Recent advances in amino acid sensing and new challenges for protein nutrition in aquaculture

The research on the nutrient status of protein, one of three macronutrients, has been ongoing for several decades. In particular, aspects of protein/amino acid digestion, absorption, transportation, and metabolism have been well studied (Albanese 1959; Dardevet 2016). All organisms including aquatic animals have the capacity to sense the availability of the protein/amino acid required to synthesize functional enzymes, generate energy, and construct fundamental cellular structure. Recent studies on the molecular regulation of amino acid sensing has expanded our understanding of how protein/amino acid nutrition is precisely controlled (Chantranupong et al. 2014; Dardevet 2016; Rebsamen et al. 2015; Wang et al. 2015; Wolfson et al. 2016, 2017). In this review, we first illustrate two amino acid-sensing pathways conserved in most eukaryotes, TORC1 and the GCN2 signaling pathway. Then, we highlight three newly discovered amino acid sensors. Furthermore, we summarize the pioneering work on protein/amino acid nutrition in farmed aquatic animals.

As fundamental elements for the growth of all organisms, the input of protein/amino acid initiated a series of subcellular responses. However, the mechanisms by which amino acids regulate organismal growth were unclear until a few decades ago. In 1975, Vezina et al. (1975) purified an anti-fungal compound called rapamycin from bacteria. Sixteen years later, Heitman et al. (1991) identified two molecular targets of rapamycin-FKBP12 (FK506-binding protein) complex, TOR1 and TOR2, by screening rapamycin-resistant yeast mutants (Heitman et al. 1991). TOR is a highly conserved serine/threonine protein kinase that belongs to the phosphatidylinositol 3-kinase (PI3K)-related kinase (PIKK) family. TOR usually interacts with other proteins to form two distinct protein complexes, TOR complex 1 (TORC1) and TOR complex 2 (TORC2), to regulate downstream effectors (Sabatini 2017). Among these two complexes, TORC1 has been approved as the regulatory center of cellular growth and metabolism by nutrients and nutrient-related signals, such as insulin and IGFs (insulin-like growth factors).

TORC1 integrated a series of upstream signaling including growth factors, amino acids, energy, oxygen, and DNA damage (Chantranupong et al. 2015). Insulin and IGFs are the two major growth factors, which trigger the signaling cascades to activate TORC1 (Chantranupong et al. 2015). After binding to the cooperating receptors, insulin and IGFs induced the phosphorylation of Akt through PI3K (phosphoinositide 3-kinases). The full activation of Akt requires phosphorylation of Akt at both Ser473 and Thr308 sites, which are highly conserved across different organisms (Oldham and Hafen 2003). Whereas the Ser473 site is phosphorylated by PI3K-PDK1 (Pyruvate Dehydrogenase Kinase 1), the phosphorylation of Thr308 is mediated by TORC2 (Senoo et al. 2019; Yang et al. 2015). Akt also is serine/threonine kinase and a central integral signaling node for a number of signaling pathways. One of its substrates is TSC (Tuberous Sclerosis Complex), which has been approved as a TORC1 inhibitor. The activation of TORC1 requires small GTPase Rheb (Ras homolog enriched in brain) (Long et al. 2005; Sancak et al. 2007), whereas TSC functions as a GAP (GTPase activating protein) for Rheb, promotes hydrolysis of Rheb-GTP, and blocks its function (Inoki et al. 2003; Tee et al. 2003). Akt phosphorylated TSC2 occurs at multisites and is dissociated from lysosome membranes where Rheb recruits and activates TORC1 (Fig. 1) (Dibble and Cantley 2015).

Figure 1.  The signaling pathways upstream of TORC1. Growth factors including insulin and IGF induce Akt phosphorylation, which in turn inhibits TSC1/2 complex and activates Rheb. Rheb recruits TORC1 to the surface of lysosome, where amino acids and growth factors cooperate to ensure proper TORC1 activation

Two well-characterized molecular substrates of TORC1 are S6K (Ribosomal protein S6 kinase) and 4EBP1 (Eukaryotic translation initiation factor 4E (eIF4E)-binding protein 1), both of which are key mediators during protein synthesis process. The phosphorylation of S6K by TORC1 phosphorylates the cap binding complex component eIF4B at S422, promotes the recruitment of eIF4B to eukaryotic initiation factor 4A (eIF4A) at the translation initiation complex, and improves protein translation (Raught et al. 2004). 4EBP1 is a translation initiation inhibitor by binding and inactivating the eIF4E. TORC1 phosphorylates 4EBP1 at multi-sites including T37, T46, S65, and T70. The phosphorylation of 4EBP1 releases eIF4E from 4EBP1 and improves the eIF4E dependent translation initiation (Heesom and Denton 1999).

