Samples (2 mg) were mixed with 200 mg KBr, ground, and then pressed into pellets for FT-IR scanning. Spectra were collected in the frequency range of 4000–400 cm⁻¹.

The molecular weight distribution of YZ-0.5A was analyzed by high performance gel-permeation chromatography (HPGPC) using a Shimadzu LC-10A system linked to a Shimadzu RI-502 (Liu et al., 2018).

The monosaccharide composition of YZ-0.5A was analyzed by high performance anion exchange chromatography (HPAEC) with a CarboPac TM PA20 (3 × 150 mm) column (Dionex, CA, USA) (Liu et al., 2017). Fucose, rhamnose, arabinose, galactose, glucose, xylose, mannose, fructose, galacturonic acid, and glucuronic acid were used as standards (Catal et al., 2008; Song et al., 2019). Methylation analysis of YZ-0.5A was determined by gas chromatography-mass spectrometry using a GCMS-QP2010 system (Shimadzu, Japan) equipped with an RXI-5 SIL ms quartz capillary column (30 m × 0.25 μm × 0.25 mm) according to a method in (Guo et al., 2019).

The NMR chemical shifts of one-dimensional (¹H, ¹³C, and ¹³C-DEPT-135) and two-dimensional (¹H-¹H COSY, NOESY, ¹³C-¹H-HSQC, and ¹³C-¹H-HMBC) YZ-0.5A were measured using a Bruker 600 MHz spectrometer (Bruker Corp, Fallanden, Switzerland) and are presented in ppm.

The formulation and composition of the basal diets are listed in Table 1 . The crude protein, crude lipid, and ash content of this formula are 47.44 ± 0.34%, 14.95 ± 0.03%, and 12.49 ± 0.17%, respectively. Five semi-purified basal diets were supplemented with 0 (control, without extra YZ0.5-A addition), 150 (YZ-0.5-150), 300 (YZ-0.5-300), 600 (YZ-0.5-600), or 1200 (YZ-0.5-1200) mg/kg YZ0.5-A. All the ingredients were ground into powder, mixed according to the proportion, and made into 2-mm pellets with a pelletizer (F-26; South China University of Technology, Guangzhou, China). Then, all diets were air-dried at 40°C until the moisture level reached 4% and stored at −20°C.

The experimental procedures were similar to those described in our previous study (Zou et al., 2019). Three hundred juvenile groupers (13.46 ± 0.08 g) were randomly divided into five experimental groups, with three replicates in each group (15 floating cages). Fish were fed to apparent satiation two times daily at 8:00 and 16:00. The feeding trial lasted for 8 weeks, during which the number of dead fish, the weight of dead fish, and feed consumption were recorded daily. At the end of the feeding trial, fish were fasted for 24 h and euthanized with 100 mg/L MS222 to detect the weight gain (WG) and condition factor (K). Blood samples were collected by caudal venipuncture using 2 ml syringes to analyze biochemical indices. After blood withdrawal, the fish were dissected for the calculation of hepatosomatic index (HSI), and viscerosomatic index (VSI). Then, liver samples were removed and stored at −20°C for further analyses. All experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Zhongkai University of Agriculture and Engineering (ethics code permit number: ZK20180518).

The viability of hepatocytes treated with YZ0.5-A was evaluated using cell counting kit-8 assays (CCK-8; ABP Biosciences, Beltsville, MD, USA) according to the method used in our previous study (Zou et al., 2018).

Hematoxylin and eosin (H&E) staining on the liver tissue and hepatocytes of hybrid grouper was conducted according to previous methods (Zou et al., 2022). In vivo experiment, dissected liver specimens were fixed at 4% neutral buffered formalin and dehydrated in graded ethanol concentrations and impregnated with paraffin. Subsequently, the paraffin-soaked specimens were frozen and cut into 5-μm thickness sections in the cryostat. The sections were stained with hematoxylin and eosin (H&E) for optical examination. In vitro experiment, isolated hepatocytes were washed with phosphate buffered saline (PBS), fixed with 4% paraformaldehyde for 20 min, and rinsed in 0.1M PBS×4 for 2 min each time. Then staining with H&E was performed after the removal of PBS.A Light microscope (200×, MshOt MS60) was used to picture the morphological features of the processed samples.

