Since its discovery1,2 in 1995, the fat-derived hormone adiponectin has garnered attention for its roles in regulating glucose and lipid metabolism. Its two membrane-spanning receptors3, adiponectin receptor 1 (AdipoR1) and AdipoR2, transduce adiponectin signals in many target tissues, including the liver, heart, kidney and pancreas4. But an understanding of the mechanisms by which these receptors propagate the vast array of physiological effects that have been attributed to adiponectin has remained elusive. Vasiliauskaité-Brooks et al.5 provide structural and biochemical evidence on page 120 that adiponectin receptors have inherent activity as ceramidase enzymes, modulating metabolism by hydrolysing the lipid ceramide to produce two other signalling molecules — sphingosine and a free fatty acid.

Ceramide is a member of the sphingolipid class of biologically active lipid that induces desensitization to the effects of insulin and promotes inflammation and cell death. Normal concentrations of sphingolipids are crucial for the structural integrity and function of the plasma membrane — aberrant accumulation can promote a condition known as metabolic syndrome, whereas targeted disruption of sphingolipid synthesis in rodents diminishes the development of a range of disorders, including insulin resistance and heart failure associated with the toxic accumulation of lipid intermediates6. Hydrolysis of ceramide by ceramidase produces sphingosine, which is then phosphorylated to sphingosine 1-phosphate (S1P) in cells. In contrast to the harmful effects of ceramide, S1P potently induces cell proliferation and inhibits programmed cell death. Ceramide, sphingosine, S1P and free fatty acids can therefore all act as mediators of metabolism-related signal transduction.

The first clues to the enigmatic mechanism by which AdipoRs act emerged when researchers noted similarities between these receptors and ceramidases7. AdipoRs are members of the progesterone and adipoQ receptor (PAQR) family, which have seven transmembrane structural domains. They contain three highly evolutionarily conserved residues of the amino acid histidine, which mirror similar residues in a bacterial ceramidase8. In the ceramidase, these histidine residues hold a zinc ion in the catalytic site — this ion is essential for enzyme activity.

A body of evidence now demonstrates that AdipoRs control lipid metabolism at least in part by mediating ceramidase activity. In yeast, a PAQR depends on ceramidase for its activity7, and human AdipoR1 and AdipoR2 can promote ceramidase activity in ceramidase-deficient yeast7,9. Our lab has found evidence4 in mice that adiponectin exerts its beneficial metabolic effects by activating a ceramidase activity that is somehow regulated by the conserved histidine residues in AdipoR1 and AdipoR2. We have also shown that overexpression of these receptors in liver and fat cells potently improves systemic lipid and glucose metabolism by lowering ceramide levels in an adiponectin-dependent fashion10. Inverse correlations between the levels of sphingolipids and adiponectin have been reported11,12 in adult and adolescent humans, suggesting that this signalling pathway is widespread.

But whether the receptors themselves have an intrinsic enzymatic activity has remained under debate, because reports have suggested that AdipoR1 may instead interact with a recruited ceramidase13. Supporting an inherent catalytic activity for the receptors, crystal structures of human AdipoRs have revealed14 that the conserved histidine residues hold a zinc ion in a hydrophobic binding pocket, and mutation of these residues abolishes activation of the enzyme AMPK — a downstream effect of adiponectin signalling15. Now, Vasiliauskaité-Brooks and colleagues have generated crystal structures of human AdipoR2 that complement these previous findings, demonstrating for the first time that the catalytic activity is a ceramidase activity.

The structures, in which the hydrophobic pocket is imaged at higher resolution than in the previous AdipoR structures14, revealed that AdipoR2 co-crystallizes with a fatty acid, which is located inside the hydrophobic binding pocket. The pocket forms part of a tunnel that spans the receptor, with one exit into the transmembrane region and two facing the cytoplasm (Fig. 1). Vasiliauskaité-Brooks et al. next provided evidence that AdipoR2 binds to and hydrolyses ceramide. Structural analysis, along with biochemical data and computational models, revealed the conformational changes that enable enzymatic activity: ceramide binding is followed by a rapid rearrangement of the zinc binding site, leading to ceramide cleavage to form a free fatty acid in the binding pocket and a sphingosine molecule in part of the tunnel exposed to the cytoplasm. The sphingosine may be free to drift into the cell, providing an explanation for why it is not observed in the researchers' crystal structures.

Figure 1: Adiponectin receptor activity crystallized.
figure 1

Vasiliauskaité-Brooks et al.5 resolved structures of the membrane-spanning adiponectin receptor 2 (AdipoR2) protein, which is activated by binding of the hormone adiponectin and mediates glucose and lipid metabolism. a, The structures reveal that an internal tunnel spans AdipoR2, with a zinc ion (Zn2+) positioned in a binding pocket where the tunnel splits to form two entry points to the cytoplasm. On adiponectin binding, the lipid ceramide enters the tunnel and is hydrolysed to form two molecules: sphingosine and a free fatty acid bound to the Zn2+. b, The molecules are released into the cell, where sphingosine is readily phosphorylated to sphingosine 1-phosphate (S1P). Changes in the cellular levels of ceramide, sphingosine, S1P and free fatty acids drive metabolic responses to adiponectin.

Full size image

The authors also generated a structure of AdipoR1 by reanalysing previous structural data14. AdipoR1 did not reveal the same tightly bound fatty acid and buried catalytic cavity as AdipoR2. Instead, the catalytic site and substrate binding domain are completely accessible to the inner leaflet of the plasma membrane. Nonetheless, biochemical data confirmed that AdipoR1 has adiponectin- dependent, inherent ceramidase activity.

These observations are of great importance, because they offer highly refined structural insights into the basic functions of AdipoRs that can be combined with physiological evidence from mouse studies to provide detailed mechanistic insights into receptor function. However, major questions remain. For instance, we still do not understand the structural basis for the interaction of adiponectin with its receptors. Furthermore, although the authors' data clearly indicate that the enzymatic activity of AdipoRs is independent of other factors in vitro, it remains to be seen whether any accessory molecules can further enhance activity in vivo.

Finally, although Vasiliauskaité-Brooks and colleagues' biochemical experiments show that the presence of adiponectin increases the ceramidase activity of either AdipoR 25-fold, they cannot exclude the possibility that these receptors also possess another lipid hydrolase activity that could govern downstream signal transduction. Analyses of the lipid make-up of cells undergoing signal transduction and thorough screening of AdipoR activity in the presence of lipids other than ceramide could shed light on this issue.

Adiponectin mimetics might unleash the full antidiabetic potential of these receptors.

Adiponectin sensitizes liver and fat cells to insulin in mice to promote nutrient uptake4,10,15. Moreover, circulating levels of adiponectin are higher in lean humans than in obese individuals, and correlate with fat-cell health and lowered risk of developing diabetes and heart disease. Thus, activation of signalling pathways downstream of adiponectin might help to prevent the development of insulin resistance and type 2 diabetes. The current study should greatly aid the quest to design drugs that promote AdipoR activity. Adiponectin mimetics might unleash the full antidiabetic potential of these receptors. Footnote 1