Hypothalamic integration of nutrient sensing in fish Free

Much of our current understanding of the regulation of feeding behaviour is based on work in mammals, particularly lab models like mice and rats, with scarce information available in wild species. In mammals, food intake is regulated within the central nervous system (CNS) according to the levels of circulating nutrients, such as glucose, fatty acids and amino acids. These nutrients reflect the composition of the consumed food, and they act as a signal of the energy status of the organism. To detect changes in the levels of these circulating nutrients, a combination of enzymes, channels, carriers, transporters and other molecules in the brain act as sensors, enabling the detection and metabolism of nutrients. This information is integrated in the hypothalamus to elicit changes in food intake through changes in specific neuropeptides. Nutrients interact with circulating hormones, all acting on the hypothalamus (see Box 1). However, the focus of this Review is on the nutrient part of the regulation, rather than on hormonal effects.

In fish, research carried out in recent years has provided evidence for the existence of comparable mechanisms to those known in mammals. In this Review, we aim to describe these processes in fish, focusing not only on specific mechanisms for sensing glucose, fatty acids and amino acids, but also discussing their impact on metabolism and the regulation of feed intake. It is our hope that this synthesis will allow us to identify key knowledge gaps that can be addressed by future research. It is important to mention that fishes form a heterogeneous group, and that, so far, knowledge on the mechanisms underlying their nutrient sensing has been obtained from a relatively small number of fish species. Therefore, it is not currently possible to suggest clear trends within teleost fish regarding these mechanisms.

In fish, there are multiple hypothalamic mechanisms for detecting glucose levels, and most of them, as characterized in mammals, are dependent on the entry of glucose into the cells and its subsequent metabolism. The best-known glucosensing mechanism is based on glucokinase (GCK) activity and GLUT2 (SLC2a2)-mediated transport (Fig. 1; Marty et al., 2007; De Backer et al., 2016). Once glucose is transported into a neuron through the low-affinity GLUT2 transporter, GCK facilitates its phosphorylation to glucose-6-phosphate. This initiates glycolysis, increasing intracellular ATP levels and leading to the closure of ATP-sensitive potassium (K⁺_ATP) channels. This closure induces membrane depolarization, allowing calcium influx through L-type voltage-dependent calcium channels, and ultimately enhancing neuronal activity. In fish, researchers have used a variety of techniques to alter glucose levels experienced by rainbow trout (Oncorhynchus mykiss) hypothalamic cells, both in vitro and in vivo (Polakof and Soengas, 2008; Polakof et al., 2007a, 2007b, 2008a,b,c; Conde-Sieira et al., 2010a,b, 2011, 2012; Aguilar et al., 2011; Otero-Rodiño et al., 2015a). All such interventions lead to modifications in different components involved in the Gck–Glut2-dependent glucosensing mechanism within the hypothalamus. The expression of the components of this system are increased when circulating levels of glucose rise and decreased when blood glucose levels fall. In addition to rainbow trout, there is evidence for central glucosensing capacity in other fish species, such as medaka (Oryzias latipes, Hasebe et al., 2016) or grass carp (Ctenopharyngodon idella, Chen et al., 2022).

As noted above, there are multiple glucosensing mechanisms, and not all rely on Gck–Glut2. In mammals, an alternative mechanism relies on the transport of glucose into the neuron by the sodium-coupled glucose cotransporter 1 (SGLT-1), dependent on the entry of Na⁺ (Fig. 1; O'Malley et al., 2006). Because glucose is electrically neutral, the entry of Na⁺ through SGLT-1 is sufficient to induce depolarization and increase neuronal activity. This response can occur either through direct changes of the membrane potential or indirectly through coupling with a G-protein (Díez-Sampedro et al., 2003). In addition to stimulating neuronal activity, elevated glucose levels lead to increased levels of Sglt1 mRNA in various mammalian tissues, meaning that this transporter effectively serves as a glucosensor (O'Malley et al., 2006; González et al., 2009). In fish, Sglt-1 is present in various tissues, including the hypothalamus (rainbow trout: Sugiura et al., 2003; Soengas and Polakof, 2013; Conde-Sieira et al., 2013; Craig et al., 2013; gilthead sea bream, Sparus aurata: Sala-Rabanal et al., 2004), although its involvement in glucosensing is not clear.

Another GCK-independent glucosensing mechanism involves the stimulation of the sweet taste receptor, which is a heterodimer of type 1 taste receptor subunits (T1Rs) formed by T1R2, T1R3 and α-gustducin. Binding of glucose to this receptor induces the activation of an intracellular signalling cascade leading to membrane depolarization (Fig. 1; Ren et al., 2009; Herrera Moro Chao et al., 2016). T1R2/3 is a metabotropic (see Glossary) G-protein-coupled (GPR) receptor that can modulate neuronal activity in the presence of glucose in the brain (Welcome et al., 2015) in a manner similar to that of T1R2/3 located in taste buds (Kinnamon, 2012). In rainbow trout hypothalamus, the mRNA abundance of α-gustducin, t1r2 and t1r3 decreases under conditions of hyperglycemia, suggesting that the sweet taste receptor may play a role in glucosensing in this brain area (Otero-Rodiño et al., 2015a). Furthermore, levels of mRNA of t1r2 in the brain of rainbow trout increase in fish nutritionally programmed to cope with elevated levels of dietary carbohydrate (Balasubramanian et al., 2016).