Liver tissues were embedded in 4% buffered formaldehyde and processed into frozen thickness sections (9-μm) as described in 2.6.2. The sections were then stained with Oil Red O solution. The hepatocytes were fixed as described in 2.6.2, immersed in 60% isopropanol for 3-5s and rinsed in distilled water. The cells were stained for 8-10 min with filtered oil red O in dark, then exhaustively rinsed with water. The nuclei were counterstained with hematoxylin for 3-5 min. After one further rinse in water, intracellular lipid droplets were observed using microscope (MshOt MS60). The relative area of vacuolation and lipid droplets in stained samples was analyzed by Image-Pro Plus 6.0. stained samples.

The initial body weight, final body weight (FBW), WG, feed efficiency (FE), HSI, VSI, K and specific growth rate (SGR) were measured using the following formulae:

Whole body and muscle moisture were determined using the Kjeldahl method. Crude protein (N×6.25) was measured after acid digestion using a Kjeltec 8400 (FOSS, Hilleroed, Denmark). Crude lipid was determined using ether extraction by Soxtec™ 2055 (FOSS, Hoganos, Sweden). The samples were dried in the oven at 150°C for 24 h to measure moisture content. Ash in the whole body, muscle, and diet was measured with a muffle furnace (FO610C; Yamato Scientific Co., Ltd., Tokyo, Japan) at 550°C for 24 h.

Biochemical indicators of antioxidation, including CAT, SOD, GR, and GSH-Px activity were evaluated using detection kits (Jiancheng Institute of Biotechnology, Nanjing, China) in a spectrophotometer following the method previously described (Zou et al., 2018).

Apoptotic cells were evaluated by flow cytometric analysis after using Annexin H-APC/7AAD detection kit (Multi sciences, Shanghai, China) following the manufacturer’s instructions. Briefly, hepatocytes (1 × 10⁶ cells) were reseeded in 6-cm dishes and proceeded with Annexin V-APC/7-AAD double staining, fixed at 28 °C for

15 min, and subsequently stored in ice. Then, cells were analyzed by flow cytometry.

Total RNA was isolated from the liver samples using TRIzol reagent (Invitrogen, Waltham, MA, USA). mRNA quality was measured by spectrophotometry (A260:280 ratio) and electrophoresis using 1.2% agarose gels. Then, cDNA was synthesized using an Easy Script First-Strand cDNA Synthesis SuperMix (Transgen Biotech, Beijing, China). The specific primers for each gene are listed in Table 2 (Tan et al., 2018; Zou et al., 2021). The real-time PCR reactions were carried out in a final volume of 20 μl, using 1×Power SYBR Green PCR MasterMix buffer (ABI, USA) on a Step-one PCR amplifier (ABI, USA). Cycling parameters were as follows: 1 min at 94°C, followed by 40 cycles of 10 s at 94°C; 40 cycles of 20 s at 60°C; and 40 cycles of 30 s at 72°C. All assays were carried out in triplicate. The mRNA expression levels were normalized to the geometric mean of the reference genes (β-Actin) using the 2^−ΔΔCt method.

The NaCl-soluble polysaccharide YZ-0.5A was extracted from pomelo fruitlets and obtained after deproteinization. The isolated YZ-0.5 was first fractionated with a DEAE-Sepharose Fast Flow chromatography column and eluted with 0.5 M NaCl solution. After Sephadex G-200 gel chromatography, the curve showed that a YZ-0.5A fraction was obtained from YZ-0.5 ( Figure 1A ). YZ-0.5A is an acidic polysaccharide, as it was eluted with NaCl solution. When the NaCl concentration was increased to 0.5 M, the production of soluble sugar concentration increased to a maximum and then decreased with increasing NaCl.

