Peroxisome proliferator-activated receptors (PPARs) belong to a subfamily of the nuclear receptors which includes the retinoic acid receptors, the thyroid hormone receptors, and the RevErbA-related orphan receptors [1]. The PPAR subfamily contains three isoforms, namely, PPAR α (PPARA, NR1C1), PPAR β/δ (NR1C2 identified here as PPAR δ), and PPAR γ (PPARG, NR1C3, PPAR γ1, and PPAR γ2 sub-isoforms), that are encoded by different genes on different chromosomes.

In humans, PPAR α is mapped on chromosome 22 on the regions 22q12-q13.1; 22q13.31 with a linkage group of six genes and genetic markers [2]. The human PPAR γ gene is located on chromosome 3 at position 3p25, close to the retinoic acid receptor beta (RAR β) and the thyroid hormone receptor beta genes [3–5]. Two different human PPAR γ transcripts are expressed in hematopoietic cells: a 1.85-kb transcript, which corresponds to the full-length mRNA (PPAR γ1), and a shorter 0.65-kb transcript (PPAR γ2) [5]. PPAR γ2 is mostly expressed in adipose tissue where the PPAR γ2/PPAR γ1 ratio of messenger RNA is directly correlated with body mass index and where a low-calorie diet downregulates PPAR γ2 messenger RNA in subcutaneous fat [6]. Several variants in the PPAR γ gene have been identified, with the Pro12Ala variant having been the most extensively examined in epidemiologic studies. A strong association between PPAR γ 12Ala polymorphism and a reduction in type 2 diabetes risk (odds ratio: 0.86, 95 % CI: 0.81–0.90) was described in an updated meta-analysis of 60 studies involving 32,849 subjects with type 2 diabetes mellitus (T2DM) and 47,456 control subjects evaluated by the Human Genome Epidemiology Network [7].

The human PPAR δ, which was cloned from a human osteosarcoma cell library, is located on chromosome 6 at position 6p21.1-p21.2 [8]. In the mouse, where the first PPAR, PPAR α, was identified in 1990 by Issemann and Green [9], PPAR α is found on chromosome 15, PPAR γ is located on chromosome 6 at position E3-F1, while PPAR δ is found on chromosome 17 [10]. In both human and mouse, the PPAR transcript is encoded by six exons (one in the A/B domain, two in the C domain, one for the hinge region, and two for the ligand binding domain).

PPAR isoforms share a common domain structure as shown in the schematic view in Fig. 19.1. Five domains designated A/B, C, D, E, and F are distinguishable, and each has a different function. The N-terminal A/B domain contains at least one constitutionally active transactivation region (AF-1) and several autonomous transactivation domains (AD) [1]. The specificity of gene transcription is granted by the isoform-specific sequence of the A/B domain of the receptor [11]. Chimeric proteins generated by fusion with the A/B domains of other receptor proteins attenuate the specificity of target gene activation [11]. The DNA binding domain (DBD, C domain) is the most conserved region, which contains a short motif responsible for DNA binding specificity (P-box) on sequences called peroxisome proliferator response elements (PPREs), typically containing the AGGTCA motif.

The D domain, called a hinge, permits the change in shape of PPARs. The C-terminal E/F domain contains the ligand binding domain (LBD) and the AF-2 region for binding coactivators and corepressors. When activated by ligands, PPARs heterodimerize with another nuclear receptor, the retinoid X receptor, and alter the transcription of target genes after binding to specific PPREs on target genes.