Besides improving protein synthesis, TORC1 inhibits protein catabolism by suppressing autophagy to promote cell growth (Noda and Ohsumi 1998). Autophagy is a survival mechanism that cells self-degrade "unnecessary" or dysfunctional components to increase the availability of amino acids for pro-survival processes. Unc-51 like autophagy activating kinase (ULK1) is important for the initiation of autophagy. TORC1 suppresses autophagy by binding and phosphorylating ULK1 at inhibitory sites under nutrient replete conditions (Chan 2009).

Eukaryotes, from yeast to mammals, live in a complex environment and face many stress including food shortages. Thus, the organisms have evolved several components and pathways to sense intracellular nutrients. For example, GPR40 (G-protein-coupled receptor 40) and GPR120 could sense free fatty acids (Itoh et al. 2003). GLUT2 (SLC2A2) is a glucose transporter, it senses the extracellular glucose levels (Thorens and Mueckler 2010). Also, a recent study found glucose sensing by AMP-responsive protein kinase (AMPK) requires TRPV (transient receptor potential V) channels (Li et al. 2019). For monitoring amino acid availability, GCN2 is playing the critical sensory role.

During the protein synthesis process, tRNAs are loaded with their cognate AAs by amino acid-specific aminoacyl tRNA synthetases (aaRS) (Ibba and Soll 2000). Under conditions of amino acid limitation, uncharged tRNAs are accumulated. S. cerevisiae initiated a process termed general amino acid control to acclimatize this condition (Hinnebusch 1988, 2005; Wek et al. 1995).

Since each amino acid cannot substitute another one during protein synthesis process, (especially for the essential amino acids which cannot be synthesized by animals themselves) the cell must be able to efficiently detect the shortage of any amino acids in order to shut down peptide chain synthesis and switch on some amino acid biosynthetic processes. The protein GCN2 has high affinity to all uncharged tRNAs (Dong et al. 2000) regardless of its amino acid specificity. The structure reveals that GCN2 is under auto-inhibition status resulting from stabilization of a closed conformation that restricts ATP binding when amino acids are available (Padyana et al. 2005). After binding with uncharged tRNAs, the conformation of GCN2 is changed, and the kinase domain is exposed (Padyana et al. 2005). One important substrate of GCN2 is eukaryotic initiator factor 2 alpha (eIF2α), a key early activator of translation initiation (Berlanga et al. 1999). Although most mRNAs are translationally repressed upon the phosphorylation of eIF2α, mRNAs containing a short upstream open reading frame (uORF) located in the 5′ untranslated region (5′ UTR) were translated by a re-initiation mechanism or internal ribosome entry sites (IRES) (Chan et al. 2013; Rzymski et al. 2010). ATF4, for example, is one of the key response factor of GCN2 activation (Karpinski et al. 1992). After upregulation by GCN2, ATF4 induces the transcription of a cascade of genes involved in amino acid synthesis and transport, intermediary metabolism, and energy metabolism (Fig. 2) (B'chir et al. 2013; Bunpo et al. 2009; Harding et al. 2000; Krokowski et al. 2013).

Figure 2.  The signaling pathway of GCN2 and its cross-talk with TORC1. The availability of amino acids controls cell growth and metabolism through GCN2 and TORC1 signaling pathway. In amino acid deficiency status, uncharged tRNA activates GCN2, eIF2a, and ATF4 to initiate the amino acid response. When the amino acids are available, TORC1 signaling pathway is activated through amino acid sensors and negates the effect of GCN2

Most of the physiology functions of GCN2 is revealed by GCN2-deficient rodent models. Knock out Gcn2 in mice exhibit normal phenotype under regular feeding condition. However, if the Gcn2^-/- female mice were fed with amino acid-deficient diets, most notably leucine deprivation, the prenatal and neonatal mortalities of their offspring are relatively higher (Zhang et al. 2002). Dietary leucine restriction resulted in loss of body weight and liver mass in wild-type mice. In contrast, knockout of Gcn2 significantly reduced the survival rate when mice fed a leucine-deficient diet (Anthony et al. 2004). Furthermore GCN2 controls animal feeding behavior.

