**Therapeutic Potential of Peroxisome Proliferator-Activated Receptor (PPAR) Agonists in Substance Use Disorders: A Synthesis of Preclinical and Human Evidence**

#### **Justin Matheson 1,2, \* and Bernard Le Foll 1,3,4,5,6,7,8**


Received: 17 April 2020; Accepted: 8 May 2020; Published: 12 May 2020

**Abstract:** Targeting peroxisome proliferator-activated receptors (PPARs) has received increasing interest as a potential strategy to treat substance use disorders due to the localization of PPARs in addiction-related brain regions and the ability of PPAR ligands to modulate dopamine neurotransmission. Robust evidence from animal models suggests that agonists at both the PPAR-α and PPAR-γ isoforms can reduce both positive and negative reinforcing properties of ethanol, nicotine, opioids, and possibly psychostimulants. A reduction in the voluntary consumption of ethanol following treatment with PPAR agonists seems to be the most consistent finding. However, the human evidence is limited in scope and has so far been less promising. There have been no published human trials of PPAR agonists for treatment of alcohol use disorder, despite the compelling preclinical evidence. Two trials of PPAR-α agonists as potential smoking cessation drugs found no effect on nicotine-related outcomes. The PPAR-γ agonist pioglitazone showed some promise in reducing heroin, nicotine, and cocaine craving in two human laboratory studies and one pilot trial, yet other outcomes were unaffected. Potential explanations for the discordance between the animal and human evidence, such as the potency and selectivity of PPAR ligands and sex-related variability in PPAR physiology, are discussed.

**Keywords:** PPAR; nuclear receptors; addiction; alcohol; nicotine; opioids; psychostimulants; animal models; human studies

#### **1. Introduction**

Substance use disorders (SUDs) continue to represent a significant global public health burden. In 2017, of the estimated 271 million people aged 16–64 years worldwide who had used drugs in the past year, nearly 35 million (~13%) were estimated to suffer from an SUD [1]. An SUD is a diagnostic entity in the Diagnostic and Statistical Manual, 5th Edition (DSM-V) that refers to the repeated use of a substance that causes significant impairment, e.g., continued use despite physical and psychological harms and failure to meet expectations at work or school [2]. The term "addiction" is often used to refer to the severe stage of an SUD characterized by compulsive drug-seeking despite negative consequences [3,4] that runs a chronic, relapsing course with poor long-term durability of abstinence from drug-taking even with treatment [5].

Research into the neurobiology of addictions over the past few decades has substantively advanced our understanding of the key facets of compulsive drug-taking [6,7]. For example, while the focus of early addictions research was the acute, positively reinforcing properties of drugs of abuse, it is now recognized that negatively reinforcing states involving anhedonia, dysphoria, and anxiety become more important in maintaining drug-taking over time [7]. As a result, motivation to use the drug shifts from seeking pleasure to avoiding negative affect. Thus, pharmacotherapeutic strategies to treat addictions need to not only reduce the reinforcing or rewarding properties of drugs, but also target the negatively reinforcing states associated with chronic drug-taking that contribute to the significant risk of relapse [7]. Agonist substitution therapies have been successful in mitigating this negative reinforcement in some SUDs, e.g., methadone or buprenorphine for managing withdrawal and craving associated with opioid use disorder [8] and nicotine replacement therapy (NRT) for managing nicotine withdrawal [9]. Other medications, such as naltrexone or acamprosate for alcohol use disorder [10] and varenicline or bupropion for nicotine dependence [9], have demonstrated some efficacy in reducing positive and/or negative reinforcing aspects of drug use. Nevertheless, long-term abstinence rates remain low across SUDs, highlighting the need for novel pharmacological treatment approaches.

Peroxisome proliferator-activated receptors (PPARs) are a subfamily of nuclear receptors that dimerize with retinoid X receptors (RXRs) to regulate gene expression by binding to specific peroxisome proliferator response elements (PPREs) in enhancer sites of select genes [11]. Three isoforms of PPARs have been identified: α, γ, and β/δ. So far, the therapeutic potential of PPAR ligands had been in non-psychiatric fields. While PPARs were initially identified as lipid sensors [11], burgeoning evidence has demonstrated a role of these nuclear receptors in a wide range of physiological functions, including central nervous system (CNS) functions such as memory consolidation and modulation of pain perception [12]. PPAR agonists have been recently considered for their potential to treat neuropsychiatric disorders, largely due to their ability to target levels of neuroinflammation thought to be involved in the pathophysiology of these illnesses [13]. In particular, mounting evidence of an important relationship between neuroimmune function and addition-related processes has generated interest in investigating the role of PPARs in drug-related behaviors [14,15].

Converging lines of evidence have also suggested a more direct role of PPARs in addiction-relevant neurocircuitry. Initial evidence came from studies demonstrating that selective inhibition of fatty acid amide hydrolase (FAAH), an enzyme responsible for degradation of the endogenous cannabinoid anandamide and the endogenous PPAR ligands oleoylethanolamide (OEA) and palmitoylethanolamide (PEA), could suppress nicotine-induced activation of dopamine neurons in rats [16,17]. Importantly, this effect was mimicked by OEA and PEA, but not anandamide, suggesting the effect was due to PPAR activation specifically [16]. Exogenous PPAR agonists have also been demonstrated to attenuate nicotine-induced [18,19] and heroin-induced [20] excitation of dopamine neurons in the ventral tegmental area (VTA) and elevations of dopamine in the nucleus accumbens (NAc) shell in rats. Further confirmatory evidence comes from rodent studies demonstrating that PPAR isoforms are indeed localized in addiction-relevant brain regions such as the VTA [21,22], an important part of the mesocorticolimbic dopaminergic system that plays a central role in drug-related reward [7], and that PPAR-γ colocalizes with tyrosine-hydroxylase-positive cells in the VTA, suggesting direct expression

in dopaminergic neurons [23]. A detailed presentation of the neurobiological substrates mediating the impact of PPAR agonists on addiction-related behaviors is beyond the topic of the present review (see [18,19] for some mechanistic studies).

The goal of the present review is to expand upon our previous review of the preclinical evidence for a role of PPARs in addiction [24] to incorporate novel preclinical findings as well as the current state of evidence from clinical and laboratory studies in humans.

#### **2. Preclinical Behavioral Evidence**

Evidence for the role of PPAR agonists in modifying addiction-like behaviors in animal models is broadly divided into two categories: drug consumption/motivation to use and withdrawal/relapse. A summary of key methodological details and relevant findings of the studies reviewed is provided in Table 1.

#### *2.1. PPAR-*α *Agonists*

#### 2.1.1. Consumption/Motivation

A significant body of evidence has consistently demonstrated that PPAR-α agonists can attenuate voluntary consumption and operant self-administration of ethanol in rodents [25–32]. Using the two-bottle choice paradigm, studies have found a decrease in voluntary consumption of ethanol following administration of the clinically useful drugs gemfibrozil [25] and fenofibrate [26,27,29,30,32], the endogenous agonist OEA [28], the experimental agonist WY14643 [28], and the dual PPAR-α/γ agonist tesaglitazar [26,29,30]. In addition, operant self-administration of ethanol was attenuated following administration of OEA and WY14643 under a one-response fixed ratio (i.e., FR1) schedule [28] and fenofibrate under FR2 and progressive ratio (PR) schedules [31]. Importantly, the effects of the PPAR-α agonists on attenuation of voluntary consumption of ethanol were reversed when animals were pre-treated with the PPAR-α antagonists GW6471 [28] or MK886 [30]. Overall, these results strongly support a role of PPAR-α agonism in reducing willingness to consume ethanol and in reducing the reinforcing properties of ethanol.

Two studies have assessed the effects of fenofibrate on the development of ethanol conditioned place preference (CPP) as a measure of the rewarding effects of ethanol, with mixed results [32,33]. Blednov et al. (2016) found no effect of oral administration of 150 mg/kg fenofibrate or 1.5 mg/kg of the dual PPAR-α/γ agonist tesaglitazar on the development of ethanol CPP in male mice [33]. However, Rivera-Meza et al. (2017) found that oral administration of 50 mg/kg fenofibrate attenuated the development of ethanol CPP in male rats selectively bred for high ethanol intake (i.e., UChB rats) [32]. The inconsistency between these two studies is unclear but could be due to the different doses of fenofibrate used or differences in ethanol-related behaviors of the two different animal models.

More limited, but robust, evidence has supported a role of PPAR-α agonists in attenuating operant self-administration of nicotine in rodents and non-human primates [18,19]. Mascia et al. (2011) found that both WY14643 and methyl-OEA reduced nicotine self-administration under an FR5 schedule in rats and an FR10 schedule in monkeys, and that these effects were reversed by co-administration of the PPAR-α antagonist MK886 [18]. WY14643 had no effect on operant self-administration of cocaine in monkeys, demonstrating specificity to nicotine [18]. Panlilio et al. (2012) found further evidence that the clinically useful drug clofibrate prevented the acquisition of self-administration in naïve rats and decreased self-administration in experienced rats and monkeys, an effect that was reversed by treatment with MK886 [19]. Neither study found an effect of PPAR-α agonists on nicotine discrimination [18,19].

**Table 1.** Overview of methodological details and primary findings of the key studies providing behavioral evidence for a role of PPAR agonists in modulating addiction-related behaviors in animal models.


#### *Cells***2020**, *9*, 1196






#### **Table 1.** *Cont*.


2BC, two-bottle choice; BEZA, bezafibrate; CIG, ciglitazone; CLO, clofibrate; CPP, conditioned place preference; FEN, fenofibrate; FR, fixed ratio; GEM, gemfibrozil; i.c.v., intracerebroventricular; i.m., intramuscular; i.p., intraperitoneal; i.v., intravenous; KO, knock-out; methOEA, methyl oleoylethanolamide; OEA, oleoylethanolamide; OlGly, N-Oleoyl-glycine; PIO, pioglitazone; p.o., per os (oral); ROSI, rosiglitazone; SA, self-administration; TESA, tesaglitazar; WT, wild-type.

Two additional studies have suggested a role of PPAR-αagonists in attenuating nicotine CPP [40,43]. Jackson et al. (2017) found that both WY14643 and fenofibrate significantly reduced nicotine preference in the CPP experiments, though fenofibrate was less potent [40]. Importantly, WY14643 did not shift the potency of nicotine in the CPP paradigm, and the effect of WY14643 was specific to nicotine as it had no effect on cocaine preference [40]. In support of these findings, Donvito et al. (2019) found that exogenous administration of the lipid transmitter n-Oleoyl-glycine (OlGly) prevented the development of nicotine, but not morphine, CPP, and that this effect was blocked by the PPAR-α antagonist GW6471 [43]. Taken together, the results of the operant self-administration and CPP experiments provide strong support for a role of PPAR-α agonism in reducing the reinforcing and rewarding properties of nicotine.

Finally, one study found that OEA reduced behavioral sensitization to cocaine and cocaine CPP, though this effect was intact in PPAR-α KO mice, suggesting this was due to a PPAR-independent mechanism [36].

#### 2.1.2. Withdrawal/Relapse

Conflicting evidence exists regarding how PPAR-α agonists influence withdrawal from ethanol [28,33]. Bilbao et al. (2016) found that i.p. injection of 5 mg/kg of the endogenous PPAR-α agonist OEA significantly reduced total ethanol withdrawal scores in male rats, and furthermore decreased each of the individual withdrawal signs evaluated (vocalizations, head tremor and rigidity, tail tremor, and body tremor) [28]. Blednov et al. (2016) found that oral administration of 150 mg/kg fenofibrate or 1.5 mg/kg of the dual PPAR-α/γ agonist tesaglitazar actually increased withdrawal severity (handling-induced convulsions score) in male (but not female) mice [33]. The results of these two studies are difficult to compare given the different choices of PPAR-α agonist, dose, and route of administration, withdrawal signs evaluated, and animal models, but do suggest some role of PPAR-α in modulating ethanol withdrawal.

In the same experiments described above, Bilbao et al. (2016) found that both OEA and WY14643 were able to attenuate cue-induced reinstatement of ethanol-seeking after a period of deprivation [28], providing preliminary evidence that PPAR-α agonism may help to prevent alcohol relapse.

Two studies have suggested a role of PPAR-α agonists in reducing nicotine withdrawal signs. Jackson et al. (2017) assessed the impact of PPAR-α agonists on symptoms of precipitated nicotine withdrawal. They observed that WY14643 attenuated anxiety-like behaviors, hyperalgesia, and somatic withdrawal signs, while fenofibrate attenuated only somatic withdrawal signs [40]. Similarly, Donvito et al. (2019) found that exogenous administration of the lipid transmitter OlGly attenuated anxiety-like and somatic signs of nicotine withdrawal [43].

Finally, two studies have provided evidence that PPAR-α agonists can block reinstatement of nicotine-responding following a period of extinction [18,19]. Mascia et al. (2011) found that WY14643 attenuated reinstatement in both rats and monkeys using a procedure that combines both nicotine- and cue-induced reinstatement, and that this effect was reversed by co-administration of the PPAR-α antagonist MK886 [18]. Similarly, Panlilio et al. (2012) found that clofibrate attenuated both nicotine- and cue-induced reinstatement of nicotine responding in monkeys, and that these effects were reversed by pre-treatment with MK866 [19]. The reduction in withdrawal symptoms and the attenuation of both drug- and cue-induced reinstatement suggest that PPAR-α agonists may be useful in preventing relapse in nicotine-dependent smokers.

#### *2.2. PPAR-*γ *Agonists*

#### 2.2.1. Consumption/Motivation

Similar to the evidence for PPAR-α agonists, the results of several studies support a role of PPAR-γ agonists in attenuating voluntary consumption and operant self-administration of ethanol [26,29,30,35,37]. In the two-bottle choice paradigm, voluntary ethanol consumption was found to be attenuated by treatment with rosiglitazone [35] and pioglitazone [29,35,37], as well as the dual PPAR-α/γ agonist tesaglitazar [26,29,30]. Stopponi et al. (2011) additionally observed that pioglitazone significantly reduced operant self-administration of ethanol under an FR1 schedule [35]. While one study found that pre-treatment with the PPAR-γ antagonist GW9662 reversed the effects of pioglitazone on voluntary ethanol consumption [35], another study found no effect of GW9662 pre-treatment on the ethanol-reducing effects of the dual PPAR-α/γ agonist tesaglitazar, suggesting that the PPAR-α isoform may be more important in modulating ethanol-related behaviors than the PPAR-γ isoform [30].

Limited evidence suggests that PPAR-γ agonists may not influence ethanol CPP. As described above, Blednov et al. (2016) found no effect of tesaglitazar on ethanol CPP [33].

One study found that pioglitazone reduced operant self-administration of heroin under an FR1 schedule and significantly decreased the breakpoint in a PR schedule [20]. Furthermore, the effects of pioglitazone on self-administration were reversed by pre-treatment with the PPAR-γ antagonist GW9662 [20]. This preliminary evidence suggests that PPAR-γ agonists may be useful in reducing the reinforcing effects of opioids such as heroin.

Two studies have suggested that PPAR-γ agonists can attenuate behavioral sensitization to stimulant drugs [34,41]. Maeda et al. (2007) found that treatment with ciglitazone or pioglitazone during a withdrawal period, but not concurrently with methamphetamine, significantly attenuated behavioral sensitization to methamphetamine, while the PPAR-γ antagonist GW9662 significant augmented behavioral sensitization [34]. Miller et al. (2018) found that treatment with pioglitazone 4 days prior to testing significantly attenuated both the development and expression of behavioral sensitization to cocaine and attenuated lever-pressing for cocaine-associated cues during a period of forced abstinence [41].

