Role of transcription factors in mediating post-ischemic cerebral inflammation and brain damage

Abstract

Inflammation is a known precipitator of neuronal death after cerebral ischemia. The mechanisms that promote or curtail the start and spread of inflammation in brain are still being debated. By virtue of their capability to modulate gene expression, several transcription factors induced in the ischemic brain can modulate the post-ischemic inflammation. While the induction of transcription factors such as IRF1, NF-κB, ATF-2, STAT3, Egr1 and C/EBPβ is thought to promote post-ischemic inflammation, activation of transcription factors such as HIF-1, CREB, c-fos, PPARα, PPARγ and p53 is thought to prevent post-ischemic inflammation and neuronal damage. Of these, PPARγ which is a ligand-activated transcription factor was recently shown to prevent inflammatory gene expression in several animal models CNS disorders. This review article discusses some of the molecular mechanisms of PPARγ induction by its agonists following focal cerebral ischemia.

1. Introduction

Stroke (focal cerebral ischemia) is the 3^rd leading cause of death in the United States, and continues to be the leading cause of long-term disability, with very few options for effective treatment paradigms. The thrombolytic compound tissue plasminogen activator (t-PA) is the only Food and Drug Administration-approved agent for stroke therapy in humans. However, t-PA is known to have a relatively short therapeutic window and is not suitable for many stroke patients (Maestre-Moreno et al., 2005; Juttler et al., 2006). Recently, the thiazolidinedione (TZD) class of non-steroidal anti-inflammatory compounds such as rosiglitazone or pioglitazone (currently being used to treat type-2 diabetes) have shown efficacy in ameliorating the negative outcome following cerebral ischemia. These drugs act by binding and activating a family of transcription factors called Peroxisome Proliferator-Activated Receptors (PPARs), and have been hypothesized to curtail inflammatory processes subsequent to cerebral ischemia. In this review, we have discussed the functional significance of transcription factors in mediating cerebral inflammatory cascades with special emphasis on the beneficial effects of PPAR agonists in animal models of cerebral ischemia.

2. Putative contributors of ischemic brain damage

In experimental animals as well as humans, focal ischemia leads to an infarct with a core (an area with ∼95 to 100% decreased blood flow) surrounded by a penumbra (an area with ∼85 to 90% decreased blood flow). It is believed that the penumbral neurons could be rescued if the therapeutic interventions are given within the first 6 h after stroke (Lipton, 1999). Acutely after focal ischemia, anoxic depolarization occurs in the core, concomitant with increased extracellular potassium and intracellular calcium, as well as rise in extracellular glutamate. In the penumbra, sporadic transient depolarization (spreading depression) results in further glutamate release. Prolonged rise in the extrasynaptic glutamate exerts excitotoxicity via overstimulation of glutamate receptors. Due to insufficient supply of oxygen and glucose, cells undergo mitochondrial failure, leading to an inability to maintain ionic gradient across the membrane. Signs of cellular death are detected both in the core and the penumbral region, either by necrosis, apoptosis or autophagy (reviewed in Lipton, 1999). Furthermore, generation of free radicals and oxidative stress contribute to the development of infarct, resulting in more damage in that region. Lactic acidosis and lowered pH are also presumed to influence the cellular processes, adding to the overall neuronal damage. The damage continues to grow in the penumbral region and after about 7 days, there is almost a complete loss of cellular elements in the infarcted region (termed pan-necrosis). Thus the infarct takes several days to mature and involves a myriad of processes which, with proper understanding, could be targeted to minimize the lesion development.

In addition to the above excitotoxic events, acute inflammatory reactions that start within hours and persist for days after the insult are currently being thought to contribute to the neuronal death following stroke (Barone and Feuerstein, 1999). A major event that triggers inflammation in the brain is the expression of adhesion molecules like intracellular adhesion molecule-1 (ICAM-1) on capillary endothelial cells and white blood cells (WBCs) which will facilitate the adhesion and transendothelial migration of neutrophils and macrophages. While in capillaries, the accumulation of WBCs impairs blood flow and when extravasated into the brain parenchyma they release a host of neurotoxic substances including pro-inflammatory cytokines, chemokines and oxygen/nitrogen free radicals that are extremely toxic to neurons (Clark et al., 1995; Denton and Dietrich, 2003). Following transient as well as permanent middle cerebral artery occlusion (MCAO) in rodents, neutrophil accumulation can be detected very early in the infarct area (Barone et al., 1992; Kochanek and Hallenback, 1992; Garcia et al., 1994; Braun et al., 1996; Yanaka et al., 1997), and reperfusion seems to hasten and intensify the degree of neurtophil infiltration (Clark et al., 1994). Activation of resident microglia also enhances the inflammatory process and drugs like minocycline reduces the infarct volume after stroke by preventing microglial activation (Yrjenheikki et al., 1998; 1999).

3. Cerebral inflammation following focal ischemia

In the brain, inflammation acts as a double-edged sword. The timing, type of inflammatory mediator and the extent of its stimulation determines the balance between the bad versus good effects of inflammation (Barone and Feuerstein, 1999). Uncontrolled numbers of the infiltrated macrophages and neutrophils increases free radical generation which is detrimental to vascular integrity leading to cell death. However, macrophage and microglial response after ischemia is necessary for scavenging the necrotic debris and to facilitate the plasticity (Danton and Dietrich, 2003). For example, the complement system that gets activated in post-ischemic brain initiates local inflammatory responses that ultimately contribute to neuronal survival and tissue remodeling. However, complement activation to an inappropriate extent promotes inflammatory mediator release and tissue injury. Hence, cerebral inflammation can exert beneficial or detrimental effects depending on the pathophysiologic context (Van Beek et al., 2003).