Rodents could recognize the unbalanced food and reduce the intake of it. However, this rejection behavior of unbalanced food was impaired in Gcn2^-/- mice (Maurin et al. 2005). In the nutrition restriction or hypoxia conditions, one adaptive response is the formation of new blood vessels, which is termed as angiogenesis. A recent study found that the GCN2/ATF4 pathway regulates VEGF expression and angiogenesis upon SAA (sulfur amino acid) restriction in endothelial cells, and this process is independent of hypoxia or HIF1α (hypoxia-induced factor 1α) (Longchamp et al. 2018). In addition to sensing signal amino acid deprivation, TORC1 and GCN2 respond to different dietary protein sources. For example, when feeding the carnivorous turbot with plant protein, the TORC1 signaling cascade is decreased (Xu et al. 2016). It is a reasonable hypothesis that organisms use TORC1 and GCN2 to sense amino acid profiles of different dietary protein sources in order to maximize the protein nutrition.

The TORC1 complex consists of TOR, regulatory-associated protein of TOR (Raptor), and regulatory-associated protein of TOR (Raptor) (Sengupta et al. 2010). In addition to these components, the activation of TORC1 by amino acids involves a series regulatory proteins, including Rag GTPases, Ragulator complex, v-ATPase, Folliculin/FNIP1/2 complex, GATOR1/2 complex, and KICSTOR complex (Wolfson et al. 2017). Rag GTPases consist of a constitutive heterodimer of RagA/B and RagC/D, and activated Rag GTPases recruit TORC1 translocation to the lysosomal surface, where it is phosphorylated and activated by Rheb (Kim et al. 2008; Sancak et al. 2008). Rag GTPases are in active conformation when the RagA/B is loaded with GTP and RagC/D is loaded with GDP. Folliculin/FNIP1/2 complex is RagC/D GAP and improves Rag GTPases function (Tsun et al. 2013). The GATOR1 complex functions as RagA/B GAP and inhibits Rag GTPases function, thus inhibiting TORC1 activity. GATOR2 promotes TORC1 activity by negatively regulating GATOR1 (Bar-Peled et al. 2013; Panchaud et al. 2013). The localization of GATOR1 on the lysosomal surface requires a scaffold complex consisting of KPTN, ITFG2, C12orf66, and SZT2-containing regulator of TORC1 (KICSTOR) (Peng et al. 2017; Wolfson et al. 2017). However, among these newly revealed TORC1 regulators, none of the proteins or protein complexes bind to amino acids directly. Recently, the discoveries of SLC38A9, Sestrins, CASTOR1, SAMTOR, Transmembrane 4 L Six Family Member 5 (TM4SF5), and leucyl-tRNA synthetase (LARS) represent the dawn of the age of amino acid sensors (Fig. 3) (Gu et al. 2017; Han et al. 2012; Jung et al. 2019; Wolfson et al. 2017; Zheng et al. 2016).

Figure 3.  Schematic representation of the key amino acid sensing players upstream of TORC1. Rag GTPases activate TORC1 when RagA/B is loaded with GTP, and its activity is negatively regulated by GATOR1. GATOR2 activates TORC1 signaling by inhibiting GATOR1. The function of GATOR2 is inhibited by CASTOR1 (arginine sensor) and Sestrin2 (leucine sensor). The binding of amino acids removes these sensors' inhibitory function on TORC1 signaling. SLC38A9 senses lysosome arginine level and promotes the activity of RagA/B. TM4SF5, LARS, and SAMTOR are cytosolic arginine, leucine, and S-adenosylmethionine sensors, respectively