An alternative glucosensor system based on the nuclear liver X receptor (LXR) has been described in mammals. Here, the activation of LXR increases in response to increased glucose, resulting in a decrease in gluconeogenesis (Anthonisen et al., 2010) and changes in the mRNA abundance of certain transcription factors (Higuchi et al., 2012; Kim and Ahn, 2004; Festuccia et al., 2014). Glucosensing in mammals can also be mediated by enhanced mitochondrial production of reactive oxygen species (ROS; see Glossary) by electron leakage, which can occur through the mitochondrial uncoupling protein UCP2a (Blouet and Schwartz, 2010; Blanco de Morentin et al., 2011), such that an increase in UCP2a activity indicates a lower level of glucose (Kong et al., 2010; Beall et al., 2012; Thorens, 2012). In fish, there is some evidence that both the Lxr- and mitochondria-dependent mechanisms of glucosensing are functional in rainbow trout hypothalamus (Otero-Rodiño et al., 2015a, 2016).

In recent years, experimental evidence from mammals has pointed to an important role for astrocytes and tanycytes in the glucosensing function of the hypothalamus (Pellerin and Magistretti, 1994; López-Gambero et al., 2019). In fish, there is little information on this mechanism, but lactate treatment can modulate central glucose metabolism in rainbow trout, which would suggest the possible presence of an astrocyte–neuron lactate shuttle that would facilitate glucosensing (Polakof and Soengas, 2008; Otero-Rodiño et al., 2015b).

In summary, in fish, there is sufficient information available on the Glut2–Gck mechanism of glucosensing to suggest that it plays a comparable role in fish to that known in mammals. However, more fish species need to be assessed to in order to determine the relative importance of this mechanism in species with different dietary habits (e.g. carnivore, omnivore or herbivore) or habitats. In contrast, in fish, the information regarding alternative mechanisms of glucosensing is very scarce, demanding further research.

Changes in glucose levels induce changes in feed intake and in appetite-related neuropeptides in fish (Polakof et al., 2011; Soengas, 2014). In general, the activation of glucosensing systems induced by the presence of high levels of carbohydrate in the diet or by increasing levels of glucose in vivo or in vitro results in decreased feed intake, in parallel with an increased abundance of mRNA encoding the anorexigenic (see Glossary) neuropeptides Pomc and Cart, and decreased levels of mRNA encoding the orexigenic (see Glossary) neuropeptides Npy and Agrp (Table 1). Moreover, in regions of the brain where neuropeptides are generated, studies utilizing histochemistry reveal the existence of proteins associated with glucose sensing, as demonstratred for Gck (Polakof et al., 2009), as well as Sglt-1, Lxr or T1r3 (Otero-Rodiño et al., 2019a,b). These findings imply a functional connection between glucosensors and neuropeptides in these brain areas.

As well as having a role in regulating appetite, glucosensing mechanisms are also involved in regulating other aspects of energy homeostasis: for example, glucosensing influences hormone secretion and energy expenditure, which modulate the activity of peripheral organs such as the liver and pancreas (Timper and Brüning, 2017; López-Gambero et al., 2019). In fish, intracerebroventricular (ICV) glucose administration affects hepatic metabolism (Polakof and Soengas, 2008). Therefore, the presence of glucose in the fish brain is a signal of energy abundance, which induces a reduction in the activity of the hepatic pathways involved in glucose production and release.

In mammals, hypothalamic detection of high glucose levels induces pancreatic counter-regulatory responses to restore normal blood glucose levels. These responses are mainly mediated by parasympathetic and sympathetic efferent nerves that innervate pancreatic α- and β-cells, inducing the release of the hormones insulin and glucagon (Blouet and Schwartz, 2010; Ogunnowo-Bada et al., 2014; Roh et al., 2016). In rainbow trout, changes in glucose concentration in the brain result in increased Gck activity and expression in Brockmann bodies (see Glossary), which are homologous to the mammalian endocrine pancreas (Polakof and Soengas, 2008). Accordingly, in fasted fish, plasma insulin levels decrease and plasma glucagon levels increase (Navarro and Gutiérrez, 1995), whereas the expression of insulin is higher in zebrafish (Danio rerio) exposed to elevated glucose levels (Jurczyk et al., 2011). Although there is a reasonable amount of information available regarding the impact of hypothalamic glucosensing on feed intake in fish, the impact of central glucosensing on peripheral metabolism is basically unknown and requires further research, especially considering the impact of metabolic changes on fish growth.

The most-accepted mechanism through which fatty acids (particularly long-chain fatty acids; LCFAs) are sensed in mammalian hypothalamic cells is metabolic in nature. Increased plasma levels of LCFAs lead to an increase in the levels of malonyl-CoA, which in turn inhibits carnitine palmitoyltransferase 1 (CPT-1; also known as carnitine acyltransferase 1; Fig. 2). This enzyme is in the outer mitochondrial membrane, and it is responsible for catalyzing the transport of LCFAs into mitochondria. Inhibition of CPT-1 prevents the mitochondria from importing fatty acid-CoA for β-oxidation (see Glossary; López et al., 2005, 2007; Gao et al., 2013). This mechanism is also likely to be functional in fish. For example, hypothalamic cells of several fish species show increased levels of malonyl-CoA and/or decreased Cpt-1 in response to the LCFA oleate (C18:1 n-9), both in vitro and in vivo [rainbow trout: Librán-Pérez et al., 2012, 2013b, 2014a; Velasco et al., 2017b; Senegalese sole (Solea senegalensis): Conde-Sieira et al., 2015; Chinese perch (Siniperca chuatsi): Luo et al., 2020].

