A Second Chance for a PPARγ Targeted Therapy?

Antidiabetic Actions of a Non-Agonist PPARγ Ligand Blocking Cdk5-Mediated Phosphorylation
Choi et al
Nature. 2011;477:477–481.

A new class of non-agonist ligands target the transcription factor PPARγ and promote expression of insulin-sensitizing adipokines. They have potent antidiabetic actions, yet they lack several of the adverse effects commonly associated with thiazolidinediones. The ligands may represent a new class of anti-diabetes medications that preserve the most beneficial effects of PPARγ activation without imparting major side effects, which have limited the clinical usefulness of thiazolidinediones.

Once thought to be a magic bullet for the treatment of type 2 diabetes, thiazolidinedione (TZD) drugs such as rosiglitazone and pioglitazone are now recognized to have adverse effects, which have limited their clinical usefulness. TZDs were originally thought to improve insulin sensitivity by robustly activating the transcriptional activity of the ligand-activated nuclear receptor PPARγ. A study by the Spiegelman laboratory published in the July 22, 2010 issue of Nature revealed that rosiglitazone also inhibits the obesity-induced phosphorylation of PPARγ by the cyclin-dependent kinase 5 (Cdk5).¹ The inhibition of CDK5-mediated phosphorylation of PPARγ could also be mediated by PPARγ ligands, which lack full agonist activity suggesting a novel action of TZDs on PPARγ. Now, in the September 22, 2011 issue of Nature, the same research team reports that a new class of PPARγ ligands, which do not act as classical agonists as they lack robust transcriptional activation, effectively block Cdk5-mediated phosphorylation of PPARγ.² Importantly, these compounds have potent antidiabetic actions, but they do not cause fluid retention nor inhibit bone formation, major adverse effects of TZDs. The study provides an important proof-of-principle that new non-TZD drugs targeting PPARγ could be designed that preserve its most beneficial actions while eliminating its most detrimental side effects.

PPARγ is the target of the TZD class of antidiabetic agents, which improves glycemic control by increasing sensitivity to insulin. An ideal type 2 diabetes drug will: 1) provide long term glycemic control and thus reduce microvascular complication risk; and 2) reduce the risk of comorbidities associated with diabetes including cardiovascular disease. Clinical studies revealed that the improvement in glycemic control with TZD was superior, especially in terms of durability and their unparalleled ability to prevent type 2 diabetes, compared with other orally active antidiabetic agents such as sulfonylureas and biguanides thus satisfying the first criterion.^3–5 Even after troglitazone was found to cause idiosyncratic hepatic toxicity and was removed from the market, the enthusiasm for rosiglitazone and pioglitazone, which are not known to cause liver toxicity remained high. Indeed, a multitude of small clinical studies and animal studies revealed beneficial effects of TZDs on blood pressure, atherosclerosis, inflammation, and endothelial function satisfying the second criterion. Importantly, the PROactive (PROspective pioglitAzone Clinical Trial In macroVascular Events) trial suggested a reduced risk of macrovascular events (the secondary endpoint) in high risk type 2 diabetes patients taking pioglitazone.⁶ However, the PROactive trial along with other large clinical trials suggested that there was increased risk of non-fatal heart failure in patients taking pioglitazone and rosiglitazone (reviewed in⁷). In recent years, associations of TZD treatment with myocardial infarction (specifically rosiglitazone), bone loss and fracture, and bladder cancer have prompted new warnings and restrictions on the use of TZDs.^8,9

PPARγ, best known as a major regulator of adipogenesis, is also expressed in other classic insulin target tissues such as liver and skeletal muscle, but it also exerts effects in tissues as diverse as blood vessels, macrophages and brain. It is a member of the nuclear receptor family of ligand-activated transcription factors which upon activation results in transcriptional activation of many genes involved in carbohydrate and lipid metabolism. The canonical pathway of PPARγ-mediated transactivation in response to ligand involves the replacement of a repressor complex of proteins on the PPARγ response element (PPRE)-bound PPARγ/RXR heterodimer with an activator complex which remodels chromatin to induce expression of the target gene(s). Gene expression profiling has revealed that as many as 750 genes in adipose tissue are PPARγ responsive¹⁰; whereas genome wide chromatin immunoprecipitation of PPARγ in differentiated 3T3-L1 pre-adipocytes and macrophages shows the presence of thousands of binding sites for PPARγ/RXR heterodimers in the genome.^11,12 Supporting the transcriptional model of PPARγ action is the presence of unique signatures in the DNA of adipocytes and macrophages where there is a close localization of active accessible PPREs with the binding sites for the cell-specific transcription factors C/EBPα and PU.1, respectively.^12,13

