Peroxisome proliferator-activated receptors (PPARs) are ligand-activated transcription factors that belong to the family of nuclear receptors (NRs). They are involved in the transcriptional control of genes regulating various physiological processes such as lipid-homeostasis, glucose metabolism, inflammation, and cellular differentiation and proliferation (1, 2). Three different isoforms have been identified, PPARα, PPARβ/δ, and PPARγ (1, 2).
Among the different PPAR subtypes, the N-terminal domain is barely conserved and it is responsible for the different biological functions. The DNA-binding domain contains two zinc-finger motifs and shows a high-homology between the subtypes. The ligand-binding domain shows less homology, which causes the ligand-specific binding of each subtype (3).
All PPAR isotypes are expressed in the brain, PPARβ/δ being the most abundant in this tissue. However, PPARγ has received a lot of attention because ligands of PPARγ have been identified as neuroprotective substances (4). PPARγ has been found in neurons as well as in glial cells (5). In vivo studies have shown that the administration of PPARγ ligands leads to a reduced pathology in many neurodegenerative diseases such as Alzheimer's disease (AD), Parkinson's disease (PD), amyotrophic lateral sclerosis, multiple sclerosis, Huntington's disease (HD), and stroke (6–9). The group of PPARγ ligands includes compounds from the group of the thiazolidinediones (TZDs), which involves synthetic agonists such as pioglitazone, rosiglitazone, and troglitazone. Natural ligands of PPARγ are fatty acids and eicosanoids, and the most recognized endogenous PPARγ ligand is the 15-deoxy-delta(12,14)-prostaglandin J2 (reviewed in ref.10). In addition, certain nonsteroidal anti-inflammatory drugs such as ibuprofen or indomethacin can act as PPARγ agonists (10). On the other hand, PPARγ can also be activated by ligand-independent pathways (11).
Upon ligand activation, PPARs form an activated heterodimer complex with the retinoid-X-receptor (RXR), which then binds to specific recognition sequences, the so-called PPAR-response elements in the promoter region of the target genes (12). The function of NRs such as PPARs is regulated by recruitment of coactivators in the presence of agonists and release of corepressors, resulting in transcriptional activation. Transcriptional transrepression can be achieved by diverse mechanisms (13). The precise spatial and temporal regulation of transcription by NRs depends on multiprotein regulatory proteins, which are characterized as coregulators. For instance, differentiation of preadipocytes in different types of adipocytes (white or brown) depends on which particular coactivator forms a complex with the PPARγ–RXR heterodimer (14). These cofactors modulate transcriptional initiation at regulated promoters by a variety of mechanisms including histone acetylation, chromatin remodeling, and direct interactions with basal transcription complexes (15). In vitro studies have demonstrated other mechanisms of transcriptional regulation, such as transcriptional interference (termed as “squelching”), which can occur when one NR represses the transcriptional activity of another NR by sequestering coactivators required by both (Fig. 1). Therefore, the density of different cofactors in certain brain areas would be a limiting factor in the transcription of genes (16).
One of the main coregulators of PPARγ is the PPARγ coactivator-1α (PGC-1α) (17). Recently, PGC-1α has been reported to be involved in numerous neurodegenerative diseases. In this review, we aim to analyze the role of the main regulators of PPARγ in the brain and during neurodegeneration, including PGC-1α, receptor-interacting protein 140 (RIP140), and NR corepressor (NCoR)/silencing mediator for retinoid and thyroid hormone receptor (SMRT).
THE PPARγ COACTIVATOR 1α (PGC-1α)
PGC-1α was discovered in the brown adipose tissue (BAT) as a PPARγ coactivator during thermogenic response to cold (17). In addition to PGC-1α, there are two other related transcriptional coactivators, PGC-1β and PGC-1-related coactivator (PPRC1, commonly known as PRC) (see ref.18). PGC-1α and PGC-1β display a great degree of homology although they are slightly differently regulated (19).
