2.1. Leptin

Leptin, encoded by the lep gene (Halaas et al., 1995; Munzberg & Morrison, 2015; Zhang et al., 1994), is a peptide hormone that promotes negative energy balance via mechanisms including reducing food intake, increasing energy expenditure, and increasing lipolysis in adipose tissue (Friedman, 2019; Grill & Kaplan, 2001; Harris, 2014; Pandit et al., 2017; Reidy & Weber, 2000). Leptin is primarily produced and secreted from adipose tissue (Frederich et al., 1995; Lonnqvist et al., 1995; Maffei et al., 1995), but is also synthesized in other organs including skeletal muscle (Wang et al., 1998), heart (Purdham et al., 2004), and stomach (Bado et al., 1998). Humans and non-human animals with mutations in the lep gene that result in decreased leptin levels [e.g., OB (human) or ob (rat, mouse)] exhibit obesity (Farooqi et al., 2001; Zhang et al., 1994), demonstrating the importance of leptin in the control of energy balance. In rodent models, leptin reduces food intake when administered peripherally or centrally, and is effective to suppress feeding when administered either chronically or acutely (Brown et al., 2006; Halaas et al., 1995; Harris et al., 1998; Keung et al., 2011; Seeley et al., 1996). Leptin also can work cooperatively with other feeding-related peptides such as insulin (Air et al., 2002) and amylin (Li et al., 2015; Mietlicki-Baase et al., 2015; Turek et al., 2010) to promote negative energy balance.

In addition to its general suppressive effects on caloric intake, leptin impacts macronutrient intake and preference in rodent models. Acute systemic or central administration of leptin in rats given a choice of carbohydrate, protein, and fat decreases fat and protein intakes (Wetzler et al., 2004; Wetzler et al., 2005). Age may be a factor in the acute effects of leptin on carbohydrate intake, as older (14 week old) but not younger (8 week old) rats significantly reduced their carbohydrate intake in response to acute peripheral leptin (Wetzler et al., 2004). In contrast, chronic systemic administration of leptin over a 1-week period produced slightly different effects than did acute administration. Rats treated with leptin delivered by osmotic minipump had a more persistent reduction of protein intake, as well as a transient decrease in carbohydrate intake, compared to their baseline (pre-minipump) intake (Wetzler et al., 2004). These data collectively support the notion that exogenous leptin influences macronutrient intake but produces different effects on macronutrient selection when administered acutely or chronically.

There is compelling evidence that endogenous leptin signaling affects feeding (Brunner et al., 1997; Hayes et al., 2010) but studies investigating specific effects on macronutrient intake are limited. Experiments using leptin-deficient ob/ob mice show that they consumed a greater percentage of their calories from fat, and a lower percentage from carbohydrate, compared to lean controls (Currie, 1993; Currie & Wilson, 1992). Other studies have used the complementary strategy of manipulating expression of the leptin receptor to examine the effects of endogenous leptin on food or macronutrient intake. For example, AAV-mediated knockdown of hindbrain leptin receptors increases intake of a 60% fat mixed-macronutrient diet as well as carbohydrate (sucrose) (Hayes et al., 2010), indicating the pivotal role of leptin action in this area of the brain for feeding. Still, studies specifically examining the role of endogenous leptin signaling in rats given access to isolated macronutrients are needed to further understand the role of this hormone in macronutrient selection.

2.2. Glucagon-like peptide-1

Glucagon-like peptide-1 (GLP-1) is a peptide that is produced peripherally by enteroendocrine open-type cells in the intestine (Steinert et al., 2017). GLP-1 is also produced centrally in the nucleus tractus solitarius (NTS) of the hindbrain (Holst, 2007; Merchenthaler et al., 1999). GLP-1 is categorized as an incretin hormone due to its ability to stimulate insulin secretion via activation of G protein-coupled GLP-1 receptors (GLP-1R) on pancreatic beta cells; this serves to decrease blood glucose levels after consumption of a meal in a physiologically relevant manner (Hayes et al., 2014; Kim & Egan, 2008; Nauck & Meier, 2018). In addition to its glycemic effects, GLP-1R activation promotes negative energy balance, producing hypophagia and weight loss in humans and in non-human animal models (Dube et al., 2020; Gunn et al., 1996; Raun et al., 2007; Scott & Moran, 2007; Sun et al., 2015; Turton et al., 1996; Vilsboll et al., 2012; Yang et al., 2014). Although some data suggest that physiological intake- and weight-suppressive effects of GLP-1 may be minimal (Punjabi et al., 2014; Steinert et al., 2014), other work supports a role of endogenous GLP-1R signaling in energy balance control (Alhadeff et al., 2017; Liu et al., 2017; Lopez-Ferreras et al., 2018) and there is clear evidence in support of the pharmacological relevance of GLP-1R activation for the control of feeding and body weight (Drucker, 2022; Muller et al., 2019). Understanding the roles of centrally- and peripherally-synthesized GLP-1 in energy balance remains an ongoing area of investigation (Daniels & Mietlicki-Baase, 2019), but an elegant series of studies has recently provided evidence that the effects of centrally- versus peripherally-produced GLP-1 on food intake are separable (Brierley et al., 2021). Although this does not rule out the possibility that some peripherally-made GLP-1 may access the CNS, for example by crossing the blood-brain barrier (Daniels & Mietlicki-Baase, 2019; Kastin et al., 2002), this underscores the need for further research to parse the roles of central versus peripheral GLP-1 in energy balance control.

