Sister Act

See allHide authors and affiliations

Science  08 Jun 2007:
Vol. 316, Issue 5830, pp. 1436-1438
DOI: 10.1126/science.1144837

All biologists know that the myriad enzymatic pathways that extract energy from metabolites and convert them to essential products are literally the biochemical basis for life. Many also view these reactions as an indigestible list of obtuse facts, which makes it all the more impressive that the late Dr. Robert Atkins, creator of the popular high-protein, high-fat, low-carbohydrate diet, was able to pass on a relatively obscure aspect of this process to a substantial fraction of the public—namely, that ketone bodies, produced from metabolizing fat, accumulate when you fast, allowing the body to use fat instead of carbohydrates for energy. Like all metabolic processes, the production of ketone bodies from fat (a process called ketogenesis) is regulated. As Atkins stressed, “burning” of fat is blocked by insulin in response to eating carbohydrates but is activated by starvation and also by a low-carbohydrate “ketogenic” diet. Two recent papers in Cell Metabolism (1, 2) have identified a remarkable and unexpected role for an obscure growth factor in this process. A third paper in the Proceedings of the National Academy of Sciences (3) reveals how another protein facilitates this growth factor's effect on metabolism.

The human genome encodes 22 members of the fibroblast growth factor (FGF) family (4). Most function in diverse processes such as development and wound healing. But three members—FGF19 (FGF15 in the mouse), FGF21, and FGF23—have recently emerged as metabolic hormones. FGF19 modulates bile acid biosynthesis, and itself is regulated by farnesoid X receptor (FXR), a nuclear receptor that is activated by bile acids (5). FGF23 regulates phosphate and calcium homeostasis, and its expression is controlled by the vitamin D receptor (6, 7). FGF21 has a variety of beneficial effects on undesirable metabolic parameters. For example, treating obese mice (genetically engineered to lack leptin, a hormone that controls appetite) with FGF21 decreases the concentrations of serum glucose and triglycerides and increases insulin sensitivity (8). Similar results were observed in obese rhesus monkeys (9). Inagki et al. (1) and Badman et al. (2) now show that FGF21 expression in the liver of fasted mice is activated by the nuclear hormone receptor peroxisome proliferator-activated receptor α (PPARα, also known as NR1C1).

Like other nuclear hormone receptors, PPARα directly regulates gene expression in response to low molecular weight signaling molecules. But unlike the well-known nuclear receptors that respond to endocrine hormones, PPARαbelongs to a new group of metabolic receptors that respond to common metabolites (10, 11). In the case of PPARα, fatty acids can function as endogenous signaling ligands, and it is well documented that PPARα functions in the liver to induce the expression of enzymes that promote fatty acid oxidation. This represents an elegantly simple regulatory circuit in which the presence of the energy source activates an enzymatic pathway that causes its own combustion.

In fatty acid oxidation, the straight carbon chains are chopped into two-carbon bits in the form of acetyl-coenzyme A (CoA). Among the many potential fates of acetyl-CoA is conversion into four-carbon ketone bodies that are released by the liver and used by other tissues, particularly the brain, as an energy source. As Inagaki et al. and Badman et al. show, FGF21 promotes fatty acid oxidation and diverts its output to ketone body production.

Both groups show that the dramatic induction of FGF21 expression in the livers of fasted mice is absent in transgenic mice lacking PPARα. Badman et al. used RNA interference to decrease endogenous FGF21 expression, which resulted in markedly increased triglyceride levels in both the liver and serum of mice fed a ketogenic (low carbohydrate, high fat) diet. This was associated with decreased expression of fatty acid-oxidizing enzymes and also 3-hydroxy-3-methylglutaryl (HMG)-CoA synthase 2 (HMGCS2) and carnitine palmitoyltransferase 1a (CPT1a), key enzymes in ketone body production that are both gene targets of PPARα. Inagaki et al. extended previous reports on the effects of both FGF21 treatment and transgenic overexpression of FGF21 in the mouse liver to demonstrate increased ketogenesis. Thus, in transgenic mice genetically engineered to overexpress FGF21, the normally low ketone body concentration in serum was increased severalfold, though the transgene did not further augment the elevated levels in fasted mice. Acute FGF21 treatment partially rescued the defective ketone body production observed in fasted mice lacking PPARα. Surprisingly, increased ketogenesis in the fed FGF21-overexpressing transgenic mice was not associated with elevated messenger RNA expression of HMGCS2 or CPT1a in the liver, although an increase in protein expression suggests possible posttranscriptional alteration in enzyme activity.