#### 2.2.2. Withdrawal/Relapse

Similar to the results for PPAR-α agonists, the current evidence for an effect of PPAR-γ in modulating ethanol withdrawal signs is split [33,35]. As previously described, Blednov et al. (2016) found that the dual PPAR-α/γ agonist tesaglitazar increased withdrawal severity in mice [33]. In contrast, Stopponi et al. (2011) found that oral administration of both 10 and 30 mg/kg pioglitazone significantly reduced total withdrawal signs (composite score of ventromedial distal flexion responses, tail rigidity, and tremors) in rats [35]. While once again significant methodological differences prevent clear comparison of these results, it is important to note that in the same set of experiments, Blednov and colleagues did not find that the effects of tesaglitazar on ethanol-related behaviors were blocked by pre-treatment with the PPAR-γ antagonist GW9662 [30]. Thus, the ability of tesaglitazar to increase ethanol withdrawal severity in their experiment may not have been due to its actions at PPAR-γ.

Two studies have provided evidence for a role of PPAR-γ agonism in blocking reinstatement of ethanol-responding [35,37]. Both studies found that pioglitazone alone significantly attenuated stress-induced reinstatement (using yohimbine as a stressor), but not cue-induced reinstatement [35,37]. However, when pioglitazone was co-administered with naltrexone, there was an attenuation of cue-induced reinstatement [37]. These results suggest that PPAR-γ agonists may be useful in preventing alcohol relapse, possibly to a greater extent when administered concurrently with naltrexone, a non-selective opioid receptor antagonist that is already approved by the United States Food and Drug Administration (FDA) to treat alcohol use disorder [10].

One recent study found that PPAR-γ activation may play a role in nicotine withdrawal. Administration of pioglitazone prior to assessment of nicotine withdrawal attenuated somatic and anxiety-like signs of withdrawal in rats and in wild-type mice with intact PPAR-γ, but not in conditional neuronal PPAR-γ KO mice [42]. In addition, the effect of pioglitazone on both somatic and anxiety-like signs of nicotine withdrawal was blocked by pre-treatment with the PPAR-γ antagonist GW9662 in WT mice [42].

Finally, one study provided evidence that PPAR-γ agonists can reduce opioid withdrawal and relapse [39]. Treatment with pioglitazone significantly attenuated both the development and expression of morphine withdrawal in mice, and the PPAR-γ antagonist GW9662 blocked the ability of pioglitazone to attenuate the expression of morphine withdrawal [39]. Furthermore, pioglitazone significantly attenuated yohimbine- and heroin-induced reinstatement of heroin-responding in rats, while having no effect on cue-induced reinstatement [39]. Previously, the same group reported that pioglitazone significantly attenuated the development of analgesic tolerance to morphine [38], which provides additional evidence for a role of PPAR-γ in the effects of repeated morphine administration.

#### *2.3. Summary of Preclinical Evidence*

The majority of the preclinical behavioral evidence suggesting a role of PPAR agonists in addiction-like behaviors has focused on ethanol. Currently, the literature strongly supports a role of PPAR-α agonists (gemfibrozil, fenofibrate, OEA, and WY14643), and PPAR-γ agonists (rosiglitazone and pioglitazone) or a dual PPAR-α/γ agonist (tesaglitazar) to a lesser extent, in attenuating the voluntary consumption and reinforcing properties of ethanol in rodents. Limited evidence suggests that the PPAR-α agonist fenofibrate may additionally reduce the rewarding properties of ethanol, as assessed in the CPP paradigm. While agonists at both PPAR-α (OEA and fenofibrate) and PPAR-γ (pioglitazone) seem to have some role in modulating ethanol withdrawal signs, the nature of this role is unclear. However, the evidence does suggest that PPAR agonists may be useful in reducing the likelihood of alcohol relapse after a period of abstinence. PPAR-α agonists (OEA and WY14643) were shown to attenuate cue-induced reinstatement of ethanol-seeking, while a PPAR-γ agonist (pioglitazone) was shown to attenuate stress-induced reinstatement (and possibly also cue-induced reinstatement when co-administered with naltrexone).

Robust evidence from a limited number of studies strongly supports a role of PPAR-α (and possibly PPAR-γ) agonists in modulating nicotine-related behaviors in both rodents and non-human primates. The PPAR-α agonists methyl-OEA, WY14643, and clofibrate were found to reduce the reinforcing properties of nicotine. In addition, WY14643, fenofibrate, and OlGly were found to reduce the rewarding effects of nicotine in the CPP paradigm. WY14643 was shown to decrease behavioral and somatic signs of nicotine withdrawal, while both WY14643 and clofibrate reduced drug- and cue-induced reinstatement of nicotine-seeking. Finally, the PPAR-γ agonist pioglitazone reduced somatic and anxiety-like signs of nicotine withdrawal.

Preliminary evidence suggests that PPAR-γ agonists may have a role in modulating opioid-related behaviors. Studies found that pioglitazone was able to reduce the reinforcing effects of heroin in an operant self-administration paradigm, decrease both drug- and stress-induced reinstatement of heroin-seeking, and reduce the development and expression of morphine tolerance and withdrawal.

Finally, there seems to be a role of PPAR agonists in psychostimulant-related behaviors, yet the evidence is mixed. The PPAR-γ agonists ciglitazone and pioglitazone attenuated behavioral sensitization to methamphetamine, while pioglitazone attenuated behavioral sensitization to cocaine. Additionally, the endogenous PPAR-α agonist OEA attenuated behavioral sensitization to cocaine and cocaine CPP, but through a PPAR-α-independent mechanism. However, it is important to note that studies of nicotine-related outcomes found no effect of PPAR-α agonists on operant self-administration of cocaine or cocaine CPP.

#### **3. Clinical or Human Laboratory Evidence**

A summary of the methodological details and relevant findings of the human studies reviewed is provided in Table 2.

#### *3.1. PPAR-*α *Agonists*

Two published placebo-controlled studies have evaluated the potential of PPAR-α agonists in treatment of nicotine dependence [44,45]. Perkins et al. (2016) recruited nicotine-dependent smokers high in quit interest for a double-blind, counterbalanced, crossover trial with a target dose of 160 mg of fenofibrate administered once daily for 4 days following a 4-day dose run-up period [44]. There was no difference between fenofibrate and placebo on quit days, the primary outcome of the trial. In addition, there were no drug effects on any of the secondary outcomes, including pre-quit smoking reinforcement (i.e., number of puffs taken from participants' preferred brand of cigarettes and self-reported rewarding effects), craving responses during a cue reactivity task, and mean daily reductions in smoking [44]. In support of these negative findings, our lab found no effect of gemfibrozil (600 mg administered orally twice daily) on total self-reported days abstinent in a sample of nicotine-dependent smokers intent on quitting [45]. Similarly, we found no effects on secondary outcomes including a forced choice procedure (i.e., reinforcing effects) and both physiological and subjective measures of cue reactivity. In sum, despite the compelling preclinical evidence, the limited human evidence has not supported a role of PPAR-α agonists in treating nicotine dependence.

#### *3.2. PPAR-*γ *Agonists*

Three placebo-controlled studies have examined the potential for PPAR-γ agonists in modulating opioid-related outcomes [46–48]. In a sample of healthy, non-medical users of prescription opioids, there was no effect of 15 or 45 mg oral pioglitazone administered daily for 2–3 weeks on self-reported positive and negative subjective effects of oxycodone in a single-blind, within-subjects design [46]. In addition, pioglitazone had no impact on self-reported drug wanting (opioids, alcohol, cannabis, and tobacco) during the maintenance phases [46]. In a follow-up study, Jones and colleagues assessed the effects of 45 mg oral pioglitazone administered daily for 3 weeks in a sample of non-treatment-seeking adults with an opioid use disorder using a single-blind, randomized, between-subjects design [47]. Pioglitazone did not alter the reinforcing effects of heroin, its abuse liability, or cue reactivity, though self-reported ratings of "I want heroin" were significantly reduced [47]. Finally, Schroeder et al. (2018) assessed the potential for pioglitazone as an adjunct pharmacotherapy for patients with an opioid use disorder undergoing buprenorphine taper [48]. Pioglitazone treatment had no effect on withdrawal severity, and may actually have increased subjective withdrawal; yet, this trial was limited by very low recruitment numbers [48].

Two additional studies have investigated the role of pioglitazone in nicotine dependence and cocaine use disorder. In a single-blind, between-subjects laboratory study of nicotine-dependent smokers not interested in quitting, compared to placebo treatment, 45 mg oral pioglitazone administered daily for 3 weeks decreased self-reported measures of nicotine craving, though had minimal or no impact on reinforcing effects, self-reported positive or negative subjective effects, or cue reactivity [49]. In a pilot study to assess the potential of pioglitazone to target craving and white matter integrity in treatment-seeking adults with cocaine use disorder, daily administration of 45 mg oral pioglitazone for 12 weeks conferred benefit over placebo in reducing cocaine craving [50].

Taken together, the limited available human evidence suggests that the PPAR-γ agonist pioglitazone may be beneficial in reducing heroin, nicotine, and cocaine craving. However, it remains unclear how PPAR-γ agonists may impact more direct measures of treatment efficacy such as quit days or reductions in drug use.

**Table 2.**Overview of methodological details and primary findings of the key clinical and human laboratory studies of PPAR agonists in drug-related outcomes.



**Table 2.** *Cont*.

COWS, Clinical Opiate Withdrawal Scale; FEN, fenofibrate; GEM, gemfibrozil; p.o., per os (oral); PIO, pioglitazone; SA, self-administration; SOWS, Subjective Opiate Withdrawal Scale.

#### **4. Synthesis of the Preclinical and Human Evidence**

Given the robust preclinical evidence that both PPAR-α and PPAR-γ play a role in addiction-related behaviors, the lack of significant findings from human studies is somewhat surprising. For example, multiple preclinical studies demonstrated that PPAR-α agonists were effective in reducing the reinforcing and rewarding properties of nicotine and reducing nicotine withdrawal and reinstatement of nicotine-seeking [18,19,40,43], yet two human trials found no effect of the PPAR-α agonists fenofibrate [44] or gemfibrozil [45] on smoking cessation outcomes. Potential explanations for the poor concordance between the animal and human evidence to data are discussed below.

Perhaps the most salient discordance between the animal and human literature is the complete lack of placebo-controlled trials of PPAR agonists for treatment of alcohol use disorder. One Phase II trial of pioglitazone for treatment of alcohol craving and other alcohol-related outcomes in adults with alcohol use disorder (ClinicalTrials.gov identifier: NCT01631630) was terminated due to feasibility problems. A similar Phase II trial of fenofibrate (ClinicalTrials.gov identifier: NCT02158273) has been completed, though the results are unpublished. The most consistently reported and robust addiction-related outcome associated with PPAR agonists in the preclinical literature is a reduction in voluntary consumption of ethanol. Yet, as of this writing, the potential for PPAR agonists in treatment of alcohol use disorders in human has not been reported in the published literature. Thus, this is an important priority for future research. Currently, most pharmacotherapies available for the treatment of substance use disorders are substance-specific (although some are able to affect different substance use disorders). Therefore, it would be important to study the role of PPAR agonists in various substances use disorders, as it is unlikely that a single drug would be able to cure all substance use disorders.

The choice of PPAR agonist and dose is likely an important source of the poor translation from animal to human studies. For example, Jones and colleagues noted that the pioglitazone dosing parameters they employed were based on clinical utility in treating type-II diabetes [46,49], which may not be sufficient to elicit an effect in attenuating the abuse liability or reinforcing effects of opioids or nicotine. Similarly, while the preclinical evidence for a role of fibrate drugs in attenuating nicotine-related behaviors came from a study administering clofibrate [19], Perkins et al. (2016) used fenofibrate instead, as clofibrate was removed from the U.S. market due to its adverse effects [44]. Fibrate drugs, in general, may be less effective in reducing the rewarding and reinforcing effects of nicotine compared to experimental compounds such as WY14643 [40]. This could be due to the poor blood-brain barrier penetrance of fibrates like fenofibrate [51,52] or the low potency and PPAR-α selectivity of fenofibrate [53]. It should be noted in general that the PPAR agonists available do not act with 100% selectivity on specific PPAR isoforms and therefore, action on multiple PPAR isoforms is a possibility that should be kept in mind while interpreting the research results. Thus, different dosing paradigms, or perhaps more potent and selective PPAR agonists, may be needed to elicit clinically meaningful outcomes.

Similarly, species differences in the distribution and signaling of PPARs could also play in a role in the negative human findings. For example, significant differences in the expression [54] and activity [55] of hepatic PPAR-α has been demonstrated in human and rat, in part due to differences in the PPREs of target genes [55]. In addition, species differences have been demonstrated in PPAR-α binding of and response to specific ligands (including clofibrate) [56]. While one recent study did suggest a similar brain distribution of PPARs in adult mice and humans [57], it is still possible that species differences in PPAR-ligand dynamics and in PPAR distribution and signaling could limit the translation of findings from animal models to human studies. The fact that there is poor inter-species comparability in the activity of PPAR agonists is not something unique for PPAR ligands. There have been multiple cases of drugs that appear to be effective in preclinical studies that have not been effective in clinical trials. For example, despite an extensive preclinical literature showing that corticotropin-releasing hormone (CRH) acting via its CRH1 receptor can affect alcohol-seeking behavior, the drug pexacerfont, a CRH1 brain-penetrant antagonist, had no clinical efficacy in a clinical trial in subjects with alcohol

dependence [58]. Although it is yet too early to determine if PPAR agonists would similarly fail in humans, this remains a possibility.

Another possibility is simply that the published human studies were underpowered and too few in number to draw conclusions. Jones and colleagues note in two of their pioglitazone studies that they did not reach their recruitment goals [47,49]. Schroeder et al. (2018) noted significant difficulty in recruiting for their study of pioglitazone effects on opioid withdrawal during buprenorphine taper, reaching less than half of their target recruitment [48]. Schmitz et al. (2017), despite finding a potentially meaningful effect of pioglitazone on cocaine craving, note that their study was a pilot trial not specifically powered to detect a difference between drug conditions [50]. Appropriately powered randomized clinical trials are required to clarify the human evidence.

Finally, one possibility that has yet to be considered is the role of sex-related factors in the behavioral pharmacology of PPAR agonists. As seen in Table 1, the overwhelming majority of preclinical studies reviewed included only male animals in their experiments. In the two papers that did report sex differences, the PPAR-α agonist fenofibrate was shown to have more consistent and robust effects on ethanol-related outcomes (voluntary consumption and withdrawal severity) in male mice compared to female mice [30,33]. Furthermore, emerging evidence has found higher expression of PPAR-α mRNA and protein in immune cells of male rodents [59,60]; a role of PPAR-α in neuroprotection [61] and hippocampal synaptic plasticity [62] in male, but not female, rodents; and sex differences in the adverse effects and pharmacokinetics of PPAR-γ agonists such as pioglitazone in humans [63]. Given that all human studies reviewed included female participants (though consistently a small minority), sex differences in the effects of PPAR agonists on drug-related outcomes could have obscured overall drug effects.

#### **5. Future Directions**

Given the robust preclinical evidence for an effect of PPAR-α agonists in particular on ethanol-related outcomes, an important first step in moving forward with translating the animal evidence will be conducting human laboratory studies to determine if PPAR agonists (such as gemfibrozil or fenofibrate) modulate the subjective and reinforcing effects of alcohol. Subsequent to this, or in parallel, pilot RCTs to evaluate the efficacy and feasibility of administering PPAR agonists in alcohol use disorder will be necessary.

