The experiment was carried out in filed culture ponds in Chonghu Fish Farm, Hubei Province, China. Water temperature was between 22°C in July, 26°C in August–October. Despite of the Tm, the other water quality parameters of each culturing pond were monitored in the middle of each month, meanwhile, 2 days before sampling, the physical and chemical parameters of the water in temporary tanks were also detected and analyzed. C. idellus used in this study were the same batch of cultivated C. idellus fingerling with average initial weight at 35 g per tail. These fish were divided into two groups with three replicate ponds for each group. At the beginning of the experiment, about 3,000 tails of C. idellus, polycultured with 550 tails of Hypophthalmichthys molitrix (about 14 g per tail) and 350 tails of Aristichthys nobilis (about 25 g per tail), were assigned to each pond (the area is 22666.67 m²/pond). Fish in the grass feed group (GFG) were fed with 100 kg grass per pond, at 9 am per day, they include Lolium perenne, Euphrasia pectinata and Sorghum sudanense (crude protein 15.30%, crude fat 2.88%, moisture 78.40%, crude fiber 25.90%, ash 3.50%). The natural grass were planted along the culture ponds of both experimental groups, which could provide adequate food sources for the fish of GFG. Whereas, the fish in the artificial feed group (AFG) were fed with commercial feed (Haid, China), which contain crude protein 28.00%, crude fat 7.06%, moisture 8.75%, crude fiber 15.00%, and ash 15.63%. 15 kg artificial feed were supplied to each pond of AFG at 9 am and 15 pm per day. All the farmed C. idellus were fed to satiation during the whole rearing period. This experiment was carried out from 8th, July to 28th, October, 2016.

At the end of this experiment, 150 fish in each group were captured and then transferred to the laboratory in the College of Fisheries, Huazhong Agricultural University. Finally, 120 fish in each group were used for growth performances data measure and samples collection.

The Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and “Guidelines for Experimental Animals” of the Ministry of Science and Technology (Beijing, China) were compiled during the experimental period. Institutional Animal Care and Use Ethics Committee of Huazhong Agricultural University had approved our study. All efforts were made to minimize suffering of sampled fish species.

During the 4-month culture period, the rearing conditions and water quality parameters were monitored and analyzed timely. Water temperature (Tm) of the two experimental groups ponds were both between 22°C in July, 26°C in August–October. Despite of the Tm, the other water quality parameters of each culturing pond, such as pH, dissolved oxygen (DO), ammonia nitrogen (NH₃/NH4+-N), nitrate nitrogen (NO3--N), nitrite nitrogen (NO2--N), total nitrogen (TN), total phosphorus (TP) and total organic carbon (TOC), were also measured in the middle of each month. Meanwhile, 2 days before sampling, the physical and chemical parameters of the water in temporary tanks were also detected and analyzed.

Among these tested water quality parameters, Tm, pH, and DO values were directly detected by Multi-Parameter Water Quality Detector. The TOC values were tested and analyzed by nondispersive infrared absorption method (GB/T13193-1991). Otherwise, the TN and TP were detected according to alkaline potassium persulphate digestion-UV spectrophotometric method (GB/T11894-1989) and ammonium molybdate spectrophotometric method (GB/T 11893-1989), respectively. The values of NH₃/NH4+-N, NO3--N, and NO2--N were also measured and calculated by their corresponding national standard methods: GB/T7479-1987, GB/T 7480-1987, and GB 7493-1987.

Upon arrival, C. idellus of the two experimental groups were respectively stocked in temporary rearing tanks for 1 week. The feeding was stopped 2 days before sampling. The sampled fish was anesthetized in 100 mg·L⁻¹ tricaine methanesulfonate (MS-222, Sigma, St. Louis, Missouri, USA) before measuring the growth indexes and samples collection.

