Revealing Alteration in the Hepatic Glucose Metabolism of Genetically Improved Carp, Jayanti Rohu Labeo rohita Fed a High Carbohydrate Diet Using Transcriptome Sequencing

Although feed cost is the greatest concern in aquaculture, the inclusion of carbohydrates in the fish diet, and their assimilation, are still not well understood in aquaculture species. We identified molecular events that occur due to the inclusion of high carbohydrate levels in the diets of genetically improved ‘Jayanti rohu’ Labeo rohita. To reveal transcriptional changes in the liver of rohu, a feeding experiment was conducted with three doses of gelatinized starch (20% (control), 40%, and 60%). Transcriptome sequencing revealed totals of 15,232 (4464 up- and 4343 down-regulated) and 15,360 (4478 up- and 4171 down-regulated) differentially expressed genes. Up-regulated transcripts associated with glucose metabolisms, such as hexokinase, PHK, glycogen synthase and PGK, were found in fish fed diets with high starch levels. Interestingly, a de novo lipogenesis mechanism was found to be enriched in the livers of treated fish due to up-regulated transcripts such as FAS, ACCα, and PPARγ. The insulin signaling pathways with enriched PPAR and mTOR were identified by Kyoto Encyclopedia of Genes and Genome (KEGG) as a result of high carbohydrates. This work revealed for the first time the atypical regulation transcripts associated with glucose metabolism and lipogenesis in the livers of Jayanti rohu due to the inclusion of high carbohydrate levels in the diet. This study also encourages the exploration of early nutritional programming for enhancing glucose efficiency in carp species, for sustainable and cost-effective aquaculture production.


Introduction
Aquaculture production is increasing significantly at the global level, contributing to 6.7% of total fishery production as compared to marine production (captured) [1]. For sustainable aquaculture, the availability of the feed and its cost is the greatest concern. During feed formulation, carbohydrate

Analysis of DEGs and Annotation
In order to identify differentially expressed transcripts/genes (DEGs) among the liver transcriptomes of control and high CHO-fed fish, comparative analyses were performed. The differential expression analysis revealed that totals of 15,232 and 15,360 transcripts were differentially expressed in the liver tissues of rohu fish fed with 40 and 60% CHO as compared to the control group (20%). Among that, 4464 transcripts were up-regulated, while 4343 transcripts were found to be down-regulated and 6425 were neutral in the control versus fish fed with 40% CHO. In the fish fed a high CHO (60%) diet, 4478 transcripts were up-regulated, while 4171 transcripts were down-regulated and 6711 were found to be neutral compared to the control ( Figure 1). Table 2 describes the number of differentially expressed genes in the control and treated groups. The volcano plot (scatter plot) depicts the distribution of up-and down-regulated transcripts, including neutral expression, in both the control and treated groups ( Figure 2). Among these DEGs, a total of 30 were identified as significantly differentially expressed transcripts/genes, and were associated with metabolism, among the control and fish fed a high-CHO diet using the criteria of adjusted P-value < 0.05 and (Log 2 (fold change)) > 1. For the validation of DEGs among control and treated groups, six transcripts/genes associated with metabolism were selected for qPCR, and their relative expression patterns were consistent with the resulting liver RNAseq data. The reads were mapped to a reference database for annotation, and the liver transcriptome data of rohu were annotated with the zebrafish public database (Nr protein, KEGG, GO and COG) using BLAST2GO with an E-value cut-off of <10 −5 . In the BLAST search, gene names and the associated protein accession numbers with respect to transcripts were retrieved, and further gene ontology (GO) terms were assigned. The gene ontology (GO) assignments were performed for classifying the functions of expressed genes/transcripts of the livers of rohu fish based on homology searching. The functional enrichment analysis indicated that all the differentially expressed transcripts of control and treated groups were significantly enriched in three GO terms (Table S1). In the liver, the 446 up-regulated DEGs found in fish fed a high-CHO (40%) diet were significantly enriched GO terms, with 1749 in the biological processes, 1275 in the cellular components and 2286 in the molecular function categories, while the enriched GO terms in fish fed a high-CHO (60%) diet included 3119 in the biological category, 1278 in the cellular category and 2298 in the molecular function category.  . Volcano plot for differential gene expression. Fish feed with (A) 40% CHO vs. 20% CHO and (B) 60% CHO vs. 20% CHO. Scattered points represent genes: the x-axis is the fold change for the ratio of treated vs. control, whereas the y-axis is the statistic or -Log 10 (P-value), which shows the probability that a gene has statistical significance in its differential expression. The green dots are thus genes significantly up-regulated after treatment, red dots are genes significantly down-regulated after treatment and black dots are genes significantly neutrally regulated after treatment.  The top gene ontology terms were found to be associated with biological process categories, such as behavior, cell adhesion molecules, threonine phosphatase, developmental process, growth, immunity, localization, locomotion, metabolic process receptor binding, response to the stimulus, insulin signaling, signaling liver development, etc. In the cellular component categories, GO terms were well represented in the cytoplasm, mitochondrion, exocyst, nucleoplasm, organelles, membrane, extracellular region, extracellular matrix, cell, cell junction, synapse, extracellular space, an integral component of the membrane, etc. In the molecular function group, GO terms such as binding, catalytic activity, molecular function regulator, RNA polymerase, motor activity, nucleic acid binding, signal transducer activity, protein binding, protein kinase, transcription cofactor activity, structural constituent of the ribosome, actin-binding and hormone activity were abundantly represented ( Figure 3). The Circos3 plot describes overlaps between the differentially expressed genes based on their functions or shared pathways ( Figure 4).

Figure 1.
Venn diagram of differentially expressed genes. The sum of numbers in each big circle is the total number of differentially expressed genes in each comparison group, and the overlapping part is the number of common differentially expressed genes among the comparison groups.

