Symposium review : Fueling appetite: Nutrient metabolism and the control of feed intake*

Conceptual models developed over the past century describe 2 key constraints to feed intake (FI) of healthy animals: gut capacity and metabolic demand. Evidence that greater energy demands (e.g., greater milk production) drive a corresponding increase in caloric intake led to the dominant concept that animals “eat to energy requirements.” Although this model provides reasonable initial estimates of FI, it lacks a proposed physiological basis for the control system, does not consider nutrient constraints beyond energy, and fails to explain differential energy intake responses to different fuels. To address these gaps, research has focused on mechanisms for sensing nutrient availability and providing feedback to hypothalamic centers that integrate signals to con-trol feeding behavior. The elimination of FI response to certain nutrients by vagotomy suggests that peripheral tissues play a role in nutrient sensing. These findings and the central role of the liver in metabolic flux led to the development of the hepatic oxidation theory (HOT). According to the HOT, liver energy charge is the regulated variable that induces dietary intake changes and consequently affects whole-body energy balance. Evidence in support of HOT includes associations between hepatic energy charge and meal patterns, increased FI in response to phosphate trapping, and reduced FI in response to phosphate loading. In accordance with the HOT, infusion studies in dairy cattle have consistently demonstrated that providing fuels that either oxidize or stimulate oxidation in the liver decreases FI and energy intake to a greater extent than fuels that bypass the liver. Importantly, this holds true for glucose, which is readily oxidized by nerve cells, but is rarely taken up by the bovine liver. Although the brain integrates multiple signals including those related to gastric distention and illness, the HOT provides a physiological framework for understanding the dominant role the liver likely plays in sensing short-term energy status. Understanding this model provides insights into how to use or bypass the regulatory system to manage FI of animals.


INTRODUCTION
Energy at its basis is defined as the ability to do work, which in a biological system often translates to heat.In nutritional biology, energy is assigned the unit of either a kilojoule or a calorie, defined as the amount of heat necessary to raise the temperature of 1 g of water from 14.5 to 15.5°C.Consequently, energy requirements have been determined through the estimation of heat expenditure and oxygen consumption of animals, based on the work of Antoine-Laurent de Lavoiser in 1784.As our understanding of chemistry and metabolism has increased, the use of respiration chambers and bomb calorimetry have resulted in the capability to estimate the energy requirements of animals under various circumstances (i.e., maintenance, lactation, pregnancy, and so on) and the energy content of diets fed to animals.However, these techniques are measurements of whole-body oxidation, which creates a "black box" situation, in that we are able to calculate the amount of energy to input into a system and subsequent energy output of the system, but have not elucidated the mechanisms occurring within the box (i.e., the physiological and metabolic controls associated with feed intake, FI).Feed intake, a function of meal size and frequency, is affected by various physiological signals continuously in flux that are driven by the metabolic and physiological state of the animal (Allen, 2014).Consequently, understanding and predicting feeding behavior is a complex issue as feeding is a 2-state variable (i.e., eating or not eating), but the factors influencing these states are likely continuous variables (Forbes and Gregorini, 2015).Conrad et al. (1964) showed that digestibility of feed influences FI and suggested that upon reaching a certain point, likely the point of gut distension, factors controlling FI shifts from "body weight (reflecting roughage capacity), undigested residue per unit body weight per day (reflecting rate of passage), and dry matter digestibility" (p.54) to "metabolic size, production, and digestibility" (p.54).As a result of this work and the work of others, the idea that animals "eat to energy requirements" became a predominant concept in explaining FI regulation.Yet, this concept still does not pinpoint any potential whole-body "energostatic" mechanism for physiological regulation of feeding behavior.For example, research has suggested that feeding behavior may be influenced by nutrient constraints beyond energy.Albornoz et al. (2019) has shown that feeding diets with the same gross energy digestibility, but 2 different levels of starch fermentability, resulted in a lower DMI and ME intake in dairy cows fed the more fermentable starch source.
The primary product of starch fermentation in the rumen is the VFA propionate, which has been shown to exhibit hypophagic effects compared with other VFA, such as acetate (Oba and Allen, 2003) and other gluconeogenic precursors (Gualdrón-Duarte and Allen, 2017Allen, , 2018)).Because propionate is primarily absorbed and used by the liver (Reynolds et al., 2003), these studies suggest that the metabolism of nutrients in the liver have some influence on feeding behavior.Therefore, cross-talk between the central and peripheral nervous systems (e.g., brain and liver) is likely occurring, and is responsible for coordinating a response to energy status and subsequent adjustments to satiety and hunger signaling that affect feeding behavior.