PPAR-α agonists showed promise for targeting nicotine-related behaviors in animal models, yet two adequately powered human trials found no benefit of fenofibrate or gemfibrozil on smoking cessation or other nicotine-related outcomes. It is possible that these agonists do not have sufficient pharmacological activity at PPAR-α to elicit clinically meaningful outcomes. Indeed, preclinical evidence has shown that more potent compounds like WY14643 confer benefit in attenuating nicotine-related behaviors over fibrates [40]. Selective PPAR modulators (SPPARMS), such as the highly potent and selective PPAR-α agonist K-877, have already shown some promise in treating dyslipidemias and insulin resistance with favorable adverse effect profiles compared to approved drugs such as fenofibrate [53]. If these compounds continue to show efficacy with limited adverse effects, it may be worth testing SPPARMS as smoking cessation drugs in RCTs.

It is possible that targeting PPAR isoforms alone may not be sufficient to treat addictions. For example, as discussed previously, pioglitazone was more effective in reducing reinstatement to ethanol-seeking when it was co-administered with naltrexone [37], an opioid receptor antagonist, suggesting some degree of synergy between PPAR activation and opioid receptor inhibition. Similarly, it has been proposed that simultaneous inhibition of FAAH and activation of PPARs may have an additive or even synergistic effect in treating cancers [64], and this approach may similarly hold promise in the context of addiction pharmacotherapy [65]. Future studies should consider possible synergistic effects that could be achieved by modulation of multiple signaling systems.

It will also be important to validate that the PPAR ligands that are used for SUD treatment are able to occupy/activate brain PPARs. Use of brain imaging approaches such as positron emission tomography could be useful for such target engagement validation. This is critical as some of the previous drug indications for PPAR ligands were likely mediated through PPAR action at the periphery [66].

The PPAR-β/δ isoform was not discussed in this review due to the lack of evidence implicating PPAR-β/δ agonists in addiction-related behaviors. However, it is important to note that PPAR-β/δ is present in the rodent brain at higher levels than the other two isoforms [67] and may play a role in regulating the expression and activity of PPAR-α and PPAR-γ [68]. Furthermore, limited evidence has suggested a role of PPAR-β/δ in neurodevelopmental and neurodegenerative disorders, possibly related to its anti-inflammatory properties [13]. Thus, future studies should investigate the role of PPAR-β/δ agonists in behavioral models of addiction.

A robust body of literature has demonstrated sex-related variability in the effects of common drugs of abuse and in addiction-related processes across animal and human studies [69–71], and emerging evidence suggests similar sex-related variability in the pharmacology of PPAR ligands and in PPAR signaling and function [59,61–63]. Considering sex as a biological variable in future animal studies of PPAR agonists and addiction-related behaviors will be another important next step.

Taken together, this review highlights the robust findings obtained in preclinical studies with agonists at both the PPAR-α and PPAR-γ isoforms that appear effective to reduce both positive and negative reinforcing properties of various drugs of abuse. However, the clinical findings are so far mixed and seem to indicate that the potential is much lower in human subjects. At this point, it is still important to perform small-scale appropriately powered proof of principle studies with PPAR drugs engaging brain PPARs to validate these findings in humans. Positive signals should then be followed by larger RCTs for further validation.

**Author Contributions:** Writing—original draft preparation, J.M. & B.L.F.; writing—review and editing, J.M. & B.L.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** Le Foll has obtained funding in the past from Pfizer (GRAND Awards, including salary support and donation of product) for investigator-initiated projects, notably on the impact of gemfibrozil on smoking cessation. All other fundings from Le Foll are unrelated to this review topic. Le Foll is supported by a clinician-scientist award from Department of Family and Community Medicine from the University of Toronto and is supported by the Center for Addiction and Mental Health.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **PPAR-Mediated Toxicology and Applied Pharmacology**

#### **Yue Xi 1,2 , Yunhui Zhang 1 , Sirui Zhu 1 , Yuping Luo 1 , Pengfei Xu 2, \* and Zhiying Huang 1, \***


Received: 5 January 2020; Accepted: 30 January 2020; Published: 3 February 2020

**Abstract:** Peroxisome proliferator-activated receptors (PPARs), members of the nuclear hormone receptor family, attract wide attention as promising therapeutic targets for the treatment of multiple diseases, and their target selective ligands were also intensively developed for pharmacological agents such as the approved drugs fibrates and thiazolidinediones (TZDs). Despite their potent pharmacological activities, PPARs are reported to be involved in agent- and pollutant-induced multiple organ toxicity or protective effects against toxicity. A better understanding of the protective and the detrimental role of PPARs will help to preserve efficacy of the PPAR modulators but diminish adverse effects. The present review summarizes and critiques current findings related to PPAR-mediated types of toxicity and protective effects against toxicity for a systematic understanding of PPARs in toxicology and applied pharmacology.

**Keywords:** PPARs; toxicology; pharmacology; ligand

#### **1. Introduction**

Peroxisome proliferator-activated receptors (PPARs), a group of nuclear hormone receptors, are composed of three isoforms which were identified as PPARα, PPARγ, and PPARβ. Each is encoded by distinct genes and has different targeting ligands, tissue distribution, and biological activities. PPAR family proteins, like other nuclear receptors, have three main functional segments, activation function 1 (AF1) and the conserved DNA-binding domain (DBD), the hinge region, and the ligand-binding domain (LBD) and AF2. The variable N-terminal regulatory AF1 domain binds co-regulators and the conserved DBD, which can bind to the peroxisome proliferator response elements (PPREs). The mobile hinge region links DBD and the conserved LBD in the middle. LBD and the variable C-terminal AF2 domain form a large ligand binding pocket [1,2]. Due to the large LBD pocket, PPARs have the capacity to bind various compounds, including endogenous or synthetic ligands and xenobiotic chemicals. In pharmacology, the ligands of PPARs are classified into full agonists, partial agonists, neutral antagonists, and inverse agonists. Recently, we summarized the 84 types of PPAR synthetic ligands for the treatment of various diseases in current clinical drug applications [3]. The LBD contains a C-terminal AF2 motif that is a ligand-dependent activation region [4]. Under physiological conditions, PPARs bind with co-repressors and form heterodimers with retinoid X receptor (RXR) [5]. In response to ligand activation, the protein conformation is changed and stabilized, which leads to dissociation of co-repressors and the recruitment of transcription co-activators and DNA-binding cofactors. This complex regulates transcription of target genes by binding specific DNA sequences, called peroxisome proliferator response elements (PPREs), on promoter regions of target genes [6,7].

PPAR activated genes play critical roles in fatty acid transportation and catabolism, glucose metabolism, adipogenesis, thermogenesis, cholesterol transportation and biosynthesis, and anti-inflammatory response [4,8]. Because of their broad-spectrum biological activities, PPARs arouse much attention, and they are studied intensively. Accumulated studies show that activation of PPARs has unique pharmacological effects on cardiovascular function, neurodegeneration, inflammation, cancer, fertility, and reproduction, and it is well established for managing dyslipidemia, diabetes, insulin resistance, and metabolic syndrome, which stimulates researchers to persistently develop more new drugs targeting PPARs [9,10]. Some PPAR agonists are approved as clinical agents such as thiazolidinediones, fibrates, and glitazars, for the treatment of diabetes, dyslipidemia, and diabetes-associated complications, respectively.

Despite the multiple biological activities of PPARs, several studies and clinical cases indicated that PPARs mediate various adverse effects of drugs, especially PPAR ligands or xenobiotic chemical-induced toxicity in different systems. Thiazolidinediones (TZDs), one class of PPARγ agonists, can cause fluid retention, heart failure, and hepatotoxicity [11,12]. Glitazones, another form of PPARγ ligand, were reported to cause peripheral edema, congestive heart failure, and body weight gain. Gemfibrozil, as a valuable agent to coronary heart disease, was shown to induce tumorigenesis, muscle weakness, and liver hypertrophy [13]. The detailed information and adverse reactions or toxicity of the 18 approved clinical agents that target PPARs are summarized in Table 1. Because of the high prevalence of tumorigenesis in PPAR activation by synthetic compounds, the Food and Drug Administration (FDA) requires that any PPAR agonists undergo a two-year rodent carcinogenicity study before being tested in clinical trials [13]. Moreover, PPARs were shown to be involved in pollutant-induced toxicity in the cardiovascular system, liver, reproductive and developmental system, gastrointestinal tract, muscle, and nervous system.

Based on the above information, this review is focused on the reports of PPAR activation-mediated toxicity and protective effects to date, aiming to provide an overview of studies evaluating the toxic role of PPARs in various systems and the molecular mechanisms of PPAR-elicited toxicity.

Fenofibrate

Fenofibrate

(Antara)

(Antara)

Choline fenofibrate

Choline fenofibrate

(Fenofibric Acid)

(Fenofibric Acid)

Bezafibrate

Bezafibrate

Bezafibrate

(Bezalip)

(Bezalip)

(Bezalip)

Choline fenofibrate(Fenofibric PPARα

PPARα agonist

PPARα agonist

(Antara)PPARα agonist360.834C20H21ClO4

PPARα agonist

PPARα agonist

PPARα agonist

PPARα agonist

360.834

360.834

421.918

421.918

421.918

361.822

361.822

361.822

H20ClNO

H20ClNO

H28ClNO

H28ClNO

5

5

4

4

H21ClO

H21ClO

4

4

C20

C20

C22

C22

22

H

C19

C19

19

H


Abbvie

Abbvie

Abbvieemia

Abbvie

Abbvie

Abbvie

Roche

Roche

Diagnostics

Diagnostics

Roche Diagnostics

Hypercholesterol

Hypercholesterol

Common: abnormal liver tests including aspartate

Common: abnormal liver tests including aspartate

Common: abnormal liver tests including aspartate

aminotransferase (AST) and alanine

aminotransferase (AST) and alanine

aminotransferase and alanine aminotransferase

aminotransferase (ALT), and headache

aminotransferase (ALT), and headache

Rare: high blood pressure, dizziness, itching,

Rare: high blood pressure, dizziness, itching,

Rare: high blood dizziness, itching,

nausea, upset stomach, constipation, diarrhea,

nausea, upset stomach, constipation, diarrhea,

upset constipation,

urinary tract infections, muscle pain, kidney

urinary tract infections, muscle pain, kidney

urinary tract infections, pain, kidney problems, and respiratory tract infections

problems, and respiratory tract infections

problems, and respiratory tract infections

Diarrhea, dyspepsia, nasopharyngitis, sinusitis,

Diarrhea, dyspepsia, nasopharyngitis, sinusitis,

upper respiratory tract infection, arthralgia,

upper respiratory tract infection, arthralgia,

Diarrhea, dyspepsia, nasopharyngitis, sinusitis, upper respiratory tract infection, myalgia, pain in extremities, dizziness

myalgia, pain in extremities, dizziness

myalgia, pain in extremities, dizziness

Stomach upset, stomach pain, gas, or nausea may

Stomach upset, stomach pain, gas, or nausea may

occur in the first several days; itchy skin, redness,

occur in the first several days; itchy skin, redness,

headache, and dizziness

headache, and dizziness

hyperlipidemiaStomach upset, stomach or nausea may occur first several days; skin, redness, headache, and dizziness

emia

emia

Hypertriglyceride

Hypertriglyceride

Hypertriglyceridemia

mia

mia

Hyperlipidemia

Hyperlipidemia

Hyperlipidemia

Hypertriglyceride

Hypertriglyceride

Hypertriglyceridemia

mia

mia

hypercholesterole

hypercholesterole

hypercholesterole

mia mixed

mia mixed

mia mixed

hyperlipidemia

hyperlipidemia

**Table 1.**The information and adverse reactions or toxicity of peroxisome proliferator-activated receptor (PPAR) targets related to 18 approved clinical drugs.**Generic NameType of PPAR Molecular Weight and Generic NameType of PPAR Molecular Weight and Generic Name (Brand Name)Type of PPAR AgonistMolecular Weight and Molecular FormulaStructureCompanyIndicationsAdverse Reaction or Toxicity**

Rosiglitazone

Rosiglitazone

maleate/metformin

Rosiglitazone

maleate/metformin

hydrochloride

(Avandamet)

maleate/metformin

PPARγ agonist;

AMPK activator

PPARγ agonist;

PPARγ agonist;

473.5(C22

473.5(C22

(C24

(C24

H34

H34

N

N

H23

4

H23N

4

O

O

N

3

3

O

O

7S)

7S)

6(C

6(C

H23

H23

H

4

4

4

N

N

N

H12ClN

H12ClN

HClN

490.62

490.62

3

O

O

3O

3

5

5S)/473.5(C22

5S)/473.5(C22

)

5

)

7S)/165.

7S)/165.

7S)/165.

AMPK activator

Sulfonylurea

Sulfonylurea

receptor

receptor

modulator/PPA

Rγ agonist

modulator/PPA

Rγ agonist

hydrochloride

(Avandamet)

Glimepiride/rosiglitazo

Glimepiride/rosiglitazo

ne maleate (Avandaryl)

Fenofibrate/simvastatin

(Cholib)

(Cholib)

(Cholib)

Fenofibrate/simvastatin

Fenofibrate/simvastatin

Fenofibrate/simvastatin

PPARα agonist/

inhibitor

inhibitor

inhibitor

HMGCR

HMGCR

HMGCR

PPARα agonist/

PPARα agonist/

PPARα agonist/

360.834

360.834

360.834

360.834

O

O

O

O

439.57

C25H29NO4S

H29NO

H29NO

439.57

439.57

H29NO

4

4

4

S

S

S

5

5

5

5

)

)

)

)

H38

H38

H38

H38

4)/418.57(C25

4)/418.57(C25

4)/418.57(C25

4)/418.57(C25

H21ClO

H21ClO

H21ClO

(C20H21ClO

C25

C25

C25

(C20

agonist439.57

(C20

(C20

HMGCR

inhibitor

PPARα/γ

agonist

(Lipaglyn)PPARα/γ

agonist

PPARα/γ

PPARα/γ

agonist

(Cholib)

Saroglitazar

(Lipaglyn)

(Lipaglyn)

Saroglitazar

Saroglitazar

Saroglitazar

(Lipaglyn)

ne maleate (Avandaryl)


**Table 1.** *Cont*.

Glimepiride/rosiglitazo(Avandaryl)receptor modulator/PPARγ <sup>5</sup>OS)GlaxoSmithKlineDiabetesCardiac major

GlaxoSmithKline

GlaxoSmithKline

GlaxoSmithKline

Mylan

Mylan

Mylan

Mylan

Zydus Cadila

Zydus Cadila

Zydus Cadila Mixed hyperlipidemia

Mixed hyperlipidemia

Mixed hyperlipidemia

Mixed hyperlipidemia

Diabetic dyslipidemia

Diabetic dyslipidemia

Diabetic dyslipidemia

Zydus CadilaDiabetic dyslipidemiaAsthenia, gastritis, chest discomfort, peripheral edema, dizziness, and tremors

GlaxoSmithKline

Diabetes

Diabetes

5GlaxoSmithKlineDiabetesLactic adverse

Diabetes

Diabetes

Hyperlipidemia

Hypertriglyceride

Hyperlipidemia

Lactic acidosis, cardiac failure, adverse

Lactic acidosis, cardiac failure, adverse

cardiovascular events, edema, weight gain, hepatic

effects, macular edema, fractures, hematologic

cardiovascular events, edema, weight gain, hepatic

effects, macular edema, fractures, hematologic

effects, and ovulation

cardiovascular weight gain, hepatic effects, edema, effects, and ovulation

effects, and ovulation

Cardiac failure with rosiglitazone, major adverse

Cardiac failure with rosiglitazone, major adverse

cardiovascular events, hypoglycemia, edema,

cardiovascular events, hypoglycemia, edema,

weight gain, hepatic effects, macular edema,

cardiovascular events, hypoglycemia, edema, gain, effects, edema,

weight gain, hepatic effects, macular edema,

fractures, hypersensitivity reactions, hematologic

effects, hemolytic anemia, and increased risk of

fractures, hypersensitivity reactions, hematologic

reactions,

risk

effects, hemolytic anemia, and increased risk of

cardiovascular mortality for sulfonylurea drugs

cardiovascular mortality for sulfonylurea drugs

cardiovascular

Raised blood creatinine levels, upper-respiratorytract infection (colds), increased blood platelet

Raised blood creatinine levels, upper-respiratorytract infection (colds), increased blood platelet

Raised blood creatinine levels, upper-respiratorytract infection (colds), increased blood platelet

Raised blood creatinine levels, upper-respiratoryinfection (colds), increased blood platelet

counts, gastroenteritis (diarrhea and vomiting)

and increased levels of alanine aminotransferase

and increased levels of alanine aminotransferase

and increased levels of alanine aminotransferase

counts, gastroenteritis (diarrhea and vomiting)

counts, gastroenteritis (diarrhea and vomiting)

counts, gastroenteritis (diarrhea and vomiting)

and increased levels of alanine aminotransferase

Asthenia, gastritis, chest discomfort, peripheral edema, dizziness, and tremors

Asthenia, gastritis, chest discomfort, peripheral edema, dizziness, and tremors

Asthenia, gastritis, chest discomfort, peripheral edema, dizziness, and tremors

Common: diarrhea, nausea

Gemfibrozil

Gemfibrozil

(Lopid)

(Lopid)

Ciprofibrate

Ciprofibrate

(Lipanor)

(Lipanor)

Pemafibrate

Pemafibrate

(Parmodia)

(Parmodia)

Pravastatin

Pravastatin

sodium/fenofibrate

PPARα agonist

PPARα agonist

PPARα agonist

PPARα agonist

PPARα agonist

PPARα agonist

3-hydroxy-3-

3-hydroxy-3-

methylglutaryl-CoA reductase

methylglutaryl-CoA reductase

(HMGCR)

250.338

250.338

H22

H22

289.152

289.152

H14Cl

H14Cl

490.556

490.556

N

N

H35NaO

H35NaO

2

2

O

O

6

6

7)/360.