Unlike the situation in mammals, where fatty acid-sensing mechanisms only respond to some LCFAs, the fish hypothalamus seems to detect monounsaturated fatty acids, medium-chain fatty acids (MCFAs) and polyunsaturated fatty acids (PUFAs). For example, increased malonyl-CoA and/or decreased Cpt-1 levels in the fish hypothalamus occur not only in response to oleate, but also to the MCFA octanoate in rainbow trout (Librán-Pérez et al., 2012, 2013b, 2014a) and the PUFA α-linolenate (ALA) in Senegalese sole (Conde-Sieira et al., 2015). The ability of fish (especially marine species) to respond to changes in PUFA levels might relate to the high amount of n-3 PUFAs in fish diets (Sargent et al., 2002) and consequently in their tissues (Mourente and Tocher, 1992; Tocher, 2003).

Alternative mechanisms of fatty acid sensing are also present in the fish hypothalamus. One such mechanism involves the fatty acid translocase (also known as cluster of differentiation 36; Fat/cd36; Fig. 2). This protein increases its capacity to bind to fatty acids in response to increased fatty acid levels, resulting in the modulation of mRNA abundance of several transcription factors, such as srebp1c and ppara (Le Foll et al., 2009). In fish, administration of oleate (which is not sensed by mammals) and/or octanoate upregulates the expression of hypothalamic mRNAs encoding Cd36, Srebp1c and Pparα in rainbow trout (Librán-Pérez et al., 2012, 2014a) and Chinese perch (Luo et al., 2020). A similar response also occurs upon feeding rainbow trout with a lipid-enriched diet (Librán-Pérez et al., 2015a). Similarly, an oleate-induced increase in abundance of cd36 mRNA and an oleate- and ALA-induced increase in ppara mRNA occur in the head of Senegalese sole post-larvae when the same nutrients are administered orally (Velasco et al., 2017a). Expression of cd36 is also upregulated in the brain of grass carp fed a high-fat diet (Tian et al., 2017) and in the hypothalamus of Chinese perch after ICV treatment with oleate, linolenate or α-linolelante (Luo et al., 2020). Intraperitoneal administration of oleate and ALA downregulate ppara and srebp1c in the Senegalese sole hypothalamus (Conde-Sieira et al., 2015). Finally, Fat/cd36 gene knockout suppresses the spontaneous linoleate preference in zebrafish (Liu et al., 2017).

Another response of mammalian hypothalamic neurons to increased levels of LCFAs is the inhibition of K⁺_ATP channels (Fig. 2). This inhibition can be caused either by the activation of specific isoforms of protein kinase C (PKC) (Benoit et al., 2009) or by the enhanced production of ROS, owing to electron leakage from mitochondria (Schönfeld and Wojtczak, 2008). In the fish hypothalamus, changes in the mRNA abundance of components of the K⁺_ATP channel and/or mitochondrial uncoupling (through UCP2a) occur in response to various fatty acids in rainbow trout (Librán-Pérez et al., 2012, 2013b, 2014a; Velasco et al., 2016a,b, 2017b) or Senegalese sole (Conde-Sieira et al., 2015; Velasco et al., 2017c), thus supporting the role of this channel.

Lipoprotein lipase (LPL) has also been identified as a lipid sensor in the mammalian hypothalamus. The activity of LPL increases in response to a rise in LCFA levels (Picard et al., 2014). It is not clear whether this lipid sensor is active in fish: in rainbow trout hypothalamus, the abundance of lpl mRNA decreases in response to ICV administration of oleate (Velasco et al., 2016a,b) or in vitro incubation with this fatty acid (Velasco et al., 2017b).

Finally, it is well established that hypothalamic neurons in mammals express GPRs – particularly GPR40 and GPR120 [also named free fatty acid receptor 1 (FFAR1) and 4 (FFAR4), respectively] – that can detect and respond to LCFAs (Dragano et al., 2017). Activation of these receptors is enhanced as LCFA levels increase, triggering the activation of the phospholipase C (PLC)/inositol 1,4,5-triphosphate (IP3) intracellular signalling pathway, and ultimately leading to an increase in intracellular Ca²⁺ levels (Usui et al., 2019). The presence of GPR84, which binds to MCFAs, has also been reported in the mammalian brain (Ichimura et al., 2009). In contrast to mammals, Ffar4 is absent in fish; consequently, Ffar1 is the only putative receptor for LCFAs in fish (Fig. 2). Recent studies have provided evidence for the presence of fatty acid-sensing mechanisms involving Ffar1 in rainbow trout hypothalamus (Velasco et al., 2020), as well as for mechanisms involving Gpr84 and Gpr119 (Fig. 2; Velasco et al., 2021). However, knowledge on Gpr-mediated fatty acid sensing in the fish brain is still very scarce and more research on this topic is required.

Lipids have an important impact on feed-intake regulation in fish: fish fed lipid-enriched diets display different feed-intake levels (usually decreased) compared with those fed a normal diet. Moreover, enhanced lipid storage is also usually associated with reduced feed intake (Shearer et al., 1997; Silverstein et al., 1999; Johansen et al., 2002, 2003). Among the lipid pool, fatty acids are the most important lipids in terms of energy use; thus, it is not surprising that the differences in feeding observed with lipid-enriched diets are predominantly caused by fatty acids (Table 1). For example, in rainbow trout, the lowest feed intake is seen in fish fed with lipid-enriched diets that result in higher plasma levels of fatty acids (Luo et al., 2014) and PUFAs (Roy et al., 2020). Moreover, a clear increase in feed intake is observed in rainbow trout displaying a pharmacologically mediated fall in circulating fatty acid levels (Librán-Pérez et al., 2014b). In addition, intraperitoneal (Librán-Pérez et al., 2012) or ICV (Librán-Pérez et al., 2014a; Velasco et al., 2016a,b) administration of oleate or octanoate results in a significant decrease in feed intake in rainbow trout, with the effect being greater for octanoate.

