The Immunostimulant Effects of Isochrysis galbana Supplemented Diet on the Spleen of Red Hybrid Tilapia ( Oreochromis spp.) Evaluated by Nuclear Magnetic Resonance Metabolomics

Signi ﬁ cant measures at improving aquaculture health have to be taken since disease outbreaks commonly occur in the fast-growing aquaculture industry. This study aimed to determine the immunomodulatory properties of an indigenous Malaysian microalgae species, Isochrysis galbana (IG), on red hybrid tilapia ( Oreochromis spp.). Then, the potential biomarkers for immunomodulatory properties were identi ﬁ ed through metabolic changes in the spleen using the proton nuclear magnetic resonance ( 1 H NMR) metabolomics approach. IG was cultivated in an indoor annular photobioreactor before being harvested and freeze-dried into dried biomass. The dried IG biomass was incorporated into four experimental diets in concentrations of 0% (diet A), 0.6% (diet B), 1.3% (diet C), 2.5% (diet D), and 5.0% (diet E) basal feed per kg. After 14 days of feeding, diet D has consistently improved the immune responses of innate immunity such as phagocytosis ( P > 0 : 05 ), respiratory burst ( P < 0 : 05 ), and lymphoproliferation ( P < 0 : 05 ) activities which might have helped in maintaining the overall health status of red hybrid tilapia. The orthogonal partial least squares (OPLS) model between diets D and A indicated three important metabolites, namely, isoleucine, glutamate, and tyrosine, have signi ﬁ cantly been upregulated when compared to the control. Meanwhile, IG supplementation also resulted in lower concentrations of various metabolites in the spleen including α / β glucose, choline, hypoxanthine, adenosine, and inosine. Hence, the application of metabolomics tools has proved the potential of IG-incorporated ﬁ sh diet by strongly in ﬂ uencing the metabolic condition in red hybrid tilapia spleen cells and further boosting the immune response and health status in improving aquaculture health.


Introduction
Nowadays, aquaculture production provides about 50% of the global fish supply while the other half is provided by fisheries [1]. In Malaysia, tilapia (Oreochromis spp.) is the second highest harvested freshwater fish, with a total annual production of 33,404 tonnes. The estimated wholesale value of this species was RM 346 million [2], which indicates the importance of tilapia farming in Malaysia, However, world aquaculture production is vulnerable, and an increase of disease outbreaks has been reported due to culture intensification, resulting in partial or total loss of production [3]. Hence, the fore of research is using immunostimulants as a suitable alternative to strengthen both the specific and nonspecific fish immune systems in aquaculture [4,5]. However, the recommended dose of each immunostimulant differs based on the plant, fish species, and route of administration [6].
Microalgae are categorized as eukaryotic organisms with unicellular photosynthetic ability and exist in a different range of morphologies and characteristics. Supplementation of several microalgae such as Chlorella vulgaris and Spirulina platensis in fish diet has been proven in stimulating the fish immune system [7][8][9]. Isochrysis galbana (IG) is a marine microalgae species classed under Isochrysidales order in the same group of Haptophyta division (Phylum). The advantage of IG is due to its less cell wall, rapid growth, and high endurance to high temperature and outdoor environment [10][11][12], which makes it possible to be cultivated under unpredictable tropical conditions [13]. IG is one of the high-quality microalgae used in the food industry due to its high lipid content (20-30%) mainly from PUFA, namely, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) [14]. It also contains a high total carotenoid mainly contributed by fucoxanthin [15,16]. These traits of IG make it one of the most highly demanded microalgae for aquaculture.
Various immune functions at the metabolic stage are bioenergy costly, demanding accurate control of cellular metabolic pathways to protect against contaminants, and sustain tissue homeostasis for the organism's lifespan. Since immune cells lack significant nutritional sources, these effector responses can only be preserved if immune cells can significantly improve their microenvironment absorption of glucose, amino acids, and fatty acids [17]. Previous studies have reported several microalgae with immune responses and disease resistance properties, but none of them reported the potential of IG as natural immunostimulants. In addition, research on the immunomodulatory impact of microalgae on fish at the metabolic stage has yet been published. Metabolomics provides a systematic understanding of metabolic changes and interference of biological mechanisms related to environmental influences on fish health and welfare and fish feeding [18,19]. NMR is the most frequently applied in metabolomics because of its high reproducibility, non-destructive, broad range of detection in a single type of analysis, and gives detailed structural knowledge for identification [20]. The current research is the first work in which a metabolomic strategy has been utilized to evaluate the metabolite level of the spleen cell suspension in red hybrid tilapia fed with IG-incorporated diet.