Natural ligands for PPARs are long chain fatty acids, saturated or not, and eicosanoids: 8-HETE (hydroxyeicosatetraenoic acid) and to some extent leukotriene B4 (LTB4) for PPAR α; 9- and 13-HODE (hydroxyoctadecadienoic acid), two 15 lipoxygenase metabolites of linoleic acid, and 15-deoxy PGJ2 for PPAR γ; and prostacyclin (PGI2) for PPAR δ [12–14]. However, tissue concentrations are probably too low for them being the active ligands [15]. A new candidate endogenous ligand for PPAR α in the liver is a glycerophosphocholine esterified with palmitic and oleic acids 16:0/18:1-GPC or POPC (1-palmitoyl,2-oleoyl-sn-glycero-3-phosphocholine hydroxyeico-satetraenoic acid), which was identified in the liver of mice by tandem mass spectrometry [16]. This phosphatidylcholine is displaced from PPAR α by the synthetic agonist Wy14643. Its portal infusion induces dependent gene expression of carnitine palmitoyltransferase I (CPT1) in wild-type mice, but not in PPAR α deficient mice. Recently, two other phosphatidylcholines, DLPC and DUPC (1,2-dilauroyl-sn-glycero-3-phospho-choline and 1,2-(cis-cis-9,12-octadecadienoyl)-sn-glycero-3-phosphatidylcholine, respectively), have been shown to improve glucose control in two mouse models of insulin resistance [17]; however, they did not affect rosiglitazone binding to PPAR γ, and their effects are linked to stimulation of another nuclear receptor liver receptor homologue (LRH)-1.

PPAR α was first cloned from a mouse liver cDNA library at ICI, the pharmaceutical company which developed clofibrate, the first fibrate [9], and subsequently in humans [2, 18]. Fibrates, which were in clinical use as lipid-lowering agents for 20 years before this discovery, are weak PPAR α agonists, effective on human PPAR in the micromolar range, explaining the observation that they are given in the range of 100–1,200 mg per day. Fibrates, such as fenofibrate, mainly act via activation of PPAR α in the liver to regulate genes involved in fatty acid oxidation [19]. They were then called PPAR α activators, and their main laboratory effects are to reduce triglycerides and increase high-density lipoprotein (HDL) cholesterol levels. The first potent and selective PPAR α agonist acting in the nanomolar range with clinical data was LY518674, the development of which was stopped in 2007 when phase 2 studies showed no advantage over existing fenofibrate [20].

The link between PPAR γ activation and the thiazolidinedione insulin-sensitizing agents pioglitazone and rosiglitazone was established by researchers at Upjohn and Glaxo in 1994 and 1995, respectively [21, 22]. PPAR γ increases adipocyte differentiation and storage of fat. The short-term marker of PPAR γ activation in plasma is an increase in levels of the adipocytokine adiponectin, which increases insulin sensitivity in liver and muscle [23, 24]. The first animal results with PPAR δ agonists L165041 and GW501516 were reported in 1999 by researchers at Merck and in 2001 at Glaxo [25, 26].

The PPAR machinery is similar to other nuclear receptors with sequential complexes of coactivators and corepressors with enzymatic activities (for review, see Rosenfeld 2006) [27] and a series of metabolic transformations that turn PPARs toward activation or direct them to degradation (Fig. 19.2). The role of these different proteins, their metabolic transformations, and the concept of selective PPAR modulator are summarized in the next sections. Without ligand the transcription of DNA into messenger RNA is usually repressed by the binding of corepressors on the heterodimer PPAR–RXR, and chromatin is compacted (Fig. 19.3). With the presence of ligand in the ligand binding domain, the structural changes in the AF-2 region permit to replace corepressors by coactivators, to associate remodeling of chromatin by acetylation of histones, in order for RNA polymerase to access the DNA and initiate transcription (Fig. 19.4). One important aspect common to PPAR activation is transrepression of inflammatory genes under the control of nuclear factor kappa B (NFκB) or activated protein (AP) 1. This transrepression is an indirect effect since there is no PPRE in the promoter. This was shown for PPAR γ on induction of tumor necrosis factor (TNF) α by phorbol myristate acetate in human monocytes/macrophages [28], for PPAR α on human aortic smooth muscle cells and interleukin (IL)-1-induced IL6 expression [29, 30], and for PPAR δ with expression of monocyte chemoattractant protein (MCP)-1 [31]. In human endothelial cells, fenofibrate and L165041, but not rosiglitazone, inhibited TNF α-induced monocyte adhesion, vascular cell adhesion molecule-1 (VCAM-1) expression, and monocyte chemotactic protein-1 (MCP-1) secretion through inhibition of nuclear P65 translocation, necessary for NFκB activation [32].