Many studies over the last decade showed the beneficial effects of anti-inflammatory therapies following focal cerebral ischemia. Therapies that deplete leukocytes, prevent proinflammatory cytokines and adhesion molecules, and inhibit cyclooxygenase-2 (COX-2) and inducible nitric oxide synthase (iNOS) were reported to be neuroprotective after ischemia (Iadecola and Alexander, 2001; del Zoppo et al., 2000; Zhang et al., 1994; Yrjanheikki et al., 1998). Targeting individual inflammatory mediators and transcription factors that induce inflammatory genes also prevents ischemic neuronal damage. Knockout mice lacking interleukin-1beta (IL-1β) converting enzyme, ICAM-1, interferon regulatory factor-1 (IRF-1) and cyclic AMP element binding (CREB) protein develop smaller ischemic infarcts (Iadecola et al., 1999; Kitagawa et al., 1998; Schielke et al., 1998; Hata et al., 1998). Anti-inflammatory drugs like minocycline and indomethacin are known to induce neuroprotection by preventing microglial activation (Yrjanheikki et al., 1998, 1999; Tikka et al., 2001; Tikka and Koistinaho, 2001; Monje et al., 2003; Sasaki et al., 2003). However, conclusions should be drawn cautiously, as it is not evident to what extent pro-inflammatory cascades worsen the outcome after cerebral ischemia. Anti-inflammatory substances might act over a short span of time and hence might only delay the infarct development. The functional recovery/outcome following cerebral ischemia can not be linearly linked to the size of infarct (Dirnagl et al., 1999). These arguments add an additional degree of complexity to the question of the beneficial versus the harmful effect of inflammation, and warrant careful dissection of the role of individual inflammatory cascades.

3.1. Role of adhesion molecules in WBC extravasation after focal ischemia

While extravasation of WBCs into the brain is considered as the starting point of cerebral inflammation, many proteins mediate various steps of this process viz., rolling, adhesion and transendothelial migration. Acutely following cerebral ischemia, cytokines and chemokines are released from the infarcted area, some of which induce the ligands and receptors that mediate leukocyte recruitment and infiltration into brain. Focal ischemia leads to significantly elevated levels of the cytokines tumor necrosis factor alpha (TNFα), interleukins IL-1β and IL-6 and the chemokines, monocyte chemotactic protein-1 (MCP-1) and macrophage inflammatory protein-1 alpha (MIP-1α) (Liu et al., 1994; Wang et al., 1994; Kim et al., 1995; Wang et al., 1995; Soriano et al., 2000; Vemuganti et al., 2002; Lu et al., 2003; Kapadia et al., 2006). All these molecules induce the expression of the adhesion molecules ICAM-1, P-selectin and E-selectin on endothelial cells and WBCs (Dustin et al., 1986; Bevilacqua et al., 1987; Stoll et al., 1998; Abbassi et al., 1993; Eppihimer et al., 1996). Interaction of chemokines induces a conformational change in the β2-integrins on the leukocyte membrane. This will facilitate the interaction between integrins and adhesion molecules on the vessel wall leading to rolling, adhesion and transendothelial migration of WBCs (Penberthy et al., 1995; Gerszten et al., 1999). These events considerably slow down the blood flow through the cerebral capillaries, leading to their occlusion by activated neutrophlils and platelets (Mori et al., 1992), clogging the vessel and lowering blood perfusion. In addition to ICAM-1, P-selectin plays a significant role in post-ischemic leukocyte recruitment (Weyrich et al., 1993; Okada et al., 1994; Davenpect et al., 1994; Garcia-Criado et al., 1995; Kubes et al., 1995). It was demonstrated that fewer neutrophils accumulate in the brains of P-selectin-deficient mice following cerebral ischemia (Connolly et al., 1996). Similarly, blocking the ligand site on endothelial cells of ICAM-1 (Clark et al., 1995; Zhang et al., 1995) and antisense knockdown of ICAM-1 (Vemuganti et al., 2004) resulted in significant reduction of infarct volume. Furthermore, ICAM-1-deficient mice develop smaller infarcts (Soriano et al., 1996), supporting the beneficial effect of targeting adhesion molecules in post-ischemic brain. In addition, treatment with antibodies directed against the integrins CD11b/CD18 complex on granulocytes and macrophages led to a significant reduction in infarct volume in rats subjected to transient MCAO (Clark et al., 1991; Chen et al., 1994; Chopp et al., 1994). Depleting circulating neutrophils also reduce infarct volume and edema in various experimental models of cerebral ischemia (Shiga et al., 1991; Matsuo et al., 1994).

Diapedesis and dissolution of vascular integrity also contribute to the cerebral injury. Following MCAO, degradation of the basal lamina occurs, owing to increased expression and activation of metalloproteinases (MMPs), plasminogen activators and serine proteases (del Zoppo et al., 2000). MMP2 and MMP9 induction was thought to be spatio-temporally associated with blood-brain barrier damage and neuronal injury following MCAO (Heo et al., 1999; del Zoppo et al., 2000). Although, leukocytes infiltration into the post-ischemic brain parenchyma and the production of proteases, nitric oxide (NO), hydrogen peroxide and superoxides precipitates the secondary neuronal damage after focal ischemia, the exact contribution of these processes in relation to the neuronal death needs further assessment (Danton and Dietrich, 2003).

3.2. Role of free radicals in post-ischemic inflammation

NO and COX products contributes to the inflammatory cascades following an ischemic insult. NO is produced by the enzyme NO synthase (NOS), which exists in at least 3 isoforms viz., neuronal NOS (nNOS), endothelial NOS (eNOS) and iNOS. Of these, iNOS is expressed at a very low level by brain cells, but induced significantly following inflammatory conditions like focal ischemia. Unlike the other two isoforms, iNOS is not regulated by intracellular calcium and continuously produces large amounts of NO (Vodovotz et al., 1994; Iadecola et al., 1995). NO being a free radical, its excess production exerts cytotoxicity by inhibiting the ATP producing enzymes and by the production of peroxinitrite, which nitrosylates and thus inactivates many proteins (Iadecola, 1997; Iadecola and Alexander, 2001). While eNOS activation is thought to be neuroprotective after focal ischemia, as the NO produced by this isoform could dilate the capillaries by acting on the endothelial cells in its microenvironment (Huang et al., 1996), nNOS activation due to heightened calcium availability is also known to be toxic and together with iNOS seems to play a negative role in the pathophysiology of ischemic injury, as shown by studies with NOS inhibitors and iNOS knockout mice (Iadecola et al., 1997; Willmot et al., 2005). The 3 COX isoforms, COX-1, 2 and 3 generate prostaglandins and thromboxanes which are known to promote inflammatory reactions (Minghetti, 2004). Furthermore, COX-2 upregulation to an inappropriate level after focal ischemia promotes reactive oxygen species (ROS) generation that is thought to contribute to ischemic brain damage (Nogawa et al., 1998). However, recent findings suggest that COX-2 induced brain damage may not involve ROS as COX-2 inhibitors as well as COX-2 knockout mice failed to result in reduced ROS following MCAO (Kunz et al., 2006). In addition, the 5'-flanking region of the COX-2 gene has been shown to contain several potential regulatory sequences for the transcription factors, activating protein-2 (AP-2), nuclear factor kappa-B (NF-κB), CCAAT/enhancer binding protein (C/EBP), CREB, and Sp1, which are known to be upstream to many inflammatory genes (Lee et al., 2006).