PGC-1α apart of acting as a coactivator for PPARs can also regulate other NRs such as the thyroid hormone receptor (THR), the estrogen receptor (ER), and the estrogen-related receptor α (ERRα) (14). In addition, it acts as a coactivator for other transcription factors such as the myocyte enhancer factor 2, hepatic nuclear factor 4, nuclear respiratory factors 1 and 2 (NRF-1 and NRF-2), the mitochondrial transcription factor A, Ying Yang 1, and the FOXO family receptors (20).
PGC-1α is highly expressed in tissues with high-energy demand, such as BAT, liver, skeletal muscle, heart, kidney, and brain (21). Extensive research has revealed its major role in gluconeogenesis and glucose transport, glycogenolysis, fatty acid oxidation, reactive oxygen species (ROS) suppression, mitochondrial biogenesis and respiration, peroxisomal remodeling and biogenesis, muscle fiber-type switching, and oxidative phosphorylation in various tissues (22, 23). Furthermore, PGC-1α regulates the expression of several ROS detoxifying enzymes, such as superoxide dismutases (SODs) (24).
PGC-1α has several domains with distinct functions, which include the binding site to transcription factors and the site of recruitment of acetyltransferases in the transcription complex located in the amino-terminal domain. Furthermore, PGC-1α has RNA-processing motifs in the carboxy-terminal domain which are actively involved in mRNA processing and elongation (14, 17, 25).
PGC-1α expression can be induced by cold exposure, fasting, and exercise, which require energy expenditure (17, 22, 26). Apart from gene expression, the activity of PGC-1α is influenced by post-transcriptional modifications, by means of protein phosphorylation, acetylation, SUMOylation, and methylation (14, 20, 27). Pharmacological activation of PGC-1α can be achieved using drugs such as resveratrol, which acts by decreasing PGC-1α acetylation, resulting in an increase in PGC-1α activity and its downstream genes (28). PGC-1α expression is decreased with aging, possibly owing to decreased sirtuin1 (SIRT1) levels (29) or by the action of p53 that is activated by telomere shortening and suppresses PGC-1α (30).
Studies in two different Pgc-1α knockout (Pgc-1α−/−) mice have revealed its importance for the regulation of several physiologic responses (31, 32). Pgc-1α−/− mice are resistant to diet-induced insulin tolerance and have defective mitochondrial functions. However, in one of the models the mice are obese as expected (31), considering the role of PGC-1α in BAT differentiation, whereas the second model is lean owing to hyperactivity (32).
In contrast to the abundance of information on PGC-1 cofactors in peripheral tissues, less is known about the role of these factors in brain plasticity and in neurodegeneration. A study in murine and rat brains revealed that PGC-1α is widely expressed in the cortex, the hippocampus, and the cerebellum, whereas it is absent from the hypothalamus. PGC-1α brain expression is not altered by leptin, caloric deficiency, obesity, or cold exposure (26).
Pgc-1α−/− mice exhibit reduced myelin and synaptic markers (PSD-95 and synaptophysin). In addition, these knockout mice are more susceptible to oxidative stress caused by the toxins 1-methyl 4-phenyltetrahydropyridine (MPTP) (dopaminergic neurons of the substantia nigra) and kainic acid (hippocampal glutamatergic neurons). Furthermore, PGC-1α whole-body knockouts display reduced levels of SOD1, SOD2, and catalase, supporting the important role of PGC-1α in ROS detoxification (24). Depletion of Pgc-1α in CamKII-positive neurons showed the same spongiform pattern in striatum as observed in whole-body knockouts, confirming the importance of PGC-1α expression in neuronal function. Neuronal inactivation of PGC-1α also resulted in resistance to diet-induced obesity, similar to what is detected in the total Pgc-1α−/− mice (33).
PGC-1α in HD
The involvement of PGC-1α in HD was brought to light when some neurological alterations found in the Pgc-1α−/− mice were similar to the ones present in HD mouse models (31, 32). HD is caused by the expansion of a CAG repeat in the Huntingtin (HTT) gene that results in a longer HTT protein which when exceeds 35 glutamine repeats, forms toxic oligomers and aggregates in the brain, the muscle, and the liver of HD patients, resulting in paralysis and death (23). HD patients display decreased expression of PGC-1α, PGC-1β, and markers for oxidative muscle fibers in muscle (34). Interestingly, adenoviral delivery of Pgc-1α in a HD mouse model restored muscle oxidative markers (34). In addition, mutant HTT knockdown in HD myoblasts resulted in increased levels of PGC-1α through an AMP-activated protein kinase-dependent mechanism (34).