Microalgae Growth Production. Isochrysis galbana
(UPMC-A0009) used in this study was procured from Microalgal Production Laboratory, Aquatic Animal Health Unit, Faculty of Veterinary Medicine, Universiti Putra Malaysia in April 2017. The pure culture was sustained at 23°C in an environmental chamber (Sanyo, Osaka, Japan). The growth of IG was initiated on a 3 mL aliquot of the pure culture with 20 mL Conway medium prepared in filtered (5 μm) and sterilized seawater [21]. It was eventually scaled up to 90 L with constant aeration in a 120 L annular photobioreactor, receiving artificial light-dark cycles  at intensity 150 mol/m2/s. The growth of the microalgae was monitored using the biomass weight and cell count. The biomass weight was obtained by dividing the dry weight of the filtered biomass by the filtrate volume [22]. Daily cell counts were done using a Neubauer hemocytometer (Assistant, Germany) on a well-mixed sample. The microalgae were harvested during their late exponential growth period of 15 days by centrifuging them at 8000 rpm, as described by Aguilera-Sáez et al. [23], with some adjustments. Then, the biomass collected was freeze-dried (Scanvac, Denmark) and deposited until used at -80°C.

Diets Preparation.
Star Feed TP-1 (Star Feedmills (M), Sdn Bhd, Selangor, Malaysia) commercial diet was used as a basal feed with an estimated 32% crude protein content, 3% crude fat, 14.6% ash, 12% moisture, and 3% fibre. Powder of harvested IG was combined with a commercial basal diet and some distilled water to obtain four homogenized concentrations: 0.6% (diet B), 1.3% (diet C), 2.5% (diet D), and 5.0% (diet E), of dry biomass microalgae per kg basal feed, and control (diet A) without microalgae. The formulated diets were then manually pressed inside a 5 mL modified syringe into pellet form and cut separately into 0.5 cm size. The formulated pellets were dried in the oven at 35°C for 24 hours before being deposited at room temperature in sterile cap bottles.

Experimental Design.
Sixty juvenile red hybrid tilapia fish, with an average weight of 55.6 ± 0.12 g (mean ± SD), were acquired from Taman Pertanian Universiti (TPU) fish hatchery, UPM and transported to the fish hatchery of the Laboratory of Marine Biotechnology (MARSLAB), Institute of Bioscience (IBS), UPM. The fish were acclimatized for ten days in a 500 liter (L) fiberglass aerated tank filled with de-chlorinated freshwater. During acclimatization, fish were fed with a basal diet throughout the experiment at 1% of body weight in two equal parts twice a day at 9.00 a.m. and 5.00 p.m. Water was changed every three days at the rate of 30%, and water quality was monitored daily throughout the experiment. The temperature was maintained at 28 ± 1°C, dissolved oxygen concentration more than 4.0 mg/L, and pH 8.0 ± 1 by YSI Professional Plus (Yellow Spring Instrument, USA). Following acclimatization, the fish were divided into five different experimental diets (A, B, C, D, and E) groups as described. During the feeding trial, the water temperature was 26.1 ± 0.1°C, dissolved oxygen was After the acclimatization period, fish were randomly divided into five experimental groups and fed with four different experimental diets (B, C, D, and E) and a control (A). Each group was divided into six identical 40 L rectangular glass aquariums (0.6 m diameter and 0.75 m height) at a density of 2 fish per aquarium (6 aquariums per diet). During the feeding period of 14 days, fish in each group were gently hand-fed with the diet at 1% of body weight at 9:00 a.m. and 5:00 p.m. per day. The feeding was carefully done and observed to make sure the feeds were all consumed. Water quality parameters were monitored throughout the experiment. During the feeding period, the water temperature, dissolved oxygen, and pH ranged were 26.1 ± 0.1°C, 4.94 ± 0.23 mg/L, and 7.52 ± 0.22, respectively.

Sample
Collection for Immunological Measurement. The experimental fish was placed in a tank with MS-222 at a lethal level (500 mg/L) for at least 3 min [24]. The dead fish was then passed to the dissection station, where freshly cleaned tools were used to excise the spleen from the body. The spleen was then divided into two parts whereby the first part of the tissue was placed in an individual, prelabelled vial, and flash-frozen in liquid nitrogen for NMR analysis. The second part of the tissue was weighed and meshed by gently pressing it through sterile cell strainers of 40 μm nylon mesh together with 5 mL cold 1× phosphate buffered saline (PBS). The supernatant was collected and centrifuged at 1800 rpm for 10 minutes to remove PBS. The cell pellet obtained was mixed with freezing media and aliquoted before storage at -80°C until used.
2.5. Phagocytosis Activity. Phagocytosis activity of spleen cell suspension was determined based on Ai et al. [25] with some modifications. A suspension of 100 μL spleen leucocytes (1 × 10 5 cells/mL) was placed in a sterile microcentrifuge before being mixed with 100 μL yeast suspension (Baker's yeast, Type II, Sigma-Aldrich, USA) at concentration 1 × 10 7 cells/mL and allowed to incubate for 30 min at 25°C. Following incubation, 5 μl of the suspension was dropped on a clean microscope slide precoated with 10% Poly L-Lysine (Sigma-Aldrich, USA) before uniformly spreading to produce a smear. After air-drying, the slides were fixed in methanol for 1 minute, then in May-Grunwald solution (5 minutes) and stained with 7% Giemsa (20 minutes) before being rinsed and redried. The percentage of phagocytes was determined by observing 100 phagocytes from a random region on the slide under a microscope (Nikon Eclipse 80i, Japan).
2.6. Respiratory Burst Activity. The respiratory burst activity (RBA) of the phagocytes was evaluated by nitroblue tetrazolium (NBT, Sigma, USA) assay based on the method of Secombes [26] with modifications. A hundred microliters of spleen cell suspension (1 × 10 5 cells/mL) in 1x PBS were deposited in 96 microplates, then mixed with 100 μl of 5 mg/ml zymosan A (Sigma-Aldrich, USA) along with a control (spleen cell without zymosan) before it stands for 30 min at room temperature. The suspension was centrifuged at 1800 rpm in a microplate centrifuge (Thermo Fisher Scientific Inc., Waltham, MA USA). The microplates were washed three times with discard zymosan before being stained with 100 μL of a nitroblue tetrazolium (NBT) solution (2 mg/mL) and stood for 30 min at room temperature. The suspension was then centrifuged to discard the supernatant before the reaction was stopped by adding 100 μL of methanol. Methanol was discarded through microplate centrifugation followed by the addition of 120 μL of 2 M potassium hydroxide (KOH) and 140 μL dimethyl sulfoxide (DMSO). The colour was subsequently measured at 630 nm with a spectrophotometer (Thermo, USA) using KOH and DMSO as a blank. The result of RBA for each sample was expressed as a stimulated activity with zymosan deducted from the basal activity without zymosan.