7)/360.

H30

H30

2

2

O

O

3

3

C13

C13

C28

C28

446.5(C23

446.5(C23

O

O

3

3

C15

C15


**Table 1.** *Cont*.Laboratoires Laboratoires

Pfizer

Pfizer

Sanofi-Aventis

Sanofi-Aventis

> Kowa

Kowa Hyperlipidemia

Hyperlipidemia

Stomach upset, stomach/abdominal pain, nausea,

Stomach upset, stomach/abdominal pain, nausea,

vomiting, diarrhea, constipation, rash, dizziness,

vomiting, diarrhea, constipation, rash, dizziness,

headache, changes in the way things taste, muscle

headache, changes in the way things taste, muscle

pain

pain

Hair loss, balding, headache, balance problems,

Hair loss, balding, headache, balance problems,

feeling dizzy, drowsiness or fatigue, feeling sick

feeling dizzy, drowsiness or fatigue, feeling sick

(nausea) or being sick (vomiting), diarrhea,

(nausea) or being sick (vomiting), diarrhea,

indigestion or stomach pains, muscle pains

indigestion or stomach pains, muscle pains

Cholelithiasis (upper abdominal pain, fever) and

Cholelithiasis (upper abdominal pain, fever) and

diabetes mellitus (dry mouth, excess intake of

diabetes mellitus (dry mouth, excess intake of

fluid, excessive urination, fatigue)

fluid, excessive urination, fatigue)

Abdominal distension (bloating), abdominal pain

Abdominal distension (bloating), abdominal pain

(stomach ache), constipation, diarrhea, dry mouth,

(stomach ache), constipation, diarrhea, dry mouth,

dyspepsia (heartburn), eructation (belching),

dyspepsia (heartburn), eructation (belching),

Ischemic heart

Ischemic heart

disorder

disorder

Hyperlipidemia

Hyperlipidemia

Dyslipidemia

Dyslipidemia

Mixed

Mixed

hyperlipidemia

hyperlipidemia

#### **2. PPARs in Cardiotoxicity**

Considering the high expression level of PPARs in cardiac muscles and their strong implication in metabolic disorders and endocrine disruption, interference with PPARs can affect metabolic homeostasis and development of the cardiovascular system.

Cardiac edemas and the impairment of cardiac development were observed in marine medaka larvae fish exposed to perfluorooctane sulfonate (PFOS) by interfering PPARα and PPARβ [14, 15]. Perfluorooctanoic acid (PFOA) exposure-induced right-ventricular wall thinning elevation in chicken embryos is also likely due to PPARα [16]. After exposure to di-ethyl-hexylphthalate (DEHP), changes in the metabolic profile via the PPARα pathway can be detected in rat cardiomyocytes [17]. Triclocarban (TCC) is a high-performance broad-spectrum fungicide, which can induce cardiac metabolic alterations in mice by suppression of PPARα messenger RNA (mRNA) expression and other enzymes involved in energy and lipid metabolism. A further study found TCC directly interacted with the active site of PPARα in both mice and human tissues [18]. Exposure to airborne particulate matter is positively correlated with cardiorespiratory mortality [19]. Some studies showed that heart abnormal energy metabolism caused by seasonal ambient fine particles (PM2.5) was related to PPARα-regulated fatty-acid and glucose transporters. Unmanaged heart abnormal energy metabolism eventually leads to cardiac damage and heart failure [20]. Considerable research suggested that PPARs play pivotal roles in myocardial energy dysfunction. Energy substrate utilization showed a marked shift from fatty acid to glucose and lactate and cardiac hypertrophy in PPARα <sup>−</sup>/<sup>−</sup> hearts [21]. PPARα-null hearts with decreased contractile and metabolic remodeling were rescued by enhancing myocardial glucose transportation and utilization [21].

Furthermore, PPARγ inhibits cardiac growth and embryonic gene expression and decreases nuclear factor kappa B (NF-κB) activity in mice [22]. Cardiomyocyte-specific PPARγ knockout mice were more susceptible to cardiac hypertrophy with systolic cardiac function [22]. CKD-501, a new selective PPARγ agonist, induced heart toxicity in db/db mice by PPARγ-dependent mechanism [23]. Rosiglitazone leads to cardiac hypertrophy partially independent of cardiomyocyte PPARγ [22]. Another study indicated that rosiglitazone caused oxidative stress-induced mitochondrial dysfunction via PPARγ-independent pathways in mouse hearts [24]. Anna et al. reported that atorvastatin ameliorated cardiac hypertrophy by improving the protein expression of PPARα and PPARβ, which regulated the gene expression involved in fatty acid metabolism and avoided NF-κB activation by reducing the protein–protein interaction between PPARs and p65 [25]. Moreover, atorvastatin reduced the paraquat-induced cardiotoxicity via the PPARγ pathway [26]. Hesperidin, a flavanone glycoside and a known PPARγ ligand, improved cardiac hypertrophy by improving cardiac hemodynamics, as well as inhibiting oxidative stress and apoptosis through increasing PPARγ expression [27]. Piperine, a phenolic component of black pepper, attenuated cardiac fibrosis via PPARγ activation and the inhibition of protein kinase B (AKT)/glycogen synthase kinase 3 β (GSK3β) [28]. Interestingly, the regulation of PPARγ by pioglitazone suppressed cardiac hypertrophy as indicated by decreased heart/body weight ratio, wall thickness, and myocyte diameter [29], but the effect of pioglitazone on limiting myocardial infarct size was a PPARγ-independent event [30]. Epoxyeicosatrienoic acids (EET), a primary arachidonic acid metabolite, blocked tumor necrosis factor α (TNFα)-induced cardiotoxicity by reducing inflammation via upregulation of PPARγ expression [31].

Some dual PPARα/γ agonists such as tesaglitazar display an increased risk of cardiovascular events. Treatment with tesaglitazar in mice caused cardiac dysfunction associated with low mitochondrial abundance [32]. In addition, tesaglitazar increased acetylation of proliferator-activated receptor gamma coactivator 1α (PGC1α) and decreased the expression of sirtuin 1 (SIRT1), which was associated with competition between PPARα and PPARγ. LY510929, another dual PPARα/γ agonist, was shown to cause left-ventricular hypertrophy in rats [33]. However, aleglitazar inhibited hyperglycemia-induced cardiomyocyte apoptosis by activation of both PPARα and PPARγ [34,35]. Activation of PPARβ signaling mediated docosahexaenoic acid (DHA), and its metabolites elicited cytotoxicity in H9c2

cells via the de novo formation of ceramide [36]. Doxorubicin (DOX) caused a remarkable decrease in cardiac dP/dT and cardiac output by inhibition of PPARβ expression in rats [37]. β

PPARβ plays an important role in angiogenesis and cancers. Activation of PPARβ in blood vessels promotes tumor vascularization and the progression of different cancer cell types through direct activation of platelet-derived growth factor receptor beta (PDGFRβ), platelet-derived growth factor subunit B (PDGFB), and the c-Kit [38]. Figure 1 summarizes regulation of PPARs in cardiotoxicity. β β β

**Figure 1.** The regulation of PPARs in cardiotoxicity.

#### **3. PPARs in Hepatotoxicity**

α α In the liver, PPARs play indispensable roles in fatty-acid and glucose metabolism, and they supply energy to peripheral tissues. Numerous studies reported that xenobiotic chemicals and environmental contaminants disrupted the normal liver homeostasis by activating PPAR subtypes that are highly expressed in hepatocytes, especially PPARα. Indeed, PPARα was recognized as a target for pollutants, which could interact with the similar nuclear receptors and subsequently induce metabolic disorders.

α Phthalates, common plasticizers in nearly all plastic consumer goods, are defined as PPAR modulators [6]. Accumulative studies showed that phthalates activated PPARα and other lipid-activated nuclear receptors in the liver, which induced metabolic disruption and endocrine disorders. The exposure concentration of phthalate metabolites such as DEHP and mono (2-ethylhexyl) phthalate (MEHP) positively correlated with insulin resistance and abdominal obesity in American male adults [39–41]. Di-*n*-butyl-di-(4-chlorobenzohydroxamato) tin (DBDCT), an organotin with high antitumor activity, was also demonstrated to induce notable toxicity in rat liver tissue via the PPAR signaling pathway [42]. DBDCT treatment aroused acute and focal necrosis and Kupffer cell hyperplasia in rat liver. The decreased expression levels of cluster of differentiation 36 (CD36), fatty acid binding protein 4 (FABP4), enoyl-CoA hydratase and 3-hydroxyacyl CoA dehydrogenase (EHHADH), acetyl-CoA acyltransferase 1 (ACAA1), phosphoenolpyruvate carboxykinase (PEPCK), PPARα, and PPARγ in DBDCT-treated liver tissue were indicated by proteomics. Furthermore, the toxic effect was alleviated by PPARγ blocking agent T0070907 [42,43]. Additionally, organotins, the major components of agricultural fungicides and pesticides, were documented to exert similar functions as PPARγ and PPARβ ligands, which promote weight gain and increase fat storage by target gene induction in liver [44]. For example, tributyltin chloride (TBT) enhanced adipogenesis and adipocyte differentiation by directly stimulating downstream transcription of PPARγ in liver and adipose tissues. In mouse models, uterus exposure to TBT disrupted hepatic architecture and caused liver steatosis by increasing lipid accumulation and adipocyte maturation [40,45]. Thus, PPARs play an important role in contaminant-induced toxicity.

The hepatotoxicity of PPARα ligands was rarely documented. Few PPARα ligands are proven to be hepatotoxicants. Fenofibrate exerts only a minimal increase of alanine aminotransferase and aspartate aminotransferase [6,46]. In contrast to hepatotoxicity mediated by PPARs, PPAR ligands also display some protective effect against hepatotoxicity. PPARα ligand activation was proven to prevent acute liver toxicity induced by alcohol, carbon tetrachloride (CCl4), acetaminophen, chloroform, thioacetamide, and bromobenzene due to the induction of fatty acid catabolism and anti-inflammatory properties [47–49]. PPARα agonists showed a reversal of fatty liver in mice even with continued ethanol consumption [50]. PPARγ agonist troglitazone and rosiglitazone are reported to induce mild liver toxicity in patients that might be PPARγ-independent because of the low expression level of PPARγ in the liver [12]. Despite the hepatotoxicity of PPARγ activation, PPARγ ligand treatment attenuated fibrogenesis by inhibiting the activation of hepatic stellate cells (HSCs) [51,52]. PPARγ ligands exhibited a suppressive effect on the expression of fibrogenic genes including collagen and α-smooth muscle actin. PPARβ activation by L-165041 enhanced the HSC proliferation and fibrogenic gene expression, and it exacerbated CCl4-induced liver fibrotic progression [53]. PPARα, PPARγ, and PPARβ display different roles in hepatotoxicity. Activation of PPARα prevents acute liver toxicity. Activation of PPARγ induces mild liver toxicity but attenuates liver fibrogenesis. Activation of PPARβ promotes the progression of liver fibrosis.

Numerous studies reported that hepatocarcinogenesis was the major toxicity induced by PPARα activation [54,55]. Unmanaged peroxisomal proliferation and hepatomegaly observed in fibrate-treated livers can ultimately lead to hepatocellular carcinoma [56]. The hepatocarcinogenesis by PPARα activation was fully investigated over 30 years. The main target of PPARα is the liver, which induces pleiotropic impacts such as hypertrophy and hyperplasia [57,58]. These unmanaged responses cause hepatocellular carcinomas in rodents. The mechanisms remain elucidated. Some studies propose that PPARα-mediated DNA replication, proliferation, and suppressed apoptosis result in PPARα agonist-induced hepatocarcinogenesis [59]. Actually, the effect of PPARα on hepatocarcinogenesis varies among different species. In human, an increased risk of liver cancer of fibrates is not yet reported. This might be due to no significant peroxisome proliferation induced by hypolipidemic agents [60] and less expression of PPARα in patient livers compared to rodent liver. Although humans show resistance to the adverse effect of PPARα-induced hepatocarcinogenesis, vigilance is still required to develop new agents.

#### **4. PPARs in Gastrointestinal Toxicity**

As indicated by emerging evidence, PPARs and their ligands also play an important role in the regulation of immune and inflammatory reactions in the gastrointestinal (GI) system.

In view of modulation of several target genes involved in metabolic processes and immune response in the GI tract, PPARs and their ligands became a research hotspot in gastroenterology [61]. Accumulative evidence showed that inflammatory bowel diseases (IBDs) and colon cancer (CC), two important GI diseases, are related to PPARs and their ligands [62,63]. PPAR agonists might serve as a new effective pharmacotherapy for IBDs and CC. PPARα mediated the anti-inflammatory effect of glucocorticoid (GC) in a chemical-induced colitis mouse model [64]. More recently, it was shown that PPARα activation diminished the therapeutic effects of rSj16 in dextran sulfate sodium (DSS)-induced

colitis mice, indicating that the PPARα signaling pathway plays a crucial role in DSS-induced colitis progression [65].