Choi et al. reported that rosiglitazone, in additional to its canonical activity to robustly increase transcription of a host of PPARγ target genes involved in adipogenesis, also prevented the phosphorylation of PPARγ on serine 273 (Ser273) by CDK5.¹ As discussed below, phosphorylation of PPARγ by CDK5 is associated with decreased expression of some important insulin-sensitizing genes. This effect of rosiglitazone occurs by a ligand-dependent mechanism because the Q286P mutation in PPARγ, which interferes with ligand binding, impaired the ability of rosiglitazone to block CDK5-mediated phosphorylation of PPARγ. Interestingly, MRL24, another high affinity PPARγ ligand, which only poorly stimulates the transcriptional activity of PPARγ but exhibits potent anti-diabetic activity in experimental models, also blocks CDK5-mediated phosphorylation of PPARγ. This suggests the hypothesis that the anti-diabetic actions of TZDs may be mediated by inhibition of CDK5 action on PPARγ rather than by the robust activation of large numbers of PPARγ target genes.

This hypothesis was supported by two important pieces of evidence. The first, derived from gene expression profiling data, obtained on rosiglitazone or MRL24 treated PPARγ-deficient fibroblasts transfected with wildtype or S273A mutant PPARγ, showed that the genes markedly induced by rosiglitazone could be primarily grouped into two clusters. The first included classic genes of adipogenesis such as fatty acid binding protein 4 (Fabp4, also known as aP2) and lipoprotein lipase (Lpl). Most of these genes were not induced by MRL24, nor were affected by the S273A mutation in PPARγ, and consequently are not altered by CDK5-mediated phosphorylation of PPARγ. The second cluster of genes was robustly induced by rosiglitazone, MRL24, and PPARγ S273A, and thus are the genes regulated by the phosphorylation state of Ser273. Twelve of 17 genes from this cluster are decreased in obese mice, among them adiponectin, an insulin-sensitizing adipokine.¹⁴ These genes form the signature of PPARγ Ser273 phosphorylation (Ser273-P). In vivo, MRL24 (and rosiglitazone) improved glucose tolerance and reduced fasting insulin in mice fed a high fat diet, and both compounds reduced Ser273 phosphorylation, but not the phosphorylation of PPARγ at Ser112, the target of Extracellular Signal-regulated Kinases (ERK). MRL24 also reversed the down regulation of several Ser273-P signature genes caused by high fat diet. Most notable was the finding that Ser273-P on PPARγ was decreased in fat biopsies from human subjects treated with rosiglitazone.

In order to capitalize on their seminal findings, and as reported in the September 22, 2011 issue of Nature, the same research team developed synthetic small molecule ligands that have high affinity for PPARγ, exhibit no canonical agonist properties, but robustly block the actions of CDK5 on PPARγ. One of these compounds, SR1664 effectively blocked Ser273-P in vitro, and in adipose tissue from high fat diet fed mice and leptin-deficient ob/ob mice. It also increased the expression of 65% of the signature genes negatively regulated by CDK5-mediated Ser273-P. Unlike rosiglitazone, SR1664 was unable to robustly induce the transcriptional activity of PPARγ in a classic PPRE reporter assay in transfected cells, and did not induce lipid accumulation or adipocyte-specific gene expression in differentiated 3T3-L1 cells, and thus it lacks the actions classically associated with a full agonist such as a TZD. Consistent with this, whereas SR1664 did not alter the occupancy of PPARγ on the PPRE upstream of the Fabp4 gene in chromatin, it was unable to recruit steroid receptor co-activator-1 (SRC1) to the promoter, suggesting a mechanistic shift in the recruitment of regulators of transcription. Indeed, structural studies revealed that SR1664 interacts differently in the ligand binding pocket of PPARγ than rosiglitazone, which may affect the interaction between PPARγ and its co-regulators.