Lymphoproliferation Assay.
A hundred microliters of spleen leucocytes (1 × 10 5 cells/mL) was deposited in a sterile 96-well microplate. Then, 100 μL of the mitogen phytohemagglutinin (PHA; Sigma-Aldrich, USA) was prepared in complete media (10% Fetal Bovine Serum: 90% Dulbecco's Modified Eagle's medium) at the concentration of 10 μg/ mL before being added into the cells to activate T-cells. Both treated and untreated cells (without PHA) were incubated for 48 hours at 25°C in a 5% CO 2 incubator. After incubation, the cells were counted manually using a haemacytometer under the microscope (Zeiss Axioscope, Germany). The same procedure was repeated for mitogen lipopolysaccharide (LPS; Sigma-Aldrich, USA) at a concentration of 40 μg/ml to activate the B-cells.
2.8. 1 H NMR Spectroscopic Analysis. The pre-weighed spleen was thawed and mixed with 200 μl of 99.9% deuterium oxide (D 2 O) containing 90 mM KH 2 PO 4 buffer solution (pH 6.0) and 0.1% trimethylsilylpropionic acid-d4 sodium salt (TSP) as the internal reference standard. The tissue was homogenized using a glass rod before being centrifuged at 10,000 rpm for 10 min at room temperature. The resultant clear supernatant was slowly pipetted out into a 3 mm standard NMR tube (Norell, USA) and sent for analysis on a Bruker 700 MHz spectrometer equipped with a TCI Cryo-Probe (Bruker Biospin, Billerica, MA, USA). The spectra were obtained using the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence to suppress the signals of proteins [27] using a 90°pulse (6.6 μs at a transmitter power of 58 dB), a spectral width of 10 504 Hz, an acquisition time of 3.1 s using 48 k data points, 128 scans, and a relaxation delay of 4.0 s. Deuterium oxide was used as the internal lock solvent, and the residual solvent peak was suppressed using a 64 scan PRESAT sequence, with a total acquisition time of 3.53 min.
The 2D NMR spectra such as J-resolved (JRES) and HSQC were acquired using a similar instrument at room temperature (25°C) using 24 scans, 1 k data points at 256 increments, a spectral width of 16 ppm in dimensions, and a relaxation delay of 2 s. 3 Aquaculture Nutrition 2.9. Pre-Processing of NMR Spectra. The pre-processing of all 1 H NMR spectra was done manually involving phasing baseline correction before referenced to the internal standard (TSP) at 0.00 ppm using Chenomx NMR Suite 5.1 Professional (Edmonton, Canada). The spectral regions from 0.3 to 10 ppm were normalized to the total sum of the spectral intensity after excluding residual water (4.7-4.9 ppm) and binned into an integrated bucket of 0.04 ppm in width. The generated binned data in a table sheet was then visualized in a multivariate analysis using SIMCA-P 14.1 (Umetrics, Umea ο , Sweden).
The dataset was set to a Pareto scale before executing the Principal Component Analysis (PCA) and OPLS models by SIMCA. In the score plot generated, each variable was represented as an individual spleen sample, which allowed the observation of the classification behaviour of spleen samples along with the two main principal components of PC1 and PC2. The metabolites responsible for the groups' separation could be identified based on the variables that appeared in the loading plots, represented as individual bins of the 1 H NMR data. The validation and fitting of the generated models were evaluated based on three-fold approach, including 100 permutation tests, computation of R2Y, Q2Y, and CV-ANOVA values [28].
2.10. Statistical Analysis. The normality of the bioactivity data was evaluated based on the Shapiro-Wilk test before statistical analysis with P-value measured greater than 0.05 indicated as the data was normally distributed. Statistical analysis was performed using SPSS 16.0 (SPSS Inc., Michigan Avenue, Chicago, IL, USA). Significant differences between the samples were determined using one-way analysis of variance (ANOVA) followed by Duncan's post hoc test at a confidence interval of 95%. A P-value less than 0.05 was considered significant.