4. Role of transcription factors in cerebral inflammation

Transcription factors play a pivotal role in controlling inflammatory gene expression. Microarray studies from our laboratory and other groups showed the induction of many transcription factors after focal ischemia (Soriano et al., 2000; Vemuganti et al., 2002; Lu et al., 2003). Of these, activation of hypoxia inducible factor-1 (HIF-1), CREB, c-fos, PPARα, PPARγ and p53 is known to prevent ischemic neuronal damage and/or promote ischemic tolerance (Tanaka et al., 2000a; 2000b; Cho et al., 2001; Maeda et al., 2001), whereas the induction of IRF-1, activating transcription factor-2 (ATF-2), signal transducer and activator of transcription 3 (STAT3), NF-κB, early growth response-1 (Egr1) and C/EBPβ promotes inflammation and neuronal death after cerebral ischemia (Table 1)(Planas et al., 1996; Bergeron et al., 1999; Iadecola et al., 1999; Johansson et al., 2000; Stephenson et al., 2000; Tanaka et al., 2000; Yan et al., 2000; Hu et al., 2000; Williams et al., 2003; Nurmi et al., 2004; Kapadia et al., 2006; Tureyen et al., 2007).

Table 1.

Transcription factors (TFs) involved in controlling inflammation and brain damage following cerebral ischemia

TF	Role	Observed actions	References
IRF1	Pro-inflammatory	IRF-1 null mice showed reduced infarct volume and neurological deficits	Iadecola et al., 1999
NF-KB	Pro-inflammatory	Inhibiting NF-KB activity resulted in smaller infarcts; mice deficient in p50 develop smaller infarcts	Nurmi et al., 2004; Schneider et al., 1999
Egr1	Pro-inflammatory	Egr1 null mice showed smaller infarcts, improved neurological function, and curtailed induction of many inflammatory genes	Vemuganti et al., 2007
STAT3	Pro-inflammatory	STAT3 siRNA prevented inflammation and brain damage in rats	Satriotomo et al., 2006
C/EBPβ	Pro-inflammatory	C/EBPβ null mice developed smaller infarcts, reduced neurological deficits and decreased apoptosis	Kapadia et al., 2006
PPARα	Anti-inflammatory	PPARα agonists decreased oxidative stress, inflammation and reduced infarction after focal ischemia	Deplanque et al., 2003; Collino et al., 2006b
PPARβ/δ	Anti-inflammatory	$\mathrm{PPAR}\beta ∕\underset{.}{\delta }$ agonists decreased ischemic brain damage	Iwashita et al., 2007
PPARγ	Anti-inflammatory	PPARγ agonists prevented post- ischemic inflammatory gene expression, induced anti-oxidant gene expression and decreased infarction and neurologic dysfunction; The neuroprotection was observed in hypertensive as well as diabetic rodents subjected to focal ischemia	Zhao Y et al., 2005, 2006; Sundararajan et al., 2005; Shimazu et al., 2005; Luo et al., 2006; Zhao X et al., 2006; Pereira et al., 2005, 2006; Collino et al., 2006; Tureyen et al., 2007

4.1. Nuclear factor kappa B (NF-κB)

Degradation of IκB induces NF-κB activation leading to the coordinated induction of multiple genes is involved in many inflammatory and immune cascades. Genes induced by NF-κB include pro-inflammatory cytokines IL-1β, TNF-α and granulocyte-macrophage colony stimulating factor (GM-CSF), and the chemokines IL-8, MIP-1α and MCP-1, that are largely responsible for attracting inflammatory cells into sites of inflammation (Nelson et al., 1993; Siebenlist et al., 1994; Mukaida et al., 1994; Ueda et al., 1994). Furthermore, NF-κB regulates the expression of adhesion molecules ICAM-1, VCAM1 and E-selectin on endothelial cells, promoting the adherence of inflammatory cells (Van De Stolpe et al., 1994; Iademarco et al., 1995). Many NF-κB down-stream gene products like IL-1β and TNF-α also re-activate NF-κB itself, resulting in a positive regulatory loop that amplifies and perpetuates the inflammatory responses (Barnes and Adcock, 1998). In rodents, activation of NF-κB occurs after MCAO and inhibiting NF-κB activity results in smaller infarcts (Nurmi et al., 2004). Furthermore, mice deficient in p50 develop smaller infarcts following focal ischemia, suggesting detrimental consequence of NF-κB activation (Schneider et al., 1999; Nurmi et al., 2004).

4.2. AP-1

AP-1 is a heterodimer of fos and jun oncoproteins, a collection of related transcription factors belonging to the Fos (cFos, FosB, Fra1, Fra2) and Jun (c-Jun, JunB, JunD) families that dimerize in various combinations through their leucine zipper regions (Barnes and Adcock, 1998). AP-1 will be activated by various cytokines, including TNFα and IL-1β via several protein tyrosine kinases and MAP kinases. Transcription factors Fos, c-Jun and JunB have been shown to be upregulated following cerebral ischemia (Kindy et al., 1991; Woodburn et al., 1993; Dragunow et al., 1994), and c-Jun was thought to play a role in the apoptotic neuronal death (Raivich and Behrenes, 2006).