With the high expression in GI tract mucosa, especially in the intestine and colon [66–68], PPARγ is closely related to GI injury and inflammatory response. The inflammatory reaction is the common pathological process of many GI diseases and trauma. Once the homeostasis of GI is disrupted by exogenous factors or endogenous metabolites and shifts to the pro-inflammatory state, the pro-inflammatory cytokines such as TNF-α, interleukin 1β (IL-1β), IL-6 are liberated by the hyperactive immune cells. Transcription factor NF-κB is one of the most important regulatory mechanisms of immune and inflammatory responses mediated by PPARs and their ligands in the GI tract. In colon, PPARγ downregulated NF-κB and mitogen-activated protein kinase (MAPK) signaling pathways, which subsequently inhibited the mucosal production of inflammatory cytokines [69]. Furthermore, in intestinal cells, activation of PPARγ resulted in decreased expression of intercellular adhesion molecule 1 (ICAM-1) and TNF-α [70], which are downstream targets of NF-kB [71]. Treatment with troglitazone attenuated colitis induced by intrarectal administration of 2,4,6-trinitrobenzene sulfonic acid (TNBS) [69]. PPARγ could function as an endogenous anti-inflammatory pathway in a murine model of intestinal ischemia–reperfusion (I/R) injury. Activation of PPARγ by its agonist BRL-49653 had a protective effect on intestinal acute I/R injury [70]. However, the protective activity of BRL-49653 was abolished in PPARγ-deficient mice. In the investigation of the prevention and treatment of radiation-induced intestinal damage, accumulating evidence supported that the administration of PPARγ agonists alleviated radiation-induced intestinal toxicity. PPARγ agonists were shown to reverse radiation-induced apoptosis and inflammation and to exert radio-protective effects on healthy bowel upon irradiation [72,73]. Further research of acute intestinal injury reported that PPARγ agonist rosiglitazone reduced the expression of the fibrotic marker transforming growth factor β (TGFβ) and phosphorylation of the p65 subunit of NF-kB triggered by pro-inflammatory cytokine TNF-α [72]. In ulcerative colitis research, promoting the nuclear localization of PPARγ weakened the activity of NF-κB signaling in both rectal tissues from dextran sulfate sodium (DSS)-induced mice and lipopolysaccharide (LPS)-stimulated macrophages [74]. PPARs inhibited the expression of macrophage-related inflammatory mediators and macrophage infiltration in the acute irradiation intestinal damage [75]. Compared with wild-type mice, PPARγ-deficient mice showed significantly severe damage after an I/R injury procedure, which indicated the anti-inflammatory and protective role of PPARγ in GI damage [70]. These studies indicated the role of PPARγ in suppression of NF-kB activation and inflammatory response in intestinal tissues.

Additionally, several reports indicated that PPARs and their ligands could lead to carcinogenesis by affecting the metabolism of glucose and lipids. Intestinal PPARα exhibited a protective effect against colon carcinogenesis by inhibiting methylation of P21 and P27 [76]. Human colorectal tumors also show lower levels of PPARα compared to normal tissue [76]. PPARγ synthetic activator rosiglitazone has a radio-sensitizing effect on human bowel cancer cells [72]. PPARγ was reported to be associated with colorectal cancer via insulin and inflammatory mechanisms [77,78]. On the other hand, PPARγ was shown to be expressed in human colonic mucosa and cancer. The ability of PPARγ activation to decrease cyclooxygenase-2 (COX-2) expression and induce apoptosis suggests that the PPARγ pathway might be a tumor suppressor in humans [79]. Another study reports that 8% of primary colorectal tumors harbor function-dead mutations in one allele of the *PPAR*γ gene and emphasizes the potential role of this receptor as a therapeutic target for cancer or in designing a mouse colon cancer model [80]. The treatment of colon cancer by suppressing the methylation of PPARγ promoter and enhancing PPARγ expression is also underway, because the hyper-methylation of promoter regions can induce PPARγ gene silence. Moreover, the risk of radiation-induced intestinal toxicity in methylated patients was also increased compared with un-methylated patients [81]. Furthermore, PPARβ is induced in intestinal stem cells and progenitor cells in high-fat diet-treated mice and enhances stemness and tumorigenicity of intestine [82]. Arachidonic acid derivative prostaglandin E2 (PGE2), which is a biologically active lipid, increases cell survival and improves intestinal adenoma formation by indirectly activating PPARβ via the phosphatidylinositol-3 kinase (PI3K)/Akt signaling pathway [83]. The activation of PPARβ also upregulates COX-2, which is a key activator for colon cancer cells [59].

We summarize the regulatory mechanism of PPARs in gastrointestinal toxicity in Figure 2. A better understanding of the role of PPARs in the GI system will help to develop novel pharmacotherapy against colon carcinogenesis and diminish intestinal toxicity.

**Figure 2.** The regulation by PPARs in gastrointestinal toxicity.

#### **5. PPARs in Reproductive and Developmental Toxicity**

Three isoforms of PPARs were found in the reproductive system including the hypothalamus, pituitary, testis, ovary, uterus, adrenal, and mammary glands. Numerous studies showed that PPARs play a role in the normal reproductive and developmental functions, and abnormal regulation of PPARs by exposure to endogenous or exogenous compounds might lead to physiological dysfunction in reproductive system [84–86]. Thus, the research on reproductive and developmental disorders focuses on PPARs and their modulators.

α

Triptolide is a major active compound in Chinese herb *Tripterygium wilfordii* multiglycoside, and it is widely used for treatment of autoimmune diseases and nephrotic syndrome [87]. However, we previously reported that triptolide causes mitochondrial damage and dysregulates fatty-acid metabolism by upregulating expression and nuclear translocation of PPARα in mouse sertoli cells [88]. A metabolomics study revealed that triptolide caused impairment of spermatogenesis accompanied by abnormal lipid and energy metabolism in male mice through downregulation of PPARs [89]. Different concentrations and times of triptolide exposure led to the different behaviors of PPARs. These findings support that PPARs are key mediators in triptolide-induced reproduction toxicity.

Phthalates, which activate PPARs, have a remarkable effect on fertility rates, ovulation, development of the male reproductive tract, spermatogenesis, and teratogenesis [6]. Early exposure to phthalates influenced perinatal and postnatal cardiometabolic programming [90]. DEHP, a phthalate ester, is commonly used in industry as a plasticizer, which activates PPARα to regulate the expression of downstream target genes. DEHP treatment had no remarkable effect on body, liver, and ovary weight in female dams (F0) and offspring (F1) in either wild-type or PPARα-knockout mice. However, it suppressed the expression of ovarian estrogen receptor α, and the repression of ovarian estrogen receptor α expression by DEHP was lost in PPARα-knockout mice [91]. PPARα transcription is related to fertility impairment in female mice exposed to high doses of DEHP (500 mg/kg of body weight per day) [92]. Moreover, it was reported that MEHP, a principle active metabolite of DEHP, decreased the activity and production of aromatase, which converted testosterone to estradiol in ovarian granulosa cells by activating PPARα and PPARγ [93]. Benzo [a]pyrene (B [a]P) is a ubiquitous environmental contaminant, and the combination of B [a]P and DEHP induced ovotoxicity in female rats and suppressed sex hormone secretion via the PPAR-mediated signaling pathway [94].

Dehydroepiandrosterone (3c-hydroxy-5-androsten-17-one, DHEA) is a ligand of PPARα, and it also stimulates the production of PPARα. Some clinical studies showed that dietary supplementation of DHEA reversed the oocyte quality in mice and aged women [95,96]. Additionally, reduced DHEA and loss of function of PPARα result in the decreased follicle quality associated with the changes of fatty-acid metabolism, transport, and mitochondrial function. Perfluorooctanoic acid (PFOA), a synthetic perfluorinated compound (PFC) which is widely distributed, significantly inhibited mammary gland growth in mice through activation of PPARα, and this effect was reversed by supplementation with exogenous estrogen or progesterone [97]. Moreover, perfluorooctane sulfonate (PFOS) is a product of metabolic degradation of PFCs and has an estrogenic activity and endocrine-disruptive properties in the marine medaka embryos, partially through the regulation of PPARs.

Additionally, 15-deoxy-delta12,14-prostaglandin J2 (15dPGJ2), which is converted by arachidonic acid via successive dehydration and isomerization, acts as an endogenous ligand of PPARγ via direct covalent binding, and it plays a key role in lipid homeostasis [98,99]. Kurtz and colleagues found that 15dPGJ2 partially restored the mRNA expression of oxidizing enzymes including acyl-CoA 1 (ACO) and carnitine palmitoyltransferase 1 (CPT1) in the lungs of male fetuses from diabetic rats, but this effect was not observed in female fetuses [100]. Moreover, it was reported that 15dPGJ2 modulated lipid metabolism and nitric oxide production in diabetes-induced placental dysfunction partially through the PPAR pathway [101]. Trichloroethylene (TCE) reduced fertilizability of oocyte and its ability to bind sperm plasma membrane proteins in rats [102]. A systematic evaluation of TCE showed that TCE could cause cardiac defects in humans when the exposure is during a sensitive period of fetal development [103]. Tributyltin chloride (TBT) activates all three types of PPARs. TBT has effects on reproductive function and induces abnormal mammary gland fat accumulation by increasing PPARγ expression [104,105]. TZDs (e.g., pioglitazone, rosiglitazone, and troglitazone) activate PPARγ to regulate the transcription of genes responsible for glucose and lipid metabolism. TZDs clinically sensitize peripheral insulin in patients with type 2 diabetes by regulating glucose and lipid metabolism [106,107]. Oral administration of rosiglitazone 4 mg once a day for three months improves hyperandrogenemia, insulin resistance, lipidemia, C-reactive protein levels, ovarian volume, and follicle number in patients with polycystic ovary syndrome (PCOS) [108]. Rosiglitazone exhibited significant protective effects on metabolic, hormonal, and morphological features of PCOS. Significant changes were also observed in the isovaleryl carnitine levels and lipid oxidation rates after pioglitazone treatment [109]. Rosiglitazone significantly improved oocyte quality in diet-induced obesity (DIO) mice, indicating the positive effect of PPARγ on ovarian function [110]. Rosiglitazone affects steroidogenesis in porcine ovarian follicles by stimulating PPARγ [111,112]. In vivo experiments demonstrated that fenofibrate inhibited ovarian estrogen synthesis [113]. A review concluded that clofibrate and gemfibrozil caused atypical changes in maternal and fetal liver during pregnancy, but there was no direct evidence of developmental toxicity or teratogenicity of clofibrate and gemfibrozil [6]. Irbesartan (IRB) is one of the most widely used angiotensin type 1 (AT1) receptor blockers (ARBs) with PPARγ agonistic activity. Rats treated with IRB showed an increase in estradiol and follicle-stimulating hormone levels, which subsequently ameliorated ovarian dysfunction [114]. These studies indicate that activation of PPARγ signaling protects ovarian function.

Genistein (49,5,7-trihydroxyisoflavone, GEN), a kind of isoflavones derived from soybeans, was investigated for its antioxidant, anticancer, and anti-inflammatory activities [115]. It is a natural ligand of PPARs, and it can improve the development and metabolism of chick embryos through the activation of PPARs [116,117]. Prostacyclin (PGI2) activated its nuclear receptor PPARβ to accelerate blastocyst hatching in mice [118]. These studies suggest that the activation of PPARs is involved in toxicant-induced reproductive toxicity.

The anti-tumor effects of PPAR agonists were documented. Rosiglitazone and troglitazone, both PPARγ activators, showed inhibitory effects on pituitary adenoma cells in mice and human, and they were considered to be a new oral drug for the treatment of pituitary tumors [119]. Moreover, troglitazone treatment stabilized the prostate-specific antigen levels in patients with advanced prostate cancer clinically by upregulating E-cadherin and glutathione peroxidase 3 [120]. Rosiglitazone showed an inhibitory effect on proliferation of primary human prostate cancer cells [121]. However, the activation of PPARβ by selective agonist GW501516 was reported to stimulate proliferation of human breast and prostate cancer cells which are responsive to sexual hormones [122]. PPARβ activation by GW501516 increased cyclin-dependent kinase 2 (CDK2) and vascular endothelial growth factor α (VEGFα) expression, indicating the improved cell proliferation and angiogenesis. This study suggested the possibility of PPARβ antagonists in treating breast and prostate cancer.

#### **6. Other Systemic Toxicity and Protective E**ff**ects Mediated by PPARs**

Fibrates, PPARα synthetic ligands, were developed for treatment of hyperlipidemia in the clinic, such as fenofibrate, bezafibrate, ciprofibrate, and so on [123–125]. However, muscle weakness, muscle pain, and even rhabdomyolysis were observed during their application [6]. Different fibrates lead to different degrees of myopathy, and that might be due to different mechanisms. The underlying mechanism is still unclear. Some studies reported that PPARα activation in skeletal muscle transactivated the genes encoding muscle proteases, and the increased expression of skeletal muscle proteases led to severe myopathy [126,127]. The muscle toxicity might result from the blood concentration of the drug, because remarkably higher incidence occurs in patients with kidney failure or hypoalbuminemia [128]. Moreover, Motojma et al. proposed that the increase in pyruvate dehydrokinase isoenzyme4 (PDK4) and the decrease in serum triglyceride (TG) level mediated by PPARα in skeletal muscle caused the degradation of protein in muscle, ultimately resulting in myopathy and even rhabdomyolysis [129]. Due to the low incidence of rhabdomyolysis, no drug was withdrawn from the market because of the muscular toxicity.

In contrast to the adverse effect mediated by PPARs, PPARs also exert protective effects against nephrotoxicity and neuron injury.

Diabetic kidney disease is one complication of type 2 diabetes. PPARα and PPARγ are famous targets for treating diabetes, especially PPARγ. Increasing studies indicated that PPARs play important roles in kidney physiology and pathology. In most cases, PPARγ serves as a therapeutic target for treating nephrotoxicity. PPARγ-null mice showed spontaneous diabetic nephropathy. PPARγ knockout mice exhibited kidney hypertrophy accompanied by increased glucosuria, albuminuria, renal fibrosis, and mesangial expansion [130,131].

PPARs also play key roles in regulating brain self-repair. Central nervous system diseases, neuron injury, and cell death are closely related to neuroinflammation [132,133]. Lovastatin (LOV) can protect vulnerable oligodendrocytes in a mouse model of multiple sclerosis (MS) by inhibiting guanosine triphosphate (GTP)-binding proteins, small Rho GTPases, via a PPARα-dependent mechanism [134]. Healthy oligodendrocytes are essential for the synaptic survival of MS neurons. PPARα activation increases the seizure threshold and controls the seizure frequency [134]. The high expression of PPARα in the brain region also prevents nicotine-induced neuronal damage by regulating tyrosine kinases and phosphokinases in neuronal current. It decreases the frequency of seizures caused by the activation of nicotine receptors in vertebral neurons [135]. Animal model studies showed that fenofibrate prevented convulsions caused by dysregulation of neurotransmitters [136]. Substantia nigra has high-density microglia which show two polarization states, M1 and M2, which have pro-inflammatory or anti-inflammatory effects, respectively [132,137]. Therefore, inhibiting the activation of M1 microglia and promoting the activation of M2 microglia are beneficial to central system diseases. In the condition of inflammation, M1 microglia are activated and release pro-inflammatory factors and neurotoxic substances, such as cytokines, reactive oxygen species, prostaglandins, and complements, which aggravate inflammatory injury [138]. Recent studies showed that PPARs (mainly PPARγ) regulate microglia-mediated inflammation in Parkinson's disease (PD) and other neurodegenerative diseases [138–140]. Pioglitazone, a PPARγ ligand, was shown to inhibit the activation and secretion of glial cells by activating PPARγ [141]. Pioglitazone also inhibits the degeneration of dopamine neurons, which induces inflammation and promotes neuron death [141]. Rosiglitazone has a protective effect on neurotoxin 1-methy-4-phenyl-1,2,3,4,6-tetrahydropyidine (MPTP)-induced PD mouse model via upregulation of M2 phenotypic-related anti-inflammatory factors and the downregulation of M1 phenotypic-related pro-inflammatory factors [142]. A recent study found that PPARα/γ dual agonist MHY908 protects dopamine neurons from MPTP-induced loss in PD mice by reducing neuroinflammation and microglia activation [141]. Moreover, L-165041, a PPARβ agonist, can inhibit the radiation-induced inflammation in microglia by inhibition of the NF-kB signaling pathway [143]. At present, Alzheimer's disease is also considered to be a neuroinflammatory disease and is characterized by abnormal accumulation of β-amyloid (Aβ). Under the condition of Aβ accumulation, M1 microglia were activated, resulting in neuronal injury and apoptosis [144]. It was shown that adiponectin can activate M2 microglia and enhance the clearance of Aβ by activating the PPARγ signaling pathway.