A decrease in abundance of npy/agrp mRNA and/or an increase in pomc/cart mRNA in response to feeding a lipid-enriched diet is observed in the hypothalamus of several fish species (Table 1). A similar increase also occurs in response to increased levels of specific fatty acids, such as oleate, octanoate or ALA (Table 1). The octanoate-mediated modulation of neuropeptide expression appears to be exclusive to fish; this fatty acid does not induce any change in mRNA abundance of neuropeptides in mammals (Hu et al., 2011).

Apart from its effects on feeding, the hypothalamic detection of fatty acids impacts several peripheral metabolic processes involved in energy homeostasis, such as hepatic glucose production and lipogenesis (Obici et al., 2002; Morgan et al., 2004; Migrenne et al., 2011). In fish, several studies have demonstrated that ICV treatment with fatty acids also alters parameters related to glucose and lipid metabolism in peripheral tissues (including the liver and the Brockman bodies), as in mammals (Migrenne et al., 2011; Conde-Sieira and Soengas, 2017). Thus, ICV-administered oleate and octanoate lead to increased levels of glucose and glycogen and decreased levels of fatty acid and total lipid, as well as decreased activities of Gck, fructose 1,6-bisphosphatase (Fbpase), fatty acid synthase (Fas) and Cpt-1 in the liver of rainbow trout. These changes counter-regulate the elevated fatty acid levels as detected in the brain (Librán-Pérez et al., 2015c). The functional connection between central fatty acid sensing and production/release of fuels from the liver is likely to be mediated through the vagus and splanchnic nerves, which innervate the liver and the gastrointestinal tract (Burnstock, 1959; Seth and Axelsson, 2010). At least in rainbow trout, the hypothalamus–inter-renal–pituitary axis (see Glossary) has also been implicated in the counter-regulatory response of the liver to a fall in circulating fatty acid levels (Librán-Pérez et al., 2015c), in a manner comparable to that described in mammals (Oh et al., 2012, 2014). Regarding the Brockman bodies, ICV administration of oleate and octanoate in rainbow trout results in the modulation of parameters related to lipid metabolism in this tissue (Librán-Pérez et al., 2015b).

In summary, lipid-enriched diets typically result in decreased feed intake in fish, likely owing to increased levels of fatty acids in the blood. Changes in feed intake are associated with changes in hypothalamic npy/agrp and pomc/cart expression, particularly in response to lipid-enriched diets or fatty acid administration. Moreover, hypothalamic detection of fatty acids influences peripheral metabolic processes, including glucose and lipid metabolism. However, further research on this subject is required to provide mechanistic detail in a wider range of fish species.

In mammals, branched-chain amino acids (BCAAs), namely leucine, isoleucine and valine, emerge as particularly relevant in terms of nutrient sensing. This significance stems from the fact that postprandial plasma levels of most amino acids, except for BCAAs, remain stable (Heeley and Blouet, 2016). Notably, among the three BCAAs, only leucine appears to be detected in the hypothalamus (Morrison and Laeger, 2015; Heeley and Blouet, 2016). Furthermore, mammalian hypothalamic neurons involved in the regulation of food intake respond specifically to changes in leucine levels (Heeley et al., 2018). In teleost fish, the available data suggest that amino acid sensing is like that observed in mammals, albeit with some notable differences; for example, valine is orexigenic in fish, but not in mammals (Fig. 3). In fish, amino acid-sensing mechanisms are predominantly activated by leucine, as observed in rainbow trout (Comesaña et al., 2018a,b, 2021b) and Atlantic salmon (Comesaña et al., 2021a). Valine has a limited effect on hypothalamic amino acid-sensing mechanisms in rainbow trout, but it exerts stronger effects in other brain areas, such as the telencephalon (Comesaña et al., 2018a,b, 2022). The mechanisms of amino acid sensing that have been characterized in the hypothalamus of these fish species are based on: (1) metabolism of BCAA, (2) metabolism of glutamine, (3) taste receptors, (4) amino acid carriers, (5) mTOR signalling pathway and (6) Gcn2 kinase. Each of these mechanisms were first characterised in mammals (although the specific mechanisms may differ in fish) and are discussed in more detail below.

The metabolic breakdown of the three BCAAs involves the enzymes branched-chain aminotransferase (BCAT) and branched-chain α-keto acid dehydrogenase complex (BCKDH) (Adeva-Andany et al., 2017). In mammals, food intake is reduced following increases in BCKDH activity or central administration of the transamination product of leucine by BCAT (Blouet et al., 2009), whereas deletion of BCAT reverses the effect (Purpera et al., 2012). In the hypothalamus of rainbow trout, this amino acid-sensing mechanism is activated by leucine through Bckdh (Comesaña et al., 2018a,b, 2021b).

Amino acid levels can also be sensed through metabolism of glutamine, as BCAA metabolism and the glutamine–glutamate cycle are linked. BCAAs are precursors of glutamine, because the BCAT enzyme produces glutamate which, in turn, is converted to glutamine by the enzyme glutamine synthetase (GLS) (Adeva-Andany et al., 2017). In mammals, leucine stimulates the activity of the enzyme glutamate dehydrogenase (GDH) which produces α-ketoglutarate (a substrate for the Krebs cycle) from glutamate (Jewell and Guan, 2013). Thus, increased levels of leucine result in enhanced metabolism of glutamine and glutamate, and ATP production. In rainbow trout, leucine fails to stimulate Gdh activity in the hypothalamus (Comesaña et al., 2018a), but this mechanism is activated through increased Gls (Fig. 3; Comesaña et al., 2018a,b, 2021b). In addition, exposure to leucine decreases glutamate levels in zebrafish brain, supporting the link between leucine and the degradation of glutamate by Gdh (da Silva Lemos et al., 2022).