The anorectic effects of GLP-1 signaling are well-established, but information on the ability of GLP-1 / GLP-1R activation to control intake of individual macronutrients is more limited. Several studies have established an intake-suppressive effect of GLP-1R agonists on individual macronutrients, but often examine intake when a single macronutrient is available in isolation – for example, suppressing carbohydrate (sucrose) intake when that is the only isolated macronutrient available (Alhadeff et al., 2012; Pritchett & Hajnal, 2012; Zhang & Ritter, 2012). Some studies have investigated the effects of GLP-1R agonists on intake of pure macronutrients when more than one type is concurrently available, although typically when an additional mixed-macronutrient food is also offered. For example, when fat (lard) and sucrose are available alongside chow as part of a choice diet, GLP-1R activation suppresses intake of all of these foods in rats (Lopez-Ferreras et al., 2019), although pure protein was not available in these experiments. Similarly, in studies investigating intake of a cafeteria diet with foods specifically enriched in particular macronutrients (e.g., low or high in fat and/or carbohydrate), systemic administration of liraglutide, a long-acting GLP-1R agonist, reduced food intake but did not specifically decrease intake of a particular food or macronutrient (Hyde et al., 2017). In humans with type 2 diabetes mellitus, food intake at a buffet-style meal was reduced by daily liraglutide treatment, but no significant changes in the percent of calories obtained from each macronutrient were detected (Flint et al., 2013). Similar results were detected after daily liraglutide treatment in healthy individuals, with energy intake and individual macronutrient intakes all suppressed by GLP-1R activation (Quast et al., 2021). However, administration of lixisenatide, another GLP-1R agonist with a half-life longer than that of native GLP-1 but shorter than that of liraglutide [reviewed by (Barnett, 2011; Yu et al., 2018)], produced selective suppressive effects on carbohydrate intake (Quast et al., 2021). This suggests the importance of variables related to timing and duration of drug effect in detecting specific intake-suppressive effects of GLP-1R activation on a single macronutrient. This notion is also supported by other work. A study using peripheral administration of the GLP-1R agonist exendin-4 showed that rats given a choice to eat a high-carbohydrate or high-protein diet displayed differential effects on intake of the diets depending on timing of measurement; early on, intake of the high-protein diet was reduced by exendin-4, while GLP-1R activation suppressed intake of the high-carbohydrate food later in the experimental period (Peters et al., 2001). Furthermore, some doses of exendin-4 selectively reduced intake of the high-protein diet with no significant effects on high-carbohydrate intake (Peters et al., 2001). An important limitation to point out in these studies is that no high-fat diet option was available during testing.

Other research points to possible effects of GLP-1R activation on macronutrient intake when considering habitual intake. Individuals who display higher fasted GLP-1 levels tended to have lower carbohydrate intake over a 3-day period (Basolo et al., 2019). This is consistent with the aforementioned ability of GLP-1R activation with lixisenatide to specifically decrease carbohydrate intake (Quast et al., 2021). In contrast, in humans with overweight and type 1 diabetes mellitus, GLP-1R activation shifted proportions of macronutrient consumption, decreasing the percentage of reported energy intake derived from fat and increasing carbohydrate intake percentage (Dube et al., 2020), although the use of food frequency questionnaires to probe effects on intake is a limitation that should be addressed in future work. These results underscore the point that the ability of hormones like GLP-1 to suppress food intake may differ in disease states, an idea that will be explored later in this review. Overall, the mixed results of studies of the effects of GLP-1R signaling on macronutrient intake point to the importance of considering experimental variables such as the types of foods offered and the method of determining macronutrient intake (e.g., measuring food intake in a laboratory setting versus recall via food frequency questionnaires), the timing of measurements, and duration of GLP-1R activation. Moreover, research examining effects on intake when animals are offered all three macronutrients in isolation is clearly needed.

2.3. Insulin

One of the most vital anabolic hormones in the body is insulin. This peptide hormone is produced by beta cells of pancreatic islets and is responsible for promoting the uptake of glucose from the blood by other organs (Wilcox, 2005). The release of insulin is largely stimulated by the increased presence of glucose in circulation (Saltiel, 2016; Thorens, 2015). After consumption of a meal, especially one rich in carbohydrates, digestion and breakdown of the nutrients will increase circulating glucose levels, resulting in insulin release to facilitate the uptake and utilization of glucose by organs such as muscles and liver (Petersen & Shulman, 2018). As such, insulin is well-known as a major controller of blood glucose levels (Roder et al., 2016).