4.3. HIF-1

Hypoxic conditions induce the oxygen-sensitive transcription factor HIF-1 (Bergeron et al., 1999; Lee and Anderson, 2006). HIF-1 is a heterodimer composed of HIF-1α and HIF-1β, and plays an essential role in hypoxia-induced transcriptional changes (Sharp and Bernaudin, 2004). For example, upregulation of HIF-1 subsequent to stroke induces iNOS and promotes neuronal death (Iadecola et al., 1995) as well as glucose transporter GLUT1 which may contribute to the adaptive mechanism to the oxygen-deprived condition resulting from ischemia (Bergeron et al., 1999).

4.4. C/EBPβ

C/EBPβ is a leucine-zipper transcription factor that regulates cell growth and differentiation in mammals. Expression of many proinflammatory genes including the cytokines and chemokines IL-6, IL-1β, IL-8, IL-12, TNFα, and MCP-1 is known to be controlled by C/EBPβ (Bradley et al., 2003). In addition, iNOS and COX-2 have C/EBPβ recognition sequences in their promoter regions and many of the IL-6 effects are mediated through the activation of C/EBPβ. Our studies showed that transient focal cerebral ischemia significantly upregulated C/EBPβ gene expression in rodent brain (Vemuganti et al., 2002), and focal ischemia in C/EBPβ null mice induced significantly smaller infarcts, reduced neurological deficits and decreased apoptosis (Kapadia et al., 2006). Furthermore, we observed curtailed inflammatory responses such as decreased ICAM-1 immunopositive vessels and less numbers of extravasated neutrophils and macrophages in C/EBPβ null mice following transient MCAO (Kapadia et al., 2006). In addition, GeneChip analysis showed that the post-ischemic induction of many inflammatory and neuronal damage inducing-genes was less pronounced in the brains of C/EBPβ null mice, suggesting a significant role for C/EBPβ in post-ischemic inflammation and brain damage (Kapadia et al., 2006).

4.5. IL6-JAK-STAT signaling

Increased levels of cytokines like IL-6 and TNF-α immediately after focal ischemia initiate cerebral inflammation and thereby secondary neuronal damage (Iadecola and Alexander, 2001; Stoll et al., 1999). The effects of these cytokines are propagated and amplified by the Janus kinase (JAK) and STAT signaling pathways (O'Shea et al., 2002). Binding of cytokines to their receptors induces transphosphorylation of the receptor-associated JAKs, which in turn phosphorylates the down-stream STAT family of transcription factors. Phosphorylated STAT dimerizes and binds to conserved genomic regulatory sequences on DNA to induce the expression of many genes (Darnell et al., 2005). Of the various isoforms of JAKs and STATs, JAK2 and STAT3 are the most-conserved isoforms and mice that lack either JAK2 or STAT3 die En Uterus (O'Shea et al., 2002). Transient focal ischemia was shown to stimulate the phosphorylation of both STAT1 and STAT3 (Planas et al., 1996; Justicia et al., 2000; Suzuki et al., 2001; Satriotomo et al., 2006). Increased STAT1 phosphorylation in neurons is known to contribute to the ischemic neuronal damage and STAT1 knockout mice show smaller infarcts (Takagi et al., 2002). Recent studies from our laboratory showed that following transient focal ischemia, JAK2 and STAT3 phosphorylation increases in the microglia/macrophages and treating ischemic rats with either a JAK2 phosphorylation inhibitor (AG490) or a STAT3 siRNA prevents inflammation as well as brain damage (Satriotomo et al., 2006). These studies indicate that JAK2-STAT3 activation might play a significant role in inducing secondary brain damage after focal ischemia. Furthermore, in the post-ischemic brain JAK-STAT activation leads to increased expression of suppressor of cytokine signaling-3 (SOCS3), which is a negative feed-back regulator of IL6 signaling. Our previous studies demonstrated that preventing SOCS3 expression with an antisense oligonucleotide infusion exacerbates inflammatory brain damage after focal ischemia in rats (Vemuganti et al., 2002).

4.6. Egr1

Egr1 is a sequence-specific transcription factor known to be induced by hypoxia (Yan et al., 1999). In peripheral organs, several inflammatory genes are induced down-stream to Egr1 (Yan et al., 2000). Our laboratory recently observed that focal ischemia increases cerebral Egr1 expression, and Egr1 null mice subjected to transient MCAO show significantly smaller infarcts, improved neurological function, and curtailed induction of many inflammatory genes (Vemuganti et al., 2007). Furthermore, we observed that adenoviral-mediated cerebral Egr1 overexpression exacerbates the focal ischemic brain damage (Vemuganti et al., 2007). These studies for the first time show Egr1 induction as a significant contributor of the inflammation and neuronal damage after stroke.

5. PPARs

Recently, much interest has been focused on the transcription factor PPAR isoforms α and γ, as their activation by specific ligands was observed to control inflammatory gene expression under various pathological conditions including focal ischemia.

5. 1. PPAR isoforms and ligands

PPARs are members of the nuclear hormone receptor super-family that include the receptors for vitamin D, thyroid hormone, retinoic acid, ecdysone, and some orphan receptors (Laudet et al., 1992). In mammals, PPAR exists as 3 isoforms PPARα, PPARβ/δ, and PPARγ, each encoded by a distinct gene. Naturally occurring fatty acids and eicosanoids bind to and activate PPARs, but the relative need for high concentration for these ligands suggests that PPARs are evolved to be activated by low-affinity ligands (Daynes and Jones, 2002). Fibrates are known to activate PPARα, although, at higher concentrations, PPARγ can be activated as well. NSAIDs such as ibuprofen, indomethacin or fenofibrate have been shown to activate PPARα 5 to 10 fold better than PPARγ (Lehmann et al., 1997). The prostaglandin J2 derivative 15-deoxy-delta12,14-PGJ2 (15d-PGJ2) is an endogenous PPARγ agonist; although as it is not present in sufficient concentrations to activate PPARγ in mammalian cells (Bell-Parich et al., 2003), it might influence other pathways, such as directly acting on IκB and the DNA binding domain of NF-κB (Straus et al., 2000). On the other hand, thiazolidinediones (TZD) like troglitazone, pioglitazone, and rosiglitazone which are synthetic insulin-sensitizing anti-diabetic ligands have emerged as potent PPARγ agonists (Lehmann et al., 1995). Among the TZDs, rosiglitazone and pioglitazone are currently approved by the US FDA for human use (Sood et al., 2000).