In mammals, three receptors belonging to the taste receptor type 1 family (T1Rr1, T1R2 and T1R3) detect sweet and umami tastes. These receptors are also present in the mammalian hypothalamus, where they contribute to amino acid sensing. The receptors heterodimerise in the presence of glucose (T1R2–T1R3) and amino acids (T1R1–T1R3), triggering a cascade of second messengers that depolarizes the membrane (Fig. 3; Behrens and Meyerhof, 2016). In fish, unlike in mammals, amino acids can be detected by both heterodimers (Morais, 2017), and an increase in leucine levels increases the levels of t1r2 and t1r3 mRNA in the hypothalamus of rainbow trout (Comesaña et al., 2018a,b).

Amino acid carriers provide an additional mechanism for sensing the levels of amino acids. Among all amino acid carriers, two are particularly noteworthy in this regard, and both fall within the solute carrier (SLC) superfamily; they are l-type amino acid transporter 1 (LAT1) and sodium-dependent neutral amino acid transporter 2 (SNAT2; Fig. 3). LAT1 and SNAT2 are present in the blood–brain barrier (see Glossary). They are key regulators of leucine intracellular concentration and are required for the activation of mTOR signalling (Hundal and Taylor, 2009; Dodd and Tee, 2012; Taylor, 2014). In fish hypothalamus, leucine upregulates snat2 expression in rainbow trout (Comesaña et al., 2018a,b) and both lat1 and snat2 in Atlantic salmon (Comesaña et al., 2021a), suggesting that these two amino acid carriers might act as independent amino acid-sensing mechanisms.

The mTOR signalling pathway in mammals is well known to be activated by nutrients, especially by leucine (Dodd and Tee, 2012; Yue et al., 2022), although the mechanism of action is unclear. Several studies have evaluated the role of mTOR signalling in amino acid sensing in the fish brain. For example, in rainbow trout hypothalamus, this mechanism is activated by leucine. In rainbow trout, ICV and IP treatment with leucine upregulates abundance of mtor mRNA and the phosphorylation status of the protein mTOR, i.e. the percentage of the protein in the phosphorylated (active) form (Comesaña et al., 2018a,b). Additionally, valine is also detected by this mechanism in the hypothalamus of rainbow trout (Comesaña et al., 2022), whereas in cultured brain cells of Chinese perch, leucine (but not valine or isoleucine) activates the downstream mTOR protein S6 (Chen et al., 2021). However, it has been suggested that the presence of insulin is also necessary for the activation of mTOR by leucine (Ahmad et al., 2021). In fact, in vitro incubation of rainbow trout hypothalamus with different concentrations of leucine in the absence of insulin does not affect mTOR or its downstream proteins but does increase levels of mRNA encoding the mTOR inhibitor Sesn2 (Comesaña et al., 2021b). The mTOR signalling pathway is also considered as an integrative pathway, linking nutrient-sensing information with the expression of neuropeptides involved in the regulation of feed intake (Soengas, 2021), as discussed below.

The amino acid-sensing mechanism that depends on GCN2 kinase (Fig. 3) is mostly activated under conditions of amino acid deficiency. When levels of circulating amino acids decline, tRNAs are no longer ‘charged’ (i.e. linked to an amino acid); uncharged tRNAs interact with and activate GCN2 kinase, which phosphorylates eukaryotic initiation factor 2α (eIF2α), thereby inhibiting it, and resulting in the selective control of translation (Battu et al., 2017). In fish, this mechanism has been studied in peripheral tissues with different amino acid deficiencies (Skiba-Cassy et al., 2016; Wang et al., 2016; Miao et al., 2021; Dou et al., 2023). The abundance of phosphorylated EIf2α protein increases with lysine deprivation in cultured brain cells of Chinese perch (Zou et al., 2022). However, this mechanism of amino acid sensing is apparently also responsive to leucine abundance because abundance of eIf2a mRNA increases in the hypothalamus of rainbow trout cultured with leucine in vitro (Comesaña et al., 2021b).

Like diets that are high in carbohydrate or lipid, diets with elevated protein levels or higher levels of leucine also lead to a reduction in feed intake in fish (Table 1). This effect involves concomitant changes in npy, agrp and pomc gene expression. Neither valine nor isoleucine affect food intake in mammals, where leucine is the only BCAA involved in food intake regulation (Lueders et al., 2022). In fish, similar results are observed with isoleucine: ICV treatment of Chinese perch with isoleucine has no effect on food intake (Chen et al., 2021). However, in contrast to mammals, ICV administration of valine in fish produces an orexigenic response but without changes in neuropeptides (Table 1). Interestingly, ICV treatment with lysine or methionine in Chinese perch also elicits increased feed intake (Zou et al., 2022)

Apart from detecting the presence of nutrients to allow the regulation of feed intake, the fish hypothalamus also uses this information to regulate peripheral metabolism, thus maintaining energy homeostasis (Soengas et al., 2018). In mammals, amino acid levels and glucose metabolism are closely related – the hypothalamic detection of leucine also modulates hepatic metabolism, with the effect of reducing glucose production (Su et al., 2012; Arrieta-Cruz and Gutiérrez-Juárez, 2016). This effect has not yet been demonstrated in fish, but some studies have shown that central detection of amino acids in fish has peripheral effects. Thus, in Chinese perch, levels of preproghrelin in the intestine and leptin in the liver change after ICV injection of leucine, valine or isoleucine (Chen et al., 2021). In addition, in the same species, ICV injection of valine acts as a nutritional signal in the brain to modulate peripheral metabolism, attenuating protein degradation (Wang et al., 2020).