Figure 2.
Volcano plot for differential gene expression. Fish feed with (A) 40% CHO vs. 20% CHO and (B) 60% CHO vs. 20% CHO. Scattered points represent genes: the x-axis is the fold change for the ratio of treated vs. control, whereas the y-axis is the statistic or -Log 10 (P-value), which shows the probability that a gene has statistical significance in its differential expression. The green dots are thus genes significantly up-regulated after treatment, red dots are genes significantly down-regulated after treatment and black dots are genes significantly neutrally regulated after treatment.  Scattered points represent genes: the x-axis is the fold change for the ratio of treated vs. control, whereas the y-axis is the statistic or -Log 10 (P-value), which shows the probability that a gene has statistical significance in its differential expression. The green dots are thus genes significantly up-regulated after treatment, red dots are genes significantly down-regulated after treatment and black dots are genes significantly neutrally regulated after treatment. The top gene ontology terms were found to be associated with biological process categories, such as behavior, cell adhesion molecules, threonine phosphatase, developmental process, growth, immunity, localization, locomotion, metabolic process receptor binding, response to the stimulus, insulin signaling, signaling liver development, etc. In the cellular component categories, GO terms were well represented in the cytoplasm, mitochondrion, exocyst, nucleoplasm, organelles, membrane, extracellular region, extracellular matrix, cell, cell junction, synapse, extracellular space, an integral component of the membrane, etc. In the molecular function group, GO terms such as binding, catalytic activity, molecular function regulator, RNA polymerase, motor activity, nucleic acid binding, signal transducer activity, protein binding, protein kinase, transcription cofactor activity, structural constituent of the ribosome, actin-binding and hormone activity were abundantly represented ( Figure 3). The Circos3 plot describes overlaps between the differentially expressed genes based on their functions or shared pathways ( Figure 4).

Pathway Analysis
KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis is mainly utilized for the systematic representation of gene/transcript functions through their network, which leads to the discovery of gene/transcript functions and their associated metabolic pathways [25]. In this study,

Pathway Analysis
KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis is mainly utilized for the systematic representation of gene/transcript functions through their network, which leads to the discovery of gene/transcript functions and their associated metabolic pathways [25]. In this study, the annotated sequences were mapped to the zebrafish reference pathways available in the KEGG database. The sequences were successfully assigned to 433 KEGG pathways in different categories, as shown in Figure 5 and Table S2. We found that metabolism-associated pathways, such as protein kinases (KID: 01001), the MAPK signaling pathway (04010), G protein-coupled receptors (04030), the insulin signaling pathway (04910), the mTOR signaling pathway (04150), G protein-coupled receptors (04030), the FoxO signaling pathway (04068), lipid biosynthesis proteins (01004), glycolysis/gluconeogenesis (00010) and the PPAR signaling pathway (03320), were the most abundant in control vs. treated groups. As per our experiment on carbohydrate metabolism, we found that 30 genes/transcripts were involved in the glycolysis/gluconeogenesis pathway in the livers of the control vs. 40% CHO and control vs. 60% CHO groups. The hexokinase, ATP dependent PFK, glycogen synthase, PHK, PGK, ACCα, PPARγ and FAS genes/transcripts were mapped against the insulin signaling pathway (04910), which depicted the involvement of genes in glucose metabolism as well as glycogen synthesis and lipogenesis ( Figure 6). The insulin signaling pathways of the liver tissues of the control and treated (high-CHO) rohu fish are depicted in Figure 7. the annotated sequences were mapped to the zebrafish reference pathways available in the KEGG database. The sequences were successfully assigned to 433 KEGG pathways in different categories, as shown in Figure 5 and Table S2. We found that metabolism-associated pathways, such as protein kinases (KID: 01001), the MAPK signaling pathway (04010), G protein-coupled receptors (04030), the insulin signaling pathway (04910), the mTOR signaling pathway (04150), G protein-coupled receptors (04030), the FoxO signaling pathway (04068), lipid biosynthesis proteins (01004), glycolysis/gluconeogenesis (00010) and the PPAR signaling pathway (03320), were the most abundant in control vs. treated groups. As per our experiment on carbohydrate metabolism, we found that 30 genes/transcripts were involved in the glycolysis/gluconeogenesis pathway in the livers of the control vs. 40% CHO and control vs. 60% CHO groups. The hexokinase, ATP dependent PFK, glycogen synthase, PHK, PGK, ACCα, PPARγ and FAS genes/transcripts were mapped against the insulin signaling pathway (04910), which depicted the involvement of genes in glucose metabolism as well as glycogen synthesis and lipogenesis ( Figure 6). The insulin signaling pathways of the liver tissues of the control and treated (high-CHO) rohu fish are depicted in Figure 7.  Clusters of orthologous groups (COG) analyses were carried out to assign or classify orthologous gene products. In this study, totals of 420,648, 409,738 and 489,475 genes/transcripts in the fish of the control 20%, 40% and 60% CHO level groups were clustered into 153, 154 and 164 COG categories, respectively. Among these COGs, most were sequence mapped or assigned to cytoskeleton, general function prediction, translation, ribosomal structure, biogenesis, signal transduction mechanisms, intracellular trafficking, secretion, vesicular transport, etc. The top 26 COG categories in the control and treated groups of the fish are shown in Figure 8  Clusters of orthologous groups (COG) analyses were carried out to assign or classify orthologous gene products. In this study, totals of 420,648, 409,738 and 489,475 genes/transcripts in the fish of the control 20%, 40% and 60% CHO level groups were clustered into 153, 154 and 164 COG categories, respectively. Among these COGs, most were sequence mapped or assigned to cytoskeleton, general function prediction, translation, ribosomal structure, biogenesis, signal  Clusters of orthologous groups (COG) analyses were carried out to assign or classify orthologous gene products. In this study, totals of 420,648, 409,738 and 489,475 genes/transcripts in the fish of the control 20%, 40% and 60% CHO level groups were clustered into 153, 154 and 164 COG categories, respectively. Among these COGs, most were sequence mapped or assigned to cytoskeleton, general function prediction, translation, ribosomal structure, biogenesis, signal  In the gene regulatory network (GRN), carbohydrate metabolism is coordinated with key candidate genes and hub genes, which were depicted in the GRN ( Figure 9). These hub genes included the synaptopodin gene, the LIM and senescent cell antigen-containing domain 1-like protein, reticulon, apolipoprotein, cytochrome P450, parvalbumin alpha, protein O-mannosyltransferase 2, supervillin-like isoform X2, transient receptor potential, melastatin 4a, myosin, light polypeptide 3, skeletal muscle, guanylate-binding protein 2, phenylalanine hydroxylase complement C4-2, TC1-like transposase, coagulation factor IXb (Zgc:136807), complement factor B/C2A, phospholipid-transporting ATPase, alpha-2-macroglobulin (fragment), S-adenosyl methionine synthase, thrombospondin 2b, notch less homolog 1 (Drosophila), very-long-chain (3R)-3hydroxyacyl-CoAdehydratase, obscurin-like 1b, and Sk-tropomodulin (tropomodulin 4 (muscle)). The detailed roles of each of the hub genes involved in the gene regulatory network are described in Table 3. Other major functional classes included the improvement of glucose homeostasis, the control of the oxidative glycolytic pathway in carbohydrate metabolism, association with fatty acid uptake by adipocytes, signaling the central hub gene, controlling glucose, and lipid metabolism by insulin. The hub genes in a given GRN have a strong tendency to exhibit pleiotropic effects. A study of these can help identify tissue-and time-specific hub gene regulators. In the gene regulatory network (GRN), carbohydrate metabolism is coordinated with key candidate genes and hub genes, which were depicted in the GRN ( Figure 9). These hub genes included the synaptopodin gene, the LIM and senescent cell antigen-containing domain 1-like protein, reticulon, apolipoprotein, cytochrome P450, parvalbumin alpha, protein O-mannosyl-transferase 2, supervillin-like isoform X2, transient receptor potential, melastatin 4a, myosin, light polypeptide 3, skeletal muscle, guanylate-binding protein 2, phenylalanine hydroxylase complement C4-2, TC1-like transposase, coagulation factor IXb (Zgc:136807), complement factor B/C2A, phospholipid-transporting ATPase, alpha-2-macroglobulin (fragment), S-adenosyl methionine synthase, thrombospondin 2b, notch less homolog 1 (Drosophila), very-long-chain (3R)-3-hydroxyacyl-CoAdehydratase, obscurin-like 1b, and Sk-tropomodulin (tropomodulin 4 (muscle)). The detailed roles of each of the hub genes involved in the gene regulatory network are described in Table 3. Other major functional classes included the improvement of glucose homeostasis, the control of the oxidative glycolytic pathway in carbohydrate metabolism, association with fatty acid uptake by adipocytes, signaling the central hub gene, controlling glucose, and lipid metabolism by insulin. The hub genes in a given GRN have a strong tendency to exhibit pleiotropic effects. A study of these can help identify tissue-and time-specific hub gene regulators. Int    3