CENTRAL AND PERIPHERAL CONTROL MECHANISMS
It is widely agreed that ultimate control of feeding behavior resides in the hypothalamus (Roche et al., 2008;Zheng and Berthoud, 2008), and a series of hypothesized mechanisms exhorting this control are listed in Table 1.A complex network of orexigenic neurons (expressing neuropeptide Y and agouti-related peptide) and anorexigenic neurons (expressing pro-opiomelanocortin and cocaine-and amphetamine-regulated transcript) receive endocrine and neural inputs that are integrated by effector neurons that stimulate feeding (or not).Given its centrality in the control of feeding behavior, it is not surprising that many investigators have hypothesized a direct role for the hypothalamus in sensing nutrient availability.In some rodent studies, intracerebroventricular delivery of quantities of nutrients that are irrelevant at a systemic level has significantly Note that it is likely that some combination of mechanisms underlies chemosensory control of feed intake, with integration in the hypothalamus; for simplicity, hypotheses are presented here as stand-alone systems.
decreased FI in short-term experiments (López et al., 2007;Lane and Cha, 2009;Frost et al., 2014).However, in addition to some evidence that the hypothalamus may directly sense metabolites to alter feeding behavior, there is other evidence that peripheral inputs are necessary for negative feedback on appetite by circulating nutrients.One mechanism for nutrient-mediated satiety signaling is through gut peptides.These are hormones secreted by L cells in the intestinal epithelial layer that sense specific nutrients and signal to the rest of the body they are available for absorption.Initially, it was believed that gut peptides primarily worked by reaching the brain through the bloodstream and acting directly on neurons in and around the hypothalamus to suppress feeding behavior, and these mechanisms are in play.However, more recent studies have demonstrated that severing the vagus nerve, a major neural pathway between the brain and digestive system, eliminated effects of some gut peptides (Brown et al., 2011;Goldstein et al., 2021).This suggests that secreted gut peptides are sensed near the site of release and that some communication to the brain is transmitted via neural connections (Borgmann et al., 2021).
In addition to mediating at least some effects of gut peptides, the vagus nerve also apparently transmits signals triggered by direct sensing of nutrients in the viscera.Although it is difficult to prove the absence of an endocrine mediator (given that some may yet be undiscovered), the firing rate of the vagus nerve changes very rapidly after portal infusion of nutrients (Goldstein et al., 2021).Portal delivery bypasses intestinal sensing, including the L cells responsible for most endocrine activity in the gut.Such observations lead to hypotheses that include chemosensory neurons embedded within the wall of the hepatic portal vein and sensing by the liver itself.
Several key studies helped to focus these proposed mechanisms on the liver.First, sectioning the hepatic vagus nerve specifically (avoiding branches innervating other parts of the viscera) eliminated satiety effects of nutrients in sheep (Anil and Forbes, 1988) and mice (McDougle et al., 2021).Although it is difficult to claim with certainty that this branch could not also reach, for example, the portal vein, it diminishes the likelihood that gastrointestinal sensing is the only mechanism in play.Second, a series of experiments that alter liver physiology (but would not likely affect direct nerve sensing of circulating nutrients) have pointed specifically to liver metabolism of nutrients as a key component of this sensory mechanism.These experiments are outlined in the next section.

HEPATIC OXIDATION THEORY
The liver plays a central role in metabolic flux and sensing of nutrients and fuels, being considered the primary sensory organ integrating the long-and shortterm mechanisms affecting FI (Allen, 2020).Research conducted on laboratory species suggests that control of feeding behavior is related to energy charge in the liver and is integrated synergistically with other metabolic inhibitory mechanisms by a common signal related to the hepatic energy status derived from the oxidation of fuels.Thus, liver energy status and sensing has been suggested as a mechanism triggering satiety or hunger signals to the brain (Friedman et al., 1999).In support of this notion, prevention of hepatic ATP production by trapping of inorganic phosphate, decreases in hepatic ATP/ADP ratio and phosphorylation potential, or inhibition of hepatic fatty acid oxidation induced feeding behavior in laboratory species, but this stimulus is blocked when neural connections between peripheral organs and the brain via the vagus nerve are removed (Rawson et al., 1994;Ji et al., 2000;Horn et al., 2001Horn et al., , 2004)).As further support, a multitude of studies have shown that infusing glucose in dairy cows does not decrease energy intake (Dowden and Jacobson, 1960;Clark et al., 1977;Chelikani et al., 2004).Unlike in nonruminant species (e.g., mice), hepatic removal of glucose and activity of hexokinase is low to negligible in mature ruminants (Ballard, 1965;Stangassinger and Giesecke, 1986), and thus, hepatic oxidation of glucose is likely negligible as well.Whereas both propionate and glycerol are 3-carbon glucose precursors with similar energy content, propionate decreased FI compared with glycerol (Gualdrón-Duarte and Allen, 2017).This is likely because propionate is an obligate anaplerotic metabolite, whereas glycerol can enter the glucogenic pathway in the cytosol without stimulating oxidation of acetyl CoA.Therefore, a lack of response to glucose and glycerol as opposed to an anaplerotic fuel such as propionate coincides with a theory that feeding behavior is influenced by the oxidation of nutrients in the liver.Ji and Friedman (1999) reported that after exposing rats to a 24-h fasting period, liver energy production concomitantly increases with compensatory hyperphagia over time upon re-feeding.The hepatic oxidation theory (HOT) proposes that when hepatic energy charge increases via fuel oxidation, the firing rate of hepatic vagal afferents decreases, subsequently signaling satiety, whereas a decrease in energy charge after a meal increases the firing rate, subsequently resulting in hunger and meal initiation (Allen et al., 2009).
Although hepatic energy charge is likely responsible for changes in the firing rate of the hepatic vagal afferents that control intake, the exact mechanism by which the signal is transmitted is not fully understood (Allen and Piantoni, 2013).