The body weight, body length and body height of total 120 tails fish per group were measured. After collecting blood samples, the visceral mass cut off from esophagus to cloacal orifice, were took out from abdominal cavity and weighed. Then, the liver tissues were totally separated from the visceral mass and weighed the fresh weight again. Afterwards, the specific growth rate (SGR) and condition factor (CF) were calculated based on the measured data.

Note: W1-Body Mass (g); W2-Initial Weight (g); T-Feeding days; L-Body Length (cm).

Serum samples were prepared according to the previous method (Shi et al., 2006). The lactate dehydrogenase (LD), alkaline phosphatase (ALP), total cholesterol (TCHO), high density cholesterol (HDLC), glucose (GLU), albumin (ALB), total protein (TP), and triglycerides (TGK) were tested by automatic biochemistry analyzer (Hitachi 7020, Hitachi High Technologies, Inc., Ibaraki, Japan). Test kits were purchased from Nanjing Jiancheng Biochemical Corporation (Nanjing Jiancheng Biochemical Corporation, Nanjing, China), and the procedures followed were in accordance with the instructions of the kit.

Muscle texture assay was performed according to the method described by Ma et al. (2014). Muscle above the lateral line was dissected from 24 fish for each culture model, and then was cut into 1 × 1 × 1 cm block for TPA test and 1 × 1 × 4 cm for shear force test. The Texture Analyser used is TA-XT Plus Micro TPA (Stable Micro Systems, Surrey, England) equipped with a flat bottomed cylindrical probe P/36R. The type of loading probe is Auto-5 g, and the samples were compressed for 2 times in TPA test. Test condition was recorded at a pre-test speed of 3 mm/s, a test speed of 2 mm/s and a post-test speed of 2 mm/s with the stay time at 5 s, the data collection rate was 200 pps. Filet heights and maximum force (g shear force) needed to compress filets 65% of their height. 65% was chosen together with rounded probe ends as to reduce damage of muscle structures (Hyldig and Nielsen, 2001). The samples of TPA and shear force tests were carried out at room temperature, with 10 fish in each culture model and five parallel samples were taken in TPA test. Three measurements were performed on each filet in shear force test (1 cm between each measure point) centrally on the flesh side and the thickest muscle layer. Texture curves were recorded and the max force was determined as an average of the three measurements.

Serial transverse 10 μm-thick sections were stained with hematoxylin & eosin (H & E staining) and intracytoplasmic lipids with oil red O (Oil O staining) according to the procedures (Rasmussen and Ostenfeld, 2000), respectively.

A total of 200–400 fibers of white muscle per fish were studied in a Leica MZ 6 microscope for their cross sectional area (CSA) and the diameter (d = 2r) of each fiber was calculated from the fiber area (A) [A = π·r²), thus, d = 2√ (A·π⁻¹)]. A size limit for identifying fibers was set at fiber diameters ≥10 μm since the optical resolution below this limit did not allow for sufficient identification and accuracy in the analyses (Luther et al., 1995). The circularity of each fiber (4π·fiber area·circumference⁻²) was also determined. A circularity of 1.0 indicated a complete circle. The free software Image J (http://rsb.info.nih.gov/ij/) was used for quantitative statistics and analyses.

The tissues were homogenized by grinding apparatus to extract total RNA using RNAiso reagent (TaKaRa, Shuzo, Japan) according to the manufacturer's instructions. The total RNA was then dissolved in 50 μL RNase-free water. The quality of total RNA was checked with the agarose gel and concentration were measured by NanoDrop 2000 (Thermo Scientific, Waltham, MA, USA). The PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa, Shuzo, Japan) was used to reverse-transcribe the first strand cDNA from the total RNA. The partial C. idellus cDNA sequences were obtained from National Center for Biotechnology Information (NCBI¹), which include MRFs (myogenin, MyoG; myogenic differentiation, MyoD; myofactar-5, Myf-5; myofactar-6, MRF4); myostatin type I and type II (MSTN-1/MSTN-2); insulin-like growth factor 1 (IGF-1); insulin-like growth factor 2 (IGF-2); collagen type I alpha-1 (COL1A-1) and collagen type I alpha-2 (COL1A-2). Special amplified primers of these genes cDNA were given in Table 3. PCR products were gel-purified, ligated into the T/A cloning vector pMD-19T (Takara, Dalian, China) and transformed into Escherichia coli DH5α competent cells. Positive clones were examined by PCR followed by direct sequencing.