Reticulon
Regulates fragmentation of ER during starvation but when glucose is in excess. [28] 4 Apolipoprotein ApoA-IV improves glucose homeostasis by promoting insulin secretion at high levels of glucose. [29] 5 Cytochrome P450 Improves glucose homeostasis. [30] 6 Parvalbumin alpha Regulates brain-liver circuit for glucose homeostasis.

Alpha-2-macroglobulin (Fragment)
Primarily associated with immunity by remodeling of liver metabolism in fish. It controls activation of the PI3K/Akt, ERK1, and MAPK pathways leading to mobilization of energetic resources away from growth and protein synthesis. [43]

Notchless homolog 1 (Drosophila)
Notch signaling central hub gene controlling glucose and lipid metabolism by insulin; also involved in inhibition of liver glucose production, including glycogenolysis and gluconeogenesis. [46] 22 Very-long-chain (3R)-3-hydroxyacyl-CoA dehydratase (EC 4.2.1.134) Involved in lipid biosynthesis especially in elongation of very-long-chain fatty acids. [47] 23 Obscurin-like 1b In fish myofibrillar component genes, obscurin is up-regulated with growth. [48] 24 Sk-tropomodulin (tropomodulin 4 (muscle)) Regulates thin filament lengths in muscles. [49] 3. Discussion Recent progress in sequencing technologies has led to improved genomic and transcriptomic resources in various model and non-model species, which has opened doors to address key biological questions. Through omics technologies, several studies reported a molecular-level underpinning of digestive physiology via the inclusion of nutrients and/or other dietary ingredients in the target organs of mammals and aquaculture species. In aquaculture fish, for enhancing production and sustainability, understanding fish nutrition and dietary requirements is a major concern [50]. Carbohydrates are the cheapest source of energy, which determines the input cost of aquaculture production. A few studies revealed that the inclusion of carbohydrates in the diets of fish has positive impacts on aquaculture. Prolonged postprandial hyperglycemia is mainly observed in several fish species (specifically in carnivorous fish) upon feeding with a high-carbohydrate diet [10,51]. The molecular level of glucose intolerance is not understood for several commercially important aquaculture species. Therefore, a study of the system biology network integrated with high-throughput sequencing technologies needs to be applied to delineate the molecular events of key metabolic genes and pathways, which could be used as indicators of carbohydrate/glucose digestibility. In this study, to delineate the molecular level responses of Jayanti rohu, L. rohita, upon feeding with different levels of gelatinized starch (20%, 40% and 60%), we performed a total liver transcriptome sequencing to identify the modulation of the key metabolic transcripts/genes and pathways involved in carbohydrate metabolism. Earlier studies hypothesized that the lower metabolic use of enriched carbohydrates/glucose in fish has been attributed to a poor molecular-level regulation of key metabolic genes, as well as their associated pathways, i.e., inefficient glycolysis, inability to produce glucose via gluconeogenesis, and low efficiency of lipogenesis [14,[52][53][54].
In this work, the differential expression profiles of all the key metabolic genes/transcripts, changes in the hepatic glucose transporters, and associated pathways, were systematically analyzed using the liver transcriptome data, and they were found to be involved in processes of carbohydrate metabolism such as glycolysis, glycogenesis, gluconeogenesis and lipogenesis ( Figure 10 and Table S3). The liver is a vital organ and the main source of energy for different activities of metabolism, and it synthesizes key enzymes/hormones and transporters [55]. It plays an essential role in glucose homeostasis via the modification/induction of glucose transporters (GT), as well as key enzymes, including the pathways associated with net glucose flux in the hepatocytes, for achieving successful glucose production from the substrate energy yield [10,56]. regulation of key metabolic genes, as well as their associated pathways, i.e., inefficient glycolysis, inability to produce glucose via gluconeogenesis, and low efficiency of lipogenesis [14,[52][53][54].
In this work, the differential expression profiles of all the key metabolic genes/transcripts, changes in the hepatic glucose transporters, and associated pathways, were systematically analyzed using the liver transcriptome data, and they were found to be involved in processes of carbohydrate metabolism such as glycolysis, glycogenesis, gluconeogenesis and lipogenesis ( Figure 10 and Table  S3). The liver is a vital organ and the main source of energy for different activities of metabolism, and it synthesizes key enzymes/hormones and transporters [55]. It plays an essential role in glucose homeostasis via the modification/induction of glucose transporters (GT), as well as key enzymes, including the pathways associated with net glucose flux in the hepatocytes, for achieving successful glucose production from the substrate energy yield [10,56].

