Couple Dynamics: PPARγ and Its Ligand Partners

Abstract

Ligand-regulated transcriptional activity is the most important property of nuclear receptors, including PPARγ. In this issue of Structure, Hughes et al. determined how the dynamic conformations of ligands and the receptor contribute to the degree of ligand-dependent activation of PPARγ, which provide further insights into design of PPARγ-based anti-diabetic drugs.

PPARγ, a key activator of adipogenesis, is also the molecular target of the thiazolidinedione (TZD) class of anti-diabetic drugs such as rosiglitazone and pioglitazone. These TZD drugs are full agonists of PPARγ that are able to promote adipocyte differentiation. Although TZDs improve insulin sensitivity and glucose metabolism, they have been associated with severe side effects including fluid retention, weight gain, and cardiovascular diseases. The major challenge of PPARγ-based drug discovery is how to retain the beneficial glucose-lowering effects of PPARγ ligands but avoid their - undesired side effects. The attempts to overcome such challenges have been blocked by the lack of basic understanding of how PPARγ activation by small molecule ligands is linked with their anti-diabetic effects. A series of recent papers, including the one by Hughes et al. (2012) in this issue of Structure, begin to shed light into complex mechanisms that link PPARγ activity with insulin sensitivity and glucose metabolism.

PPARγ and two related receptors PPARα and PPARδ/β comprise a subfamily of nuclear receptors, whose transcription activity is tightly controlled by the conformation of an activation helix (AF-2), which resides in the C terminus of the ligand binding domain. PPARγ is known to have a ligand-independent basal activity (Xu and Li, 2008). As shown in Figure 1, in the absence of any ligand, the AF-2 helix of PPARγ is in equilibrium between closed (active) and open (inactive) conformations (Nolte et al., 1998). The binding of activating ligands, such as TZD or fatty acids, locks the AF-2 in the active conformation through a tight interaction between the AF-2 helix and the bound ligand (Gampe et al., 2000; Nolte et al., 1998). In this active conformation, the AF-2 helix forms a charge-clamp pocket to interact with short LXXLL motifs of coactivators that are required for transcriptional activation (Nolte et al., 1998). In contrast, the binding of antagonists, as seen in PPARα, destabilizes the AF-2 helix from the active conformation and opens up the charge clamp pocket for binding of larger LXXXIXXXL motifs of corepressors (Xu et al., 2002). For a long time, it was widely believed that transcriptional activity of PPARγ by TZDs was responsible for TZD’s effects on insulin sensitivity and glucose regulation as the in vitro affinity of PPARγ correlate with the potency of their in vivo glucose lowering activity (Willson et al., 1996). This view is further supported by recent evidence of adipocyte specific-knockout of nuclear corepressor NCoR, which increases PPARγ transcriptional activity and improves glucose metabolism (Li et al., 2011).

Figure 1. Dynamic Structures of PPARγ in Apo and Various Ligand-Bound States.

Apo PPARγ is shown with AF-2 helix, a hand-shaped ligand binding pocket (PC) and the β sheet (β) near Ser-273, with the AF-2 helix in a balance between open and closed conformations. The degree of Ser-273 phosphorylation is shown with a red bar. Rosiglitazone (TZD) binding stabilizes the AF-2 and the β sheet, reducing Ser-273 phosphorylation, while DA does not have an effect on PPARγ structure or Ser273-phosphorylation. MRL20 and MRL24 adopt multiple conformations in the PPARγ pocket with different stabilization of the AF-2 helix and the β sheet, thus affecting different levels of agonism and Ser-273 phosphorylation.

However, the direct correlation between PPARγ transcriptional activity of ligands and their anti-diabetic effects has been challenged from the very beginning. The first puzzle came from the genetic knockout PPARγ in mice, where heterozygous mice, which have half activity of wild-type, displayed more robust insulin sensitivity and glucose metabolism than wild-type mice (Barak et al., 1999). Second, although specific high affinity endogenous ligands have not been identified (Xu and Li, 2008), fatty acids are considered to be general PPARγ ligands that are present in sufficient concentrations to activate PPARγ in vivo (Xu et al., 1999). It is not clear why exogenous ligands would have significant anti-diabetic effects. The final puzzle lies in several PPARγ ligands, including MRL24, which have very poor agonist activities but have very good anti-diabetic effects (Choi et al., 2010). These puzzles remain as the dark clouds in the field of PPARγ and diabetic research.

Recent works from the Spiegelman and Griffin groups, which reveal the link of inhibition of Cdk5 phosphorylation of PPARγ by rosiglitazone and MRL24 with their anti-diabetic effects, start to provide answers to the third puzzle above (Choi et al., 2010). Cdk5 phosphorylates PPARγ at Ser-273. Both rosiglitazone and MRL24 inhibit this phosphorylation (MRL24 does not directly stabilize the AF-2 helix), and this inhibition is correlated with its anti-diabetic outcome of rosiglitazone in small clinical samples. Furthermore, SR1824, a non-agonist PPARγ ligand that blocks Cdk5-mediated phosphorylation, has similar anti-diabetic effects as rosiglitazone, consistent with the fact that Cdk5-mediated phosphorylation may play a major role in the development of insulin resistance (Choi et al., 2011). However, blocking the Cdk5-mediated phosphorylation of PPARγ may not be the only path to develop PPARγ-based anti-diabetic drugs. PPARγ has a large ligand binding pocket comprising of a small hydrophobic thumb pocket and a larger palm-like pocket (Figure 1). Both rosiglitazone and MRL24 dock into the larger palm pocket and stabilize the β sheet region of Ser-273, thus blocking Cdk5 phosphorylation. However, decanoic acid (DA), a 10-carbon fatty acid derived from natural middle chain-length triglycerides, is a weak PPARγ partial agonist that binds to the thumb pocket (Figure 1). DA does not stabilize PPARγ, including the β sheet and AF-2 regions, thus it does not inhibit Cdk5-mediated phosphorylation of PPARγ. Interestingly, DA was able to improve metabolism of glucose and lipids without induction of adipogenesis and body weight gain (Malapaka et al., 2011), suggesting complex mechanisms of PPARγ-mediated anti-diabetic effects beyond inhibition of Cdk5 phosphorylation.

The paper by Hughes et al. (2012) adds another layer of complexity in PPARγ-ligand interactions. Through studies of NMR and hydrogen/deuterium exchange, the authors discovered that AF-2 moves between all possible conformations on the intermediate NMR exchange time scale and the degree of stabilization of the active AF-2 conformation by MRL24, MRL20, and rosiglitazone correlates with their agonist activity (Figure 1). The most unexpected discovery is the occurrence of the multiple bound conformations of MRL20 and MRL24 in the PPARγ pocket, contradicting the single conformation observed in the earlier crystal structures (Bruning et al., 2007). Correspondingly, the receptor also adopts multiple conformations to accommodate the changes of ligand conformations. This observation challenges the prevalent view of one-on-one ligand-receptor interactions with respect to their bound conformations. The multiple docking modes arisen from promiscuous coupling between ligand and the receptor has also added significant difficulties in structure-based design of PPARγ-based anti-diabetic drugs. The continued quest for better PPARγ-based anti-diabetic drugs will have to be rooted in the field of PPARγ biology and pathology of diabetes, where much remains to be learned.