Once nutrient-sensing systems are activated/inhibited, the relevant information is integrated in the hypothalamus through changes in the levels of agrp/npy and pomc/cart mRNA, which all encode neuropeptides involved in the regulation of feed intake (Fig. 4). This occurs through changes in signalling pathways, causing the activation/inhibition of transcription factors that ultimately regulate feed intake. The relevant mechanisms of integration are well understood in mammals (for reviews, see Blouet and Schwartz, 2010; Morton et al., 2014). Below, we discuss the current state of knowledge around the hypothalamic integration of nutrient-sensing information in fish.

In mammals, one of the hypothalamic signalling pathways that is sensitive to changes in nutrient-sensing systems is the AMP-activated protein kinase (AMPK) pathway. The levels of hypothalamic AMPK and the phosphorylation status of the protein decrease as the levels of glucose, fatty acids or amino acids increase (Cai et al., 2007; Beall et al., 2012; Fromentin et al., 2012; Oh et al., 2016). Ampk is present in fish brain (Soengas, 2021), but few studies have addressed changes under conditions relating to the control of feed intake. The results of those that have are mostly in agreement with the mammalian model (Fig. 4). For example, feed deprivation results in increased levels and phosphorylation status of the protein Ampkα in the hypothalamus of rainbow trout (Conde-Sieira et al., 2019) (although note that this is not seen in the brain of channel catfish, Ictalurus punctatus; Abernathy et al., 2019). Furthermore, when rainbow trout are fed a lipid-enriched diet (Librán-Pérez et al., 2015a), the phosphorylation of Ampkα in the hypothalamus decreases. Similar changes are seen in fish with hyperglycaemia or those fed a diet high in carbohydrates (Nipu et al., 2022; Kamalam et al., 2012; Xu et al., 2017). In rainbow trout, the phosphorylation of Ampkα is decreased in response to higher levels of nutrients such as oleate (Velasco et al., 2016a, 2017b; Blanco et al., 2020), octanoate (Velasco et al., 2017b), glucose (Otero-Rodiño et al., 2017) or β-hydroxybutyrate (Comesaña et al., 2019). In contrast to the mammalian model, the phosphorylation of Ampkα is not increased in rainbow trout hypothalamus when the levels of leucine are increased (Comesaña et al., 2018a,b, 2020, 2021).

The involvement of Ampk in the integration of nutrient sensing is further supported by the fact that the phosphorylation of Ampkα decreases in rainbow trout hypothalamus following treatments that suppress appetite (i.e. treatments that mimic the activation of nutrient-sensing pathways; Velasco et al., 2016b, 2017a, 2019, 2020, 2021; Blanco et al., 2020). In contrast, it should be noted that Siberian sturgeon treated with the anorexigenic hormone adiponectin show enhanced levels of Ampkα2 (Tang et al., 2022). There are several different isoforms of Ampkα, and it seems that Ampkα2 is involved in regulation of feed intake, whereas Ampkα1 appears to modulate peripheral metabolism to maintain homeostasis, as observed in rainbow trout (Conde-Sieira et al., 2019, 2020). Finally, it is known that there is an interaction between hormones and the integration of nutrient-sensing pathways through Ampk, because the presence of oleate counteracts the stimulatory effect of ghrelin on Ampk in rainbow trout hypothalamus (Velasco et al., 2016a).

As discussed above, mTOR is thought to play an important role in integrating nutrient-sensing pathways. In fish, mTOR is involved in regulation of feed intake. In zebrafish, feed deprivation results in a decrease in mTOR levels (Craig and Moon, 2011), whereas mTOR levels and/or phosphorylation status increase under the anorectic (i.e. appetite-suppressing) conditions elicited by the presence of nutrients, as demonstrated in rainbow trout (Velasco et al., 2017b; Comesaña et al., 2018a,b; Blanco et al., 2020). Accordingly, ICV treatment with the appetite-stimulating amino acid valine in rainbow trout results in decreased levels of mTOR (Comesaña et al., 2022).

The response of mTOR to changes in nutrient levels is also supported by changes in its downstream proteins. For example, in Chinese perch, the levels of phosphorylated S6 increase after ICV treatment with the anorexigenic amino acid leucine, whereas a decrease occurs after treatment with the orexigenic amino acid valine (Chen et al., 2021). Interestingly, these changes are abolished in the additional presence of rapamycin, an inhibitor of mTOR (Chen et al., 2021). It should be noted that the involvement of mTOR in the integration of nutrient sensing might depend on the different species assessed, as well as on the specific amino acid involved. For example, in rainbow trout, leucine has an anorectic effect; however, in vitro, rainbow trout hypothalamus does not show any response to leucine in terms of the levels and phosphorylation status of mTOR or any effects on downstream proteins (Comesaña et al., 2021a,b). This suggests that the anorectic effect of leucine in rainbow trout is independent of mTOR signalling in the hypothalamus. In contrast, a comparable treatment with the orexigenic amino acid valine does result in changes in mTOR and downstream proteins (Comesaña et al., 2022).

Additional studies support a role for hypothalamic mTOR in feed-intake regulation in fish, supporting its role as an integrator of nutritional signals. Multiple treatments that induce anorectic conditions in fish cause an increase in mTOR levels and/or phosphorylation status (Penney and Volkoff, 2014; Librán-Pérez et al., 2015a,b; Dai et al., 2018; Velasco et al., 2017a, 2018, 2019, 2021; Blanco et al., 2020). Furthermore, hypothalamic mTOR activation modulates the abundance of pomc and npy mRNA in Japanese sea bass (Liang et al., 2019).