Results and Discussion
3.1. Phagocytosis Activity. Innate immunity is of utmost importance in teleosts since their adaptive immune response is less evolved than in mammals. Phagocytosis is a common innate immune response mechanism by phagocyte cells involving monocytes, macrophages, neutrophils, mast cells, dendritic cells, and nonspecific cytotoxic cells [29][30][31]. Phagocyte cells are responsible for defending the host by ingesting toxic foreign particles, including dead or dying cells [30,32,33]. Phagocytosis begins as phagocyte cells engulf recognized particles, including pathogens, aged red blood cells, and foreign substances, by attaching them to their surface and internalizing them into phagosomes formed around the objects engulfed [34]. In the present study, phagocytic activity is defined as the percentage of leukocytes that engulfed yeast cells within a total of 100 leukocytes counted in the smear.
The impact of IG-supplemented diets on the phagocytosis activity is not dose-dependent and not significant except for group B, as displayed in Figure 1. The highest percentage of phagocytosis was observed in group B of 0.63% IG, followed by D (2.5% IG) with 55% and 41% accretion, com-pared with the control (group A) after 14 days of feeding treatment. Other groups showed no significant difference with less than 10% increment of activity compared to the control. It was mentioned that successful administration of any immunostimulant demands the use of an accurate dietary inclusion level over a sufficient period of time [35].
There was yet no study reported on the effect of dietary IG on the phagocytosis activity in tilapia. However, the potential of other microalgae in phagocytosis activity has been revealed. For instance, Galal et al. [8] found that 10% Chlorella vulgaris incorporated diet significantly (P < 0:05) increased phagocytic activity in tilapia after 15 and 30 days of the feeding period. Oral administration of 10 mg Spirulina platensis suspension to tilapia also enhanced the phagocytic activity more than vaccine-treated [36]. In another study, significant dose-dependent stimulation of phagocytic activity was observed on head kidney leucocytes of gilthead seabream fed with Phaeodactylum-incorporated diet. However, not much difference was observed with Tetraselmis-and Nannochloropsis-incorporated diets than the control [37].
The manual count of phagocytic cells using an electron microscope as applied in the present study has several disadvantages. During the counting process, the glass slide's scanning movement introduced overlapping scanning regions resulting in counting errors, in which the cells might not be uniformly distributed across the smear. These issues can be overcome by using the flow cytometry method, which is more accurate and less time consuming [38]. Unfortunately, the flow cytometer was not available to be used for the current study.