Expression patterns of these genes (MyoG, MyoD, Myf-5, MRF4, MSTN-1, MSTN-2, IGF-1, IGF-2, COL1A-1, and COL1A-2) were analyzed using quantitative real-time PCR (qRT-PCR). The β-actin and 18s rRNA were selected as reference genes. All primer sequences were described in Table 1. The qRT-PCR assay was performed using SYBR Premix Ex Taq™ (TaKaRa, Dalian, China) on a Roche Light Cycler 480 machine (Roche, Sussex, UK). The qRT-PCR conditions were as follows: denaturation at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s, annealing at 58°C (MyoG)/60°C (MyoD)/60°C (Myf-5)/56°C (MRF4)/58°C (MSTN-1)/58°C (MSTN-2)/60°C (IGF-1)/58°C (IGF-2)/55°C (COL1A-1), and 55°C (COL1A-2) for 20 s, and elongation at 72°C for 15 s. The relative quantification of the target and reference genes was evaluated according to standard curves. Each experiment was conducted in triplicate. The relative stability measures (M) of the reference genes were calculated by GeNorm² (http://medgen.ugent.be/genorm/). Based on the results, β-actin was chosen as the reference gene in subsequent analyses.

The white muscle fiber circularity data were ln-transformed by software Image J. Fiber diameter was additionally fitted as a covariate to fiber circularity. For comparison of the distributions of muscle fiber sizes, the non-parametric Kolmogorov-Smirnov two sample test was applied. This test is very sensitive in detecting differences in dispersions and skewness compared to other statistical tests (Zar, 1996). One-way ANOVA was used for analysis of differences in the two frequencies of each fiber diameter category. Differences were considered significant at P < 0.05 and highly significant at P < 0.01.

For statistical analysis, data from qRT-PCR were all presented as the mean ± S.E. for each genes (MRFs, MSTN-1, MSTN-2, IGF-1, IGF-2, COL1A-1, and COL1A-2). The optimized comparative Ct (2^−ΔΔCT) value was used to compute the relative expression level (Livak and Schmittgen, 2001). One-way ANOVA and t-test were used to compare the significance of genes among different positional muscle samples and the expression of each gene in the same muscle position between the two culture models C. idellus. Meanwhile, Duncan's test was also applied to multiple comparisons by IBM SPSS Statistics 19.0 (SPSS, Chicago, IL, USA).

The other analyses of variance were performed either on means of triplicates (growth performances, blood biochemical indexes and texture indexes). The 0.05 probability level was used to denote data that were statistically significant, while 0.01 probability level was used to denote data that were highly significant.

During the 4-month breeding cycle, the results of water quality parameters were showed Figure 2. The results of pH were higher in AFG since September, in especial, significances were tested in October and November (P < 0.05). Most of the DO values were higher in the GFG with no significance (Figure 2A). Among the detections of various forms of nitrogen, most of the measured NH₃/NH4+-N and NO2--N values were higher in AGF, particularly, the detected concentrations of NO2--N in July and October were significantly higer in AFG (P < 0.05). In addition, as time goes on, the concentrations of NH₃/NH4+-N increased in both groups, however, the change trend of NO3--N concentrations were opposite. The other aspect, no significance was found in all the test results of TN, TP, and TOC measurement (Figure 2C). In general, no matter which test indicator, their variation tendency in GFG were consistent with the change trendency of the corresponding test indicator in AFG.