A bigger surprise seen with the FGF21-overexpressing transgenic mice was a dramatic increase in hepatic expression of several lipases usually found only in the pancreas. This was coupled with increased expression of more conventional lipases (and thus, lypolysis) in white adipose tissue. Inagaki et al. suggest that these increases in lipase activity may promote ketogenesis by augmenting the supply of fatty acids in the liver.

Here the story takes a bizarre turn. Mice maintained in constant darkness also express the same pancreatic enzymes in the liver and other peripheral tissues (12). Li and colleagues ascribe this phenomenon to increased concentration of 5′-adenosine monophosphate (5′-AMP) and showed that treatment with 5′-AMP induces torpor in mice (12). Torpor is an extreme example of decreased energy output that is associated with drastically reduced body temperature. Inagaki et al. show that when FGF21-overexpressing transgenic mice were fasted for 24 hours, their body temperature declined dramatically and locomotor activity ceased. Decreased body temperature was also observed in fasting wild-type mice when they were infected with an FGF21-expressing adenovirus or treated with a synthetic PPARα agonist. These observations clearly link the FGF21-PPARα pathway to torpor, but the specific relationship with 5′-AMP signaling remains to be explored. More broadly, in combining the current results on fasting and ketogenic response with previous results on glucose metabolism, it is evident that FGF21 is emerging as a regulator of overall energy balance.

Patterns in metabolic pathways.

Three different signaling pathways that regulate homeostasis may share a common design, involving a nuclear receptor, a fibroblast growth factor (FGF), an FGF receptor, and a member of the Klotho family of proteins. Each nuclear receptor forms a dimer with the retinoid X receptor (RXR); the Klotho proteins interact with distinct FGF receptor isoforms. Each pathway contributes to the broader mechanisms by which each nuclear receptor influences homeostasis.


It is intriguing that the link between FGF21 and PPARα completes a tidy pattern. Each of the three metabolic FGFs is now closely tied to a functionally complementary nuclear receptor signaling pathway. Determining whether this pattern extends to downstream signaling molecules is complicated by the responses of the relatively limited number of FGF receptors to multiple FGFs (4). A major advance in this area was recently provided by yet another unexpected connection. Klotho was initially identified as a mouse locus that, when mutated, causes progeria, or premature aging (13). Remarkably, interaction with Klotho protein converts two specific isoforms of FGF receptor 1 from general FGF receptors to specific FGF23 receptors (14), a role that is nicely supported by striking similarities in the phenotypes of mice lacking either Klotho or FGF23. In Greek mythology, Klotho, along with Lakhesis and Atropos, was one of three sisters responsible for spinning, measuring, and cutting the thread representing the fate of each individual. And just as might be predicted, Klotho's awkwardly named sister βKlotho was very recently found to combine with either of two FGF receptor isoforms to confer a specific response to FGF21 (3). The possibility of a simple three-way nuclear receptor-FGF-Klotho sister pattern (see the figure) is supported by the existence of another close relative, Klotho-LPH related protein (KLPH) (15). But as Macbeth learned, the pronouncements of the three weird sisters must be interpreted with caution. Thus, mice lacking βKlotho (16) show bile acid defects similar to those in mice lacking either FGFR4 or FGF15, rather than the triglyceride or glucose metabolic defects that would be anticipated from disrupting FGF21 signaling. Perhaps such phenotypes will emerge. At this early stage, however, the basis for downstream signaling is one of a number of tangles in the FGF21 and FGF15/19 threads that remain to be teased apart.


View Abstract

Stay Connected to Science

Navigate This Article