Respiratory
Burst Activity (RBA). The ability of leukocytes to destroy invading pathogenic agents is one of the important defence mechanisms. A mechanism known as leukocyte respiratory burst activity involves professional phagocytes such as neutrophils, macrophages, dendritic cells, and monocytes, which also work as reactive oxygen species (ROS) producers [39]. ROS is predominantly formed in the mitochondria, peroxisomes, plasma membrane, and cytoplasm. Oxygen is less reactive due to biradical orbitals. However, the electron transfer or energy absorption can trigger the activation of ROS from molecular oxygen. ROS is comprised of oxygen-free radicals such as hydroxyl radical (HO•), superoxide anion (O 2 •-), peroxyl (ROO•), alkoxyl (RO•), and nitric oxide (NO), as well as nonradical molecules such as oxygen, hydrogen peroxide (H 2 O 2 ), and hydrochloric acid (HCl), and transition metals such as copper and iron [40,41]. The leukocyte respiratory burst activity was used to indicate innate immunity since it was first discovered in mammals in the early 1930s. Some researchers noticed that high oxygen consumption happened during phagocytosis. Leukocytes increase their intracellular oxygen utilization during the phagocytosis of pathogens and subsequently produce diverse ROS. All ROS generated during phagocytosis with their by-products are very reactive and utilized as an immune system strategy to destroy invading pathogens. Respiratory or oxidative burst is a resting metabolic pathway in cells which is quickly activated to recognize the microbial agent. It is responsible for generating potent oxidizing 4 Aquaculture Nutrition molecules or burst activity that functions to destroy microorganisms [39,42,43].
In the present study, zymosan particles made up of yeast cell walls were detected as a foreign pathogen and rapidly engulfed by phagocytic cells in the spleen leucocytes suspension. Nitroblue tetrazolium (NBT) chemical was used to determine the superoxide anion generated in various phagocytic cells. Current results in Figure 2 have exhibited a significant enhanced respiratory burst activity in the diet groups except for diet C when compared to the control. Diet D (2.5% IG) was marked as the highest respiratory burst activity with a value two times higher than the control.
Other studies also showed significant results of respiratory burst activity such as in juvenile great sturgeon fed with 10% S. platensis incorporated for eight weeks [44]. In another study, 0.13% of Spirulina in a basal diet demonstrated a positive immunity promoter for Nile tilapia in Egypt with significant enhanced respiratory burst activity after 12 weeks of feeding [45]. Similarly, Watanuki et al. [46] discovered the effects of 1 mg Spirulina in 0.1 ml suspension on carp expressing the highest superoxide anion production after three days of posttreatment.
However, the current results were not dose-dependent since diet D of 2.5% IG enhanced more superoxide production than diet E containing 5.0% IG. A similar pattern was also observed in Asian sea bass after being fed with different garlic (Allium sativum) supplemented diets for two weeks [47]. Treatment dose garlic at 10 g/kg feed gave the highest reading of superoxide anion production compared to 15 and 20 g/kg feed. In a different study, chitosan was a stimulating agent for phagocytosis and respiratory burst activity in cobia fish. Unfortunately, a high chitosan level of 6.0 g/kg caused inhibition of the activities, suggesting that the cells had become exhausted for some duration after being stimulated with a high concentration of chitosan [48]. It was supported by other studies which concluded that high chitosan  5 Aquaculture Nutrition levels significantly inhibited phagocytosis and respiratory burst activity [49,50]. However, the current study shows better results than long-term administration of dietary content of seaweed powders after demonstrating no significant differences regarding respiratory burst activity between control and different feeding period [35].
3.3. Lymphoproliferation Activity. Another essential approach for assessing the fish immune system is through activation of lymphocytes to proliferate the initiation of mitosis via chemical mediators known as a mitogen. Lymphocytes, which comprised most of total white blood cells, acts as primary component responsible of various immunological responses [51]. Mitogen is mostly made up of bacteria and plants as carbohydrate-binding lectins [52]. Leucocyte proliferation, also defined as mitogenesis, begins after exposure to mitogens, such as concanavalin A (ConA) and phytohemagglutinin A (PHA), which primarily stimulate T-cells. At the same time, lipopolysaccharide (LPS) activates B-cells, and mitogenic pokeweed (PWM) can enhance both T-and B-cells. The main target of mitogens is the plasma membrane, whereby the mitogens attach to the carbohydrate substituents of membrane glycoproteins. For example, ConA attaches to α-D-mannosyl, while PWM and PHA bind to N-acetyl-D-glucosamine. The interaction with mitogen develops membrane receptors to stimulate adenylate cyclase, resulting in signal transduction from the membrane to the lymphocyte nucleus [53]. In vitro lymphocyte activation by respective mitogens enables the cells to generate specific immune response cytokines [54,55]. The evaluation of proliferation can be determined easily by counting the growth in cell density, using a haemacytometer, 6 Aquaculture Nutrition or more advanced techniques to measure the absorption of radiolabelled DNA precursors 3H-thymidine [56]. Thus, lymphoproliferation assay has become a critical parameter to evaluate the efficacy of dietary supplementation on immune response as it has been applied in numerous studies on tilapia [57][58][59]. Figure 3(a) demonstrates the number of T-cells stimulated by mitogen PHA and without mitogen calculated manually using a haemacytometer after 48 hours incubation. Diet C shows the highest number of cells with a clear stimulatory effect on leukocyte proliferation (P ≤ 0:05) in the presence of mitogen in 8-fold accretion, followed by diets D (7-fold) and E (5-fold) compared to non-supplemented fish (diet A: control). Similarly, the highest number of Bcells stimulated by LPS was observed in diet C, followed by diet D significantly (P < 0:05) with fold change enhancement of 0.36 and 0.35, respectively, compared to control (Figure 3(b)). Interestingly, spontaneous lymphocyte proliferation in the absence of mitogens indicated a high proliferation of T-and B-cells but not significantly different from mitogen-stimulated proliferation.
This finding is in agreement with previous studies that demonstrated the immunomodulatory properties of several microalgae. For example, Chuang et al. [60] revealed that microalgae Dunaliella salina in 369 and 922.5 mg/kg doses has enhanced the significant proliferation of both T-and B-cells under untreated and LPS conditions (P < 0:001) after orally administered to leukemic mice. Based on the increment of T-and B-cells, it is believed that the actions of D. salina in leukemic mice might be involved in triggering a cascade of cellular and humoral immune responses. In another study, astaxanthin, a common carotenoid present in microalgae, had successfully stimulated mitogen-induced lymphoproliferation in humans. The stimulation index of T-and B-cells from peripheral blood mononuclear cells was consistently higher (P < 0:05) after being treated with dietary astaxanthin for eight weeks [61]. The administration of chrysophanic acid diet at 8 mg was also shown to significantly enhance the lymphocyte proliferation index in Indian major carps after 4 weeks of treatment; however, the index started to reduce after 8 weeks of feeding [62]. Therefore, this present finding could also be seen to prove the potential of IG as one of the promising candidates in boosting the innate and adaptive immune responses in red hybrid tilapia after 2 weeks of feeding.  Table 1. Twenty-two metabolites were detected in the up to middle-field region (0.3-5.5 ppm) of the spectrum comprising overlapping signals of amino acids and carbohydrates with different intensities.
Meanwhile, in the downfield region of the spectrum (6.0-9.0 ppm), 13 metabolites were successfully detected, consisting of different types of nucleosides and amino acids mostly associated with an aromatic ring. Most of the assignments were made based on metabolites reported in a spleen sample from a rat model acquired on an NMR Bruker-AV600 spectrometer [63], except for leucine which was referred to De Pascali et al. [64]. Additional data provided by the 1 H-13 C HSQC correlation improved the identification of assigned metabolites but was limited to several compounds due to their low concentrations.
Identification of 8 compounds (a to h) was confirmed based on 2D-NMR of HSQC as shown in Figure 5.   Isoleucine (a) was confirmed by the triplet peak detected at 0.94 ppm and cross-peaked to C-4 of the terminal methyl proton at 13.93 ppm. The cross-peaks of two doublets at 23.7 and 24.9 ppm in the HSQC spectra have verified the presence of terminal methyl protons of leucine (b). Valine (c) was confirmed from the correlation of its doublet at 1.05 ppm to 20.77 ppm, which is one of its terminal methyl carbons (CH 3 ). Meanwhile, the presence of glutamate was confirmed based on the correlation of the multiplet signal seen at 2.07 ppm with its methylene carbon (CH 2 ) at 29.88 ppm. The contrast between αand β-glucose could be differentiated by the position of the anomeric hydrogens or the hydroxyl group attached to the anomeric carbon [65]. In the present study, β-glucose could be characterized by the appearance of a doublet at 4.66 ppm; meanwhile, αglucose can be recognized by its doublet at 5.24 ppm. Unfortunately, no signal could be observed in the HSQC spectra probably due to its minute peak intensity. Taurine (e) was justified with the HSQC correlation of its triplet at 3.27 ppm with its methylene carbon (CH 2 ) at 50.47 ppm. Another metabolite is betaine (f) that was confirmed through the cross-peak of its singlet at 3.27 ppm and of the terminal methyl (CH 3 ) at 56.19 ppm ( Figure 5). Meanwhile, the downfield region of the spectra (6.0-9.0 ppm) shows the presence of adenosine monophosphate (AMP: g) that was confirmed when its doublet of doublet peak at 4.01 ppm correlated to one of methylene carbon (CH 2 ) at 63.05 ppm. AMP contains a purine skeleton connected to a ribose by monophosphate. The last confirmed metabolite through 2D HSQC correlation is inosine (h) which was assigned based on the correlation of the small doublet of doublet signal at 4.27 ppm with methine carbon (CH) at 68.90 ppm. However, the signal hardly could be seen in Figure 5. Inosine relates to a family of purine nucleosides made up of a ribosyl or deoxyribosyl moiety attached to a purine skeleton. Other metabolites were not identified using HSQC correlation due to their poor signals in the 1D NMR spectra. Nevertheless, their presence was confirmed by 2D JRES spectra of their multiplicity.

