripheral clock system. This indicated that feed with low- to medium-sugar content was more beneficial for the growth and development of H. kuda. Generally, carnivorous fish have a much lower ability to utilize sugars than herbivorous fish, but appropriate sugar content also promotes the growth of carnivorous fish (Hemre et al. 1993; He et al. 2011). The metabolism of sugar in fish included fermentation, oxidation, energy supply, and conversion storage. Among them, conversion storage involved two pathways: synthesis of liver glycogen and synthesis of fat (Luo and Xie 2010; Li and Zhang 2015; Wilson 1994). Therefore, an increase in sugar content in feed promoted the synthesis of lipids in the body. Many fish studies had also confirmed it, as the whole fish fat content increased with increasing sugar levels when the feed sugar level for Siganus gutatus was between 5 and 25% (Li et al. 2012); when the feed sugar level of GIFT Oreochromis niloticus increased from 10 to 45%, the whole fish fat also showed a gradual increase trend (Jiang et al. 2013); the crude fat content of Amphiprion ocellaris also increased with the increase of feed sugar levels (Zhao et al. 2017). However, it has been reported that excessively high sugar levels can lead to a decrease in fat accumulation in fish. The total fat content of the allogynous silver carp (Carassius auratus var. gibelio) remained stable at sugar levels < 28% and decreased at sugar levels > 28% (Tan et al. 2009); when the sugar level was above 30%, both the whole fish and muscle fat content of Japanese yellow croaker (Nibea japonica) decreased (Tan et al. 2009). All of these indicated that excessive intake of sugars by fish can affect normal lipid synthesis. However, the impact of high sugar was not as significant as that of high-fat diet, as all peripheral clock genes and lipid metabolism genes in high-fat diet completely lost their expression rhythms.

Conclusion

Excessive dietary lipids (> 8.80%) and carbohydrates (> 23.52%), combined with insufficient protein levels (< 38.67–42.34%), significantly disrupt circadian rhythmicity in the seahorse Hippocampus kuda. High-lipid diets (> 8.80%) induce rhythmic desynchronization in peripheral clock genes (e.g., Bmal1, Per1), impairing synchronization between the central and peripheral clock systems and delaying or disrupting peak phases of lipid metabolism genes (e.g., Srebp1, Hmgcr). High-carbohydrate diets (> 23.52%) activate hepatic lipogenic pathways (e.g., upregulation of fatty acid synthase genes), promoting lipid accumulation while suppressing rhythmic expression of peripheral clock genes. Low-protein diets (< 38.67%) further inhibit β-lipoprotein synthesis, reduce cholesterol metabolic efficiency, and indirectly compromise clock system coordination. Experimental validation demonstrated that optimal dietary ratios (8.80% lipids, 38.67–42.34% protein, and 19.10–23.52% carbohydrates) maximize central-peripheral clock synergy, simultaneously enhancing circadian rhythmicity of lipid metabolism genes and growth performance in H. kuda. These findings provide critical parameters for precision nutrition in carnivorous fish aquaculture, highlighting that maintaining peripheral clock-metabolic gene coordination significantly improves feed conversion efficiency in seahorse cultivation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Xu provided guidance and suggestions, while Bo, Xie, and Lou conducted experiments and authored the paper.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.