Two seminal studies in HD mouse models discovered a direct role of mutant HTT in the suppression of PGC-1α. The first study found that mutant HTT impairs the activity of PGC-1α, resulting in downregulation of its target genes (35). The second study showed that mutant HTT directly binds to the Pgc-1α promoter, thus inactivating transcription of the gene (36). Together with the findings from St-Pierre et al (24), these studies support a direct role of HTT in the suppression of PGC-1α, resulting in mitochondrial impairment and increased ROS production, which is observed in HD patients. Moreover, the HD mouse models show impaired thermogenic response and weight loss, the latter also common in HD patients (35). There is further evidence that suggests that PGC-1α also plays a role in postnatal myelination and that deficient PGC-1α activity in oligodendrocytes may contribute to abnormal myelination in HD (37).
PGC-1α in PD
Another neurodegenerative disease in which PGC-1α has been implicated is PD. In PD, the accumulation of α-synuclein in dopaminergic neurons of the substantia nigra results in their death and the loss of dopamine in the brain (23). Mutations in α-SYN, PINK1, PARKIN, LRRK2, and DJ-1, all found in familial PD, also cause mitochondrial dysfunction (38). Studies related to the role of PGC-1α in PD have provided conflicting results, regarding the beneficial/detrimental effects of PGC-1α activation or overexpression in PD (39).
In a recent study, Ciron et al. (40) showed that adenoviral delivery of PGC-1α in the nigrostriatal system increased dopaminergic death. This effect could be owing to excessive overexpression of PGC-1α, resulting in mitochondrial hyperactivity, anoxia, and increased production of ROS (39). This is in contrast with other studies, showing that transgenic overexpression of PGC-1α or activation of PGC-1α by resveratrol protects dopaminergic neurons in the MPTP mouse model of PD (41).
Other recent reports have found that mutations in PD genes affect PGC-1α levels. For instance, PINK1 mutations result in impairment of Parkin recruitment in neurons, increased mitochondria numbers, and upregulation of PGC-1α (42). Interestingly, a substrate of the E3-ligase parkin, the parkin-interacting substrate (PARIS), which is selectively active in dopaminergic neurons, has been found to act as a repressor of PGC-1α by binding to insulin-response sequences in the PGC-1α promoter. In a healthy neuron, PARIS is ubiquitinated and degraded. Upon neurodegeneration, parkin is inactivated, PARIS is not degraded and it represses PGC-1α, subsequently affecting the activity of the NRF-1, which is a target gene of PGC-1α (43).
In addition, it has been shown that PGC-1α knockdown results in enhanced α-synuclein accumulation and downregulation of the AKT/GSK-3β signaling pathway in human neuronal cells (44). A meta-analysis of 17 genome-wide expression studies revealed that in the early stages of sporadic PD, two gene sets were altered and both had, in common, the mitochondrial master gene PGC-1α. The first set of genes included the nuclear genes encoding electron transport chain protein subunits. The second set of genes encoded enzymes involved in glucose metabolism. The authors confirmed that PGC-1α protected against α-synuclein-mediated cell death in vitro (45).
PGC-1α in AD
AD is the main cause of dementia in the aging population and is characterized by the accumulation of aggregated Aβ peptide in neuritic plaques and hyperphosphorylated tau protein forming neurofibrillary tangles. We and others have shown that the brains of AD patients display reduced levels of PGC-1α compared to healthy controls (18, 46) and that PGC-1α overexpression could be protective against AD (18, 46).