#### **7. Conclusions**

Better understanding of the role of PPARs in toxicology and pharmacology and the underlying molecular basis is necessary for PPARs-related clinical drug discovery and development. Unfortunately, there are limited studies reviewing the integrated network of relationships in these aspects. Lots of PPAR ligands have beneficial effects on applied pharmacology, but they are also accompanied by various toxicities. Here, we mainly summarized the regulation of PPARs in toxicology and protection against toxicity in various systems, such as cardiotoxicity, hepatotoxicity, gastrointestinal toxicity, and reproductive and developmental toxicity (Figure 3). We hope that a comprehensive understanding of PPAR-mediated toxicology and applied pharmacology will contribute to the safety of PPAR-targeted therapies in the future.

**Figure 3.** Concept map of the PPARs in various systemic toxicities.

**Author Contributions:** Conceptualization, P.X. and Y.X.; writing—original draft preparation, Y.X., P.X., Y.Z., S.Z. and Y.L.; writing—review and editing, Y.X., P.X. and Z.H.; supervision, P.X. and Z.H.; project administration, P.X. and Z.H.; funding acquisition, Z.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Natural Science Foundation of China (Grant No. 81773992), the General Projects in the Natural Science Foundation of Guangdong Province (Grant No. 2018A0303130170), and the National Engineering and Technology Research Center for New drug Druggability Evaluation (Seed Program of Guangdong Province, 2017B090903004). All support is gratefully acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **PPAR Beta**/**Delta and the Hallmarks of Cancer**

### **Nicole Wagner \* and Kay-Dietrich Wagner**

Université Côte d'Azur, CNRS, INSERM, iBV, 06107 Nice, France; kwagner@unice.fr

**\*** Correspondence: nwagner@unice.fr; Tel.: +33-493-377665

Received: 20 April 2020; Accepted: 1 May 2020; Published: 4 May 2020

**Abstract:** Peroxisome proliferator-activated receptors (PPARs) belong to the nuclear hormone receptor family. Three different isoforms, PPAR alpha, PPAR beta/delta and PPAR gamma have been identified. They all form heterodimers with retinoic X receptors to activate or repress downstream target genes dependent on the presence/absence of ligands and coactivators or corepressors. PPARs differ in their tissue expression profile, ligands and specific agonists and antagonists. PPARs attract attention as potential therapeutic targets for a variety of diseases. PPAR alpha and gamma agonists are in clinical use for the treatment of dyslipidemias and diabetes. For both receptors, several clinical trials as potential therapeutic targets for cancer are ongoing. In contrast, PPAR beta/delta has been suggested as a therapeutic target for metabolic syndrome. However, potential risks in the settings of cancer are less clear. A variety of studies have investigated PPAR beta/delta expression or activation/inhibition in different cancer cell models in vitro, but the relevance for cancer growth in vivo is less well documented and controversial. In this review, we summarize critically the knowledge of PPAR beta/delta functions for the different hallmarks of cancer biological capabilities, which interplay to determine cancer growth.

**Keywords:** peroxisome proliferator-activated receptor; angiogenesis; proliferation; metastasis; immortality; resistance to cell death; growth suppressors; immune system; cellular metabolism

#### **1. Introduction**

Peroxisome proliferator-activated receptors (PPARs) belong to the group of nuclear receptors. They exist in three different isoforms: PPARα (NR1C1), PPARβ/δ (NR1C2) and PPARγ (NR1C3). They heterodimerize with RXR; and upon ligand binding act mainly as transcriptional regulators of specific target genes. Dependent on the tissue distribution, cofactors and availability of ligands, PPARs exert multiple functions (reviewed in [1]). PPARα is mainly expressed in liver, heart, brown adipose tissue, kidney and intestine and regulates energy homeostasis by activation of fatty acid catabolism and stimulation of gluconeogenesis [2]. PPARβ/δ is more or less ubiquitously expressed with some species differences, while PPARγ is expressed in white and brown adipose tissue, the gut and immune cells [1]. Endogenous ligands for PPARs are fatty acids, triglycerides, prostacyclins, prostaglandins and probably retinoic acid. Although varies different binding sites for PPARs in target genes have been reported, they share in general as a response element a direct repeat of the sequence AGGTCA, spaced by a single nucleotide, which was originally identified for PPARα (reviewed in [1]). Thus, in case more than one of the receptors is expressed in a certain cell-type, one could expect cross talk in response to endogenous or pan-PPAR pharmacological agonists. Specific agonists for PPARα are used classically for the treatment of dyslipidemia and agonists for PPARγ are insulin sensitizers to treat patients with type 2 diabetes. Currently, no PPARβ/δ activators or antagonists are in official clinical use. A recent review summarized novel developments regarding patents for PPAR modulators and possible novel clinical indications [3]. Clinical evidence for the use of PPAR agonists and antagonists is reviewed in [4]. Toxicological aspects and side effects of PPAR modulators have been reviewed

recently [5]. Increasing interest focuses on potential implications of PPARs in cancer. The major clinical trials database (https://clinicaltrials.gov) lists one clinical trial for a PPARα antagonist for treatment of multiple kinds of cancer, 24 trials for modulators of PPARγ for cancer treatment, but none for PPARβ/δ. The human protein atlas (https://www.proteinatlas.org/ENSG00000112033-PPARD/pathology) lists low cancer type specificity, but detection of PPARβ/δ in all cancer types. A current major limitation for the investigation of PPARβ/δ expression in human cancer samples compared to healthy tissues is the quality of commercially available antibodies. In agreement with this, large differences for PPARβ/δ RNA and protein levels in tumors are noted in the human protein atlas. The protein expression is globally described, but not annotated to certain cell types in the different tumors. Correlations of tumor PPARβ/δ expression with patients outcome have been reviewed recently [6].

Earlier experimental results concerning the role of PPARβ/δ activation for cancer growth were completely controversial with one study showing that pharmacological activation with GW501516 enhanced tumor growth in Apc(min) mice [7], while another study in the same year in the same journal showed enhanced tumor growth in Apc(min) mice crossed with PPARβ/δ knockout mice [8]. Many studies using different cell models have been published afterwards. Several aspects of PPARβ/δ function with relevance for cancer growth have been reviewed recently [1,5,6,9–11].

On a global view, tumor progression is determined by the interplay of cancer cell proliferation, angiogenesis, resisting cell death, evading growth suppressors, activating invasion and metastasis, enabling replicative immortality, deregulating cellular metabolism and avoiding immune destruction, which was defined by Hanahan and Weinberg as the didactic concept of the "hallmarks of cancer" [12,13]. We will follow here this concept and review the knowledge of PPARβ/δ function for the different hallmarks of cancer capabilities.

#### **2. PPAR**β/δ **and Cell Proliferation**

Most published papers focused on tumor growth-promoting or tumor-inhibiting actions of PPARβ/δ. Unfortunately, only few manuscripts distinguished between direct effects on cell proliferation and secondary effects, which might affect tumor growth. Thus, for simplification, we will summarize in this chapter the published results on cell proliferation as well as on general tumor growth. Table 1 summarizes published effects of PPARβ/δ on cell proliferation and tumor growth.


**Table 1.** Effects of PPARβ/δ on cell proliferation and tumor growth.


**Table 1.** *Cont.*

↑ Increase; ↓ decrease; ≈ not significantly different; CLL: chronic lymphocytic leukemia; Apc: Familial Adenomatous Polyposis gene mutated; GW501516, GW0742—specific PPARβ/δ agonists; DG172, NXT1511, GSK3787— PPARβ/δ antagonists.

It has been shown that the PPARβ/δ agonist GW501560 increased VEGF expression in tumor cell lines [16] and promoted tumorigenesis in Apc(Min/+) mice [7,16] linking tumor cell growth and angiogenesis in the gut. Genetical disruption of PPARβ/δ in colon epithelial cells resulted in a lower incidence of azoxymethane-induced colon tumors and reduced VEGF expression [22]. Unfortunately, vessel formation has not been analyzed in detail in these models [16]. In the same azoxymethane-induced colon tumor model, also reduced tumor growth in response to GW0742, which was abolished by PPARβ/δ knockout [23] and a general reduced colon tumor growth in PPARβ/δ knockout mice [8] have been reported. This discrepancy remains unexplained. PPARγ and PPARβ/δ activation of VEGF and cyclooxygenase-2 (COX2) was confirmed in in the colorectal tumor cell lines SW480 and HT29 [49]. Curiously, that opposite regarding cancer growth and VEGF regulation was described using the KM12C colon cancer cell line with silencing of PPARβ/δ in mouse xenograft models [24]. Whether this reflects an unusual behavior of this specialized cell line remains to be determined. The tumor promoting action of GW501516 in Apc mice was confirmed recently and extended by the findings that the PPARβ/δ antagonist GSK3787 suppressed tumorigenesis. PPARβ/δ expression was significantly higher in human colorectal cancers compared to adenomatous polyps and normal mucosa [50] and also in the malignant cells—invasive front versus their paired tumor centers and adenomas—and proinvasive pathways (connexin 43, PDGFRb, AKT1, EIF4G1 and CDK1) were upregulated in response to PPARβ/δ stimulation [17]. Again, the opposite result has also been published for human and mouse tumor samples [51], while another important report using human colorectal cancer samples confirmed high expression of PPARβ/δ and COX2, which was correlated with the incidence of liver metastasis and identified as significant independent prognostic factor [52]. In colitis-associated colon cancer mouse models, PPARβ/δ overexpression promoted tumorigenesis in mice [20] and increased IL-6 expression and STAT3 phosphorylation, whereas concomitant 15-Lipoxygenase-1 expression in colonic epithelial cells suppressed these effects [19]. In an elegant study using different mouse lines, Beyaz et al. showed that high fat diet (HFD) via activation of PPARβ/δ augments the numbers and function of intestinal stem and progenitor cells. Pharmacological activation of PPARβ/δ using GW501516 recapitulated the effects of HFD on these cells. PPARβ/δ activation in the setting of a loss of the APC tumor suppressor gene allowed stem and progenitor cells to initiate tumorigenesis [21], which is in agreement with the studies mentioned above.

Regarding mammary neoplasia, Yuan et al. [36] showed in transgenic animals that activation of PPARβ/δ in the mammary epithelium resulted in progressive histopathologic changes that culminated in the appearance of estrogen receptor- and progesterone receptor-positive and ErbB2-negative infiltrating ductal carcinomas after 12 months in transgenic animals, while treatment with GW501516 shortened the interval until tumor appearance to 5 months. Histologically, Ki-67 expression was increased demonstrating enhanced proliferation of the epithelial cells, and several metabolic changes were observed (see below). Additionally, in animals with 3-phosphoinositide-dependent kinase-1 (PDK1) overexpression in mammary epithelium, GW501516 accelerated tumorigenesis, which was more pronounced in mice with PDK1 overexpression [38]. This is in agreement with many other reports as PDK1 overexpression resulted in an increase in PPARβ/δ expression and profound metabolic changes. Furthermore, GW501516 increased PPARβ/δ and PDK1 expression in mammary tumors [40]. In MMTV-ErbB2/HER2 onco-mice, knockout of FABP5, which shuttles ligands from the cytosol to nuclear PPARβ/δ was sufficient to reduce mammary tumorigenesis highlighting the importance of this molecule and endogenous PPARβ/δ ligands for cancer growth [39]. On the molecular level, epidermal growth factor receptor ligands signal through the ERK and the phophatidylinositol-3-kinase cascades to activate the transcription factor NF-kappaB. NF-kappaB increases via direct transcriptional activation the expression of FABP5 in MCF-7 breast cancer cells, which stimulates proliferation [53]. In Cox-2 overexpressing mice, mammary tumorigenesis was increased, which could be reverted by crossing them with PPARβ/δ knockout mice [37]. In severely immunocompromised mice, MCF-7 breast cancer cells with overexpression of PPARβ/δ produced bigger tumors and more metastasis compared to wild-type cells. Treatment of MCF-7 cells with PPARβ/δ antagonists in culture reduced significantly the number of these cells [35,54].

Martín-Martín et al. showed an opposite result for prostate carcinoma. PPARβ/δ mRNA was downregulated in prostate cancer specimens compared to benign prostate hyperplasia samples; and prostate epithelium-specific knockout of PPARβ/δ increased cellularity. Additional supporting evidence was obtained by the generation of different overexpression or silencing clones from different human prostate cancer cell lines. Mechanistically, PPARβ/δ exerted its activity in a DNA binding-dependent and ligand-independent manner, which involved regulation of the secretory trefoil factor family member 1 [25]. To which extend the stable cell clones and mice exposed during the entire lifespan to the Cre corresponding to tumor development in humans in vivo remains to be determined. In contrast, silencing of PPARβ/δ in prostate cancer cell lines inhibited tumor cell proliferation and tumor growth, which was attributed to activation of the ABCA1 cholesterol transporter-Caveolin1-TGFβ receptor signaling axis [26]. A similar observation in prostate cancer cells was published by Morgan et al., identifying fatty acid binding protein 5 (FABP5) as a direct target gene of PPARβ/δ [27].

Overexpression of PPARβ/δ decreased the cell number in neuroblastoma NGP, but not in SK-N-BE(2) and IMR-32 cell clones. In xenograft models, PPARβ/δ overexpression reduced tumor growth in NGP cell clones, but to a lesser extent in SK-N-BE(2) and IMR-32 cell clones [29]. As the level of overexpression of PPARβ/δ was highest in NGP cells, it is difficult to judge whether the different outcome is due to the different cell lines used or a response to the different levels of PPARβ/δ overexpression. Whether the results correspond to neuroblastoma pathogenesis in vivo remained an open question. A comparable observation was published by the same group when using testicular embryonal carcinoma cell clones with PPARβ/δ overexpression and the agonist GW0742 [41].

In transgenic hepatitis B virus (HBV) mice, long term treatment with the PPARβ/δ agonist GW0742 reduced the number of hepatic tumor foci. Based on reduced expression of cyclin D1 and c-Myc, a reduction in tumor cell proliferation has been proposed [30]. In human hepatocellular carcinoma cell lines, GW501516 increased proliferation, while RNAi against PPARβ/δ inhibited cell growth. PPARβ/δ activation up-regulates the expression of cyclooxygenase (COX)-2, a rate-limiting enzyme for prostaglandin synthesis and tumor growth in hepatocellular cancer lines [31].

Chronic exposure to ultraviolet light (UV) induced PPARβ/δ activity in the skin of mice. Increased PPARβ/δ activity directly stimulated Src expression, increased Src kinase activity and enhanced the EGFR/Erk1/2 signaling pathway, resulting in increased epithelial-to-mesenchymal transition (EMT) marker expression. PPARβ/δ-null mice developed fewer and smaller skin tumors. Furthermore, topical application of the PPARβ/δ antagonist GSK0660 prevented UV-dependent Src stimulation; and the

expression of PPARβ/δ positively correlated with the expression of SRC and EMT markers in human skin squamous cell carcinoma (SCC) highlighting the clinical relevance of these findings [42]. Another report claimed that the agonist GW0742 delayed chemical induced skin carcinogenesis; combination of GW0742 and the COX2 inhibitor nimesulide resulted in a further decrease of tumor multiplicity in wild-type mice, but not in PPARβ/δ-null mice [55]. Given that the graphs in the different groups for tumor incidence, multiplicity and size look comparable and no statistical information is provided, it is difficult to follow this line of evidence, which is in sharp contrast to many other published papers. Even more surprising, the same authors reported earlier for a comparable model no effect of GW0742 on chemical induced skin carcinogenesis [56], or no combined effects for GW0742 and the COX2 inhibitor nimesulide in induced colon cancers [57] or independence of the COX2 inhibitor effects on PPARβ/δ [58].