The FT-IR spectrum of YZ0.5-A is shown in Figure 1B . The broad peak at approximately 3600–3200 cm⁻¹ was due to the stretch vibration of the -OH bond, which was the characteristic polysaccharide peak. The band at 2935 cm⁻¹ was the characteristic C-H absorption peak. The absorption peak at 1610 cm⁻¹ represented N-H stretching vibration (PoMohammadur Dounighi et al., 2015). The absorption bands at 1427, 1338, 1240, and 1016 cm⁻¹ represented C-O, C=O, and O-H vibrations (PoMohammadur Dounighi et al., 2015; Amutha et al., 2017). Further, the characteristic absorption band at 890 cm⁻¹ was assigned to the pyran structures, which might be caused by C-H stretching vibration. YZ0.5-A may be a β-terminus differential isomer, but the β-terminus configuration needed to be further determined by nuclear hydrogen and carbon spectra. According to our previous research, the crude polysaccharide extracted from pomelo fruitlets is a residue polymer containing acidic monosaccharides with pyran rings, which is consistent with this result.

According to the HPGPC analysis, there was only one symmetrical signal peak in the molecular weight profile of the YZ0.5-A fractions ( Figure 1C ), indicating high sample purity. The estimated average molecular weight of the YZ0.5-A fractions, based on calibration with the standard, was calculated to be 15,332 Da. Previous studies have suggested that biological activity is closely related to molecular weight (Sheng and Sun, 2014). If the molecular weight of the polysaccharide is too small or too large, the molecule usually has no biological activity (Huang et al., 2020). Thus, proper molecular weight distribution can denote good biological activity (Zhang et al., 2016).

The anion chromatography analysis of 12 standard monosaccharides is shown in Figure 1D and the chromatogram ( Figure 1E ) revealed five peaks indicating that YZ0.5-A was heterogeneous. Based on the retention time of each monosaccharide, the monosaccharide composition in YZ0.5-A included Rha, Ara, Gal, Glu, and GalA in a molar ratio of 0.088: 0.157: 0.331: 0.123: 0.301. As the previous study regarding the structure-activity relationship of polysaccharide reported, polysaccharides with an MW range of 4000–100,000 Da and a high proportion of GalA and Ara could show appreciable bioactive, like antioxidant activity (Li et al., 2016). Hence, YZ0.5-A with 15,332 Da and containing nearly 45% of GalA and Ara could be postulated to have bioactive potential.

Methylated YZ0.5-A was analyzed by gas chromatography-mass spectrometry (GC-MS) to elucidate the types of glycosyl linkages. As summarized in Table 3 , GC-MS demonstrated that YZ0.5-A has 11 main types of sugar linkages: terminal Araf, 1,5-linked Araf, terminal Glcp, terminal Galp, 1,2,4-linked Rhap, 1,2-linked Galp, 1,4-linked Glcp, 1,4-linked Galp, 1,6-linked Glap, 1,4,6-linked Glcp, and 1,3,6-linked Galp, with molar percentage ratios of 3.43:22.4:2.63:4.22:8.96:6.8:14.42:17.68:7.65:6.16:5.65, respectively. According to the results in the monosaccharide composition identified by HPAEC analysis, the GC-MS results demonstrated that the major backbone of YZ0.5-A may contain 1,5-linked Araf (22.40%), 1,4-linked Glcp (14.42%), and 1,4-linked Galp (17.68%).

The 1D NMR spectrum of YZ0.5-A, including ¹H, ¹³C, and Dept135, is shown in Figures 2A–C . The proton spectrum signals between δ3.0–5.5 ppm and δ3.2–4.0 ppm were the sugar ring proton signals (Li et al., 2017), and the main terminal matrix proton peaks, namely, δ4.40, 4.42, 4.57, 4.58, 4.62, 4.89, 4.97, 5.02, 5.07, 5.09, 5.22, 5.28, and 5.34 ppm, were concentrated in the δ4.3–5.5-ppm area. The main anomeric carbon signal peaks of δ108.77, 108.31, 105.52, 105.28, 105.0, 104.58, 104.35, 101.41, 101.14, 101.13, 100.04, 99.7, and 98.44 ppm appeared at the ¹³C NMR spectrum (201 MHz, D2O), which were distributed in the δ93–105-ppm area. According to the HSQC spectrum ( Figure 2D ), the aforementioned signals could be classified as δH/C: 5.09/108.31, 5.02/108.77, 4.62/105.28, 4.42/104.35, 4.58/105.52, 4.57/105.00, 4.40/104.58, 5.22/99.70, 4.97/98.44, 5.07/100.04, 5.34/101.14, 4.89/101.41, and 5.28/101.13 ppm labeled as A–M, respectively. The Dept135 spectrum of YZ0.5-A showed inverted peaks at δ70.6, 70.13, 67.81, 66.77, 62.3, 62.22, 62.22, 61.99, 61.99, and 61.95 ppm, indicating the C6 chemical shift.