In the mammalian hypothalamus, Akt levels and phosphorylation status are increased in the presence of higher levels of nutrients (Hu et al., 2016; Park et al., 2011). Current evidence suggests that Akt is also involved in the hypothalamic mechanisms that regulate feed intake in fish (Fig. 4). For example, exposure to various nutrients in vivo and in vitro activates Akt signalling through increased phosphorylation of the Akt protein (Velasco et al., 2017b; Otero-Rodiño et al., 2017; Comesaña et al., 2018a; Blanco et al., 2020). Furthermore, a comparable increase in Akt protein levels occurs in the brains of fish fed diets enriched in specific nutrients, including carbohydrates (Dai et al., 2014; Jin et al., 2014; Jörgens et al., 2015) or lipids (Librán-Pérez et al., 2015a,b; Dai et al., 2018; Xu et al., 2019). Akt activation also occurs under other conditions that elicit anorectic responses (Gong et al., 2016; Velasco et al., 2016b, 2017a, 2020, 2021; Blanco et al., 2020). The involvement of hypothalamic Akt in regulation of feed intake is also supported by the fact that an opposing response (decreased phosphorylation status) is observed under orexigenic conditions (Velasco et al., 2017a). In mammals, the activation of Akt signalling in the hypothalamus results in changes in fatty acid metabolism through the activation of the transcription factor SREBP1c and its targets ATP citrate lyase (ACLY) and FAS (Kim et al., 2007). Accordingly, in the rainbow trout hypothalamus, enhanced phosphorylation of Akt occurs in parallel with increased levels of acly, fas and srebp1c mRNA (Velasco et al., 2016b), thus indirectly supporting the involvement of Akt in signalling related to nutrient sensing.

In mammals, brain homeobox transcription factor (BSX) is one of several transcription factors that are activated in response to the signalling pathways discussed above. It interacts with cAMP response-element binding protein (CREB; see below) in the mammalian hypothalamus, leading to an increase in mRNA encoding BSX, NPY and AgRP (Fig. 4; Nogueiras et al., 2008; Varela et al., 2011; Lee et al., 2016). BSX levels are reduced under anorectic conditions and increase under orexigenic conditions (Nogueiras et al., 2008; Lage et al., 2010). Only a few studies in fish have evaluated changes in Bsx that could relate to the regulation of feed intake. Thus, we know that feed deprivation in goldfish increases bsx mRNA levels in the hypothalamus (Vinnicombe and Volkoff, 2022). Accordingly, in rainbow trout, exposure to oleate (Conde-Sieira et al., 2018), leucine (Comesaña et al., 2021a,b) or glucose (Conde-Sieira et al., 2018; Blanco et al., 2020) (all of which reduce feed intake) results in decreased levels of Bsx. However, ICV treatment with the orexigenic amino acid valine also results in a decrease in levels of Bsx protein in rainbow trout (Comesaña et al., 2022). The effects of glucose on Bsx in the rainbow trout hypothalamus disappear after insulin treatment (Blanco et al., 2020), suggesting that there is an interaction between this nutrient and hormone that has an impact on integrative hypothalamic mechanisms. Other conditions that are known to induce an anorexigenic response in rainbow trout also decrease the levels of Bsx protein (Velasco et al., 2019, 2020, 2021). However, in goldfish, treatment with the anorexigenic hormone Cck does not alter levels of bsx mRNA (Vinnicombe and Volkoff, 2022).

CREB is another transcription factor that is thought to be involved in the connection between brain metabolism and the expression of neuropeptides that regulate appetite in mammals. A decrease in CREB levels in the mammalian brain induces a decrease in the abundance of Npy and Agrp mRNA, resulting in decreased food intake (Fig. 4; Belgardt et al., 2009; Varela et al., 2011; Blanco de Morentín et al., 2011). In fish, the available information regarding the possible involvement of Creb in regulation of feed intake is scarce and is mostly restricted to rainbow trout. In this species, Creb phosphorylation decreases in response to raised levels of oleate (Velasco et al., 2017b; Conde-Sieira et al., 2018), octanoate (Velasco et al., 2017b) and glucose (Conde-Sieira et al., 2018; Otero-Rodiño et al., 2019b). In response to leucine, the phosphorylation of Creb is decreased in rainbow trout in vivo after ICV (Comesaña et al., 2018a) or IP (Comesaña et al., 2018b) treatment, but no changes occur in response to leucine under in vitro conditions (Comesaña et al., 2021a,b). Interestingly, ICV treatment of rainbow trout with valine also fails to elicit a Creb response (Comesaña et al., 2022), as does supplementation of feed with methionine in zebrafish (Pisera-Fuster et al., 2021). The presence of an inhibitor of Creb blocks the response of Creb to fatty acids in rainbow trout hypothalamus (Velasco et al., 2017b), indirectly supporting its role in regulation of feed intake. The changes in Creb that occur in response to changes in nutrient levels are comparable to those observed under other anorectic conditions in rainbow trout, such as treatment with the anorexigenic hormones Cck or Glp-1 (Velasco et al., 2019). Moreover, in zebrafish, increased levels of Creb also occur under the orexigenic conditions elicited by feed deprivation (Craig and Moon, 2011).