In 2015, three independent groups identified that SLC38A9 exerts a key role in amino acid induced activation of TORC1 (Jung et al. 2015; Rebsamen et al. 2015; Wang et al. 2015). Before these discoveries, the function of SLC38A9 was unknown. SLC38A9, which is a member of the solute carrier family 38 (SLC38) of sodium-coupled amino acid transporters, is a lysosomal 11-transmembrane segment protein. The full activation of TORC1 requires the binding of SLC38A9 to the Rag GTPase-Ragulator complex (Jung et al. 2015; Rebsamen et al. 2015; Wang et al. 2015). SLC38A9 transports arginine with a high K_m, but further study has revealed that SLC38A9 mediates the transport of many essential amino acids out of lysosomes in an arginine-regulated fashion (Wang et al. 2015; Wyant et al. 2017). Overexpression of SLC38A9 cytosolic N-terminal region of 119 amino acids activates TORC1 signaling even in amino acid deprivation conditions indicating that this region is the functional Ragulator-binding domain. Its activity is controlled by other regions, which may be sensitive to the amino acids' concentration (Wang et al. 2015). Arginine-induced TORC1 activation is repressed by loss of SLC38A9 (Wang et al. 2015). Mutagenesis studies found 2 amino acid residues are critical for SLC38A9's function. The I68A mutant does not bind Rag-Ragulator but still transports arginine similarly to the wild-type. In contrast, T133W mutant has lost the function of arginine transportation, but is still able to traffic to lysosomes or associate with Ragulator or Rag GTPase (Wang et al. 2015; Wyant et al. 2017). Besides transporting amino acids, SLC38A9 functions as the guanine exchange factor (GEF) to convert RagA from GDP- to the GTP-loaded state in an arginine binding manner. Therefore, this activates the Rag GTPase heterodimer (Shen and Sabatini 2018). Interestingly, the TORC1 bound Rag GTPases cannot interact with SLC38A9 indicating that the activated Rag GTPases are released from SLC38A9 via an unknown mechanism to activate TORC1 (Shen and Sabatini 2018). The crystal structure of SLC38A9 provides detailed information of how SLC38A9 senses arginine. By using the zebrafish homolog-drSLC38A9, Lei et al. (2018) found that arginine interacted with Thr117, Met119, Thr121 and Ser122 from the transmembrane helix 1a (TM1a).

Besides SLC38A9, there are other arginine sensors, i.e. CASTOR1 and TM4SF5. CASTOR1 has been identified from the GATOR2 interacted protein database, which was generated using IP/MS analyses (Chantranupong et al. 2016). Whereas SCL38A9 is a lysosome located transmembrane protein, CASTOR1 is a cytosolic arginine sensor. Crystal structure of arginine-bound CASTOR1 indicates that arginine binds to the homodimeric CASTOR1 at the interface of two ACT (Aspartate kinase, Chorismate mutase, TyrA) domains. This binding causes an allosteric conformational change of the adjacent GATOR2-binding site and releases GATOR2 from CASTOR1. The released GATOR2 activates TORC1 by inhibiting GATOR1 activity (Saxton et al. 2016b). TM4SF5 has four transmembrane domains and is related to the tetraspanins but with N‐glycosylation modifications (Wright et al. 2000). TM4SF5 directly binds free _L-arginine via its extracellular loop, which possibly mediated the efflux of arginine. Upon arginine sufficiency, TM4SF5 translocates from plasma membrane to lysosome membrane, where it interacts with TOR and the SLC38A9 in an arginine-regulated manner. This interaction further causes arginine efflux and actives TORC1 (Jung et al. 2019).

