Since the debut in 2005 (1), fibroblast growth factor 21 (FGF21) has been of growing interest due to its dramatic beneficial effects at pharmacological levels on weight reduction and alleviation of obesity, diabetes, and fatty liver disease (2, 3). These effects include promoting clearance of systemic glucose and lipids (cholesterol, fatty acids, and triacylglycerides) and enhancing insulin sensitivity, adiponectin action, mitochondrial function, thermogenesis, and energy expenditure (1, 2, 4). Unlike most of the FGF family members that require extracellular heparan sulfate proteoglycan and act locally via autocrine and paracrine mechanisms, FGF21 as a member of the endocrine FGF19 subfamily travels through the circulation to sites distal from its origin and acts predominantly as an endocrine hormone (5). Instead of heparan sulfate proteoglycans, the activity of FGF21 is determined by a transmembrane co-receptor betaKlotho (KLB) that is present as a binary complex with FGF receptor tyrosine kinases (FGFR) in several endocrine and metabolic tissues. This allows selective action of circulating FGF21 on the target tissues expressing FGFR1-KLB complex without affecting those expressing FGFR1 alone.

Basal serum levels of FGF21 vary widely among individuals and are generally low in pure strains of laboratory mice (6). Deficiency or overexpression of FGF21 does elicit alterations in metabolic gene expression in the liver and adipose tissue, but without dramatic metabolic phenotypes under normal physiological conditions (7, 8). In contrast, pharmacologic applications of FGF21 in obese and diabetic animals suggest a wide range of effects across multiple tissues with marked anti-obesogenic and anti-diabetic efficacy (1, 2, 4, 9). However, beyond such drastic pharmacological effects in the obese, the possible functions of FGF21 as a circulating hormone in organ-specific physiopathologies have remained elusive.

Studies in mice indicate that FGF21 is inducible and the liver is a major source of circulating FGF21 (9, 10). Parallel to serum levels, expression of hepatic FGF21 under normal physiological conditions is low; however, during prolonged fasting and starvation, both hepatic and serum FGF21 become dramatically elevated. This triggered early proposals that the physiological role of FGF21 was a starvation hormone that regulates ketogenesis essential for brain function during severe carbohydrate deficits (9, 10). A growing number of studies with animals and patients indicate that FGF21 is induced, in addition to starvation and obesity, by diverse pathogenic conditions such as liver injury, viral infection, chemical insult, specific nutritional deficiency, the hepatic regenerative response as well as liver diseases as hepatosteatosis, steatohepatitis, cirrhosis, and liver cancer (11–17). The common feature of all these conditions is that they impose stress on the liver that compromises its role in maintenance of organism metabolic homeostasis. Therefore, hepatic FGF21 appears to be an inducible hepatic stress signal.

In contrast to the liver that is a switching station for processing metabolites in support of other organs and tissues and the adipose tissue that serves as a storage depot of energy and metabolic precursor of lipids, muscle is a major fuel consumer. To meet high aerobic catabolic demands for ATP, muscle secretes myokines that call on liver and adipose tissue to supply adequate fuel. Consistent with this concept, several recent studies indicate that FGF21 is induced in skeletal, heart, and gastrocnemius muscle under conditions that cause local and systemic metabolic stress (18–25). These include patients with a mitochondrial respiratory chain deficiency in myocytes and mice defective in muscular autophagy/mitophagy (20, 26, 27). FGF21 may be an insulin and AKT-regulated myokine and its expression is associated with chronic muscular hyperinsulinemia or lipodystrophy (22, 23). Increases in muscle contraction such as in chronic exercise also upregulates muscular FGF21 production (21, 25). Fe-S cluster-deficient muscles in patients showed a dramatic upregulation of FGF21 expression and elevated levels of circulating FGF21 (20). This indicates that muscle stressed by perturbations in mitochondrial energy metabolism responds by increasing the secretion of FGF21 into the circulation.

In addition to liver and muscle, it has been suggested that FGF21 is inducible and secreted into the circulation in other tissues and organs under stress conditions. Brown adipose tissue (BAT) may be a source of FGF21 in response to cold stress and fetal-to-neonatal transition (28, 29). White adipose tissue (WAT) has been reported to secrete FGF21 under some conditions (30). However, a physiological contribution of stress-induced adipocyte FGF21 to systemic FGF21, or an autocrine activity of FGF21 within WAT, BAT, and adipocytes in the local microenvironment, remains to be validated. Since stressed pancreas has been reported to be a beneficiary of external FGF21 (31, 32), it is also a candidate for induction of FGF21 as a signal for aid from adipose tissue.

These studies suggests that the activation of FGF21 in liver, muscle, and other tissues is a response of these tissues and organs to stress triggered by external challenge and internal cellular pathogenesis that lead to defects in metabolic homeostasis. In support of this notion, several nuclear receptors and transcription factors that are involved in metabolic stress responses regulate FGF21 expression. These include PPARα/γ, RAR, ChREBP, SREBP1c, LXR, STAT3/5, p53, and ATF4 (9, 16, 26, 33–39). The question is why these organs when under stress use FGF21 as a systemic alarm signal.