Overexpression of PGC-1α and PGC-1β in a neuroblastoma cell line and in primary neurons from the Tg2576 AD mouse model resulted in decreased levels of Aβ, attenuation of FOXO3A, and increase in the neurotrophic-soluble APP (18, 46). We have also shown that the PGC-1α-mediated reduction in Aβ was owing to a decrease in BACE1 transcription and expression. BACE1 is the main enzyme responsible for Aβ generation. In our study, we demonstrated that this effect was mediated through PPARγ activation as it was not observed in cells in which PPARγ had been knocked down (18). On the other hand, other reports have suggested that the decrease in BACE1 may be owing to increased proteasomal degradation mediated by the neuronal-specific E3-ligase SFCFbx2, which is a downstream target of PGC-1α (47).
Additionally, we demonstrated that transfection of PGC-1α cDNA or incubation with resveratrol increased neprilysin activity (which is the main enzyme involved in Aβ degradation), without affecting its expression. However, these effects were not mediated via PPARγ activation (18).
Activation of PGC-1α by resveratrol has been reported to promote neuronal survival and reduced neurodegeneration in the hippocampus, prevent learning impairment, and decrease the acetylation of PGC-1α and p53 in a transgenic model of AD (48). Moreover, resveratrol appears to promote the clearance of Aβ by increasing neprilysin activity in vitro (49). However, it remains controversial whether the effects of resveratrol are mediated through SIRT1-mediated deacetylation (50).
PGC-1α in Amyotrophic Lateral Sclerosis
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease that affects motor neurons in the spinal cord and the cortex and results in gradual muscle weakening, paralysis, and death. Several mechanisms have been proposed for the aetiology of the disease, among them oxidative stress, endoplasmic reticulum stress, glutamate excitotoxicity, and inflammation (51). PGC-1α affects all these pathways and three recent studies showed that it has a protective role in the progression of ALS (52–54).
PGC-1α expression appears to decline in transgenic ALS mouse models with the progression of the disease (53). Transgenic animals overexpressing PGC-1α in neurons (54), muscle (52), or in whole body (53) were crossed with animal models of ALS, such as the SOD1-G93A and the double-transgenic mice showed improved motor performance, slowed progression of ALS, reduced weight loss, and decreased motor neuronal death. Neuronal overexpression of PGC-1α also resulted in reduced phosphorylated MAPK p38 and JNK and in restoration of electron transport chain activities (54). The effect on survival and disease improvement was greater in higher-expressing PGC-1α mice.
THE RECEPTOR INTERACTING PROTEIN 140 (RIP140)
RIP140 (encoded by the gene Nrip1) was identified as a cofactor for NRs and found to function as a corepressor for a number of NRs that regulate metabolic pathways, namely ERRs, liver X receptor, RAR/RXR, THR, glucocorticoid receptors (GRs), and PPARs (55–57). However, RIP140 was found to be a coactivator of a number of transcription factors in inflammatory pathways, for example NF-κB (58) and CREB/Jun (59) in macrophages thereby stimulating the expression of a number of inflammatory cytokines (58). The importance of RIP140 in these signaling pathways was confirmed by examining the phenotype of RIP140 null mice which store markedly less triglycerides in white adipose tissue and exhibit protection from hepatic steatosis, resistance to diet-induced obesity, and increased clearance and insulin sensitivity (56).
RIP140 exerts its transcriptional regulation by competing with transcriptional coactivators or via recruitment of histone deacetylases and C-terminal-binding proteins (55, 60–62). On the other hand, it has been shown that in some cases, RIP140 can interact directly with PGC-1α (63).
RIP140 was found to be expressed in the mouse brain, particularly in neurons from cortex, hippocampus, and pituitary gland (64). In addition, RIP140 levels seem to be highly expressed during neurodevelopmental stages and reduced during aging (65, 66). Interestingly, the gene coding for RIP140 is located in chromosome 21, and consequently an increase of RIP140 has been found in the brains of Down-syndrome patients (67). Recently, RIP140 has been reported to be involved in the regulation of memory formation. RIP140 knockout mice (RIP140−/−) exhibited marked memory impairments compared to wild-type controls. In addition, these mice showed alterations in the response to stress. These changes were not related to morphological alterations in cortex and hippocampus of RIP140−/− mice (64). As RIP140 regulates the transcriptional activity of several NRs, such as GR, RXR, and ERα, which are linked to memory processes as well as stress, this could explain the learning and memory deficits in mice in which RIP140 has been depleted.