Table 2 shows the growth performance of the two experimental groups C. idellus fed with artificial feed and grass feed, respectively. The average initial weight of the two groups fish was about 35 g per tail. After 113 days of separate feeding, the body mass, body length, body height, visceral weight and liver weight, even SGR were all significantly higher in AFG (P < 0.01). Only the CF was significant higher in GFG (P < 0.05).

Table 3 shows the serum biochemical data of C. idellus in the two experimental groups. The comparison results indicate that different feeding diet had significant effect on LD, ALP, TCHO, HDLC, and TP concentrations (P < 0.05). The great significances were observed in the concentrations of LD between the two culture models (P < 0.01). No significant differences were observed in the concentrations of GLU, ALB, and TGK between GFG and AFG.

The data of texture measurements are represented in Table 4. Significant differences were detected in adhesiveness, springiness, cohesiveness, gumminess, chewiness, resilience, and shear force between the two groups (P < 0.05) except the hardness. Adhesiveness, springiness, cohesiveness, gumminess, and chewiness were significantly higher in AFG (P < 0.05), while resilience and shear force were significantly higher in GFG (P < 0.01).

The histological changes were recorded using diameters of white muscle fibers, interstitium between myofibers, and filet lipid distributions (Figure 3). The average fiber diameters of back, rib, and tail muscle in AFG were significantly lower than those in GFG (P < 0.01), however, the opposite result occurred in abdomen muscle (Figure 4A). Compared with fish in GFG, the interspaces between myofibers of the four tested muscle parts were all evidently larger in AFG fish (P < 0.01; Figure 4B). There was no significant difference in fat content of rib muscle tissue between two groups. For back and tail muscle tissues, the filet lipid contents was significantly higher in GFG (P < 0.01). The trend is reversed for abdomen muscle tissue (Figure 4C).

The gene expression levels of MRFs varied either in nine positional white muscle tissues and one red muscle tissue in each group or in the same part of muscle between different groups. The expression levels of MyoG in various positional muscle samples, except for in tail muscle tissue, were 1.2–7.6 times higher than those in AFG (P < 0.01). MyoD expression levels were higher in back, rib, red, tail, paired fins, opercular, and eyes muscles of C. idellus in GFG. In the other muscular tissues, higher expression levels were observed in AFG (P < 0.01).

The expression patterns of Mfy-5 and MRF4 differed from MyoG and MyoD. These gene expressions were all influenced by both diet and sampling positions (P < 0.05). Mfy-5 and MRF4 expressed the most in red muscle of the two experimental groups. And their expression levels were significantly higher in back, eyes and tail muscle tissues of AFG, while higher expressions occurred in muscle tissues of eyes, tail, opercula, odd fin, and paired fins of fish from GFG (P < 0.05) (Figure 5).

Gene expressions of MSTNs (MSTN-1 and MSTN-2) and IGFs (IGF-1 and IGF-2) were plotted in Figure 6. The expression level of MSTN-1 was higher in back, tail, paired fins, jaw, and eyes muscle samples of fish in AFG, and expressed lower in the other positional muscles (P < 0.01). MSTN-2 expressed much more in the most of sampled muscles in fish of GFG, except for lower expression level detected in the muscle of back, rib, and eyes (P < 0.01).

Gene expressions of IGF-1 and IGF-2 varied in the deferent sampling positions of muscle. IGF-1 expressed significantly higher in GFG (P < 0.01), with a few exceptions of expressions in back and eyes muscles. Gene expression level of IGF-2 was significantly higher in AFG (P < 0.01).

Collagen type I alpha were coded by two subtypes genes CoL1A-1 and CoL1A-2, which had the similar expression patterns in two groups (Figure 7). The both genes had the highest expressions in red muscle and higher expressions in eyes muscle, opercular muscle, and tail muscle. CoL1A-1 and CoL1A-2 expressed lower in the muscle of back, rib, abdomen and paired fins. There was significant deference in the expressions of two genes in each sampled muscle tissue between groups (P < 0.01).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer SL and handling Editor declared their shared affiliation.