Metabolite Changes in Different Fish Group Diets.
As an unsupervised and nonbiased method of class separation, the  10 Aquaculture Nutrition primary purpose of PCA is to acquire preliminary information on the dataset without relying on any previous information referring to the data distribution. The prominent feature of PCA is that the differences are exposed, and thus, it is possible to measure the accuracy of the results, as it is not available in supervised methods. Therefore, supervised classification approaches (PLS-DA/OPLS-DA) guided by the appropriate PCA cluster are preferable for obtaining satisfactory biological outcomes [66]. The metabolite changes experienced after 14 days feeding regime of IG-supplemented diets in fish were observed through the PCA analysis on spleen 1 H-NMR data. The goodness of fit and predictability of the model showed four components with R2X value of 0.723 and Q2X value of 0.341. It is normal to obtain lower R2 and Q2 values in a biological model as shown in other fish metabolomics studies [67,68]. The first two main components of PC1 and PC2 contributed 31.2% and 22.6%, describing a total of 54% of the variance. Each dot in the score plot (Figure 6(a)) defines as a spleen sample from a fish.
The score plot displays three clusters with group C and control group (diet A) were clustered together indicating both groups shared close similarity in chemical profiles. The IG-supplemented groups D and E, particularly E, were discriminated away from the control group, indicating their different metabolite content. However, no clear cluster was observed for B as its variables were scattered between two quadrants by PC1 due to a high intravariation. These clusters suggested that the fish in groups B, D, and E have experienced some metabolite changes causing them to move away from the control group (A).
The metabolites responsible for the separation are displayed in the loading plot (Figure 6(b)) wherein defined by variables with respective bins of NMR chemical shift regions with acceptable 95% jack and knife bars. The variables assigned as betaine, taurine, choline, and αand β-glucose located on the positive side of PC1 were more prominent in group C and control. In contrast, metabolites on the negative side of PC1 were identified as leucine, isoleucine, and lysine presented higher in group E. Glutamate, serine, glycine, and aspartate were dominated in diet D, on the positive side of PC2.
3.6. Biomarker Identification Using OPLS Model. The potential biomarkers for the immunomodulatory effects of IG were further analyzed using a supervised OPLS model to correlate between metabolites in the spleen and the conducted immunomodulatory bioactivities (phagocytosis, RBA, and lymphoproliferation). The respective model was generated between 1 H NMR data of diet A as control and diet D (2.5% IG), as displayed in Figure 7(a) since this group consistently gives significant results among the conducted bioactivities.
Four components were generated (Figure 7

12
Aquaculture Nutrition with a difference between them not more than 0.3 [69]. Hundred random permutations of the y-variable conducted for each bioactivity verified the minimum validity of the current model after y-axis intercepts below zero as stated in SIMCA. However, the validity of each bioactivity was sup-ported with observed vs predicted plots with excellent correlation between all bioactivities (Y-data) and metabolites (X -data) represented by regression lines with a value between 0.74 and 0.97. Their permutation tests and observed vs predicted plots are provided in Figures 8 and 9.