NR COREPRESSOR (NcoR)
Another corepressor for PPARγ is the NCoR. NCoR is closely related to the SMRT. NCoR and SMRT contain an NR interaction domain that interacts with an NR region between the DNA-binding domain and the ligand-binding domain. However, they can also affect chromatin conformation into an inactive state for transcription as their N-terminal domains can interact with other corepressors, such as Sin3 and histone deacetylases (16).
Constitutive deletion of NCoR is embryonically lethal in mice and causes defects in multiple organs (68). In these mice, it was shown that SMRT cannot compensate for the lack of NCoR, supporting distinct biological roles for NCoR and SMRT. A recent report in mice knockout for NCoR in adipocytes resulted in a phenotype similar to PPARγ activation (69). NCoR knockout mice were obese and showed reduction in most proinflammatory genes, including TNF-α, IFN-γ, IL-1β, iNOS, IL12p40, COX2, and MCP-1, whereas expression of the noninflammatory M2 marker genes, arginase and Mgl2, was increased. In addition, adipocyte NCoR deletion led to an activated PPARγ transcriptional state, which was not further enhanced by PPARγ ligand (TZD) treatment (69). Deletion of NCoR in skeletal muscle resulted in a substantial increase in oxidative metabolism, mitochondrial content, and muscle fiber size. Compared to the wild-type animals, the mutant mice also display an increase in reprogramming to oxidative muscle and better exercise capacity (70).
NCoR and SMRT transcripts have been found ubiquitously detected and unevenly distributed in brain tissue of young adult male rats (71). NCoR hybridization signal was found high in the hippocampus with the granular cell layer of the DG exhibiting the strongest signal, whereas SMRT's signal was also very high in all hippocampal subregions. In general, the distribution of NCoR and SMART immunoreactivity paralleled that of their mRNA expression, being predominantly expressed in the forebrain. Interestingly, the results by confocal microscopy indicated that NCoR and SMRT colocalize in vivo, possibly reflecting complexes identified in vitro (71).
Studies in NCoR null embryos have revealed that it is required for normal neural development. Interestingly, the loss of NCoR repression function permits non-neuronal peripheral tissues to express neuron-specific genes. It was suggested that disruption of NCoR reduces the number of neural progenitor cells and promotes astroglia differentiation (72). Interestingly, a correlation between NCoR and SMRT expression and the astrocyte-derived cancer glioblastoma multiforme, the most common and aggressive type of primary brain tumor, has been observed (73). NCoR expression and subcellular localization has been linked to certain types of cancer (13).
Overactivation or dysfunction of several NR coregulators is related to several human brain diseases such as brain tumors or neurodegenerative diseases. Recently, a great number of studies regarding the role of the PPARγ cofactor PGC-1α have shed important light on its role in neurological diseases such as AD, PD, ALS, and HD. This cofactor seems to be neuroprotective through different molecular and subcellular mechanisms, that is by affecting mitochondrial activity, oxidative stress, neuronal survival, and the regulation of the expression of proteins involved in those pathologies, such as Aβ, in the case of AD. This has important consequences as a potential target for therapy for these neurodegenerative diseases.
Reports on knockout mice for PPARγ corepressors have revealed the relevance of these proteins in the mechanism of PPARγ regulation. In addition, deletion of cofactors such as RIP140 and NCoR resulted in alterations in neuronal development and physiological processes such as memory deficits and stress.
However, as these cofactors are common to diverse transcription factors, there are some concerns related to the specificity of drugs targeting these proteins without unwanted side effects. In addition, the extent of expression of this proteins could have different outcomes, thus dosing is becoming an important issue. Therefore, further studies will be essential to analyze the role of these cofactors in health and disease.
The authors thank Dr. Mark Christian (Imperial College London) for critical reading of the manuscript and the Alzheimer's Research UK (grants ART/PG2009/5 and ART/PhD2011/16 to M. S.) for financial support. There is no conflict of interest by any of the coauthors.