The anomeric hydrogen signal of residue D was designated as δ4.42 ppm, and its signals of H2-H6a could be obtained from the COSY spectrum ( Figure 2E ), which are δ4.42, 3.57, 3.68, 4.05, 3.87, and 3.96. Meanwhile, as a reference to the HSQC spectrum ( Figure 2D ), the corresponding carbon signals of H2-H6a could be inferred to be: 72.23, 81.54, 70.60, 75.25 and 70.60. Therefore, based on these results and the literature, residue D was considered as →3,6)-Galp-(1→(Hu et al., 2015). In this way, the hydrogen and carbon signals of all other residues could also be obtained and inferred to the possible glycosidic bond. Hence, all glycosidic bonds were determined and assigned, as shown in Table 4 .

Some legible cross-peaks between the ¹H and ¹³C signals of different sugar residues could be observed in the HMBC spectrum ( Figure 2F ), which are: h H1 and j C2, j H1 and j H4, e C1 and e H4, e C1 and h H4, b C1 and l H6, f C1 and d H3 and d C1 and h H4. These cross-peaks corroborated the presence of →2,4)-α-L-Rhap-(1→4)-α-D-GalAp-(1→, →4)-α-D-GalAp-(1→4)-α-D-GalAp-(1→, →4)-α-D-Glcp-(1→4)-α-D-Glcp-(1→4,6)-α-D-Glcp-(1→4)-β-D-Galp-(1→4)-β-D-Galp-(1→2,4)-α-L-Rhap-(1→, →5)-α-L-Araf-(1→3,6)-α-D-Glcp-(1→, →2)-β-D-Galp-(1→3,6)-β-D-Galp-(1→, →3,6)-β-D-Galp-(1→2,4)-α-L-Rhap-(1→. While in the NOESY spectrum ( Figure 2G ), k H1 and k H4, k H1 and l H4, l H1 and e H4, b H1 and b H5 and g H1 to d H6 were observed, indicating the presence of →4)-α-D-Glcp-(1→4)-α-D-Glcp-(1→, →4)-α-D-Glcp-(1→4,6)-α-D-Glcp-(1→, →4,6)-α-D-Glcp-(1→4)-β-D-Galp-(1→, →5)-α-L-Araf-(1→5)-α-L-Araf-(1→ and →6)-β-D-Galp-(1→3,6)-β-D-Galp-(1→.

Based on the monosaccharide composition and methylation and NMR spectroscopy analyses, we inferred that the main chain of YZ0.5-A is mainly composed of →2,4)-α-L-Rhap-(1→4)-α-D-GalAp-(1→4)-α-D-GalAp-(1→ with three branches substituted at O-4 of →2,4)-α-L-Rhap-(1→. The putative structure of YZ0.5-A is outlined in Figure 2H .

The effect of dietary YZ0.5-A on the composition of the whole body and muscles of hybrid groupers is shown in Table 5 . Dietary YZ0.5-A significantly decreased the crude lipid content throughout the whole body (P< 0.05). Dietary YZ0.5-A had a significant effect on crude protein and crude lipid levels in the muscles of hybrid groupers (P< 0.05). The addition of 300–600 mg/kg YZ0.5-A to feed significantly reduced the overall and muscle lipid content. These results indicate that supplementing YZ0.5-A in feed reduces lipid deposition in fish.