It is significant that SR1664 was just as effective as rosiglitazone in improving fasting insulin and glucose levels, and glucose tolerance in ob/ob mice. However, SR1664 did not affect bone mineralization or the expression of osteoblast genes in cells in culture, and did not cause the weight gain, increase in body fat or fluid retention, which are adverse characteristics of TZD. This suggests that SR1664 preserves benefits but eliminates some of the common detrimental effects of TZD treatment. Collectively, these data suggest the hypothesis that the beneficial actions of TZDs as a treatment for type 2 diabetes may be related in part to a decrease in CDK5-mediated phosphorylation of PPARγ.

PPARγ is thought to act as a sensor of fatty acid metabolites, although it responds to other metabolites as well.¹⁵ One can conjecture that as a sensor it must respond to signals generated by the environment, a high fat diet for example, and that its activity should be tightly regulated like a rheostat (Figure). Perhaps, TZDs may cause the rheostat to be increased above its ability to properly respond to metabolic changes (e.g., like affixing the volume too high on a radio), in a manner that would not occur with endogenous ligands. A converse effect is induced by Ser-273 phosphorylation which silences the rheostat, at least as it pertains to beneficial gene targets such as adiponectin. In contrast to these two extremes, the actions of SR1664 in blocking Ser273-P may be sufficient to preserve sufficient PPARγ activity without pegging it to either extreme under conditions where it normally would be impaired by high fat or obesity or overactivated by TZD. What remains unclear is how endogenously produced PPARγ ligands function. Presumably, they act along a continuum that includes on the one hand, like SR1664, activation of the Ser273-P-inhibited signature genes (i.e. adiponectin), and on the other, activation of the broad spectrum of genes activated by a TZD. Where each ligand acts in this continuum may presumably depend on its unique interactions with the ligand binding pocket, and how these effects influence its interaction with those structural moieties that regulate CDK5 binding or cofactor release or recruitment. Also largely unstudied is whether Ser273 phosphorylation functions similarly in non-adipose tissues important to PPARγ action. Choi et al., demonstrate improved insulin sensitivity in white adipose tissue and liver, however, changes in insulin sensitivity in the skeletal muscle – the main site of TZD insulin sensitization – were not reported.²

• Download figure
• Open in new tab
• Download powerpoint

Figure.

High fat diet (HFD), obese or leptin-deficient ob/ob mice exhibit increased free fatty acids (FFA) and cytokines, which activate CDK5. CDK5 phosphorylates Ser273 on PPARγ resulting in dysregulated expression of a series of genes considered to be insulin sensitizers such as adiponectin. This leads to glucose intolerance, insulin resistance and ultimately diabetes. The phosphorylation of PPARγ by CDK5 can be blocked by TZDs or SR1664 thus promoting preserved expression of adiponectin. In the paradigm show here, CDK5 through Ser273-P on PPARγ switches off the rheostat, an effect blocked by SR1664, TZDs, and perhaps some endogenous ligands. TZDs also cause a transition from a state of basal activity, to one that is fully activated leading to PPRE-mediated transactivation of many genes involved in adipogenesis, fluid retention, and alter gene expression in osteoblasts. Here, the volume on the rheostat is turned to its maximum setting. It remains unclear if endogenous ligands can modulate the activity of the rheostat thus regulating the level of PPARγ activation, and if so, why this is impaired in obesity.

Will SR1664 or similar ligands be clinically beneficial? Answering this will require extensive testing in both animals and humans. The authors admit that SR1664 has poor pharmacokinetic properties suggesting it will not be tested in patients. Nevertheless, this sets the stage for the development of similar molecules, which retain the beneficial physiological effects of SR1664 but exhibit better pharmacokinetics. Obviously, there are many questions that arise, including critical cardiovascular uncertainties. Will compounds like SR1664 provide protection from the cardiovascular sequelae associated with diabetes? Will they share the same vascular protective effects as TZDs? We have shown that interference with PPARγ in either vascular muscle or endothelium severely impairs vascular function and causes hypertension,^16–18 a phenotype similar to the hypertension observed in subjects with certain mutations in PPARγ.¹⁹ This leads us to ask whether PPARγ is phosphorylated by CDK5 in the blood vessels, and if so, does this occur in obesity or hypertension? As in adipose tissue and macrophages, rosiglitazone influences the expression of many genes in the blood vessel wall as well.²⁰ Therefore it will be important to determine whether the beneficial effects of PPARγ activation in the blood vessel require changes in the expression of these genes or whether the preservation of the signature genes of CDK5-phosphorylation is sufficient.