5.2. PPAR structure

Considerable amount of information is now available on the various domains of PPAR protein and their function. The PPAR domains include (a) the N-terminal A/B domain called activation function-1 (AF-1) which comprises the ligand-independent transactivation function, (b) DNA-binding domain (DBD), (c) ligand-binding domain (LBD), (d) D-domain linking the LBD and DBD (Escher and Wahli, 2000, Fig. 1a). Retinoic acid-X-receptor (RXR) contains a highly conserved central DNA-binding domain and a less conserved ligand-binding domain and serves as common heterodimeric, permissive partner for many nuclear receptors, including PPAR. When a ligand binds to PPAR, it dimerizes with RXR to form a heterodimeric complex that binds to the cis-acting sequences (peroxisome proliferator response elements; PPREs) on DNA to initiate transcription of target genes (Escher and Wahli, 2000; Chang and Szabo, 2000; Blanquart et al., 2003).

Figure 1.

PPARγ gene organization (A) and a simplified representation of action of PPAR:RXR heterodimer and ligands in transcriptional process (B). Under the presence of ligands, corepressor is released from PPAR:RXR heterodimer, enabling it to bind to to cofactor HAT complex. With further recruitment of general transcription complexes, transcription process can resume. AF-1; activation function-1. DBD; DNA binding domain. LBD; ligand binding domain. D; hinge domain linking DBD and LBD. PPRE; peroxisome proliferator activated receptor responseive element. HAT; histone acetyltransferase.

5.3. Mechanism of action of PPAR

Upon ligand binding, PPARs are liberated from inhibitory nuclear receptor co-repressor/silencing factor and, with recruitment of co-activators, act as transcription factors of target genes. Binding of ligand on either receptor of the PPAR:RXR heterodimer complex can activate the complex, but the simultaneous binding of ligands on both PPAR and RXR is known to be more potent (Desvergene and Wahli, 1999). Action of nuclear factors in regulating gene expression is quite complex. PPARs can influence the expression of various genes either positively (by transactivation) or negatively (by transrepression). Upon ligand binding, PPAR:RXR complexes with cofactor complexes and transactivates target genes that are under PPAR control (Fig 1b). For example, TZD binding to PPARγ results in the recruitment of cofactors and increased transcription activity. In the absence of the ligand, the PPAR/RXR complex can bind to the PPRE element with a co-repressor complex and hinder target gene transcription. PPAR can also transrepress certain genes that are under the control of other transcription factors such as NF-κB, AP-1 and Egr1, by inhibiting the ubiquitylation/degradation of co-repressor proteins via ligand-dependent addition of small ubiquitin-like modifier (SUMO) (Pascual et al., 2005). In addition to these, the PPAR:RXR heterodimeric complex competes with the co-activator complexes for other transcription factors, as well as directly interacts with transcription factors to regulate their function. To complicate matters further, in the case of PPARγ, TZDs can have PPARγ-independent actions that include mitochondrial dysfunction-related stress-response, involvement of proteins such as MitoNeet and increased astrocyte glucose uptake and lactate production (Dello Russo et al., 2003; Feinstein et al., 2005).

5.4. Regional and cellular distribution of PPAR

PPARα is expressed in cells with high catabolic rates of fatty acids and high peroxisome-dependent activities such as hepatocytes, cardiomyocytes, proximal tubules of kidney cortex, and intestinal mucosa. This explains its mRNA being most abundant in the liver, heart and kidney (Escher and Wahli, 2000). PPARβ/δ is expressed in relatively high levels in the adrenal gland, heart, intestine and kidney (Kliewer et al., 1994). Widespread expression of this isoform in both embryo and in adult has been shown (Kliewer et al., 1994; Cullingford et al., 1998), and its expression is comparatively higher than the other two isoforms in many tissues (Kliewer et al., 1994; Braissant et al., 1996; Cullingford et al., 1998). PPARγ shows the most restricted pattern of expression; high expression is observed in the adipose tissue, and modest expression is observed in the large intestine and immune system (Braissant et al., 1996). In the adult brain, PPARs show relatively low abundance with the exception of PPARβ/δ form, suggesting limited function of PPARs in the CNS under normal physiological conditions (Kliewer et al., 1994; Braissant et al., 1996).

Immunohistochemical studies demonstrated the distribution of PPARs in the adult rodent CNS (Braissant et al., 1996; Moreno et al., 2004). Except for the inner/outer layer of the retina where all the 3 isoforms are abundantly present, PPARβ/δ was shown to be localized in a relatively uniform fashion in the CNS, whereas PPARα and γ showed a low level of protein throughout the brain. PPARα transcript was expressed in the CA1 area and granule cells of the dentate gyrus, but not in the CA3 area of the hippocampus. PPARγ signal was weak and limited to granule cells of the hippocampal dentate gyrus (Braissant et al., 1996). PPARβ/δ expression was detected in the dentate gyrus, CA1 and CA3, as well as hilus. In the cerebellum, PPARβ/δ was present in Purkinje cells, granule cells, golgi cells and interneurons of the molecular layer, whereas PPARα and γ were scarcely present. Reticular formation was the exception where both α, β/δ and γ were present in similar levels. Moreno et al (2004) showed that PPARγ was also present in caudate putamen and globus pallidus of the basal ganglia as well as thalamic nuclei and piriform cortex. In murine brain, PPARβ/δ was shown to be high in the entorhinal cortex, hippocampus and hypothalamus (Woods et al., 2003), which was not in complete agreement with the other reports (Braissant et al., 1996; Moreno et al., 2004).

With respect to cell types of the CNS, all PPAR isoforms are known to be present in both neurons and astrocytes (Braissant et al., 1996; Moreno et al., 2004). Primary cultures of cortical neurons and astrocytes show mRNA of all isoforms of PPARs, where different isoforms seem to contribute to neuronal and astrocytic differentiation (Cimini et al., 2005; Cristiano et al., 2005). A recent study showed that exposing oligodendrocytes to PPARβ/δ agonists leads to stimulation of oligodendrocyte differentiation in cultured oligodendrocytes, where PPARβ/δ expression has also been reported (Saluja et al., 2001; Woods et al., 2003). The subcellular localization of PPARs seems to be mostly nuclear, but some cytoplasmic localization for PPARα and γ has been reported in subsets of neurons (Moreno et al., 2004).