Alteration in Alpha Amylase and Glucose Transporter Activity
Most fish possess α-amylase (EC 3.2.1.1) activities, which mainly break down the polymers of glucose molecules (α-glycosidic linkages), such as starch/glycogen, but are unable to cleave the βglycosidicanomeric bonds of some polysaccharides, for example cellulose or hemi-cellulose, etc., which determine CHO digestion efficiency [32]. In the present study, we used a gelatinized starch in the feed preparation, and upon feeding we observed an increase in α-amylase expression activities in the fish fed at a 40% CHO level as compared to the control. This was also increased in the fish fed a 60% CHO diet. Several studies emphasized that the inclusion of a gelatinized starch polysaccharide is the best method to achieve carbohydrate digestibility [32,33]. However, carnivorous fish, such as Figure 10. Heat map for hierarchical clustering of differential gene expression for selected genes. Fish feed 20% CHO was used as the control, and we also used 40% CHO and 60% CHO. Different colors indicate different P-value (−1 to 1).

Alteration in Alpha Amylase and Glucose Transporter Activity
Most fish possess α-amylase (EC 3.2.1.1) activities, which mainly break down the polymers of glucose molecules (α-glycosidic linkages), such as starch/glycogen, but are unable to cleave the β-glycosidicanomeric bonds of some polysaccharides, for example cellulose or hemi-cellulose, etc., which determine CHO digestion efficiency [32]. In the present study, we used a gelatinized starch in the feed preparation, and upon feeding we observed an increase in α-amylase expression activities in the fish fed at a 40% CHO level as compared to the control. This was also increased in the fish fed a 60% CHO diet. Several studies emphasized that the inclusion of a gelatinized starch polysaccharide is the best method to achieve carbohydrate digestibility [32,33]. However, carnivorous fish, such as catfish, exhibited lower amylase activity as compared to herbivorous and omnivorous fish (for example, carp, tilapia) [37].
Several isoforms of the glucose transporters (GLUT) are reported to act as glucosensing systems, and play a vital role in the transport of glucose molecules into hepatocytes or muscle cells during carbohydrate metabolism mediated by insulin [53,[57][58][59]. There are 14 members of the glucose transporters/GLUT protein family [60,61]. In our studies, we observed that hepatic glucose transporter 2 and solute carrier family 2 (facilitated glucose transporter), member 8 (slc2a8), activities were increased in the livers of fish fed a high-CHO diet as compared to the control, while the sodium-dependent glucose transporter 1 expression was neutral in both groups. However, recent studies have shown that modulations of the gene expression activities of glucose transporter 1 occur in grass carp, Ctenopharyngodon idellus, fed with high dietary carbohydrate levels [62]. Contrastingly, hepatic GLUT1 and GLUT2 were dysregulated in tilapia after 1-3 h due to the inclusion of high CHO in the diet, while an increase in the plasma glucose level was observed [63]. In carnivorous fish, such as rainbow trout and Atlantic cod, the expression level of hepatic glucose transporters 2 (GLUT2) did not rely on the feeding regime (i.e., feeding or fasting status), but hepatic GLUT2 and muscle-specific GLUT4 gene expression were stimulated in the rainbow trout during re-feeding [64][65][66][67].
Recent studies have implied that the down-regulation of hepatic GLUT2 gene activity occurred due to the inclusion of high CHO in the diet of Japanese flounder and rainbow trout, and thereby reduced the activity of the PI3K-AKT signal transduction pathway and glycolysis [68,69]. The present study is also in line with an earlier report, which demonstrated that high dietary CHO assimilation was inhibited due to the failure of mechanistic glucose transportation across hepatic cells [6]. In higher vertebrates, the hepatic GLUT4 gene expression in hybrid giant grouper was increased due to the inclusion of digestible CHO, such as maize starch, in the diet [70]. In our study, we did not observe any expression transcripts of GLUT4 in any of the treated groups compared to the control.

Modulation of the Glycolytic Mechanism
The metabolic activities in the liver are tightly controlled and mediated by the insulin signaling pathway, which in turn maintains glucose homeostasis [55,71]. In the liver, glucose is taken up by hepatocytes with the help of glucose transporters, and further converted into pyruvate via glycolysis. ATP energy generation, followed by the TCA cycle and further excesses of CHO, induces the process of glycogen synthesis and de novo lipogenesis. The key regulatory enzymes, such as glucokinase, (GK/GCK) or hexokinase (HK), pyruvate kinase (PK) and phosphofructokinase (PFK), play a vital role in glycolysis in the liver [72][73][74]. The inclusion of high levels of CHO/glucose in the diet enhanced GK gene expression in several aquaculture species, such as rainbow trout [64], which is the first key enzyme in the glycolysis process.
In our study, we also observed a heightened expression of the hexokinase 1 gene in the 40% CHO-fed fish as compared to the control; contrastingly, HK1 expression was down-regulated in the 60% CHO-fed fish as compared to the control. Similarly, the phosphofructokinase (PFK) expression level showed similar trends to hexokinase in all treatment groups. GK and PFK are two key indicators for effective hepatic glycolysis in several fish, such as tilapia [53,63]. Recent studies showed increases in the GK, PK, PFK and G6PDH gene activities in the livers of Megalobrama amblycephala due to the incorporation of high CHO levels in the diet [62,75,76], and similar trends were observed in other species, such as gilthead sea bream (Sparus aurata) [71], rainbow trout (Oncorhynchus mykiss) [10,12,77,78], tilapia [63], giant grouper (Epinephelus lanceolatus) [70], juvenile largemouth bass, and Micropterus salmoides. Subsequently, we observed an increased level of expression of the key enzymes of glycolysis in the control vs. fish fed a 40% CHO diet, while they were down-regulated in the control vs. fish fed a 60% CHO diet. The enzymes associated with gluconeogenesis were found to be down-regulated in both the groups of fish.
The pyruvate kinase (PK) gene is the major key regulatory enzyme of the glycolytic pathway, and studies have reported an increase in its expression profile due to high CHO levels in the diets of fish [79] such as perch (Perca fluviatilis) [80], gilthead sea bream (Sparus aurata) [81], European sea bass [82], gibel carp (Carassius auratusgibelio) and Chinese long snout catfish (Leiocassis longirostris gunther) [83]. The hepatic PK gene expression activities were enhanced significantly in grass carp with an increase in the percentage of CHO levels in the diet [74]. Interestingly, our results showed a neutral expression profile for the PK gene in all the treatments, including the control. In common carp (Cyprinus carpio), pck transcription was strongly inhibited as a result of feeding [84], while g6pc expression was sharply down-regulated after feeding in gilthead sea bream [64,84].