When independent intakes of each macronutrient are considered, several studies demonstrate that insulin can potently suppress intake of fat. For example, when rats were given sources of pure fat, pure carbohydrate, and pure protein, intracerebroventricular administration of insulin reduced fat intake (Chavez et al., 1996). Similarly, when separate foods enriched in each macronutrient were available, insulin injection directly into the arcuate nucleus of the hypothalamus reduced intake of the high-fat diet as well as total fat intake (van Dijk et al., 1997). This is not to say that insulin cannot affect the intake of other macronutrients; for example, carbohydrate intake was decreased in men after administration of insulin via the intranasal route (Krug et al., 2018). However, this study investigated intake of mixed-macronutrient foods in a buffet meal, so it is possible that the nature of the test foods and/or other experimental variables including the species and the route of administration may have played a role in the differential effects observed in these experiments.

Impaired insulin signaling associated with dysregulated glycemic control is the hallmark of diabetes mellitus (Roder et al., 2016; Scheen, 2003). This disease state can also impact macronutrient intake. The association of macronutrient intake with insulin in the context of diabetes will be further explored in Section 3.1 below.

2.4. Amylin

The peptide hormone amylin is produced by the pancreas (Westermark et al., 1987) and in the brain (Dobolyi, 2009; Li et al., 2015; Szabo et al., 2012). Pancreatic amylin is cosecreted with insulin to promote glycemic control (Young, 2005b), and in vitro studies show that glucose can stimulate release of amylin from the pancreas (Inoue et al., 1990, 1991). Amylin reduces gastric emptying, which aids in lowering blood glucose levels (Young, 2005a). In addition to the important role of amylin in glycemia, it also has a critical role in energy balance control. Amylin is well-established as a satiation signal that suppresses feeding and body weight via actions in the CNS (Lutz, 2005, 2022; Mollet et al., 2004; Rushing et al., 2000; Rushing et al., 2002). Numerous studies have established that peripheral or central administration of amylin can reduce intake of mixed-macronutrient diets such as chow (Arnelo et al., 1996; Lutz et al., 1994; Rushing et al., 2000; Rushing et al., 2002). However, research into the effect of amylin on intake of specific macronutrients is more limited. One of the major central sites of action for amylin is the area postrema (AP) of the hindbrain (Coester et al., 2020; Lutz et al., 2001; Lutz et al., 1998; Mollet et al., 2004; Potes et al., 2010). The AP is a circumventricular structure in the hindbrain that is well-positioned to detect circulating feeding-relevant signals and communicate this information to the adjacent nucleus of the solitary tract (NTS), an important neural hub for energy balance control (Grill & Hayes, 2009, 2012). The AP is responsive to glucose, and interestingly, glucose-activated neurons in the AP are predominantly also excited by amylin (Riediger, 2012; Riediger et al., 2002), suggesting that the AP may be a site at which information on dietary nutrients and amylin signaling can be integrated to exert control over energy balance.

A limited number of experiments has specifically examined how amylin may suppress intake of particular macronutrients. A few studies suggest that amylin may interact with dietary protein, and specifically, that the ability of amylin to suppress food intake and to activate AP neurons may be blunted in rats given a protein mash versus a carbohydrate, fat, or non-caloric mash (Michel et al., 2007). Similar effects were observed in rats chronically maintained on a higher-protein diet compared to rats chronically maintained on a lower-protein diet (Zuger et al., 2013). Given that amylin-induced c-Fos was attenuated in the AP of rats on the higher-protein options in these studies (Michel et al., 2007; Zuger et al., 2013), this points to the hindbrain as a potential site of interaction for amylin and dietary protein. However, other potential central sites of action were not investigated in these studies. Other experiments have focused on actions of amylin within nuclei of the mesolimbic reward system. Mesolimbic sites including the ventral tegmental area (VTA) and nucleus accumbens (NAc) have prominent roles in the control of palatable food intake. Studies on the effects of mesolimbic amylin receptor activation on intake of particular macronutrients have focused largely on fat and carbohydrate. In both the VTA and NAc shell, amylin receptor activation via direct infusion of amylin receptor agonists reduces intake of sucrose, suggesting effects of amylin on carbohydrate intake – or at least palatable carbohydrate – via action at these sites (Baisley & Baldo, 2014; Mietlicki-Baase et al., 2013). VTA amylin receptor activation has also been examined for its role in the control of fat intake, and in one-bottle tests, intra-VTA administration of the amylin receptor agonist salmon calcitonin (sCT) robustly suppressed fat intake (Mietlicki-Baase et al., 2017). Furthermore, in two-bottle tests in which both fat solution and sucrose solution were available, the effects of VTA amylin receptor activation produced more robust and/or selective effects on fat intake with more minimal effects on intake of sucrose (Mietlicki-Baase et al., 2017). This may indicate that this nucleus is particularly important for amylin to reduce fat intake but again that it can also reduce intake of palatable carbohydrate solution.