CONTROL OF HEPATIC OXIDATION IN RUMINANTS
According to HOT, fuels derived from the diet or mobilized from body reserves with potential to be oxidized in the liver may affect FI in ruminants (Allen et al., 2009;Allen and Bradford, 2012).Mitochondrial oxidation requires fuels to enter the tricarboxylic acid (TCA) cycle as acetyl CoA via β-oxidation (ketogenic fuels), metabolism of pyruvate by the pyruvate dehydrogenase complex (glucogenic fuels), or after deamination of AA.The oxidation of fuels and subsequent synthesis of ATP via oxidative phosphorylation relies on the interconnection between the TCA cycle and the electron transport chain.Consequently, these metabolic pathways are tightly regulated, particularly the TCA cycle that acts as a nexus for various metabolic pathways.The control of these systems is influenced by entry of TCA cycle intermediates (anaplerosis), availability of acetyl CoA, redox state, ATP/ADP ratio, and the proton motive force (Williamson and Cooper, 1980;Stock et al., 1999;Allen and Piantoni, 2013).For example, CO 2 produced from acetyl CoA oxidation in the TCA cycle can be a source of mitochondrial bicarbonate that can subsequently stimulate soluble adenylyl cyclase, and ultimately, oxidative phosphorylation (Acin-Perez et al., 2009).Furthermore, the enzymes citrate synthase, isocitrate dehydrogenase, and α-ketoglutarate dehydrogenase within the TCA cycle are considered rate-limiting enzymes that are inhibited by increased ATP concentrations as well as increased concentrations of citrate (citrate synthase), NADH (isocitrate dehydrogenase), and succinyl CoA (α-ketoglutarate dehydrogenase; Williamson and Cooper, 1980).Moreover, the flux in the ratio of NADH/NAD + is a crucial regulator of the TCA cycle which affects enzymes that use NAD + or NADH as a cofactor in the mitochondrion such as isocitrate dehydrogenase, α-ketoglutarate dehydrogenase, and malate dehydrogenase.However, a lag between the production and oxidation of reducing equivalents by the electron transport chain may result in an accumulation of these equivalents in the mitochondrion and delay in ATP synthesis (Allen and Piantoni, 2013).Consequently, predicting flux and metabolism in a physiological setting where multiple metabolites are simultaneously entering and leaving metabolic pathways has historically been challenging.Yet, with the advancement of -omic and other techniques, understanding and incor-porating this behavior in future models may further aid in predicting feeding behavior.
Despite these challenges, research suggests that certain fuels may contribute to or promote hepatic oxidation and hypophagia in adult ruminants (Stangassinger and Giesecke, 1986;Frobish and Davis, 1977;Gualdrón-Duarte and Allen, 2018).Research by Gualdrón-Duarte and Allen (2017Allen ( , 2018) ) have shown that the metabolism of gluconeogenic precursors affects DMI and in some instances, total ME intake, as isoenergetic abomasal infusions of propionic acid, glycerol, lactic acid, and glucose resulted in a more significant decrease in DMI and total ME intake when cows were infused with propionic acid than the other metabolites.These results suggest that fuels taken up by the liver that are anaplerotic of the TCA cycle, such as propionate, promote oxidation of acetyl CoA and hypophagia (Gualdrón-Duarte and Allen, 2017), likely via increased hepatic energy charge and satiety signals relayed to the brain.In addition, this metabolic control of FI can occur within the timeframe of a meal as intraruminal infusions of propionate within 5 or 15 min during meals exhorted hypophagic effects; however, effects on feeding behavior (e.g., meal size and frequency) were different for cows in the early postpartum period compared with cows later in lactation, seemingly dependent on the physiological stage of the cow (Bradford and Allen, 2007;Maldini and Allen, 2018).This is likely explained by the differential supply of acetyl CoA available for oxidation in the liver consistent with the lipolytic state of the cow at different stages of lactation.In support of this notion, intraruminal infusions of propionate resulted in a larger decrease in DMI in cows with elevated concentrations of hepatic acetyl CoA during early lactation (Stocks and Allen, 2012).Practical implications for the nutritional management of dairy cattle derived from the HOT framework are addressed in companion publications from the same symposium.

CONCLUSIONS
Feed intake is controlled by multiple signals integrated in the brain feeding centers, and altered feeding behavior in response to hepatic oxidation suggests that the liver is a key sensor of energy status integrating long-and short-term controls.The HOT provides a physiological mechanism for explaining how metabolism of fuels in the ruminant liver can regulate FI and feeding behavior.Understanding hepatic metabolism of specific fuels and their effects on satiety or hunger response can help with better formulation of diets for dairy cows.

Table 1 .
Albornoz et al.:RUMINANT NUTRITION SYMPOSIUM Summary of key hypotheses proposed to explain how intake of nutrient-dense diets is constrained in mammals Albornoz et al.: RUMINANT NUTRITION SYMPOSIUM