Aquaculture Nutrition
The significance of any variable was determined by analyzing the variable importance in a projection (VIP) plot with jack-knifing uncertainty bars. The variables with VIP scores of more than 1 with the error bar not crossing the baseline in the loading column plot were retained as significant. The selected bars could be classified as chemical markers or biomarkers as depicted in Figure 7(c) with separation by PC1 between groups A and D. Six of them on the positive side in red column were upregulated and belong to diet D, and thirteen downregulated metabolites are on the negative side in green column for diet A (Figure 7(d)). These variables were identified as corresponding potential biomarkers by comparing their variable IDs with a list of identified chemical shifts in NMR spectra based on the previous literature and open access metabolomics database and subsequently verified by 2D J-resolved and HSQC analysis.
The relationship of the variables between their distribution in the score plot and separation in the loading plots was summarized in a biplot as shown in Figure 10. The biplot of the OPLS model discriminated diet A (control) from D (2.5% IG) with all bioactivities of phagocytosis, respiratory burst activity, and lymphoproliferation was clustered together with diet D. The variables were identified as isoleucine, glutamate, ethanol, and tyrosine which upregulated in the IG-supplemented (D) tested groups compared to the control (A). The other 13 variables such as glucose, choline, inosine, and adenosine were downregulated in the control group compared to the IG-supplemented group.  Table 2. The most important upregulated metabolite in D based on VIP value is isoleucine, with a significant fold change value compared to the control group. Isoleucine is an essential branched-chain amino acid that fishes obtain from an external diet to fulfil certain biochemical reactions and growth requirements. It includes lipolysis, lipogenesis, glucose metabolism, glucose transportation, intestinal amino acid transport, benefiting embryo growth, and immunity [70]. It has been discovered that isoleucine deficiency in the diet leads to biochemical malfunctions including growth interruption [71,72]. In the present study, isoleucine is one of the metabolites with a significant increment in fish spleen cells after being fed with an IG-supplemented diet when compared to the commercial diet. The isoleucine increase could be related to its presence with other amino acids in IG which successfully identified via NMR in polar solvents as reported before [73]. A recent study found isoleucine enhanced growth performance and intestinal immunology of hybrid catfish. After eight weeks of dietary isoleucine (12.5 g/kg diet), the activities of lysozyme, acid phosphatase, and alkaline phosphatase, and the contents of complement 3 (C3), C4, and immunoglobulin M (IgM) significantly increased [74]. In a different study, dietary isoleucine up to 16 g/kg intake by fingerling Channa punctatus for 12 weeks has significantly improved serum protein, lysozyme, and superoxide dismutase activities, improved growth and haematological parameters [75]. Superoxide dismutase (SOD) is an essential component of the enzyme mechanism and protects cells from the harmful effects of endogenic superoxide radicals [76,77]. SOD activates the transformation of radical superoxide into hydrogen peroxide and molecular oxygen. ROS formation in fish could be influenced by 16 Aquaculture Nutrition isoleucine. The elimination of ROS in fish is accomplished through antioxidant protection and seems to be primarily dependent on antioxidant compounds such as glutathione (GSH) and antioxidant enzymes. The improvement of O -2 and ·OH-scavenging ability can be seen in an isoleucineincorporated diet. In addition, sufficient isoleucine (6.6-12.5 g/kg diet) could enhance the elimination of free radicals by minimizing the oxidative damage in fish muscle mainly through the NF-E2-related factor 2 (Nrf2) signalling pathway [78]. Another significantly upregulated metabolite of the IGsupplemented diet is glutamate, which was identified based on three different chemical shifts at 2.34, 2.06, and 2.02 ppm with a significant fold change value compared to the commercial diet ( Table 2). As one of the nonessential amino acids, L-Glutamate is a functional amino acid present in both animal and plant proteins responsible for cell metabolism and signalling [79,80]. Glutamate is also an important energy source in the small intestine [81], and a precursor of essential compounds such as glutathione (GSH) [82]. In the past fish research, glutamate dietary dissolved in Dulbecco's Modified Eagle Medium (DMEM) has been proven to enhance the antioxidant ability and modulate the antioxidant-related signalling molecule expression in the intestine as well as in enterocytes [83,84]. This study suggests the biosynthesis of glutamate that probably originated from α-ketoglutarate [85]. The reversible reactions involved in converting α-ketoglutarate to glutamate catalyzed by glutamate dehydrogenase as illustrated in Figure 11. Recent studies have found dietary glutamate in certain fish species with improved growth efficiency and immune responses. For instance, low-phosphorus diet supplementation of 10 g/ kg glutamate boosted the behaviour of intestinal antioxidant enzymes and enhanced the immune system of juvenile mirror carp [86].
Ethanol was also found to be enhanced in the IGsupplemented diet. However, the increment measured was not significant, probably due to high variations between rep-licates. The increment of ethanol could be associated with glycolysis due to high glucose content in IG as reported previously in the relative quantification of metabolites identified [73]. Previous studies have proven the impact of ethanol on immunomodulatory properties [87,88].