Glycogen Synthesis Enrichment due to High CHO in the Diet
It has been reported that excess glucose available from the diet is stored in the form of glycogen in the liver through the process of glyconeogenesis [56,85,86]. We observed an increase in the glycogen synthase (GYS) and UDP-Pyrophosporylase expression levels, possibly due to induction in the process of glycogenesis in the liver of rohu for glycogen storage. This trend was similarly reported in Megalobrama amblycephala, gilthead sea bream and golden pompano (Trachinotus ovatus) [75,77,87] when fed a diet with increased CHO levels. In tilapia, glycogen accumulation in the liver is induced with the help of the key gene glycogen synthase (GYS) [63].

Impact of High-CHO Diet on Gluconeogenesis and Lipogenesis
Gluconeogenesis is the reverse process of glycolysis, which produces glucose using pyruvate as a substrate. Several studies reported that this process is only activated during fasting challenges in fish for ATP energy production. Excess glucose availability in the liver represses the enzymatic activities of gluconeogenesis. The regulatory enzymes include fructose bisphosphatase (FBPase) and phosphoenolpyruvate carboxylase (PEPCK-C), which mainly control the key steps of gluconeogenesis. A higher level of FBPase gene expression was observed in fish fed with a starch diet, namely, European sea bream and gilthead sea bass [10,82]. A down-regulation of the key enzyme G6Pase's activity in the gluconeogenesis pathway has been observed in Megalobrama amblycephala fed with dextrin [88], and this was up-regulated during fasting in the livers of rainbow trout [89][90][91]. The hepatic gluconeogenesis pathway (PEPCK mRNA expression and activity, and G6Pase activity) in zebrafish, D. rerio adults, was repressed by feeding them with early high CHO after CHO intake [92]. In the juvenile golden pompano (Trachinotus ovatus), the endogenous glycolysis pathways were enhanced, while the gluconeogenesis process was down-regulated by an increase in the dietary CHO level [93].
In the liver, excess amounts of carbohydrates (pyruvate) convert into fatty acid via the de novo lipogenesis pathway [7]. The acetyl-CoA produced through the citrate acid cycle converts into malonyl-CoA using acetyl-CoA carboxylase (ACC), and then palmitic acid is produced by fatty acid synthase (FAS) with NADPH as a precursor. It has been reported that de novo lipid synthesis is modulated by enhancement of the dietary carbohydrate level or increases in the glucose load in the diets of fish [75,[94][95][96][97][98][99][100]. Here, fatty acid synthase (FAS) plays a critical role in the process of de novo lipid synthesis. This study also showed an increase in the expression levels of fatty acid synthase (FAS), acetyl-CoA carboxylase α and malonyl-CoA in fish with higher CHO levels in the diet compared to the control group. In blunt snout bream, Megalobrama amblycephala, the hepatic FAS, acetyl-CoA carboxylase α, delta-6 fatty acyl desaturase, and peroxisome proliferator-activated receptor γ (PPARγ) showed elevated expression levels, which indicates induction of the lipogenesis pathways in the liver due to increases in CHO level in the diet [75]. Recently, a study on the liver of tilapia showed the up-regulation of transcripts such as accα, fas and dgat2 due to the incorporation of high CHO levels in the diet, but hepatic lpl and fatp5 expression showed no significant consequences [98,[101][102][103].
The heightened level of cd36 gene expression was also observed in large yellow croakers fed with high-CHO (starch) and low-lipid diets, which implied that cd36 is crucial for lipid homeostasis via regulation of the PPAR pathway [104]. Elevated levels of hepatic accα, fas and dgat2 gene expression have been demonstrated in rainbow trout (Oncorhynchus mykiss) due to high levels of dietary CHO carbohydrates [105,106]. The induction of lipogenesis was observed in our study, with evidence of associated elevated gene expression, which may not be directly linked with glucose metabolism, but the enriched pathways implicated their role in glucose homeostasis via converting excess glucose into fatty acids.