6. Role of PPARs in brain inflammation

The major function of PPARs seems to be to act as regulators of glucose and lipid metabolism and adipocyte differentiation in several subtypes of cells throughout the body. But PPARs also influence immune and inflammatory functions in macrophages, T cells, B cells, dendritic cells and endothelial cells (Gosset et al., 2001; Jones et al., 2002; Glass and Ogawa, 2006). PPARα and γ agonists have been shown to display anti-inflammatory effects (Devchand et al., 1996; Setoguchi et al., 2001), though it is not yet clear whether all the anti-inflammatory effects are receptor-dependent (Daynes and Jones, 2002).

TZDs inhibit the expression of various inflammatory proteins such as iNOS, TNFα, and MMP9 in macrophages (Li et al., 2000), and prevent the induction of pro-inflammatory transcription factors such as NF-κB, AP-1, and STATs (Ricote et al., 1998; Welch et al., 2003). In macrophages, 15d-PGJ2 or synthetic PPARγ ligand BRL49653 was shown to induce a dose dependent inhibition of iNOS, MMP9, and scavenger receptor-A expression (Recote et al., 1998). Whereas, neither 15d-PGJ2 nor troglitazone was effective in inhibiting phorbol 12-myristate13-acetate (PMA)-induced TNFα release as well as IL-1β and IL-16 expression in human monocytes (Jiang et al., 1998).

In primary mouse microglia and astrocytes, 15d-PGJ2 as well as TZDs inhibited the production of NO, cytokines IL-1β, IL-6, TNF-α and the chemokine MCP-1 (Storer et al., 2005). Furthermore, in activated astrocytes and microglia, 15d-PGJ2 and rosiglitazone inhibited the phosphorylation of STAT1, STAT3, JAK1 and JAK2 (Park et al., 2003), suggesting a role as a negative modulator of cytokine signaling. Similarly, in LPS or IFN-γ-stimulated primary microglia, 15d-PGJ2 attenuated iNOS expression and NO production, and donwregulated TNF-α and MHC class II expression (Bernardo et al., 2000). These studies provide evidence for the potential involvement of PPARγ in inflammatory processes in CNS.

7. PPAR agonists and focal ischemia

As already mentioned, pioglitazone and rosiglitazone (TZD class of PPARγ agonists) have proven to be beneficial in type-2 diabetes mellitus patients. Diabetics are at an increased risk of stroke incidence and stroke causes more damage in diabetics compared to normoglycemic individuals (Kagansky et al., 2001). Recently, enhanced functional recovery was reported in a small group of stroke patients with type-2 diabetes treated with pioglitazone or rosiglitazone (Lee and Reding, 2006). In another clinical study, increased plasma level of 15d-PGJ2 (the natural ligand for PPARγ) was correlated with smaller infarcts and good neurological outcome (Blanco et al., 2005). In addition, the beneficial role of PPAR agonist treatment has been demonstrated in many animal models of focal cerebral ischemia. Table 2 shows a summary of the findings (highlighting the animal model, drug treatment, route of delivery and the effects of PPAR agonist treatment on infarction) which are discussed below.

Table 2.

Beneficial role of PPAR agonists treatment in animal models of cerebral ischemia.

	Authors and journal	Animal model	Route of treatment and doses	Effects of drug treatments and results
1	Shimazu et al., 2005	Permanent and transient MCAO in S.D. rats	Pio, oral, 20 mg/kg/day	↓ infarct volume in transient (but not permanent) ischemia, better neurological score
2	Pereira et al., 2005	Permanent MCAO in Fischer rats	L-796,449, i..p., 1mg/kg	↓ infarct size, better neurological score, ↓ apoptotic and anti-inflammatory processes
3	Zhao et al., 2005	Transient MCAO in Wistar rats	Pio, 0.16mg/kg/day, i.c.v.	↑ sensory neurological score, ↓ infarct volume, ↓ edema, ↓ ED-1 positive cells
4	Sundararajan et al., 2005	Transient MCAO in Wistar rats	Tro (>10mg/kg), Pio (1mg/kg), i.p.	↓ infarct size, ↓ microglia/macrophages, COX-2 mRNA/protein, iNOS and IL-1β mRNA
5	Zhao et al., 2006	Transient MCAO in Wistar rats	Pio, 0.16mg/kg/day,	↓ infarct size, ↓ TNF-α, COX-2+ cells. ↑PPARγ + cells following ischemia in neurons/microglia
6	Luo et al., 2006	Transient MCAO in C57/B6 mice, cell culture	Rosi, up to 12mg/kg, i.p.	↓ infarct size, better neurological score, Rosi treatment increased binding, ↓ ICAM-1, MPO, activated microglia
7	Victor et al., 2006.	Transient MCAO in Wistar rats	Pio (1.0 mg/kg), Rosi (0.1 mg/kg), T0070907 (1.5 mg/kg), i.p.	Increased PPARγ + cells following ischemia in neurons, ↑binding by Rosi treatment
8	Chu et al., 2006	Transient MCAO in S.D. rats	Rosi, oral, 3mg/kg	↓ infarct size, better neurological score, ↓ TUNEL + cells, ↓ Caspase-3+ cells, ↓ MPO/OX-42, ↓ caspase-3 activity, ↑ vasculature
9	Ou et al., 2006	Transient MCAO/CCAO in Long Evans rats	15d-PGJ2, 20 μg, i.c.v.	↑PPARγ + cells following ischemia in neurons/microglia, and ↓ infarct volume
10	Lin et al., 2005	Transient MCAO/CCAO in Long Evans rats	15d-PGJ2 (1-50pg), i.c.v., Rosi (50ng)	↓ infarct size, HO-1 increase, ↓ IKBα degradation, ↓ caspase-3 activation, 15-d-PGJ2 and Rosi treatment ↑ PPARγ expression
11	Collino et al., 2006	Transient CCAO in Wistar rats and C57/B6 mice	Rosi, Pio, i.v., 1-6 mg/kg.	↓ oxidative stress, ROS, and HNE. Restored GSH and SOD level, reduced COX-2 increase, and inhibited NF-kB nuclear translocalization
12	Turuyen et al., 2007	Transient MCAO in S.D., SH rats, C57/B6 and db mice	Rosi, Pio, 9-cis-RA, 15d-PGJ2, i.p., 0.5- 6mg/kg	↓ infarct size, ↓ ICAM-1 and infiltrating macrophages/microglia, ↑ catalase and Cu/ZnSOD, ↑ SOCS3, ↓ STAT phosphorylation
13	Deplanque et al., 2003	transient MCAO in C57/B6 mice. Also used APOE−/−, and PPARα−/− mice	Fenofibrate, 50- 250mg/kg, oral	↓ infarct size, better MCA endothelial vasodilation, no change in iNOS, more GSH due to ↑ Cu/ZnSOD, ↓ VCAM-1 and ICAM-1
14	Collino et al., 2006	Transient CCAO in Wistar rats	WY14643 (0.1- 6mg/kg), MK886 (6mg/kg), i.v.	↓ ROS, restored glutathione level, ↓ HNE, ↓ HO-1 and p38 MAPK level, ↓ nuclear NF-kB, ↓ iNOS, ↓ ICAM-1, ↓ S100β