While animal studies often provide individual macronutrients to choose from, identification of the particular effects of amylin receptor activation on macronutrient consumption when mixed-macronutrient foods are available is less common, and research on the effects of amylin receptor activation on macronutrient intake in humans is limited. Pramlintide, an amylin analog used in the treatment of diabetes (Singh-Franco et al., 2007), reduced energy intake in adults with obesity or with type 2 diabetes when given systemically but did not significantly change the macronutrient composition of a buffet meal taken after a preload meal in one study (Chapman et al., 2005). However, more research is clearly needed to identify whether a different dosing regimen, such as either a different dose of pramlintide or a chronic administration paradigm, might reveal changes in macronutrient intake.

2.5. Ghrelin

In contrast to the anorectic feeding-related peptides discussed thus far, the peptide hormone ghrelin promotes positive energy balance (Asakawa et al., 2001; Horvath et al., 2003; Kirchner et al., 2012). Produced primarily in the stomach (Kojima et al., 1999), ghrelin has been shown to increase food intake in humans and in non-human animal models in numerous studies (Druce et al., 2005; Keen-Rhinehart & Bartness, 2005; Nakazato et al., 2001; Wren, Seal, et al., 2001; Wren, Small, et al., 2001; Wren et al., 2000). In addition to our understanding of the general hyperphagic effects of ghrelin, several studies have examined how ghrelin impacts the intake of particular macronutrients. In rats given a choice between high-fat and high-carbohydrate diets, intracerebroventricular (ICV) administration of ghrelin in rats produced stronger stimulatory effects on intake of the diet high in fat compared to the high-carbohydrate diet (Shimbara et al., 2004). Other data supports the notion that ghrelin increases fat intake; for example, when rats had a choice of consuming fat (lard), sucrose, and chow, ICV ghrelin increased intake of lard and also of chow, with no significant effect on sucrose intake (Schele et al., 2016). Direct administration of ghrelin into the VTA produced similar effects, with significant increases in chow intake and a trend for increased lard intake (Schele et al., 2016). Interestingly, chronic intra-VTA infusion of a ghrelin receptor antagonist in rats given a choice of diets high in protein, carbohydrate, and fat produced selective reduction in intake of the high-fat diet, although direct chronic administration of ghrelin itself into the VTA in this choice paradigm had no significant effects on energy intake (King et al., 2011). Together, these data suggest that central ghrelin administration has particularly potent effects on fat intake, but it is important to consider whether this is due to the energy density of the different foods available in each study. Central ghrelin may shift food intake toward more calorically dense options (Bomberg et al., 2007). In some studies, the high-fat diet option is more energy-dense than high-carbohydrate and/or high-protein options (King et al., 2011), but in other studies, the energy density of the high-fat diet is similar to other available options (Shimbara et al., 2004).

3.1. Diabetes Mellitus

Diabetes mellitus is a disease characterized by the body’s inability to properly regulate insulin, leading to abnormal blood glucose levels (Sapra & Bhandari, 2022). Type 1 diabetes mellitus (T1DM) is an autoimmune disorder in which the pancreatic beta cells are destroyed, leading to an inability to produce insulin (Lucier & Weinstock, 2022). Some work suggests that individuals with T1DM do not differ in their macronutrient intake in comparison to healthy controls (Turton et al., 2020). In contrast, individuals with Type 2 diabetes mellitus (T2DM) have functional pancreatic beta cells but exhibit insulin resistance (Goldstein, 2002). T2DM is a multifactorial disease associated with risk factors including non-modifiable characteristics like age and genetic background, as well as modifiable factors such as diet and exercise (Chen et al., 2011; Zheng et al., 2018). Because obesity is a major risk factor for T2DM (Barnes, 2011; Maggio & Pi-Sunyer, 2003), macronutrient-related factors that drive obesity (discussed in Section 3.2 below) can indirectly impact the development of T2DM. In this section, we focus on other facets of the relationship between T2DM and macronutrient intake.