Immune-Associated and Stress-Related Transcriptional Changes in the Liver due to the Inclusion of a High-CHO Diet
It was plausible to enhance the stress response and immunity-related gene modulation in fish fed with a formulated diet [107]. It has been reported that the cortisol level increased in M. amblycephala and Erythroculterilishaeformis bleeker [3] due to the high CHO level in the diet, which indicated stressful conditions for the fish. Similarly, plasma lysozyme activity decreased in several fish, with an increase in dietary CHO, of the species Epinephelus malabaricus and M. Amblycephala [108]. The down-regulation of constitutive immune genes was observed in the liver of Atlantic salmon under starvation using a transcriptome study [107]. At the molecular level, heat shock proteins (HSPs) are stress indicators that are involved in stress response, cell protection and improved tolerance level [109]. In rainbow trout, higher levels of CHO increased the HSP70 gene expression [73] and levels of hepatic HSP70 in Wuchang bream (M. amblycephala) fed with 53% CHO compared to control [110]. In contrast to this, we could not observe any differential expression pattern of HSP70 among control and treated rohu fish. The HSP70 transcript was detected only in fish fed with the 40% and 60% CHO-level diets. Interestingly, fish fed with high percentages of CHO (40% and 60%) showed an increase in HSP90 and associated transcription factors, which act as chaperones and folding catalysts, such as hsp90b1, dnajb1b, hspa4a, hspb7/8, etc. Specifically, we detected higher expression levels of Activator of HSP90 ATPase homolog 1, DnaJ (HSP40) homolog and HSP60 in the fish fed a 60% CHO diet compared to the control group counterpart, which indicated the emergence of stress in liver cells.
We detected increased expressions of cytokine receptors such as IL-6 and suppressor of cytokine signaling 3 (SOCS3) in the livers of fish fed with high CHO as compared to the control. This is in line with earlier studies, in which the augmented levels of the expressions of IL-6 and SOCS3 resulted in chronic stress, as well as impacteing the glucose transport system of Japanese flounder (Paralichthys olivaceus) fed with high dietary CHO [68]. Recent studies revealed that immune-associated transcriptional changes in the liver of Epinephelus akaara occurred due to a high level (30%) of carbohydrates in their diet, which in turn resulted in the over-expression of immune-related genes such as CXCR4, CCL4, IL8, TLR9 and NFκB inhibitor alpha, etc. [111]. In this study, the genes associated with the chemokine signaling pathway, such as C-C motif chemokine 4 (CCL4), C-X-C chemokine receptor type 4 (CXCR4) and C-X-C motif chemokine 10 (CXCL10), were significantly expressed in the group fed a high-carbohydrate diet.
A high CHO level can lead to the over-expression of immune-related genes, which may activate acute inflammatory processes in the liver. Toll-like receptors (TLR), tumor necrosis factor (TNF), interferon regulatory factor 3 (IRF3) and tumor necrosis factor receptor (TNFR) belonging to the Toll-like receptor signaling pathway showed differential levels of expression patterns in fish fed with different levels of CHO in their diet. TLR3, TLR5, IRF3, IRF5 and IRF5 were up-regulated in the livers of fish fed with high CHO levels compared to the control group. This is in line with earlier reports, which demonstrated that modulation of the liver transcripts was associated with the Toll-like receptor signaling pathway in fish fed a high-carbohydrate diet [111]. Interestingly, some of the TLR isoforms, such as TLR22, TLR1 and TRAF3, were found to be down-regulated in the livers of fish with high CHO levels in their diet. Subsequently, our analysis of fish fed with high levels of CHO reported the up-regulation of transcripts such as Death-domain-associated protein, Caspase-6, Lamin B3, Cathepsin Z, DAB2-interacting protein b, and Membrane protein, palmitoylated 5b (MAGUK p55 subfamily member 5), which are associated with the apoptosis pathway, possibly inducing hepatic cell death through the extrinsic pathway of programmed cell death (PCD) via the modulation of the p53 signaling pathway. The incorporation of dietary carbohydrates may lead to stress as well as lower immunity through the innate and adaptive immune system. Overall, the up-regulation of a large number of immunity-and stress-related genes in the immune system pathway revealed that the rohu fish's liver may experience an inflammatory response. The stress in the hepatic cells occurred due to the excess levels of glucose, via modulation of the isoform of the enzymes. Overall alteration in metabolic pathways may occur in liver tissues due to the effects of temperature or oxygen levels on the metabolism of fish, which is a vital area of research and is lacking in the present work.
3.6. Insulin Signaling Pathways, mTOR, and the PPAR Signaling Pathway The insulin signaling pathways are mainly involved in maintaining glucose homeostasis via the uptake of glucose for substrate energy generation, as well as excess glucose storage in the liver (in the form of glycogen via glycogenesis) and the induction of de novo lipid biosynthesis. After the consumption of carbohydrates, insulin promotes the entry of blood glucose using glucose transporters via modulating various series of genes, including GTs, insulin receptors, IRS and hexokinase, and activates the IRS-PI3k-Akt pathways. In our study, we observed active insulin signaling pathways in the livers of fish fed a high-CHO diet, which is in line with earlier reports (Figure 7). In this pathway, INS actions are mediated via insulin receptors (receptor tyrosine kinase family), and then by further modulating pathways [112]. High CHO levels in the diet result in hyperglycemia in rainbow trout, which activates the components of the insulin-signaling cascade. The modulation of the insulin signaling pathway (Akt/TOR) and a postprandial expression pattern of the genes related to metabolism were observed in rainbow trout fed with different carbohydrate to protein dietary ratios [78].
We observed increased levels of expression of key genes of the insulin signaling cascade, such as phosphatidylinositol 3-kinase (PI3K), which subsequently phosphorylates other proteins such as AKT, also known as protein kinase B. This cascade (IRS-PI3K-AKT) further modulates the mTOR pathway, which activates other cascade molecules regulation, such as ribosomal S6 kinase protein (S6K1), and this leads to an increase in metabolism and cell growth. In our study, heightened expression levels of insulin receptor substrate (IRS) were observed. We highlighted the up-and/or down-regulation of important genes involved in the insulin signaling pathway. Our study implied that the up-regulation of GLUT2 mediated glucose uptake in the livers of fish. Earlier studies also indicated the modulation of Akt for insulin signaling, and S6k for mTOR, which regulates the metabolic pathways in the liver [69]. In contrast, in our study, the Akt and S6k expressions were not significant in fish fed high-CHO diets. Recently, researchers have depicted the up-regulation of S6K1 and IRS-1 in the livers of blunt snout bream due to an increase in arginine level (2.70%), which indicated a controlled expression of insulin [113]. In this study, we found a neutral expression of IRS and PI3K in the fish fed a high-CHO diet, and in control fish.
Lipid homeostasis is maintained through key enzymes under the control of PPARα (regulation of fatty acid catabolism) and PPARγ (regulation of fatty acid storage), which belong to the nuclear receptor super family. We observed increased levels of expression of PPAR enzymes, which enhances the lipid biosynthesis mechanism. Researchers have suggested that treatment with bovine insulin mediated the up-regulation of PPARγ and PPARα genes in yellow catfish [114,115]. In tilapia, the modulation of peroxisome proliferator-activated receptor (PPAR) and the insulin signaling pathways was observed for de novo lipogenesis due to high dietary CHO levels [101]. The AKT-mTOR pathway also represents an important pathway during metabolism, as reported in Senegalese sole fed with different dietary protein to carbohydrate ratios [116,117]. It has been reported that the de novo lipogenesis mechanism was enhanced due to the modulation of genes such as sterol regulatory element-binding protein 1c (SREBP1c) and TORC1 [118,119], and that transcripts linked with lipogenesis and SREBP1 were up-regulated via the mTOR-dependent pathway. In this study, the FAS, ACCα and PPARγ genes associated with de novo lipogenesis were found to be up-regulated significantly in the livers of fish fed a high CHO diet compared to control fish, but we could not detect transcripts for the sterol responsive element binding protein (SREBP1) in the livers of either control or treated fish. This clearly indicates that the inclusion of high CHO levels in the diet can promote de novo fatty acid biosynthesis and fat accumulation in the livers of fish via activation of the mTOR pathway.