Several recent studies using direct ablation of FGFR1 or KLB in adipocytes, administration of activating antibody targeting specifically the FGFR1-KLB complex, or induction of dysfunctional adipose tissue (40–44) provide compelling evidence that adipose tissue and adipocyte FGFR1-KLB is the primary tissue and molecular target of inter-organ messenger FGF21, respectively. Although many types of tissues and cells express FGFR1 alone, mature adipocytes co-express FGFR1 with KLB, the transmembrane co-receptor for FGF21 and FGF19 (45). Similar to the general ablation of FGF21 (8), adipocyte-specific FGFR1 ablation resulted in few overt changes in adipogenesis, adipose secretory function, and systemic metabolic status under normal physiological conditions (46). However, under the metabolic stress of starvation, the absence of adipocyte FGFR1 caused increased adipocyte lipolysis. This occurred concurrently with elevation of serum triglycerides and fatty acids while indirectly causing an increase in hepatic lipogenesis and steatosis (46). This indicated that under such conditions adipocyte FGFR1 signaling dampens adipocyte lipolysis while concurrently dampening hepatic steatosis. Paradoxically yet most dramatically, the adipocyte FGFR1 deficiency abrogated nearly all the anti-obese and anti-diabetic effects elicited by pharmacological levels of FGF21 (40, 43). The alleviation of muscular abnormalities caused by muscular mitochondrial energy dysfunction and defects in autophagy/mitophagy appear to be in large part due to effects of induced FGF21 on adipose tissue (26, 27). Adaptation to cold stress may be aided by elevated levels of FGF21 that acts on BAT to promote pro-thermogenic mitochondrial activity (28, 29).

Taken together, these studies indicate that stress messenger FGF21 and adipose FGFR1-KLB team up to respond to stress caused by a wide range of metabolic abnormalities and pathogenesis. The stress-responsive communication mediated by extra-adipose FGF21 and adipocyte FGFR1-KLB occurs predominantly between adipose tissue and at least two tissues whose roles in metabolism are quite different, but when stressed have major consequences on the organism: (1) liver whose role is to maintain lipid, carbohydrate, and organism metabolic homeostasis through synthesis, degradation, and temporary storage; and (2) muscle which is a major fuel consumer predominantly powered by the oxidation of fats and carbohydrates (Figure 1). Although so far liver and muscle have received the most attention as two major organs that send out FGF21 as a stress signal, other organs may utilize the same mechanism to call on adipocytes in major fat depots (WAT and BAT) or in the microenvironment of diverse tissues. Within the microenvironment of tissues that contains ectopic or interspersed adipose tissue or adipocytes, such as the breast, bone marrow, and the perirenal and epicardial regions, FGF21 may serve to modify adipocyte signaling that affects parenchymal cell functions through a paracrine mechanism. Within stressed adipose tissues which are predominantly adipocytes, both paracrine and autocrine mechanisms may be at play. The ultimate effect of the communication initiated by the FGF21 messenger is to elicit beneficial signals from adipocytes that aid in reduction of the effects of stress and prevention of tissue-specific damage and pathologies resulting from prolonged stress.

As a first responder to the SOS message carried by FGF21, how do adipocytes through FGF21-mediated activation of FGFR1-KLB aid stressed tissues and in particular liver and muscle? In addition to as a reservoir of energy, adipose tissue is an endocrine organ capable of systemic signaling through its lipid metabolites and adipokines (47, 48). A number of effects of FGF21 on adipose tissue adipocytes have been reported. These include the induction of expression of nuclear receptors and coactivators, diverse enzymes and their cytoplasmic regulators involved in pathways for glucose, lipid, and energy metabolism, and the broad changes in systemic and local tissue metabolic substrates and control pathways (1, 2, 4, 9, 40, 43).

Two major effects of FGF21 directly on adipose tissue stand out that may contribute to relief of metabolic stress on liver and muscle. Activation of the adipocyte FGFR1-KLB complex modulates the rate of adipocyte lipolysis and fatty acid oxidation that indirectly relieves hepatocyte steatosis or muscular lipid overload. Upregulation of expression of genes involved in fatty acid oxidation, browning of white fat and thermogenesis is also thought to collectively contribute to the anti-obesogenic effects (1, 29, 40, 49) in the obese. In contrast, activation of the adipocyte FGFR1-KLB complex during starvation dampens the rate of adipocyte lipolysis, which may help to extend energy usage. Secondly, recent studies show that FGF21 regulates the endocrine function of adipocytes including the secretion of adipokines that act on diverse tissues throughout the body. Notably the adipokine adiponectin is upregulated and leptin downregulated by FGF21. They appear to mediate many of the systemic and local effects of FGF21 (50, 51). Reduction of effects of a lipotoxic environment and oxidative stress is a common feature of adiponectin in liver, muscle, heart, kidney, pancreas, and endothelium (52). Thus, both the regulation of fatty acid metabolism and modifications of adipokine profile by FGF21 are favorable for the relief of liver’s steatotic burden that, when extreme and chronic, causes irreversible hepatic damage (Figure 1). Similarly, these FGF21-induced adipocyte responses are favorable for reduction of lipid overload and lipotoxicity in stressed or diseased muscle and possibly other tissues when autophagy/mitophagy, the mitochondria-ER-Golgi network or homeostasis in insulin function and glucose, lipid, and energy metabolic networks is compromised (34, 53).

Are adipocytes via the FGFR1-KLB partnership the primary or even specific target of systemic or local FGF21 signal observed in diverse body tissues? FGF21-stimulated changes in systemic metabolites and adipokines whose receptors are distributed widely have potential to indirectly affect practically all body tissues. Moreover, adipocytes that are characterized by expression of FGFR1-KLB and present in the microenvironment of most, if not all, tissues have potential to affect functions of adjacent tissues via FGF21-controlled paracrine products in the microenvironment. A careful dissection of both FGFR isotypes and KLB and signals elicited by the partnership in parenchymal and adipocytes of diverse organs that show a response to systemic FGF21 is required to clarify whether non-adipocyte cells constitute additional direct targets for FGF21. It has been suggested that FGF21 may signal directly in hepatocytes (54), pancreatic islet beta cells (31, 32), and cells in the suprachiasmatic nucleus of the hypothalamus and the dorsal vagal complex of the hindbrain (55, 56).