Rodent models have been used to investigate macronutrient intake in diabetes and how insulin treatment can affect intake. Rats made diabetic by treatment with streptozotocin (STZ), which destroys pancreatic beta-cells [reviewed by (Furman, 2021)], or by pancreatectomy generally had higher intake of fat and consumed fewer kcal from carbohydrate compared to controls (Bartness & Rowland, 1983; Kanarek & Ho, 1984). The STZ dose mattered in these effects, with rats receiving a higher STZ dose choosing to consume more protein (Bartness & Rowland, 1983); this could be related to the severity of glucose intolerance induced by the different doses of STZ (Bartness & Rowland, 1983), as higher concentrations of STZ produce more beta cell destruction in vitro (Saini et al., 1996) and, in vivo, higher STZ doses produce this effect more rapidly and cause more severe hyperglycemia than do lower STZ doses (Arison et al., 1967; Fazio & Pin, 2007; Yale et al., 1986). STZ-induced diabetic animals also display reduced preference for higher concentrations of sugars than do nondiabetic controls (Tepper & Friedman, 1991). Generally speaking, the food intake of rats with STZ-induced diabetes is tied to the fat content of the diet; they will eat more food as the fat content in their food is reduced, but will not alter their intake if carbohydrate is reduced in the food by a similar amount (Friedman et al., 1985). This appears to be driven by the fact that the fat in the food is a usable source of fuel for the diabetic animals (Edens & Friedman, 1988; Friedman, 1978). Insulin treatment is effective to normalize carbohydrate intake in rodent models of diabetes (Bartness & Rowland, 1983), suggesting that the changes in carbohydrate intake are related to the altered insulin secretion / signaling in diabetes mellitus. Although numerous studies of animal models of diabetes focus mainly on effects on fat and carbohydrate intake, it should be noted that some studies also demonstrate increased protein intake in STZ-diabetic rats (Booth, 1974; Tepper & Kanarek, 1989), typically in conjunction with a reduction in carbohydrate intake (Tepper & Kanarek, 1989). Interestingly, diabetic rats do not always defend protein intake when protein in the diet is adulterated with cellulose to provide less protein per gram, but do defend fat intake during fat dilution and also defend total energy intake (Tepper & Kanarek, 1989). This contrasts with the rather robust defense of protein intake generally observed in healthy animals [reviewed by (Khan et al., 2021)] and underscores the need to understand controls of macronutrient intake not only in healthy individuals but also in various disease states.

Changing macronutrient composition of the diet can be a useful strategy in managing diabetes (Koloverou & Panagiotakos, 2016; Wheeler et al., 2012). Reductions in carbohydrate intake could be beneficial in managing the glycemic dysregulation of diabetes, because rats with STZ-induced diabetes that were maintained on a lower carbohydrate diet for several months had improved blood glucose levels compared to those on a higher carbohydrate diet (Siegel et al., 1980). Importantly, however, the quality of the carbohydrates makes a difference in diabetes outcomes in animal models; sucrose-rich diets produced worse outcomes, while fiber-rich diets helped to improve glycemic control (Marques et al., 2020). Low carbohydrate diets are also thought to be beneficial in the management of diabetes in humans (Wheatley et al., 2021), although not all individuals with diabetes are able to adhere to all macronutrient recommendations (Barclay et al., 2006; Helmer et al., 2008). Some work suggests that elevated carbohydrate intake may be a risk factor for the development of T2DM (Alhazmi et al., 2012), albeit possibly specifically in individuals with obesity (Sakurai et al., 2016). Interestingly, when comparing people with diabetes to healthy controls, some studies report no differences in macronutrient intake (McClure et al., 2020; Murray et al., 2013).

Because dietary changes are often recommended in the course of diabetes management, it is perhaps not surprising that knowing that one has diabetes mellitus can impact subsequent intake. Indeed, individuals with a diabetes diagnosis had increased protein intake and, in men, significantly lower carbohydrate intake than counterparts who had undiagnosed diabetes (Bardenheier et al., 2014; Wang et al., 2016), suggesting the important role of knowing one’s diabetes status in selecting one’s macronutrient intake. However, many individuals are not able to successfully follow macronutrient recommendations after a diabetes diagnosis (Barclay et al., 2006; Helmer et al., 2008), which could blunt the effectiveness of this type of dietary strategy in managing diabetes. Furthermore, obesity and T2DM are often comorbid conditions (Guh et al., 2009; Khaodhiar et al., 1999). Individuals with overweight/obesity who reduce their body weight exhibit increased insulin sensitivity (Clamp et al., 2017), and therefore dietary macronutrient changes can tie into both treatment of T2DM as well as reducing body weight in obesity.

3.2. Obesity

Obesity, a disease characterized by higher weight status than is considered healthy for one’s height and accumulation of excess adipose tissue (Apovian, 2016; Engin, 2017), is another prevalent metabolic condition that involves alterations in food intake and macronutrient consumption. The development of obesity is thought to be caused in part by overconsumption of energy dense foods, which are often highly palatable and rewarding (Leigh et al., 2018; Meye & Adan, 2014; Rolls, 2007). These foods are often high in fat and/or carbohydrate, pointing toward a role of macronutrient intake in the etiology of obesity. Several studies link excess dietary fat intake to elevated body mass index and the development of obesity in rodent models and in humans (Hill et al., 2000; Reed et al., 1997), although there is some debate in the literature (Ahluwalia et al., 2009). One challenge in understanding macronutrient intake in obesity is that many studies rely on dietary recall for tracking intake, which can be inaccurate. Some studies indicate that individuals with overweight and obesity underreport their energy intake by approximately 40% in dietary recalls, particularly with regard to underreporting of fat intake, and overreport low-fat food intake (Heitmann & Lissner, 1995; Lichtman et al., 1992). This can obscure the detection of associations between fat intake and weight status.