High PPARβ/δ expression was detected in human melanoma compared to normal skin [32]. PPARβ/δ activation using GW0742 or GW501516 inhibited proliferation of different melanoma cell lines [32,33], which was due to direct transcriptional repression of the Wilms' tumor suppressor WT1 and its downstream target genes zyxin [59] and nestin [59–61].

In non-small cell lung cancer (NSCLC) cell lines, PPARβ/δ activation increased proliferation and survival, while PPARβ/δ knock-down reduced viability and increased apoptosis. As reported for colon cancer, PPARβ/δ agonists induced VEGF transcription in NSCLC cell lines. Furthermore, increased expression of PPARβ/δ and VEGF in human non-small cell lung cancer samples compared to normal lung tissues has been detected [43,62]. In contrast, a study using only two lung cancer cell lines in vitro, did not find any effects on cell proliferation in response to PPARβ/δ activation [44]. In a transgenic mouse model of RAF-induced lung adenoma, tumor growth in mice lacking one or both alleles of PPARβ/δ was reported to be increased [45]. However, the histological analysis performed in this model was superficial and statistical information lacking.

We showed in liposarcoma cell lines that PPARβ/δ activation increases proliferation, which is abolished by a PPARβ/δ–siRNA or a specific PPARβ/δ antagonist. These effects were mediated via direct transcriptional repression of leptin by PPARβ/δ. PPARβ/δ was highly expressed in liposarcoma compared to lipoma and correlated with increased proliferation in human tumor samples [46].

PPARβ/δ was increased in benign and malignant human thyroid tumors and correlated with the proliferation marker Ki67. Overexpression of PPARβ/δ in thyroid cells and treatment with GW501516 increased cell proliferation in a cyclin E1-dependent manner. Specificity of the findings was proven by reduction of cyclin E1 expression and cell proliferation in response to RNAi against PPARβ/δ [47].

Epithelial ovarian cancer cell lines expressed high levels of PPARβ/δ. Inhibition of PPARβ/δ reduced epithelial ovarian cancer cell proliferation and reduced tumor growth in vivo. Mechanistically, aspirin, a nonsteroidal anti-inflammatory drug that preferentially inhibits COX-1, compromised PPARβ/δ function and cell growth by inhibiting extracellular signal-regulated kinases 1/2 [48].

Although still some controversies exist, PPARβ/δ expression has been documented in a broad variety of different tumor samples and cancer cell lines. In the majority of published reports, PPARβ/δ activation or overexpression was associated with increased cancer cell and tumor growth, some opposite results may be explained by use of different clonal cell lines or different genetic backgrounds and models in mice.

#### **3. PPAR**β/δ **and Angiogenesis**

In contrast to PPARα and PPARγ, PPARβ/δ is a proangiogenic member of the PPAR family [63]. Vascular cell expression of PPARβ/δ has first been reported in the late 90s by Xin et al., 1999, using mRNA analysis [64] and Bishop-Bailey and Hla, 1999, employing Northern blot techniques [65]. In addition, PPARβ/δ expression in vascular smooth muscle cells had been observed by Bishop-Bailey in 2000 [66].

However, no specific functions of PPARβ/δ in the vasculature were discovered at that time, due to the lack of specific ligands. The first synthetic PPARβ/δ and PPARγ non-thiazolidinedione agonist L-165041 was established in 1999 [67], followed later by the highly selective PPARβ/δ agonists GW0742 and GW501516 [68].

A first report shading light on the function of PPARβ/δ in vascular cells appeared in 2001. Hatae and colleagues observed that prostacyclins induce apoptosis via PPARβ/δ activation in HEK293 cells whereas endothelial cells, which express cytoplasmic prostacyclin receptors are protected from apoptosis. They concluded that prostacyclin-dependent receptor activation results in increased cAMP levels in endothelial cells, which protects from apoptosis while direct prostacyclin activation of PPARβ/δ in cells lacking cytoplasmic prostacyclin receptors is proapoptotic [69]. A second investigation focusing on endothelial cell apoptosis demonstrated a protective action of L-165041 as well as of carbaprostacyclin (cPGL2) upon H2O<sup>2</sup> induced apoptosis. Both substances increased expression of PPARβ/δ; knockdown of PPARβ/δ abrogated the apoptosis diminishing effects of both agents. As the molecular mechanism of this apoptosis protective function of PPARβ/δ in endothelial cells, the authors proposed the direct transcriptional activation of 14-3-3alpha protein, a cytosolic protein involved in apoptosis protection, by PPARβ/δ [70]. A later study further added activation of endothelial 14-3-3epsilon protein by PPARβ/δ agonists to the antiapoptotic role [71]. Non-steroidal anti-inflammatory drugs (NSAIDs) can induce endothelial cell apoptosis by disconcerting these transcriptional pathways [72].

PPARβ/δ agonists became of particular interest in vascular biology as they were shown to potently inhibit vascular inflammation and reduce atherosclerosis [73]. They inhibit tumor necrosis factor alpha (TNFα) mediated endothelial inflammation, evidenced by decreased expression of vascular cell adhesion molecule-1 (VCAM-1), monocyte chemotactic protein-1 (MCP-1) expression and inhibition of monocyte binding of TNFα stimulated endothelial cells treated with the PPARβ/δ agonist L-165041 [74]. It has been proposed that PPARβ/δ further controls inflammation via a ligand-dependent interaction with the transcriptional repressor BCL-6. In the absence of other ligands, PPARβ/δ binds BCL-6. When activated with a PPARβ/δ ligand, BCL-6 is released and can suppress proinflammatory pathways [65,75]. Later reports confirmed the anti-inflammatory effect of PPARβ/δ in endothelium [76,77]. PPARβ/δ also inhibits vascular smooth muscle inflammation by transcriptional activation of transforming growth factor (TGF)β1. The decreased MCP-1 expression induced by PPARβ/δ was shown to be mediated by the effector of TGF-β1, Smad3 [78].

Activation of PPARβ/δ has also been reported to prevent endothelial dysfunction by reducing oxidative stress [79]. In diabetic mice, PPARβ/δ activation mediated through phosphatidylinositol 3-kinase (PI3K) and Akt an increase of endothelial nitric oxide synthase (eNOS) activity and nitric oxide (NO) production and improved endothelium-dependent relaxation parameters [80]. In high glucose induced impairment of insulin signaling, PPARβ/δ activation restores endothelial function in part through pyruvate dehydrogenase kinase (PDK) 4 activation, thus preserving the insulin-Akt-eNOS pathway impaired by high glucose [81].

Despite its anti-inflammatory and anti-atherosclerotic functions in the vasculature, PPARβ/δ is a major factor for acute vascular hyperpermeability and vasodilatation, key features of allergic reactions, which can lead to lethal systemic anaphylaxis. The group of Michalik recently demonstrated that selective vessel-specific deletion of PPARβ/δ is sufficient to inhibit VEGF or IgE- induced acute vascular hyperpermeability and vasodilatation, most likely due to activity modulation of kinase pathways and destabilization of cell-to-cell adherens junctions. Inhibition of PPARβ/δ should be considered as a therapeutic approach in acute allergic and inflammatory diseases with disturbed endothelial integrity [82].

The first detailed report about the proangiogenic function of PPARβ/δ appeared in 2007. The selective PPARβ/δ ligand GW501516 was tested at this time in phase II clinical trials for the treatment of dyslipidemia. Using a variety of in vitro and ex vivo approaches, the authors clearly demonstrated that PPARβ/δ induces endothelial cell migration, proliferation and tube formation. They further described an increase of vascular endothelial growth factor (VEGF) expression upon activation of PPARβ/δ and already cautioned against possible negative side effects of agonist treatment in patients

susceptible for "angiogenic diseases", such as elderly persons prone to cancer incidence or diabetic individuals with retinopathies [83].

In vivo studies further showed that pharmacological activation with GW0742 as well as muscle specific transgenic overexpression of PPARβ/δ resulted in a rapid increase of capillary density and oxidative fiber numbers in skeletal muscle, resembling the muscular phenotype induced by regular physical training. It had been proposed that the observed effects were the calcineurin-nuclear factor of activated T cells (NFAT) pathway dependent, as inhibition of calcineurin by cyclosporine A (CsA) totally abolished the observed effects of pharmacological activation of PPARβ/δ [84]. Our group further demonstrated the function of PPARβ/δ in physiological vascularization. Treatment of mice with the agonist GW0742 resulted in rapid cardiac growth and vascularization without functional impairment as reflected by normal echocardiographic parameters. The cardiac hypertrophy accompanied by intensive vascularization resembled the cardiac phenotype obtained by long-term voluntary exercise. As the underlying molecular mechanism of this PPARβ/δ action, we identified the calcineurin-nuclear factor of activated T cells (NFAT) pathway [85]. However, it was unclear if the observed increased vascularization was a secondary effect of the myocardial hypertrophy or if the induction of cardiac growth was due to the increased angiogenesis. We therefore generated conditional mice with inducible vessel specific overexpression of PPARβ/δ and observed that vascular overexpression of PPARβ/δ was sufficient to induce a rapid cardiac hypertrophy. Nevertheless, the increased angiogenesis did not ameliorate cardiac function after myocardial infarction [86]. Similar observations were made using pharmacological activation of PPARβ/δ after myocardial infarction; also in this setting the increase in angiogenesis did not ameliorate the clinical outcome [87]. The proangiogenic function of PPARβ/δ was also exploited in other therapeutic approaches in ischemic cardiovascular diseases. Bone marrow derived endothelial progenitor cells (EPCs) represent an interesting path in the therapy of ischemic diseases, but due to their low number their clinical use is limited. Han and colleagues investigated the effects of PPARβ/δ agonists GW501516 or L-165041 on EPCs and found an increase of angiogenic EPC properties including increased migration, proliferation and tube formation in response to activation of PPARβ/δ. These effects were phosphatidylinositol 3-kinase/Akt pathway dependent. Systemic administration of PPARβ/δ agonists led to an increase of hematopoietic stem cells in bone marrow and blood as well as to an enhanced vascularization in ischemic hindlimb models and corneal neovascularization in vivo [88].

The therapeutic potential of PPARβ/δ modulation on aspects of ocular neovascularization, a common feature of premature or diabetic retinopathy, as well as age-related macular degeneration, the leading causes of irreversible blindness, was studied using human retinal microvascular endothelial cells (HRMEC) and in vivo models of oxygen-induced retinopathy (OIR). The authors demonstrated a stimulation of ocular vascularization with PPARβ/δ activation. Furthermore, using the selective PPARβ/δ antagonist GSK0660 [89], the potential therapeutic utility of PPARβ/δ inhibition was proven. GSK0660 decreased HRMEC migration, proliferation, and tube formation and neovascularization in OIR [90].

The effects of PPARβ/δ on tumor angiogenesis were first investigated in 2007. Employing B16 melanoma and LLC1 (Lewis lung carcinoma) tumor cell inoculated in PPARβ/δ <sup>−</sup>/<sup>−</sup> mice, Müller-Brüsselbach and colleagues demonstrated cancer vascularization defects and diminished tumor blood flow, resulting in reduced tumor growth in animals lacking PPARβ/δ. In contrast to the report from Piqueras and colleagues [83], the authors observed a hyperproliferative state of endothelial cells, leading to the formation of immature and dysfunctional microvessels upon deletion of PPARβ/δ. On a molecular level, decreased expression of the antiproliferative cyclin-dependent kinase inhibitor 1C (Cdkn1c) was observed in PPARβ/δ <sup>−</sup>/<sup>−</sup> cells isolated from Matrigel plugs, which might explain the proliferative immature state of PPARβ/δ <sup>−</sup>/<sup>−</sup> endothelium [91].

However, in a second study from this group, diminished expression of chloride intracellular channel protein 4 (Clic4) and increased expression of cellular retinol binding protein 1 (Crbp1) were observed in PPARβ/δ <sup>−</sup>/<sup>−</sup> fibroblasts and endothelial cells as compared to wildtype cells [92]. Clic4 promotes endothelial cell proliferation, capillary network and lumen formation [93], whereas Crbp1 binding retinoids in contrast favors growth arrest and differentiation [94]. This is in discrepancy to the observed hyperproliferative state of PPARβ/δ <sup>−</sup>/<sup>−</sup> endothelial cells observed by the group of Müller-Brüsselbach [91] and fits to the conclusions made by Piqueras and colleagues that PPARβ/δ stimulates endothelial cell proliferation [83].

An important study further confirmed the strong implication of PPARβ/δ in proangiogenic stimulation favoring tumor progression. Abdollahi and coworkers aimed to identify genes involved in the "angiogenic switch", the shift of an angiogenic balance to a proangiogenic state, one hallmark of cancer progression. Human microvascular cells were submitted to proangiogenic stimuli and subsequent cDNA arrays performed to identify differentially expressed genes upon proangiogenic stimulation. Further selection of genes based on their involvement in the angiogenic network identified PPARβ/δ as a "hubnode" in the "angiogenic switch". The authors confirmed their findings in vivo using B16 melanoma and LLC1 Lewis lung carcinoma inoculated in PPARβ/δ <sup>−</sup>/<sup>−</sup> mice, in which they observed dramatically reduced tumor angiogenesis and growth. PPARβ/δ expression levels in human cancer samples further correlated with advanced stages of tumor progression and metastasis [95,96].

Recently, PPARβ/δ activators (L165041 and GW501516) were shown to induce interleukin 8 (Il-8) expression in endothelial cells by transcriptional and posttranscriptional mechanisms [97], and enhanced production of IL-8 due to PPARβ/δ activation caused not only elevated tumor angiogenesis, but also metastasis formation in vivo [98].

Our group further confirmed the general tumor-angiogenesis and cancer growth promoting effect of PPARβ/δ [14]. Although we observed a decrease of LLC1 cancer cell proliferation in vitro upon treatment with GW0742, tumor growth and metastases formation in LLC1 cancer bearing animals was enhanced upon administration of the PPARβ/δ agonist. Tumor vascularization was strongly increased, which supports the hypothesis that enhancement of angiogenesis by PPARβ/δ dominates the eventually growth-inhibiting function on cancer cells. To further determine the functional relevance of PPARβ/δ for tumor vascularization and identify angiogenic signaling pathways, we made use of mice with conditional inducible vascular overexpression of PPARβ/δ subcutaneously implanted with LLC1 cells. Vessel-specific overexpression of PPARβ/δ was sufficient to increase cancer growth, progression and metastases formation. Tumor-sorted endothelial cells were submitted to RNA-sequencing; 283 genes were found to be differentially expressed and cluster analysis revealed mostly up-regulation of genes upon overexpression of PPARβ/δ in endothelial cells. This argues for an angiogenesis boosting effect of PPARβ/δ rather than a repression of antiangiogenic molecules to enhance angiogenesis. We identified six potential target genes of PPARβ/δ, all of them known to be involved in tumor angiogenesis, by combining the top ten network analysis with a search for PPAR responsive elements: Vegf receptors 1 (Flt1), 2 (Kdr) and 3 (Flt4), [99,100], and platelet-derived growth factor receptor beta (Pdgfrβ) [101], platelet-derived growth factor subunit B (Pdgfb) [102] and the tyrosinkinase KIT c-kit [103,104]. Finally, we confirmed that PPARβ/δ directly transcriptional activates Pdgfrβ, Pdgfb, and c-Kit. PPARβ/δ tumor-angiogenesis promoting effects are mediated via activation of the PDGF/PDGFR pathway, c-Kit and probably the VEGF/VEGFR pathway [14].