Previous studies showed that the appropriate lipid content of the juvenile hybrid grouper feed is 7%–13% (Jiang et al., 2016). When the fat content in the diet of hybrid grouper reaches 15%, there is significant fat deposition in the liver or visceral fat tissue in farmed fish (Jia et al., 2017; Zhang et al., 2019a). The effects of YZ0.5-A on the growth performance, feed utilization, and morphological parameters of hybrid groupers fed by a high-fat diet (HFD) are shown in Table 6 . The WG of hybrid groupers in the YZ-0.5-600 groups was significantly higher than those in the YZ-0.5-0 group (P< 0.05). Except for that in the YZ-0.5-1200 group, the FE of the YZ-0.5-0 group was lower than that of the other experimental groups (P< 0.05). HSI, VSI, and K were all decreased, albeit with varying degrees, in the YZ0.5-A treated groups (particularly in YZ-0.5-600 group) compared with the control group (P< 0.05). The decreases of VSI and HSI might be due to the loss of lipid in the liver, which was confirmed by the performance of the Oil Red O-stained liver sections ( Figure 3B ). The lipid droplet clusters in the YZ-0.5-300, -600, and -1200 groups was reduced relative to the YZ-0.5-0 group. Collectively, these results suggest that YZ0.5-A has a positive effect on growth performance and prevents lipid accumulation and liver enlargement in fish.

We further assessed the lipid-lowering effect of YZ0.5-A in HFD-fed hybrid grouper using RT-qPCR. As shown in Figure 3C , the expression of diacylglycerol O-acyltransferase 2 (DGAT2), glucose-6-phosphate dehydrogenase (G6PD), malic enzyme 1 (ME1), and fatty acid synthase (FAS), genes related to lipid synthesis (Zou et al., 2019), exhibited no difference in the YZ-0.5-150 to -300 groups but significantly declined in the YZ-0.5-600 group compared with that in the control (P< 0.05). Contrarily, the expression of the above factors was reversely elevated in the fish supplemented with 1200 mg/kg YZ0.5-A (P< 0.05), inferring that a specific concentration range of YZ0.5-A could inhibit lipid generation, while a high dose treatment may result in an adverse effect. In contrast, the mRNA expression of the lipolysis-related genes: lipoprotein esterase, Aconitase 1 (ACO1), peroxisome proliferator-activated receptor alpha (PPARα), and carnitine palmitoyltransferase 1 (CPT1) (Tan et al., 2019) was also investigated, and the expression was mostly enhanced after the addition of YZ0.5-A (P< 0.05). These results imply that treatment with YZ0.5-A improves lipid utilization in fish under HFD feeding.

To determine the optimal YZ0.5-A treatment time and concentration for hepatocytes and to investigate the YZ0.5-A concentration that mitigates hepatocyte viability after treatment with 2 mL/L 20% LE, we incubated hepatocytes with different YZ0.5-A concentrations for 24, 48, and 72 h. As shown in Figure 4A , all YZ0.5-A concentrations (except 150 and 300 μg/mL) decreased hepatocyte survival after 24 h. However, with extended culture time, the effect was changed. Cell viability was stable in the YZ0.5-A groups by 48h. While the obvious reduction of viability was observed in cells treated YZ0.5-A with concentration beyond 300 μg/mL compared with that in the control group by 72 h (P<0.05). Thus, we chose these three concentrations as the low-, medium-, and high-dose groups, respectively, for 24-h cell incubation. Compared with that in the control group, the results showed that hepatocyte viability of the model group significantly decreased (P< 0.05), while hepatocyte viability in the recovery group was significantly increased compared to that in the model group (P< 0.05). Viability was also remarkably decreased compared to that in the YZ-0.5-150 group (P< 0.05). Compared with that in the model group, the survival rate of the recovery group was increased, but compared to that in the YZ0.5-150 group, the survival rate was significantly decreased (P< 0.05). Therefore, 75, 150, and 300 μg/mL YZ0.5-A was used for subsequent tests.