SLAC38A9 and CASTOR1 are two newly characterized amino acid sensors, and no functional studies were reported until quite recently. However, sestrins were identified decades ago with controversial functions, including antioxidant capacity and TORC1 activity regulation (Buckbinder et al. 1994; Budanov et al. 2004; Saxton et al. 2016a, b). Sestrin was first identified as a stress-responsive gene, and its expression was regulated by multiply stresses, including prolonged hypoxia, oxidative stress, and DNA damage (Budanov et al. 2002; Ho et al. 2016). It has been reported that Sestrin1 and Sestrin2 are p53 target genes. Sestrins phosphorylate AMPK results in the phosphorylation of TSC2, thereby blocking the activity of TORC1 (Budanov and Karin 2008). Also, metabolic and stress inputs induce drosophila Sestrin to inhibit chronic TORC1 activation (Lee et al. 2010). Recently, several studies found Sestrin inhibits TORC1 activity through the GATOR complex (Chantranupong et al. 2014; Kim et al. 2015; Parmigiani et al. 2014; Peng et al. 2014). Now, Sestrin has been characterized as a cytosolic leucine sensor (Wolfson et al. 2016). Sestrin2 binds leucine with an affinity of ~ 20 μM, which correlates with leucine concentration in the media that is sensed by the TORC1 signaling pathway (Wolfson et al. 2016). When bound with leucine, Sestrin2 is dissociated from GATOR2, which in turn inhibits GATOR1 and activates TORC1. When the leucine is deprived, Sestrin2 binds GATOR2 and inhibits GATOR2-mediated TORC1 activation (Wolfson et al. 2016). The crystal structure of Sestrin2 in complex with leucine has been revealed. Moreover, Sestrin2 was found to have a leucine-binding pocket, which is necessary for proper structural folding (Saxton et al. 2016a, b). However, the apo-structure of the leucine sensor Sestrin2 is still elusive (Saxton et al. 2016a, b).

The leucine-binding pocket of Sestrin2 is crucial for the conformation change of Sestrin2 (Saxton et al. 2016a, b). However, more studies are still required to reveal the apo-structure of the leucine sensor Sestrin2 (Saxton et al. 2016a, b).

LARS is another intracellular leucine sensor (Han et al. 2012). There are several amino acid residues of LARS which are important for leucine binding. Mutation of these amino acids blocks the sensitivity of TORC1 to intracellular amino acids. Similarly to SLC38A9, LARS activates the Rag GTPase heterodimer by converting RagA from GDP- to the GTP-loaded state (Han et al. 2012). Unlike leucine and arginine, which are directly bound with their sensors, methionine is sensed indirectly through S-adenosylmethionine (SAM). SAM binds directly to SAMTOR to trigger dissociation of SAMTOR from GATOR1, thus signaling methionine sufficiency to TORC1 (Gu et al. 2017).

The newly identified members of TORC1 expand our understanding of how TORC1 functions as the key component for multi-cellular responses. With the discoveries of these sensors of TORC1, we are able to glimpse how this complex is regulated by nutrients. However, only arginine, leucine and S-adenosylmethionine sensors of TORC1 have been elucidated so far. The question of whether or not there are other amino acid sensors and the nature of their roles remains to be answered.

One of the major protein sources for human is seafood. Fish accounts for over 40% of animal protein intake in some developing countries (Quaas et al. 2016). According to FAO, the total fish production peaked in 2016 at 171 million tonnes globally, and 47% was contributed by aquaculture. This emphasizes the importance of aquaculture in food security in many regions of the world (FAO 2018). However, as a protein provider for human, aquaculture is facing a protein-intake problem because of the restricted supply and increasing price of its dietary protein source, i.e. fishmeal (FM).

In relatively advanced vertebrates, the topic of protein nutrition research is focused on the mechanism of metabolism regulation and amino acid sensing. In comparison, the research of protein nutrition in farmed aquatic animals covers several topics, including the precise protein requirements, the nature of the indispensable amino acid (IAA, also called essential amino acid, EAA) requirements, amino acid influx and transportation, and protein deposition (Kaushik and Seiliez 2010). However, for amino acid sensing only a comparatively few studies have been reported (Conde-Sieira and Soengas 2017).

Over the past three decades, a number of experiments have been conducted to assess protein requirements in farmed fish (Bowen 1987; Cowey 1994). Generally, the ability of fish—especially marine species—to utilize dietary carbohydrates as energy sources is relatively low (Moon 2001). This means that fish need to take more protein as their energy source (Stone 2003). However, by comparing several farmed animals, Cowey and Luquet (1983) found that fish only require relatively higher dietary protein levels to maximize their growth rate; however, the actual protein requirement of fish is similar to that of terrestrial animals.