Forkhead box O1 (FoxO1) is likely to be involved in the hypothalamic integration of nutrient-sensing pathways to regulate feed intake in mammals (Gross et al., 2009). In mammals, conditions that increase the levels of FoxO1 protein cause an increase in the expression of Agrp mRNA while decreasing the levels of Pomc mRNA; changes that lead to a decrease in food intake (Belgardt et al., 2009; Blanco de Morentin et al., 2011). In general, in fish, increased levels of nutrients result in increased abundance and phosphorylation of the protein Foxo1, as observed in rainbow trout in response to increased levels of oleate (Conde-Sieira et al., 2018; Blanco et al., 2020), octanoate (Velasco et al., 2017b) and glucose (Conde-Sieira et al., 2018; Blanco et al., 2020). The lack of response of rainbow trout hypothalamus to oleate in the presence of a Foxo1 inhibitor supports the specificity of this response (Velasco et al., 2017b). There is also evidence for an interaction between nutrient levels and hormones in the dynamics of Foxo1 in the hypothalamic response: in rainbow trout, the presence of insulin counteracts the effects of oleate on the abundance of foxo1 mRNA (Blanco et al., 2020). In contrast to its response to glucose or fatty acids, hypothalamic Foxo1 does not appear to respond to changes in the levels of amino acids in fish, as demonstrated in rainbow trout (Comesaña et al., 2018a,b, 2020, 2021a,b, 2022). A range of anorectic conditions other than those produced by increased nutrient levels also result in increased levels of Foxo1 protein in rainbow trout (Velasco et al., 2016b, 2017a, 2019, 2020, 2021; Blanco et al., 2020).

In this Review, we present an overview of how the fish hypothalamus integrates metabolic and endocrine information to elicit changes in feed intake and peripheral energy metabolism. The underlying mechanisms are similar in some respects to those known in mammals, but are not identical (Delgado et al., 2017; Soengas et al., 2018). The relevant metabolic effects result from the activation/inhibition of nutrient-sensing systems.

Glucosensing mechanisms in the fish hypothalamus involve a complex interplay of multiple systems, including the well known Gck–Glut2 pathway, Sglt-1, sweet taste receptors, Lxr and potential mitochondria-based mechanisms. Together, these mechanisms enable the hypothalamus to finely regulate responses to changes in glucose levels. Glucosensing in the hypothalamus influences feeding behaviour, the expression of appetite-related neuropeptides and metabolic responses in fish. Future research in this area should focus on: (1) investigating changes in the discharge frequency of neurons in response to circulating glucose concentrations; (2) the precise roles of the glucosensing systems not dependent on Gck–Glut2; (3) the importance of astrocytes and tanycytes; and (4) investigating how chronic changes in glucose levels, induced through diet or environmental conditions, impact overall metabolic health and energy balance in fish.

Compared with glucose-sensing pathways, fish and mammals are more divergent in terms of their fatty acid-sensing mechanisms. Unlike that of mammals, the fish hypothalamus responds not only to LCFAs but also to MCFAs, reflecting the unique metabolic adaptations in fish. Mechanisms involving malonyl-CoA, enzymes such as Cpt-1, Fas and Acly, as well as the fatty acid translocase (Fat/cd36), have been identified as crucial components of fatty acid sensing in fish. Moreover, the impact of fatty acid sensing on feeding regulation and metabolic processes in fish is evident. Further research is needed to delve deeper into the specific signaling pathways involved, particularly: (1) the identification and characterization of Ffar1, Gpr84 and Gpr119; (2) investigating the crosstalk between central (hypothalamus) and peripheral (liver, Brockman bodies) fatty acid-sensing mechanisms; and (3) exploring species-specific adaptations and responses to diverse types of fatty acids, considering the diversity in fish physiology and ecology.

Amino acid sensing has only recently been investigated in fish hypothalamus. However, leucine has already emerged as a crucial amino acid influencing regulation of feed intake in both mammals and fish, albeit with species-specific responses. The effects of amino acids such as valine, lysine and methionine on feed intake in fish also differ from the responses in mammals. Understanding these differences will shed light on the diverse metabolic adaptations across species. Because most of our information on amino acid sensing in fish is restricted to a few species, it is essential to enhance our understanding of the universality or specificity of the relevant mechanisms in fish species other than salmonids. Furthermore, the roles of amino acid carriers, such as Lat1 and Snat2, in the fish hypothalamus need further exploration. Considering the carnivorous nature of many fish species, understanding their amino acid sensing will have practical applications in optimizing dietary formulations for improved growth and health in farmed fish.

The hypothalamic integration of neuroendocrine signalling and nutrient sensing in the regulation of feed intake is a complex and intricate process, of which our current understanding is limited. We have identified some of the signalling pathways and transcription factors involved in this integration, but changes in feed intake stemming from alterations in neuropeptide expression are also the result of a complex interplay between nutrient and hormonal signal transduction. This interaction remains largely unexplored; for example, in mammals we have only limited evidence pertaining to the interactive effects of hormones such as leptin or ghrelin on fatty acid sensing (López et al., 2007; Blanco de Morentin et al., 2011; Lockie et al., 2019). In fish, studies conducted in rainbow trout have demonstrated the existence of interactive effects between nutrient-sensing mechanisms and hormones such as ghrelin (Velasco et al., 2016a) and insulin (Blanco et al., 2020).

In conclusion, what we know of glucosensing, fatty acid sensing and amino acid sensing mechanisms in the fish hypothalamus underscores the remarkable adaptability of fish to varying nutritional conditions. These diverse sensory pathways – involving well-established systems such as Gck–Glut2, as well as newly assessed components, such as fatty acid translocase and amino acid carriers – play pivotal roles in regulating feeding behaviour and metabolic responses. The complex integration of neuroendocrine signalling, transcription factors and nutrient sensing in the hypothalamus highlights the need for further research to unravel the species-specific nuances. As our understanding deepens, addressing the long-term effects of altered sensing mechanisms on energy homeostasis in fish becomes essential. This holistic perspective not only enhances our knowledge of fundamental physiological processes but also holds promise for optimizing dietary formulations and promoting sustainable and efficient aquaculture practices.