Abbreviations used : S.D, Sprague-dawley; i.p.; intraperitoneal; i.c.v., intracerebroventricular; Rosi, rosiglitazone; Pio, pioglitazone; Tro, troglitazone; SHR, spontaneously hypertensive rat; db, diabetic; MPO, myeloperoxidase; HNE, hydroxynonenal; GSH, N-(N-L-gamma-glutamyl-L-cysteinyl)glycine ; SOD, superoxide dismutase; MAPK, mitogen activated protein kinase; SOCS, suppressors of cytokine signaling ; STAT, signal transducers and activators of transcription

Rats fed on a pioglitazone-fortified diet for 3 days prior to the induction of transient focal ischemia were observed to develop smaller infarcts and better neurological scores compared to pair-fed controls (Shimazu et al., 2005). These authors showed that upregulation of the antioxidant enzyme Cu/ZnSOD might be a potential neuroprotective mechanism underlying the neuroprotective effect of pioglitazone after focal ischemia. However, the same treatment did not show protection when rats were subjected to permanent ischemia, suggesting that PPARγ-dependent protection is specific to the free radical-generating and/or inflammatory nature of the reperfusion injury. Interestingly, when a non-TZD PPARγ agonist L-796,449 was injected (i.p., 1mg/kg) 10 min prior to permanent MCA occlusion, the treatment resulted in reduced infarct size and better neurological outcome (Pereira et al., 2005). The reason for this disparity is still unclear, but may be due to the PPAR-independent actions of the drug. L-796,449 treatment also increased the neuroprotective enzyme heme oxygenase-1 (HO-1) while blocking NF-κB translocation into the nucleus, with decreased caspase-3 activity, reduced levels of inflammatory mediators iNOS and MMP-9, as well as decreased amounts of NO₂^- and NO₃^- (Pereira et al., 2005). A later study showed that intravenous injection of either pioglitazone or rosiglitazone reduced oxidative stress, COX-2 expression and activation of MAPK and NF-κB following a transient common carotid artery occlusion (Collino et al., 2006a). Chu et al. (2006) suggested that rosiglitazone-fed rats had better neurological scores, smaller infarcts, reduced number of TUNEL-positive cells, reduced caspase-3 activity, and curtailed infiltration of myeloperoxidase positive (MPO⁺) neutrophils and OX-42⁺ macrophages following transient focal ischemia. Interestingly these authors also reported an increased vasculature in the rosiglitazone-treated group with increased number of endothelial cells positive for BrdU, suggesting there may be enhanced angiogenesis following PPARγ activation.

The beneficial effect of TZD treatment may be due to the direct action of TZDs on brain rather than on the peripheral organs, as pioglitazone administered into the brain via osmotic mini-pump also resulted in significant neuroprotection following MCAO (Zhao et al., 2005). When pioglitazone was infused intraventricularly for 7 days (5 days prior and 2 days after the MCAO) there was an increase in the sensory (but not motor) neurologic scores, with reduced infarct volume and brain edema compared to vehicle treated controls (Zhao et al., 2005). Furthermore, macrophage extravasation and microglial activation was also significantly reduced in the pioglitazone-infused group supporting its anti-inflammatory properties in the ischemic brain (Zhao et al., 2005).

Further supporting the notion of enhanced anti-inflammatory influence of TZDs after focal ischemia, Sundararajan et al. (2005) showed the beneficial effect of troglitazone and pioglitazone injected i.p. 1 day prior to the ischemia. Troglitazone as well as pioglitazone treatment reduced infarct volume and resulted in better neurological score, reduced number of microglia/macrophages and decreased expression of COX-2, IL-1β and iNOS mRNA/protein in the ischemic brain. This group recently showed that focal ischemia significantly enhances PPARγ mRNA and protein levels in the neurons within the peri-infarct area of the rats subjected to transient MCAO (Victor et al., 2006).

Using a mouse model of transient MCAO, Luo et al. (2006) reported the beneficial effect of rosiglitazone delivered i.p. at differing concentrations and demonstrated that drug treatment at 3mg/kg and higher reduced infarct volume and improved neurologic score. Rosiglitazone treatment also increased the binding of PPARγ to PPRE, suggesting that the drug enhances the translational activity of PPARγ. When animals were treated with PPARγ antagonist GW9662, this protective effect of rosiglitazone was lost. Rosiglitazone treatment further reduced ICAM-1 expression, MPO levels and activity, amount of activated microglia, and cytokine levels (TNF-α, IL-6, MCP-1), supporting the link between anti-inflammatory effects of TZDs acting via PPARγ (Luo et al., 2006).