Similar to dietary strategies to help manage T2DM, altering macronutrient intake is one possible strategy to reduce body weight in overweight/obesity. However, whether and what specific changes should be made to the macronutrient composition of the diet to promote weight loss remains somewhat controversial. Also, altered macronutrient intake could impact body weight through different mechanisms, as changes in the macronutrient composition of the diet could directly affect intake and/or could change nutrient partitioning to alter adiposity [reviewed by (Hall et al., 2022)]. An in-depth examination of this topic is beyond the scope of this review, but in general, isocaloric dietary strategies that lower intake of a single macronutrient (i.e., low-fat or low-carbohydrate) are capable of producing similar reductions in body weight (Axen & Axen, 2006; Brinkworth et al., 2009; Brinkworth et al., 2016; Hall et al., 2015), and a recent review/meta-analysis showed that low-fat and low-carbohydrate diets generally produce similar body weight reductions (Ge et al., 2020). Increasing protein in the diet can also effectively induce weight loss (Weigle et al., 2005). However, regardless of the macronutrient focused on in the diet, the suppressive effects on body weight are typically driven by an overall reduction in caloric intake (Howell & Kones, 2017). Some studies suggest that a low-fat diet may promote fat loss more effectively than a low-carbohydrate diet (Hall et al., 2015; Hall et al., 2021), but others have found no differences in fat mass after low-carbohydrate versus low-fat diets (Veum et al., 2017). It is important to note that some of these diets can have undesired side effects and safety must be carefully considered (Barber et al., 2021; Freire, 2020).

Available treatments for obesity include pharmacotherapies and bariatric surgery, which can also induce changes to macronutrient intake. Bariatric surgery such as Roux-en-Y gastric bypass (RYGB) is one treatment option for obesity, although this is only recommended in certain cases (Mitchell & Gupta, 2022). Rodent studies show that RYGB can alter subsequent macronutrient selection in rats on a cafeteria diet by suppressing intake of fat (Blonde et al., 2021; Mathes et al., 2016), and also shifts food choices towards a lower-fat option (Zheng et al., 2009). However, the long-term impacts of RYGB on food/macronutrient selection in humans are unresolved (Behary & Miras, 2015; Mathes & Spector, 2012; Nielsen et al., 2017). Also under investigation is whether particular profiles of macronutrient intake are associated with weight loss or with weight regain after bariatric surgery. Some research suggests that individuals who had bariatric surgery and consumed more protein in the diet relative to fat or to carbohydrate had greater weight loss than those consuming more fat or more carbohydrate relative to protein (Kanerva et al., 2017). However, other research detected no significant differences in macronutrient intake among individuals who regained weight after bariatric surgery versus those who did not (Nymo et al., 2022), so this remains an open empirical question. Pharmacotherapeutic strategies for treating obesity are less invasive than bariatric surgery and also can change macronutrient intake. The GLP-1R agonists liraglutide and semaglutide are FDA-approved for the treatment of obesity (Latif et al., 2022) and, as discussed above in Section 2.2, GLP-1R activation can alter macronutrient intake. Other strategies, not yet FDA-approved but considered promising candidates for obesity treatment, focus on targeting some of the other feeding-relevant hormones discussed previously in this review, such as amylin (Boyle et al., 2018; Hay et al., 2015; Srivastava & Apovian, 2018) and ghrelin (Alvarez-Castro et al., 2013; Liang et al., 2021; Nagi & Habib, 2021). Given the ability of these hormones to alter macronutrient intake, these potential pharmacotherapeutic strategies may also impact intake.