Downregulated Metabolism.
In the present study, IG supplementation led to the lower concentration of α/β-glucose, choline, inosine, hypoxanthine, adenosine, and unknown as simplified in Table 2. These changes could be correlated with perturbations in glycolysis/gluconeogenesis; purine metabolism; glycine, serine, and threonine metabolism. The most downregulated with the highest VIP value metabolites were α/β-glucose which involved in glycolysis/ gluconeogenesis as suggested by KEGG. Their concentrations were both 0.61 folds decreased ( Figure 11). Glucose is an essential fuel for ATP production to fulfil the bioenergetic requirements of the cell. The first phase of glycolysis requires the phosphorylation of glucose via hexokinase to create glucose-6-phosphate (G6P). The mechanism then continues as a sequence of enzymatic reactions that eventually generate pyruvate. Pyruvate may have several fates in the cell-based cellular function, one of which is oxidized to CO 2 inside mitochondria for the optimized yield of ATP. During aerobic glycolysis, pyruvate is also transformed to lactate, a metabolic product adopted by cells that participated in vigorous growth and multiplication [89]. Glycolysis modulation is an important step in activating innate and adaptive immune cells. It provides a way to generate macromolecules and trigger the antimicrobial respiratory burst through flux enrichment in the pentose phosphate pathway [17]. During 18 hours after immune stimulation, T-cells regulate their metabolic processes, enhancing the release of glucose transporters and glycolytic enzymes. T-cells then trigger aerobic glycolysis in preparation for the activation of extensive clonal growth [90]. Therefore, a sufficient source of glucose is necessary for the sustainability of high glycolysis levels. The biplot of the OPLS model ( Figure 8) depicts all biological 17 Aquaculture Nutrition activities being discriminated against together with the IG supplementation diet (diet D) due to the high glucose content which has provided more energy for the glycolysis pathway compared to the control group (diet A). Table 2 shows the hypoxanthine, adenosine, and inosine levels in the IG supplementation group have decreased up to 0.58, 0.61, and 0.88 folds, respectively. These three purine nucleotides involve in the purine metabolism pathway, as suggested by KEGG. Purines are the most abundant metabolites in all biological entities as they are a key source for DNA and RNA building blocks. Apart from that, purines contribute as a fuel source and cofactor to support cell growth and replication. Purines and their components are also normally involved in biochemical mechanisms, including immune functions and host-tumour interactions [91]. The metabolism of purines is defined as the biochemical pathway of building and breaking down the purines inside an organism. This metabolism involves the biosynthetic process of de novo purine, the mechanism of purine salvage, and purine breakdown. The transformation of phosphoribosylpyrophosphate (PRPP) into inosine 5′-monophosphate (IMP) occurs via the de novo purine biosynthetic pathway catalyzed by several enzymes. Meanwhile, purine salvage recycles hypoxanthine, inosine, and adenine as substrates to produce purine nucleotides. Inosine and hypoxanthine in the purine degradation pathway can be further oxidized into xanthine and uric acid [92]. Furthermore, the metabolite at 3.70 ppm remained unknown as its singlet peak could not be matched with any previously reported study or referred open access metabolomics databases.

Conclusions
The current study describes the first work in which a metabolic approach was used to analyze the red hybrid tilapia spleen cells administered with an IG-supplemented diet. Our findings revealed that IG supplementation at a concentration of 2.5% has enhanced the immune response of innate immunity, which may be beneficial for protecting the health condition of the fish. The OPLS analysis demonstrated that three significant metabolites, namely, tyrosine, glutamate, and isoleucine, were substantially upregulated in the 2.5% IG-supplemented (diet D) fish compared to the control group (diet A). The IG supplementation also resulted in lower concentrations of various metabolites in the spleen including α/β-glucose, choline, hypoxanthine, adenosine, and inosine, which are involved in glycolysis and purine degradation. The metabolite alterations in the red hybrid tilapia spleen reflect both the activity and immune response in fish, suggesting that IG supplementation has strongly influenced the metabolic condition in fish cells and further boosted the immune response and health status. These findings have also proved to be successful the application of metabolomics tools in the discovery of alternative healthy feed from microalgae species to improve fish health in meeting the aquaculture's future targets. However, further

18
Aquaculture Nutrition investigation is required to explore the impact of IG supplementation on these biomarkers during microbial infections.

Data Availability
Data available on request.

Conflicts of Interest
The authors declare no conflict of interest.