Experimental Diet Preparation and Feeding Experiment
The experimental feed was formulated in our laboratory and contained three different levels (20%, 40%, and 60%) of dietary carbohydrates (gelatinized starch) ( Table 4). The 20% carbohydrate diet was considered as the control and the 40% and 60% carbohydrate levels were designated as high-carbohydrate diets (CHO). Fish meal and casein served as protein sources and corn starch was used as the main source of carbohydrates. The corn starch was gelatinized with the addition of water in an autoclave at 120 • C for 1 hr. All the ingredients were weighed and mixed properly with the addition of oil and water. The prepared dough was pelleted using a hand pelletizer and dried in hot air over night at 60-80 • C. The dried pellets were broken and sized using a grinder into pieces with a 2-3 mm diameter and stored in polyethylene bags at room temperature (30 ± 2 • C) for further use. The experimental fish (40-50 g) were collected from nursery ponds of the Division of Fish Genetics and Biotechnology, Central Institute of Freshwater Aquaculture, Bhubaneswar, Odisha, India. All the experimental procedures were performed with the approval of the Institute Bio-Safety Committee (IBSC) on Ethics. The procured fish were acclimatized in the circular tanks (1000 L) of the wet-laboratory for a period of one week. After acclimatization, the fish were randomly distributed in experimental tanks (500 L) with 10 fish per tank. The experiment was performed in triplicate and the respective diets were fed twice at 2-3% body weight. The water was partially changed and refilled with fresh water. During the feeding trial, the experimental fish were reared with optimum water quality parameters, with water temperature 27-31 • C, pH 7.3-7.8 and dissolved oxygen 5.0-6.0 mg/L, measured as per standard procedure (APHA 1985). After 45 days of the experiment, liver tissue samples were collected from three experimental fish per each treatment (total 12 fish from each treatment). The fish were anesthetized in a water tank containing 100 mg/L of MS-222 (tricainemethanesulfonate, Sigma, Milwaukee, WA, USA). The liver tissue samples were immediately collected and frozen in liquid nitrogen (LN2) and subsequently stored at −80 • C for further processing for RNA isolation.

RNA Isolation and cDNA Library Preparation, NGS Sequencing
Total RNA from liver tissues was isolated using an RNAiso kit (TakaRa, Clontech, Mountain View, CA, USA) according to the manufacturer's instructions. The three fish livers from each treatment tank were pooled for total RNA extraction. The concentrations of total RNA were checked using a Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA, USA). The quality of RNA was assessed using the Agilent2100 Bioanalyzer system (Agilent Technologies, Santa Cruz, CA, USA) and those samples with an RNA Integrity Number (RIN) > 7.0 were utilized for Illumina RNA library preparation. Nine cDNA Illumina libraries were generated from the total RNAs of liver samples. RNA was fragmented using nebulization techniques using the NEB Next First Strand Synthesis Reaction Buffer (5 ×), followed by first strand cDNA synthesis with the help of hexamer primers and M-MuLV Reverse Transcriptase, then subsequently synthesized into a second strand of cDNA. The Illumina adaptors were used for the preparation of the library and purified using Agencourt Ampure XP SPRI beads, then further enriched using PCR amplification for cluster generation. The cDNA libraries were generated using the NEBNext Ultra directional RNA Library Preparation kit (Illumina) as per the manufacturer's protocol and the quality of the library was assessed using the Agilent 2100 Bioanalyzer system (Agilent Technologies, Santa Cruz, CA, USA). Then, all the treatment and control group libraries were pooled for sequencing. The adaptor-ligated libraries of rohu liver samples processed for sequencing using an Illumina Hiseq 500 platform and paired-end reads (150 bp) were generated. The generated raw reads of rohu liver samples were submitted through the NCBI SRA, and are accessible via the NCBI BioProject accession.

Quality Check, Assembly and Mapping
The generated raw reads were further checked for quality using FastQC software (http://www. bioinformatics.babraham.ac.uk/projects/fastqc). Initially, the raw reads were pre-processed to remove the adapter sequences and the low-quality bases (<q30). The Cutadapt2 algorithm was employed to remove reads with adapters using Python scripts. The quality check and filtered reads were further subjected to HISAT software for alignment, and the clean reads were aligned to the reference genome (PRJNA437789) with default parameter sets. Based on that, reads were classified into aligned reads (aligned to the reference genome) and unaligned reads. The cleaned aligned reads were mapped to multiple reference genes.

Gene Regulatory Network
Cytoscape (version 3.7.2) was used for the analysis of the gene regulatory network between the 20% (control), 40% and 60% (carbohydrate regime) groups. Carbohydrates related to up-and down-regulated genes were selected for network construction. Algorithms like ARACNE (algorithm for the reconstruction of accurate cellular networks) and the Network Analyzer plug-in were used for analyzing the gene regulatory network of all three sets of DEGs. On the basis of high degree and betweenness, hub genes were selected. The genes within a network with a higher connectivity played a central regulatory role, and were important in detecting highly connected genes in a network.