4.1. Strain and Species

When thinking about the study of macronutrient intake in rodent models, one important consideration is the choice of strain. Baseline macronutrient intake patterns have been shown to vary among different mouse strains, with some strains displaying a strong carbohydrate preference and others displaying a strong fat preference (Smith et al., 2000; Tordoff et al., 2014). The strain chosen may depend on the outcomes being evaluated. For example, the effects of a treatment that is expected to decrease fat preference may be different in a strain displaying a strong carbohydrate preferring phenotype versus a fat-preferring strain due to baseline differences in fat intake. Additionally, each macronutrient category is not homogenous. A high carbohydrate test food may be composed of any combination of mono-, di-, and/or polysaccharides, and various strains display different preferences for these subtypes of carbohydrate. This is an important factor for consideration when determining composition of a test food. C57BL/6J (B6) mice are sweet sensitive and prefer Polycose, a polysaccharide primarily composed of glucose polymers, over sucrose at high concentrations, and prefer the opposite at low concentration (Ackroff & Sclafani, 2016). 129/P3J (129) mice, which are sweet subsensitive, display an opposite preference profile which may be explained by this difference in sweet sensitivity (Ackroff & Sclafani, 2016). This difference may be due to genetic variation in the T1R3 receptor, which is a component of the sweet taste receptor. There are genetic variations between strains in the gene encoding the T1R3 protein resulting in two variants of the receptor. Some strains, including B6 mice, display one variant and are more sensitive to sugars than other strains, including 129 mice, which display the other variant (Inoue et al., 2004; Reed et al., 2004). Additionally, mouse models utilizing knockout of the T1R3 receptor show a decreased intake of mono- and disaccharide sweeteners such as sucrose, glucose, and maltose (Damak et al., 2003; Treesukosol & Spector, 2012), but not Polycose (Treesukosol & Spector, 2012). The effects of post-oral reinforcement may also be a factor due to the compositional differences between Polycose and sucrose. Polycose contains double the amount of glucose than is contained in sucrose, a disaccharide composed of both glucose and fructose. Glucose is known to have strong post-oral conditioning effects in B6 mice (Ackroff & Sclafani, 2016). Indeed, post-oral conditioning in response to different monosaccharides has been shown to differ among mouse strains using intragastric infusion models, with B6 mice showing strong response to glucose but not fructose (Sclafani & Ackroff, 2012), and other strains such as FVB showing a strong response to fructose but not glucose (Sclafani et al., 2014).

Although these previously mentioned studies were examining preferences for various sweeteners/saccharides, they are relevant when considering macronutrient intakes because this baseline difference in preference for different saccharides, whether due to difference in taste receptors, post-oral reinforcement, or other mechanisms, has the potential to affect carbohydrate intake. This may be particularly important for studies examining “palatable” macronutrient intake. As genetic differences among various rodent strains may have unknown effects or produce subtle variations in taste preferences or macronutrient intakes, it is possible that any changes in macronutrient intake outcomes could be strain specific. It is crucial that observations be replicated among multiple strains in order to determine the generalizability of results of macronutrient intake studies.

Species is also a consideration in studies of macronutrient intake, particularly when trying to compare between species or, in non-human animal models, to enhance the translational relevance of the results. We acknowledge that animal research in the neural controls of feeding has reported on effects in a variety of species, but here we limit our discussion to the consideration of using mice and rats, as these are two commonly used animal models in this field. Mice and rats differ in many ways and there are pros and cons to the use of each of these models in scientific research (Bryda, 2013; Ellenbroek & Youn, 2016; Iannaccone & Jacob, 2009; Parker et al., 2014), including the study of ingestive behavior. For example, in thinking about the effects of various hormones on macronutrient intake, differences in the GLP-1 system have been identified between rats and mice (Huo et al., 2008; Lachey et al., 2005; Mietlicki-Baase et al., 2018; Perez-Tilve et al., 2010; Terrill et al., 2019). Similar to the concern regarding choice of strain, the choice of species is an important consideration in this type of research, and one cannot assume that the results observed in mice will be the same as those observed in rats (and vice versa), nor assume that results obtained in rodent models will fully recapitulate the physiology or behavior of humans.

4.2. Palatability

As discussed in the preceding section, issues of palatability and food reward can also complicate studies of macronutrient intake and preference. Researchers must consider the possibility that when an experimental manipulation alters the intake of a particular macronutrient, it could be due to the intervention impacting palatability, not necessarily a mechanism strictly affecting consumption of that macronutrient. Indeed, various types of food or sources of food can be representative of a single class of macronutrient, so testing whether the effects of a treatment on “macronutrient intake” actually extend to various forms of that macronutrient can provide critical insight to this issue. For example, a study of the effects of FGF21 on carbohydrate intake and preference revealed that FGF21 reduces consumption of some sugars including sucrose and glucose, but that this suppressive effect did not extend to maltodextrin (von Holstein-Rathlou et al., 2016), demonstrating the importance of investigating multiple types of a macronutrient before concluding that there is a general effect on intake of that category of nutrient.

Considering issues of palatability and reward is particularly important for sugars and fats, which are known to be highly palatable (Drewnowski, 1997). Accordingly, as researchers we must ask ourselves – if a treatment is shown to decrease sucrose intake, for example, is it actually because of a suppression of carbohydrate intake, or is it because the treatment reduces the rewarding value and/or perceived palatability of that food and in turn decreases intake? This is an issue that must be carefully considered before concluding that a particular hormone, signal, or experimental manipulation changes intake of a specific macronutrient.