Compared with the protein requirement, determination of 10 IAA requirements is relative harder to assess due to the large number of farmed finfish and shrimp species. A series of studies have been conducted to assess the IAA requirements, but the results revealed a high degree of variation (Cowey 1994; Gao et al. 2019; Mai et al. 2006a, b; Mambrini and Kaushik 1995; Wilson 2002). This may be partially explained by methodology issues (Cowey 1995). However, by analyses of published IAA requirements, Kaushik and Seiliez (2010) found the ideal amino acid profile in diets was similar as the whole body IAA profile of the corresponding species.

Fish require more protein as energy source and are strongly dependent on dietary protein/amino acid levels (Wilson 1986). Since proteins account for the significant proportion of aquaculture diets, the availability of protein sources becomes one of the major issues for the growing of the aquaculture industry. Despite the limited production and high price, fishmeal may well be the best protein source for aquaculture. However, fishmeal production reached an all-time high in 1994 at 30 million tonnes (live weight equivalent), but gradually declined since then (FAO 2018). The aquaculture industry used 69% of global fishmeal production. Meanwhile, the global aquaculture production of food fish in 2016 was 80 million tonnes. In comparison, the total worldwide production in 1994 was only 20 million tonnes. Furthermore, the FAO 2030 Agenda articulated that the fisheries and aquaculture sector is important for improving food security and fighting against hunger (FAO 2018). Recently, much effort has been focused on the discovery of alternative protein sources. However, there are many problems, including amino acids' unbalance, anti-nutrients, and pro-inflammatory factors (Lim et al. 2008).

The amino acid profiles of alternative protein sources, especially plant proteins, are usually different from that of fishmeal. The unbalanced amino acid composition triggers a series cellular responses. Thus, after administering soyabean meal-incorporated diets, in which 45% of fishmeal was replaced by soyabean meal, the postprandial influx of free amino acids in turbot (Scophthalmus maximus L.) was dramatically reduced (Xu et al. 2016). Moreover, similar results were observed with meat and bone meal replacements (Song et al. 2016). The molecular mechanism of amino acid sensing by GCN2 and TORC1, which was mainly discovered in yeast and mammalian cells, is conserved also in fish spices. Usually, plant-based diets are short of methionine, deprivation of which suppresses TORC1 signaling and increases the phosphorylation of eIF2a in primary muscle cells of turbot (Jiang et al. 2017). Similar results were found in soyabean meal-incorporated diets fed to juvenile turbot (Xu et al. 2016) indicating that the fish may initiate the amino acid response pathway to adapt to the unbalance in amino acids (Xu et al. 2016). The dysregulated free amino acid pool is influenced by amino acid absorption and transport efficiency of different protein sources (Song et al. 2016; Xu et al. 2016). Also, fishmeal replacement reforms the metabolism in aquatic animals. Soy protein-rich diets decrease the levels of plasma triglycerides and cholesterol. Moreover, the activities of hepatic lipogenic enzymes were recorded to be reduced in European seabass (Dias et al. 2005). Similar results were found in blackspot seabream, rainbow trout, and turbot (Figueiredo-Silva et al. 2010; Skiba-Cassy et al. 2016; Xu et al. 2016). These effects may be explained by the activity change in the GCN2 and TORC1 signaling pathways, which have been reported to regulate fatty-acid homeostasis (Dai et al. 2015; Guo and Cavener 2007).

Furthermore, some studies showed that different protein sources alter fish body mass (Song et al. 2016; Xu et al. 2016). For example, soyabean meal-incorporated diet resulted in decreased protein, fat and energy retention in turbot, and the whole-body fat content was reduced (Xu et al. 2016). Dietary proteins also affect postprandial free amino acid influx. Similar results were found in turbot fed with meat and bone meal-incorporated diet (Song et al. 2016).

After searching the genomic information of zebrafish, which is a model organism used worldwide, we found that most amino acid sensors (Gcn2, Tor, GATOR1, GATOR2, Sestrin, Slc38a9, Tm4sf5, Lars, CASTOR1) existed in the species. More interestingly, the crystal structure of SLC38A9 was revealed using zebrafish homolog-drSLC38A9. These results indicated that the amino acid sensors may be functionally conserved in teleost, Thus, the addition of dietary agonists of amino acid sensors may improve the utilization of alternative protein sources. For example, supplementary leucine increased TORC1 activity and specific growth rate in turbot (unpublished data). Another strategy for designing the best TORC1 stimulating diet in fish is to combine different alternative protein sources to mimic the amino acid profile of fishmeal.