Differential Expression Analysis of Transcripts, Gene Ontology and KEGG Pathway
Here, the transcripts were annotated using a homology approach to assign functional annotation with BLAST2GO. Transcripts were assigned with a homolog protein from other organisms if the match was found with an E-value 1.0 × 10 −5 and minimum similarity greater than 30%. HTSeq v0.6.1 was utilized to measure the number of clean reads mapped to each gene, and estimate or calculate their abundance. The transcripts were normalized using the standard method-the number of count genes divided by the number of FPKM (fragments per kilo base per million reads) of transcripts for each sample. The absolute counts for each transcript were identified and were used in differential expression calculations. In order to check the differential expressions of those transcripts in fish fed a high-carbohydrate diet (40% and 60% CHO) compared to control groups (20%), DESeq was used to calculate the differentially expressed transcripts, and transcripts were categorized into up-, down-and neutral-regulated based on the log2fold change cut-off of 1 value. The transcripts/genes recognized by DESeq with an adjusted P-value < 0.05 were categorized as differentially expressed genes (DEGs). The P-values of differentially expressed transcripts were adjusted using the Benjamini-Hochberg method to control the false discovery rate. Here, the difference in the expression levels of transcripts was estimated between the control group and fish fed a high-carbohydrate diet.
The functional annotations were performed by homology searching against the public databases, such as the non-redundant nucleotide (Nt) and non-redundant protein (Nr) database of NCBI (http: //www.ncbi.nlm.nih.gov/), Swiss-Prot (http://www.ebi.ac.uk/uniprot/), clusters of orthologous groups (COG, http://www.ncbi.nlm.nih.gov/COG/), Gene Ontology (GO, http://www.geneontology.org/), and the Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/). The results obtained through BLAST2GO were further subjected to WEGO software for classification of Gene Ontology (GO) categories such as biological processes, molecular functions, and cellular components (http://wego.genomics.org.cn/cgi-bin/wego/index.pl). The GO and KEGG analyses were carried out to find the differential expressions of transcripts/genes involved in various enriched metabolic pathways. The GO enrichment analysis was performed using the R-package (Bioconductor) software EdgeR (Empirical Analysis of Digital Gene Expression in R, http://www.bioconductor.org/packages/2.12/bioc/ html/edgeR.html). The categorized GO terms were ranked as per their corrected P-values < 0.05. The metabolism-associated KEGG pathway analysis was carried out using online KEGG Automatic Annotation Server (KAAS) software (http://www.genome.jp/tools/kaas/). The Bi-directional Best Hit (BBH) method was employed in KEGG pathway identification. Due to the absence of an annotated genome of rohu in the database, we used Zebrafish (Danio rerio) as a reference organism for pathway identification. The pathways per transcript were mapped to differentially expressed transcripts. The relevant metabolism associated pathways were significantly enriched based on the false discovery rate (FDR) ≤ 0.05 of regulated genes. The COG database (clusters of orthologous groups) was used to detect the putative function of transcripts based on known or conserved orthologous genes or their products.

Quantitative Real-Time PCR (qPCR)
In order to validate the generated liver transcriptome data, we selected six significant transcripts/genes associated with carbohydrate metabolism of rohu for performing quantitative real-time PCR (qPCR) analysis. The primers were designed using FastPCR (version 5.2.113) software for the selected transcripts and one housekeeping gene (β actin). The cDNA of the total RNA pool of the liver tissues of fish was generated using a BluePrint TM first Strand cDNA Synthesis kit (TakaRa, Clontech, Mountain View, CA, USA). The primer sequences were found in the total RNA from the liver tissues (in triplicate from each treatment) of the control fish and fish treated with high CHO levels, extracted using the RNAiso kit (TakaRa) according to the manufacturer's instructions. The RNA samples were taken as three technical replicates for performing qPCR. The quantitative PCR was carried out using a Light Cycler-480 SYBR Green I kit (Roche Diagnostics, Mannhein, Germany) as per the manufacturer's instructions. The PCR mixture consisted of 20 µL, which included a 2 µL DNA template, 0.4 µL each of forward primer and reverse primer, 10 µL SYBR ® Premix and 0.4 µL ROX Reference Dye (50×), and the total volume was made up with RNase-free water. The qPCR analyses were performed with the following conditions: initial denaturation at 95 • C for 30 s, then followed by 35-40 cycles of denaturation at 95 • C for 10 s, annealing at 60 • C for 30 s, and extension at 72 • C for 30 s. The difference between the cycle threshold (ct) value of selected genes and that of the internal control housekeeping gene was calculated and then the relative gene expression was measured using the 2 −∆∆Ct method. The box plot was used to analyze the variation distribution in the normalized count data. A Pearson's correlation coefficient analysis was carried out to check the co-linearity of count data among control and treated fish. The significant difference between control and high-CHO-fed fish was analyzed using a Student's paired t test and a one-way ANOVA test (significance value; P ≤ 0.05).

Conclusions
In order to understand the regulatory mechanisms of carbohydrate metabolism in the liver, Jayanti rohu fish fed a high-carbohydrate/glucose diet were analyzed using a series of key genes/transcripts associated with hepatic glucose transport and assimilation. The liver is a major organ in fish, and it indicates the nutritional and physiological status of fish. Hepatic genes were modulated due to the increased dietary carbohydrate levels in the livers of fish fed with a higher proportion of CHO in the diet. The present study suggested that the inclusion of high levels of CHO in the diet promotes glucose transport activities, glycolysis, glycogenesis, lipogenesis, and the pentose phosphate pathway. The transcripts associated with glycogen synthesis, such as hexokinase, glycogen synthase, PHK and PGK, were found to be up-regulated in the livers of fish fed with high levels of CHO as compared to the control, indicating the enrichment of the glycogenesis mechanism in the liver. In addition to this, the increased levels of expression of ACCα, PPARγ and FAS in the livers of fish fed with high levels of CHO indicated the activation of de novo fatty acid biosynthesis. Altogether, we have highlighted the enhanced glycolytic process as facilitated by the uptake of glucose and further activated or mediated by the insulin signaling pathway. We also identified enrichment of the insulin signaling pathways and the peroxisome proliferator-activated receptor (PPAR) pathway for de novo lipid biosynthesis. This study emphasized the fine balance of the ratio of dietary carbohydrates to protein in fish, and that the best available energy sources should be used. This work also encourages the investigation of the impact of temperature or oxygen levels on the metabolic pathways of rohu. Overall, this work suggests the incorporation of a higher level of digestible carbohydrate (gelatinized starch) in the diets of rohu, and encourages the undertaking of nutritional programming research in carp aquaculture.