Recent studies from our laboratory showed that pre-treatment as well as post-treatment with the TZDs, rosiglitazone and pioglitazone significantly decreases the infarct volume and neurological deficits in normotensive and normoglycemic rodents as well as hypertensive and hyperglycemic rodents (Tureyen et al., 2007). Our studies further showed that rosiglitazone neuroprotection will not be enhanced by RXR agonist 9-cis-retinoic acid, but could be prevented by PPARγ antagonist GW9662. Rosiglitazone significantly decreased the post-ischemic expression of ICAM-1 mRNA and protein leading to decreased extravasation of macrophages and neutrophils into the brain parenchyma. Rosiglitazone treatment also curtailed the post-ischemic expression of the pro-inflammatory cytokines (IL-1β and IL-6), chemokines (MIP-1α and MCP-1) as well as the pro-inflammatory enzymes COX-2 and iNOS. Our studies further showed that PPARγ activation by TZDs prevents the post-ischemic cerebral expression of pro-inflammatory transcription factors Egr1, C/EBPβ and NF-κB and at the same time enhances the expression of the anti-oxidant enzymes catalase and Cu/ZnSOD (Tureyen et al., 2007). Most importantly, our studies showed that rosiglitazone-induced neuroprotection after ischemia involves expression of the cytokine signaling regulator SOCS3 and the prevention of STAT3 phosphorylation. Thus, these recent studies indicate that PPARγ activation with TZDs is a potent therapeutic option for preventing inflammation and neuronal damage after stroke with promise in diabetic as well as hypertensive subjects.

It has been shown that transient focal ischemia alone can increase brain PPARγ protein content in peri-infarct area. Increased PPARγ mRNA was detected in the infarcted brain as early as 6h following focal ischemia (Ou et al., 2006), and PPARγ-immunpositive neurons were detected between 4 h and 14 days (Victor et al., 2006), whereas in neurons and microglia only transiently at 12 hr in the post-ischemic brain (Zhao et al., 2006). This increased PPARγ expression was shown to be associated with reduced infarct volume, TNFα level, and number of COX-1 and -2 positive cells in the peri-infarct cortical regions when the animals were treated intraventricularly with pioglitazone (Zhao et al., 2006). In addition, in cultured neurons pioglitazone treatment was able to inhibit COX-2 increase induced due to H₂O₂, whereas PPARγ antagonist GW9662 treatment abrogated this effect, suggesting a direct action of PPARγ on COX-2 in neurons (Zhao et al., 2006). Victor et al. (2006) also showed that increased PPARγ mRNA expression and protein levels in neurons induces reduced infarct volume if the animals were treated with rosiglitazone and that the PPARγ antagonist T0070907 treatment abrogates this effect. Interestingly, the antagonist T0070907 alone resulted in significantly more damage, suggesting that the endogenous PPARγ increase seen in the penumbral neurons might be beneficial. Ischemia reduced the PPARγ binding in the ipsilateral hemisphere of the brain, and rosiglitazone treatment increased the binding in general, where it resulted in a much more noticeable increase in binding activity in the contralateral side to the injury (Victor et al., 2006). This increased neuronal PPARγ expression was also supported by Ou et al. (2006), who found PPARγ-positive cells in the ipsilateral hemisphere following transient MCAO/CCAO injury with probes recognizing PPARγ. Furthermore, 15d-PGJ2 treatment resulted in increased binding of PPARγ to PPRE and reduced the area of infarct (Ou et al., 2006). Interestingly, intracerebral infusion of 15d-PGJ2 and rosiglitazone were able to increase PPARγ protein level in the normal brain as well (Lin et al., 2006). These authors showed that infusion of 15d-PGJ2 and rosiglitazone into brain given less than 2h post MCAO reduces infarct volume, decreases caspase-3 activation and apoptotic/necrotic cascades occurring after an ischemic insult.

There are also some studies supporting the beneficial effects of PPARα in cerebral ischemia. The PPARα agonist fenofibrate when given orally for 14 days prior to MCA occlusion reduced cortical infarct volume as opposed to very little effect of acute i.p injected fenofibrate (Deplanque et al., 2003). This discrepancy might be due to the slow rate of penetration of fenofibrate into brain. This protection by fenofibrate is suggested to occur via improved sensitivity of endothelial cells, decreased cerebral oxidative stress by increased activity of Cu/ZnSOD and prevention of pro-inflammatory events such as VCAM-1 and ICAM-1 expression in the post-ischemic brain (Deplanque et al., 2003). In further support to the beneficial effects of stimulating PPARα, treating rats with the PPARα antagonist WY14643 increased the oxidative stress, iNOS, and ICAM-1 expression following transient ischemia (Collino et al., 2006b). Recently, the beneficial role of PPARβ/δ agonists L-165041 and GW501516 in transient MCAO was demonstrated, perhaps via their anti-apoptotic effect as shown using in vitro conditions (Iwashita et al., 2006).

8. Future directions

While more evidences accumulate supporting the significance of transcription factors in promoting and curtailing inflammation thus contributing to brain damage after stroke, the therapeutic targeting of individual transcription factors is still at its infancy. Developing small molecules that can prevent the activation of pro-inflammatory transcription factors such as C/EBPβ, IRF-1 and NF-κB without undesired side effects may be beneficial as future stroke therapeutics. On the other hand, the agonists for anti-inflammatory transcription factors PPARα and PPARγ show much promise in the experimental models of stroke. Large scale clinical trials with thiazolidinediones (PPARγ agonists) and fenofibrate (PPARα agonist) will be the necessary next step to evaluate their safety and effectiveness in humans. In parallel, defining more specific mode of action by identifying the endogenous co-activators and modulators of these transcription factors in animal models will help to device more efficient therapeutic strategy. Studies in these directions are underway in several laboratories.

9. Conclusion

Inflammatory cascades are known to contribute to the pathophysiology of focal cerebral ischemia. Transcription factors induced in the post-ischemic brain plays a significant role in promoting as well as curtailing neuroinflammation; for example, transcription factors like C/EBPβ, Egr1, PPARs and NF-κB might promote and prevent inflammatory brain damage. By virtue of their prevention of many inflammatory reactions, the agonists that bind to the various isoforms of PPAR prevent ischemic brain damage. The neuroprotective effects of PPARγ agonists involve multiple mechanisms including preventing inflammatory gene expression, inducing anti-oxidant enzymes and other neuroprotective transcripts. PPARγ agonists might be acting in brain by stimulating PPARγ as well as directly without involving the PPARγ. The mechanisms of neuroprotection induced by PPARα and β/δ are not very clear and further studies are needed to elucidate them in depth. Overall, PPAR isoform activators might be developed into future stroke therapeutics.