H&E staining was used to analyze the hepatocyte structure ( Figure 4C ). In the control group, the hepatocyte nuclei were normal and neatly arranged. In the model group, the cell structure was destroyed, and vacuoles formed. In the recovery group, the cell state recovered slightly, and the structural damage improved. YZ0.5-A addition reduced the amount of hepatic cytoplasmic vacuolation, and the disrupted cell morphology was alleviated. Further, YZ0.5-A reduced the number of lipid droplets in hepatocytes. The Oil Red O-stained area in the YZ0.5-150 group were smaller than that in other groups ( Figure 4D ). Moreover, the morphological changes shown with H&E and Oil Red O staining were similar to those in the in vivo experiments ( Figures 3A, B ), giving further support to the therapeutic function of YZ0.5-A in morphological lesions inflicted by excessive lipid administration. Additionally, these results were in agreement with our previous hybrid grouper study that investigated another plant extract, Radix Bupleuri (Zou et al., 2019).

Lipids play a key role in growth and development. In fish, excessive lipid deposition in the body, especially in the liver, reduces growth rate and causes serious health problems, which poses a serious threat to the sustainable development of aquaculture (Yu et al., 2021). In the present study, as exhibited in Figure 4F , YZ0.5-A treatment significantly offset the LE-induced increase in the mRNA expression of sterol regulatory element-binding protein-1c (SREBP-1c), a transcriptional factor that is conducive to synthesizing lipids, and its functional downstream target, FAS (Ma et al., 2020). Moreover, the transcriptional levels of other genes related to lipid synthesis (G6PD and ME1) were also markedly suppressed in the YZ0.5-A treatment groups relative to the model and recovery group (P<0.05). PPARα, an important transcriptional factor which could reportedly upregulate the gene expression of the lipid oxidation-related enzyme CPT1, thus promoting lipolysis (Tao et al., 2019). Compared with those in the non-YZ0.5-A treated group, the expressions of PPARα and CPT1 were markedly enhanced in the YZ0.5-A treated group (P< 0.05). These findings indicated that YZ0.5-A shows its therapeutic effect against hepatic steatosis by intensifying fatty acid synthesis while promoting fatty acid utilization. The above performance with H&E and oil red O staining morphologically confirmed this conclusion.

AMPK, a heterotrimer comprised of an α-, β-, and γ-subunits, is a well-documented target of many therapeutic regimens for fatty liver disease (Smith et al., 2016). Specifically, AMPK activation plays a critical role in regulating lipid metabolism by suppressing SERBP-1c (Sharma et al., 2020). It has been also reported that activation of AMPK could trigger PPARα and thus improve fatty acid oxidation (Cheng et al., 2015). In this study, the LE-induced reductions in the mRNA expression of AMPK-α, -β, and -γ in the model group were significantly replenished in the YZ0.5-A-treated groups, particularly at 75 and 150 µg/mL ( Figure 4F ; P< 0.05). Considering the gene expression changes noted in the previous paragraph, these results suggest that YZ0.5-A may influence the activity of AMPK and secondarily its downstream target lipid metabolism-related genes, hence mitigating LE-induced hepatic steatosis.

Recent research has shown that lipid accumulation propels the onset of oxidative stress and thus impairs the antioxidant system, which is mainly manifested as a suppressed indicator of antioxidative enzymes(Hong and Lee, 2009). Accordingly, our experiment exhibited a similar effect, as LE challenges cause a reduction in the enzymatic activities of CAT, SOD, GR, and GSH-Px ( Figure 4B ; P< 0.05). Furthermore, apoptosis is regarded as a mechanism involved in the pathophysiologic progress of liver injury induced by excessive lipid deposition, which has occurred in various hepatocyte models, including mice (Xiao et al., 2013), chicken (Ma et al., 2020) and fish (Zhang et al., 2019b). In line with previous investigations, an obvious elevation in the apoptotic rate detected by flow cytometry could be observed in the model and recovery groups relative to that in the control ( Figure 4E ; P< 0.05). Noticeably, the aforesaid pathological performance was significantly reserved in the subsequent experiment with the supplementation of YZ0.5-A (especially at 150 μg/mL), indicating its protective effect on the onset of oxidative stress and apoptosis induced by irregular lipid metabolism.