This review is limited to PPAR activators or agonists which entered clinical development in diabetes and/or dyslipidemia (Table 19.1). Few PPAR antagonists were synthesized and they were not developed for the treatment of diabetes [60]. GW6471, a potent PPAR α antagonist, is used as a pharmacological agent to test whether an effect is PPAR dependent or PPAR independent. GW9662 is a PPAR γ antagonist which promotes the recruitment of NCoR. Finally, GSK0660 and GSK3787 are PPAR δ antagonists for pharmacological use.

The organs implicated in glucose control are listed in Table 19.2. With their direct effects on gene expression and their indirect effects on inflammation, and according to their tissue distribution, PPARs affect most of these organs, beyond the liver for PPAR α, the adipose tissue for PPAR γ, and the skeletal muscle for PPAR δ. In the kidney, they have different locations: PPAR α is located mainly in the proximal tubule, the medullary thick ascending limb, and in the mesangium; PPAR γ in the distal medullary collecting duct and glomeruli; and PPAR δ in a diffuse fashion as in other organs [62]. In the brain, the interplay of PPAR subtypes has been shown in cultures of astrocytes, where the three subtypes are present. Combined application of PPAR γ and PPAR δ activators increased cyclooxygenase 2 expression induced by lipopolysaccharide, whereas the additional application of a PPAR α agonist abolished this effect [63].

In the pancreas, the three PPARs are expressed in pancreatic β cells. PPAR α modulates fatty acid oxidation and PPAR γ directs them toward esterification. Although PPAR δ is the most abundant PPAR in the pancreas at the mRNA and the protein level, until recently its effects on fatty acid oxidation have been less well studied [64]. PPAR δ activation increases fatty acid oxidation and to a larger extent than PPAR α activation. In the pancreas, fatty acids acutely potentiate glucose-stimulated insulin secretion (GSIS), but their chronic exposure elevates basal insulin secretion and alters GSIS, a phenomenon called lipotoxicity.

Discordant results are reported in the literature with PPAR α or PPAR γ agonists. PPAR α was described to potentiate and PPAR γ to attenuate GSIS in INS-1E cells, an immortalized insulinoma rat cell line [65]. On the contrary, in vivo, the PPAR α agonist fenofibrate impaired GSIS in neonatal rats receiving monosodium glutamate to induce obesity, while pioglitazone, a PPAR γ agonist, increased it in db/db mice [66, 67]. This discordance might be explained by the low expression level of PPAR γ in INS-1E cells.

Activation of PPAR δ by unsaturated FAs or a synthetic ligand enhanced GSIS in primary rat islets or INS-1E cells without affecting basal insulin secretion [64]. In order to maintain β cell function, PPAR δ would play a role of lipid sensor to adjust the mitochondrial fatty acid oxidation. It was recently suggested that 4-hydroxynonenal (4-HNE) was one endogenous activating ligand of PPAR δ [68]. The level of reactive oxygen species (ROS), such as 4-HNE, is essential to β cell function, as low-level ROS production increases glucose-induced insulin secretion, whereas high levels of ROS can induce β cell apoptosis.

GSIS is also linked to influx of calcium ions to the cytosol induced by depolarization from the voltage-dependent Ca⁺⁺ channel. In INS-1 cells, the sarco-endoplasmic reticulum Ca⁺⁺ ATPase (SERCA2) pump maintains intracellular Ca⁺⁺ homeostasis, in particular, a high Ca⁺⁺ level in the endoplasmic reticulum. The expression of this pump is decreased in animal models of diabetes and in diabetic human islets. Pioglitazone directly increases expression of SERCA2 through transcription of the gene and indirectly through prevention of CDK5-induced phosphorylation of PPAR γ [69]. This experiment suggests that blocking CDK5 could permit to dissociate positive effects on glucose homeostasis from other effects from PPAR γ agonists.