Besides amino acid unbalance, anti-nutrients and pro-inflammatory factors are impact factors for the utilization of alternative protein sources. Anti-nutrients are substances existing in food sources; they interfere in food utilization and affect the health of animals. These effects are either caused by anti-nutrients themselves or by their metabolic products (Makkar 1993). Some common anti-nutrients in plant-derived materials are protease inhibitors, saponins tannins, lectins, antigenic compounds, gossypols, and antivitamins. Soybean saponin (SA) which exists in soybean meal is a major anti-nutrient causing metabolic disturbances, growth reduction and inducing enteritis in Japanese flounder and Atlantic salmon (Chen et al. 2011; Krogdahl et al. 2015). At the molecular level, SA reduces TORC1 activity, but increases the mRNA expressions of growth axis related genes including GH, GHR, and IGF1 (Tian et al. 2018). Gossypol, which exists in cottonseed meal, inhibits TORC1 activity and induces ER stress (Bian et al. 2017). Also, soybean agglutinin and wheat germ lectin inhibit TORC1 activity in zebrafish liver cells (Wang et al. 2019a, b).

Many attempts have been made to enhance the utilization of alternative protein sources. Resveratrol and silymarin attenuate the inflammatory response and oxidative stress induced by soybean meal in turbot (Tan et al. 2019; Wang et al. 2019a, b). Preprocessing plant-based protein sources by fermentation could ameliorate their negative effects. For example, fermentation reduces significantly indigestible oligosaccharides (stachyose, raffinose, and sucrose) and anti-nutrients (glycinin, trypsin inhibitors, tannin, and β-conglycinin) in soybean meal (Wang et al. 2016). Nevertheless, it is certainly clear that multiple approaches should be integrated to increase the utilization of alternative protein sources (Fig. 4).

Figure 4.  The strategy of improving the utilization of alternative protein source in aquaculture. The expanding aquaculture industry is highly dependent on the supply of fishmeal, which is the major protein source in aquatic animal diets. The limited availability and increasing price of fishmeal has led to the search for or development of alternative protein sources. However, the concomitant issues including anti-nutrients, pro-inflammatory factors, and unbalanced amino acid profiles of alternative protein sources have limited the utilization of those proteins. Some concerted efforts were conducted to eliminate these issues including using preprocessing technologies to remove anti-nutrients, applying advanced amino acid sensing theory to reduce the effect of unbalanced amino acid profiles, and adding anti-inflammatory supplements (which reduces inflammation) to attenuate the pro-inflammatory effects

Protein/amino acid nutrition is a very complex biological process consisting of protein digestion, absorption, transportation, and metabolism. Several signaling pathways, including GCN2 and TORC1 signaling pathways, are involved. Whereas GCN2 plays a "brake" role for cell growth, TORC1 functions as an "accelerator". There is increasing evidence showing that TORC1 is the key regulator for sensing nutritional status and regulating the molecular and cellular response. Although the molecular mechanism of amino acid sensing by GCN2 and TORC1 was mainly discovered and reported in yeast and mammalian cells, it is also conserved in fish species, including farmed aquatic animals. This makes it possible to answer serious questions (e.g. amino acids unbalance, anti-nutrient effect, and pro-inflammatory effect) caused by alternative protein sources for aquaculture. In recent years, new amino acid sensors have been identified. However, whether or not these sensors exist in aquaculture animals and if their functions are conserved have not been not fully answered. Further studies should be conducted to reveal the molecular nutrition of proteins in aquaculture.

This study was supported by the National Key R & D Program of China (2018YFD0900400), National Natural Scientific Foundation of China Grant (31772860), Aoshan Talents Cultivation Program Supported by Qingdao National Laboratory for marine science and technology (2017ASTCP-OS12), Fundamental Research Funds for the Central Universities (201822017) to GH, and China Agriculture Research System (CARS-47-G10) to KM.

GH and KM designed this review. CL, XW, and HZ wrote the article. All authors read and approved the final manuscript.

The authors declare no conflicts of interest.

This article does not contain any studies with human participants or animals performed by any of the authors.