4.3. Novelty and Learned Preference

The novelty of a test food and/or previous experience with a particular food are also important considerations when examining macronutrient intakes. Indeed, previous experience with a particular food, or a food with a similar macronutrient content, has the potential to bias preference towards that food, at least initially. For example, Tordoff and colleagues noted when testing macronutrient preferences of different mouse strains that the majority of strains displayed an initial preference for carbohydrate test food, which had a similar carbohydrate content to rodent chow (Tordoff et al., 2014). As the mice were maintained on standard chow prior to the experiments, they hypothesized that prior experience with a similar food was the driving force behind this initial preference as they are both high carbohydrate foods (Tordoff et al., 2014). Additionally, Hoch et al. (Hoch et al., 2014) noted that in rats given a choice between a test food containing chow mixed with potato chips, or a test food designed to have similar fat/carbohydrate composition to the potato chip test food, rats initially preferred the potato chip mix over the food of similar macronutrient composition for the first 3 days of testing, but gradually switched to equal preference between the two. It is important to note that the rats had previously been exposed to the potato chip test food in other experiments examining individual macronutrients, but had not been exposed to the test food with the comparable macronutrient composition. This prior experience to one food but not the other may explain these findings. It is well established that rodents exhibit neophobia to novel stimuli, whereby intake of food may be limited due to its novelty, but after a period of exposure the food becomes known and intakes increase (Modlinska et al., 2015). Care should be taken in study designs to avoid pairing known and unknown foods, as lower intakes of the unknown food may in fact be a result of neophobia.

Despite the potential for neophobia to a novel food, there is also some evidence that novel, highly palatable foods can interact with other factors to affect intakes in response to treatments. As shown by Boyle and colleagues (Boyle et al., 2018), rats exposed to chocolate Ensure for three weeks decreased intake in response to amylin treatment, however this decrease was not seen in rats treated with amylin after only three days of exposure to chocolate Ensure. In contrast, rates with short term exposure to unflavored Ensure or a high fat diet with similar macronutrient composition did show a decrease in intake after amylin administration, leading the authors to conclude that the chocolate flavor was the driving factor for this effect. It is possible that this novel chocolate flavor was too strong of a motivating force to consume the food, making it impossible to see any effects of amylin after a short exposure. However, after a longer habituation this was attenuated. This potential for flavoring agents or other aspects aside from macronutrient composition to affect observed macronutrients intakes (and intakes in response to treatment) must be considered when choosing test foods, as sometimes flavoring is added to test foods to make them more palatable. Care should be taken when drawing conclusions, as the flavoring agent may affect results, and sufficient habituation to these foods is imperative.

4.4. Sex Differences

The influence of sex on feeding is clear, with different energy intake in males and females driven by a variety of factors including sex hormones like estrogens and testosterone as well as sex differences in energy need (Zhao et al., 2020). Sex differences in energy intake is a topic that has been covered in the literature and the reader is pointed toward several excellent reviews of this area of research (Asarian & Geary, 2013; Massa & Correa, 2020; Sample & Davidson, 2018). However, the impact of sex differences on macronutrient intake is less frequently discussed. We discuss sex differences here as many studies have compared differences between males and females (i.e., different chromosomal sex), but it is important to note that biological sex and gender identity may differ. The role of gender in human research is underinvestigated but can be an important consideration in understanding neurobiology and feeding-related behavior (Culbert et al., 2021; Ristori et al., 2020).

Research comparing macronutrient intake between adult males and females has produced mixed results. For example, while some studies suggest that females consume a higher percentage of energy from carbohydrates than do males (Lieberman et al., 2020; Zhao et al., 2020), other findings have not detected this difference (Paul et al., 2004; Sun et al., 2021). Protein intake as a percentage of energy intake has been reported in females to be lower (Lieberman et al., 2020) or higher (Sun et al., 2021) than in males. These divergent findings may be due to methodological differences, such as the population studied or the method of identifying macronutrient consumption.

Another factor is the cyclical fluctuations of sex hormones in females. A large body of literature, including human and rodent work, has shown that estrogens reduce energy intake in females (Asarian & Geary, 2013; Blaustein & Wade, 1976; Dye & Blundell, 1997; Leeners et al., 2017). Some research has investigated macronutrient intake over the course of the menstrual or estrous cycle. Some reports suggest that protein intake increases during the luteal phase of the menstrual cycle (Chung et al., 2010; Gorczyca et al., 2016) but here too, there are discrepant findings (Ihalainen et al., 2021; Johnson et al., 1994). Other work suggests an increase in fat intake during the luteal phase instead (Johnson et al., 1994), and elevated carbohydrate intake during the follicular phase (Chung et al., 2010). Studies on the estrous cycle in rodents have also found no differences in protein intake over the cycle, instead finding effects on the other two macronutrients, with decreased fat intake and increased carbohydrate intake during diestrus and vice-versa in estrus (Bartness & Waldbillig, 1984), contrasting with the aforementioned results in humans. Similar to the challenges of interpreting the literature on differences in macronutrient intake between males and females, studies on differences by estrous / menstrual cycle phase can be difficult to compare because of technical differences in data collection [see for review (Becker et al., 2005; Dye & Blundell, 1997; Hampson, 2020)]. These challenges have made it difficult to get a clear picture of how macronutrient intake differs between the sexes, but furthermore, deeper analysis is required due to differences in reproductive hormones in cycling females at different phases of the cycle which can influence intake within sex.

This work was supported by NIH grant number DK128030 (EGM-B).