Despite their beneficial effects on vascular inflammation and atherosclerosis, the therapeutic use of PPARβ/δ agonists could be critical in cancer patients and should therefore in general not be considered as a therapeutic option.

#### **4. PPAR**β/δ **and Cell Death**

The first study demonstrating an inhibitory function of PPARβ/δ in cancer cell death appeared in 1999. He and colleagues revealed that the adenomatous polyposis coli (APC) tumor suppressor represses PPARβ/δ expression through inhibition of β-catenin/Tcf-4 regulated transcription (CRT). APC/β-catenin mutations can therefore lead to increased PPARβ/δ activity. Nonsteroidal anti-inflammatory drugs (NSAIDs) Sulindac and Indomethacin promoted apoptosis of colorectal cancer cells, which could be inhibited by overexpression of PPARβ/δ. The authors demonstrated that NSAIDs suppressed activity of PPARβ/δ through the direct inhibition of DNA binding activity. As fatty acids and eicosanoids are ligands and modifiers of PPAR activity, NSAID-dependent changes in eicosanoid metabolism could also contribute to inhibition of PPARβ/δ activity. NSAIDs were therefore considered as an important therapeutic approach in colorectal carcinoma as they inhibited apoptosis-preventing PPARβ/δ activity also in the context of frequently occurring APC/β-catenin mutations [105]. Other groups demonstrated that cyclooxygenase-derived prostaglandin E2 (PGE2) inhibits colon cancer cell apoptosis through the indirect transactivation of PPARβ/δ. Of note, the authors showed that PGE2 specifically regulates PPARβ/δ, not the other PPARs. The apoptosis inhibiting effects of PGE2 are mediated through indirect mediation of PPARβ/δ by activation of the PI3K/Akt signaling pathway [106]. Gupta et al. confirmed the antiapoptotic effect of PPARβ/δ activation in wildtype and PPARβ/δ-deficient HCT116 colon carcinoma cells. Pretreatment of wildtype HCT116 cells with GW501516 reduced serum withdrawal induced apoptosis, which was not the case in PPARβ/δ-deficient HCT116 cells, suggesting a specific effect of PPARβ/δ activation [7]. Nonsteroidal anti-inflammatory drugs (NSAIDs) were also shown to induce colorectal cancer cell apoptosis through other PPARβ/δ mediated mechanisms. NSAIDs inhibited 14-3-3ε protein expression, leading to apoptosis, accompanied by a decrease of cytosolic and an increase of mitochondrial Bad [107]. The authors had already shown that PPARβ/δ transcriptionally activates 14-3-3ε [70], and further confirmed their hypothesis in this study by overexpression of PPARβ/δ, which rescued colorectal cancer cells from NSAID induced apoptosis and upregulated 14-3-3ε protein levels. This additionally implicates the PPARβ/δ 14-3-3ε pathway in colon cancer cell survival [107]. Again, in the setting of colorectal cancer, it has been shown that PPARβ/δ overexpression or activation antagonizes PPARγ-induced apoptosis of cancer cells. PPARγ agonists induce apoptosis in these cancer cells through reduction of survivin, which in turn leads to apoptosis through increased caspase-3 activity. PPARβ/δ agonists inhibit induction of this apoptotic pathway by increasing survivin expression levels [108]. The apoptosis inducing effects of NSAIDs in colon cancer were also linked to 15-lipoxygenase-1 (15-LOX-1) upregulation. 13-S-hydroxyoctadecadienoic acid (13-S-HODE), the primary product of 15-LOX-1 metabolism of linoleic acid, was found to decrease activity and downregulate expression of PPARβ/δ in colon cancer cells, thereby inducing apoptosis [109]. An interesting study of Cutler and colleagues showed that fibroblasts isolated from the mucosa of hereditary non polyposis colorectal cancer (HNPCC) patients produced 50-fold more PGE2 than normal fibroblasts [110]. PGE2 inhibits apoptosis of colonic carcinoma cells through the activation of PPARβ/δ [106]. As HNPCC patients are more susceptible to develop colorectal cancer (CRC), the authors hypothesized that the overproduction of PGI2 from the stroma of HNPCC patients prevents apoptosis of neoplastic lesions through activation of PPARβ/δ and therefore facilitates progression into a malignant state of CRC [110]. In contrast to all these studies, indicating an antiapoptotic function of PPARβ/δ in colon cancer cells, one report suggested a proapoptotic function of PPARβ/δ in the setting of colon carcinoma. In a model of chemically induced colon carcinogenesis using wildtype and PPARβ/δ knockout mice, treatment of mice with the agonist GW0742 resulted in higher colonic cell apoptosis in wildtype animals as assessed by TUNEL staining and subsequent quantification of cell counts from colon sections, which does not really assure cancer cell specificity. No changes in apoptotic cell counts were observed in colons from PPARβ/δ knockout mice upon agonistic activation of PPARβ/δ [23].

Maggiora et al. investigated the effects of linoleic (LA) and conjugated-linoleic acids (CLA) on the growth of several human tumor cell lines, comprising prostate, bladder, liver, glioblastoma and breast cancer cells. In contrast to Las, CLAs had a strong growth inhibitory effect in the cancer cell lines tested and were able to induce apoptosis in the more deviated cells. PPARβ/δ levels decreased strongly in apoptotic cancer cells upon CLA treatment, but not in cell lines where only an inhibition of cell proliferation without subsequent cell death could be observed [111].

Our group investigated the effects of PPARβ/δ activation on human and mouse melanoma cells. Although we could observe a reduction of melanoma cell proliferation upon PPARβ/δ activation

either with GW0742 or with GW501516 at nanomolar concentrations, we did not observe changes in melanoma cell apoptosis [32].

In one lung cancer cell line, PPARβ/δ activation with the agonist L165041 or treatment with the NSAID Indomethacin alone had no effect on apoptosis, however, a combination of these molecules induced apoptosis in this cancer cell line [112]. In contrast, activation of PPARβ/δ with the more specific agonist GW501516 has been demonstrated to inhibit cisplatin-induced apoptosis in different lung cancer cell lines [62]. In line with this latter finding, Genini et al. reported enhanced apoptosis in different human non-small cell lung cancer (NSCLC) lines upon knockdown of PPARβ/δ [43]. We investigated the effects of PPARβ/δ activation or antagonism on mouse Lewis lung carcinoma cells and observed no differences in apoptosis for neither modulation of PPARβ/δ activity [14].

In contrast to the studies mentioned above, which mostly describe an antiapoptotic function of PPARβ/δ in cancer cells, Foreman and colleagues postulated a proapoptotic action of PPARβ/δ in a mouse mammary gland cell line. Treatment with very high concentrations of the PPARβ/δ agonist GW501516 (10 µmolar) for 24 h increased early apoptosis in this cell line, as analyzed by annexin V staining. However, prolonged treatment for 48 h at the same concentrations had no effect on apoptosis, which could raise some doubts concerning the conclusions given in this study [113]. A study from the same group could neither confirm the observation that NSAIDs decrease PPAR activation and expression in colon cancer cells, nor that PPARβ/δ exerts an antiapoptotic function in the setting of colon cancer. Using different human colon cancer cell lines treated with hydrogen peroxide to induce apoptosis and NSAIDs and different concentrations of the PPARβ/δ agonist GW0742, the authors did not observe a decrease of early (evidenced by annexin V labeling) or late (analyzed by PARP cleavage) apoptosis upon PPARβ/δ activation [51]. Bell and colleagues demonstrated that inhibition of PPARβ/δ using siRNA mediated knockdown or the antagonist GSK0660 sensitized neuroblastoma cells to all-trans retinoic acid induced cell death [114]. In line with this proapoptotic function of PPARβ/δ, Péchery and colleagues reported enhanced apoptosis in tumor cells derived from high-grade bladder tumor upon activation with the PPARβ/δ agonist GW501516 [115]. Using only one prostate cancer cell line it has been postulated that the inhibition of PPARβ/δ with the antagonist GSK0660 partially inhibited ginsenoside Rh2 induced apoptosis [116]. In line with this study, another group recently reported a proapoptotic role of PPARβ/δ in prostate cancer cells. Treatment of one prostate cancer cell line with Telmisartan, an angiotensin receptor blocker, induced apoptosis, which could be partially inhibited by pharmacological or genetic down-regulation of PPARβ/δ activity or expression [117]. Additionally, in a nasopharyngeal carcinoma cell line, proapoptotic functions of PPARβ/δ could be demonstrated. Using in vitro and in vivo xenograft assays, high concentrations of GW501516 (10 or 30 µmolar) induced apoptosis of the nasopharyngeal cancer cells. The authors proposed as underlying mechanisms the activation of adenosine monophosphate-activated protein kinase (AMPKα) and downregulation of integrin-linked kinase (ILK), as the AMPK inhibitor compound C was able to inhibit the reduction of ILK expression induced by GW501516 [118]. Employing the same cell line, the authors further implicated the microRNA miR-206 in the apoptosis promoting effects of PPARβ/δ activation, as they observed an induction of miR-206 upon GW501516 mediated PPARβ/δ activation, which could be antagonized by the PPARβ/δ antagonist GSK3787 or the AMPK antagonist dorsomorphin [119].

In conclusion, it is not perfectly clear if PPARβ/δ prevents or stimulates cancer cell death. Although the majority of studies suggest that PPARβ/δ has an antiapoptotic function in cancer cells, some reports evoke the contrary and others do not observe implication of PPARβ/δ in apoptotic cancer cell death at all. This might be due to cancer cell type specific differences, but also to discrepancies in experimental set ups.

#### **5. PPAR**β/δ **and Tumor Suppressors**

In addition to positive regulation of growth-promoting signals, cancer development also requires inhibition of negative growth regulators, i.e., escaping the action of tumor suppressor genes [12]. Although a large number of publications described the overall effects of PPARβ/δ modulation on tumor

growth, knowledge on PPARβ/δ and tumor suppressor genes is relatively limited. Mice with mutations in the adenomatous polyposis coli (APC) tumor suppressor are frequently used as a tool for PPAR research in colon cancer, but also a direct function of the APC tumor suppressor on PPARβ/δ expression has been described. APC represses PPARβ/δ expression through inhibition of β-catenin/Tcf-4 regulated transcription in colon cancer cells [105]. Besides colon cancer cells, inactivating mutations in APC or the Axin tumor suppressor proteins or activating mutations in β-catenin resulting in positive effects on T-cell factor (TCF)-regulated transcription have been described in several cancer types. Zhai et al. reported mutations leading to β-catenin deregulation in half of ovarian endometrioid adenocarcinomas. They found elevated expression of the MMP-7, CCND1 (Cyclin D1), CX43 (Connexin 43), ITF2 and also PPARβ/δ genes in ovarian endometrioid adenocarcinomas with deregulated β-catenin [120]. Transformation of intestinal epithelial cells with the K-Ras oncogene led to increased expression and activity of PPARβ/δ. Mechanistically, PPARβ/δ up-regulation was due to increased mitogen-activated protein kinase activity; and PPARβ/δ activation required the endogenous production of prostacyclins via the cyclooxygenase-2 pathway [121]. An initial important report from mice with inactivation of the APC tumor suppressor showed that treatment with the PPARβ/δ agonist GW501516 resulted in a significant increase in the number and size of intestinal polyps [7]. In contrast to the reports mentioned above, another study confirmed APC/beta-catenin-dependent expression of Cyclin D1, while expression of PPARβ/δ was not different in colon or intestinal polyps from wild-type or Apc(min) heterozygous mice or in human colon cancer cell lines with mutations in APC or beta-catenin [122]. This study based exclusively on the use of a polyclonal antibody in Western blots. The quality of the available PPARβ/δ antibodies is still a matter of concern.

Regarding the Wilms' tumor suppressor WT1, we showed that PPARβ/δ activation in melanoma cells inhibits its expression via direct transcriptional repression [32]. WT1 was originally identified as a tumor suppressor based on its mutational inactivation in nephroblastoma [123,124], but later studies provided evidence that WT1 might act as an oncogene [60,101,104,125,126]. Wt1 was up-regulated instead of downregulated in endothelial cells with PPARβ/δ overexpression [14], which suggests cell-type dependent differential regulation of Wt1 by PPARβ/δ. Whether PPARβ/δ is a direct activator of WT1 in endothelial cells and other cell-types remains to be determined.

Epidermal growth factor receptor (EGFR) signaling promotes breast cancer cell proliferation and tumorigenesis. It has been shown that EGFR ligands signal through the ERK and the phophatidylinositol-3-kinase cascades, resulting in activation of the transcription factor NF-kappaB. The NF-kappaB transcription factor directly activates the promoter of fatty-acid binding protein 5 (FABP5) resulting in increased FABP5 protein expression, which in turn shuttles endogenous ligands to PPARβ/δ [53]. In contrast, Krüppel-like factor KLF2 inhibits FABP5 protein expression and subsequent PPARβ/δ activation and thus, might act as a tumor suppressor in breast cancer cells [53].

Transducer of ErbB-2.1 (Tob1) is another tumor suppressor protein, which is inactivated in different cancer types including gastrointestinal cancers. Overexpression of Tob1 in gastric cancer cell lines induced the expression of Smad4 and p15. Tob1 decreased the phosphorylation of Akt and glycogen synthase kinase-3β (GSK3β), resulting in reduced expression and the transcriptional activity of β-catenin, which in turn decreased the expression of PPARβ/δ, cyclin D1, cyclin-dependent kinase-4 (CDK4) and urokinase plasminogen activator receptor (uPAR) in gastric cancer cells [127]. These data are in agreement with the general regulation of PPARβ/δ by β-catenin and provide an additional complex signaling pathway for stimulation of PPARβ/δ activity in cancer progression.

In neuroblastoma cell lines, all-trans-retinoic acid reduced expression of the stem cell factor Sox2 in cell lines with low expression of the tumor suppressor p53, while this was not the case in cells with wild type p53. However, PPARβ/δ activation with GW0742 reduced SOX2 expression independent on the p53 status of the cells. The authors concluded that activating PPARβ/δ induces cell differentiation through p53- and SOX2-dependent signaling pathways in neuroblastoma cells and tumors [29]. However, the exact interaction between retinoic acid and PPARβ/δ signaling on SOX2 expression and the possible role of p53 therein remains to be determined.

In smooth muscle cells, the PPARβ/δ agonist L-165041 inhibited dose-dependently proliferation by blocking G(1) to S phase progression and repressing the phosphorylation of retinoblastoma protein (Rb). In a carotid artery injury model in vivo, L-165041 inhibited neointima formation [128]. To our knowledge, this is the only report linking retinoblastoma protein and PPARβ/δ activation. Whether these findings are relevant for cancer cell proliferation or tumor angiogenesis remains to be determined.

#### **6. PPAR**β/δ **and Invasion and Metastasis**

Abdollahi et al. were the first to correlate PPARβ/δ expression levels with advanced pathological tumor stage and increased risk for distant metastasis. Statistical analyses of PPARβ/δ expression in published large-scale microarray data from cancer patients with prostate, breast, and endometrial adenocarcinoma revealed significantly increased PPARβ/δ expression levels in cases of higher malignant grade and distant metastasis formation [95]. Similar observations were made by Yoshinaga and colleagues who found an increased risk for colorectal cancer patients with high expression of PPARβ/δ and cyclooxygenase (COX)2 in the primary tumor to develop distant liver metastasis, consequently leading to a poor prognostic outcome [96]. In contrast to these studies, one group reported decreased invasion capacity of pancreatic cancer cells in vitro upon PPARβ/δ activation with GW501516 as well as downregulated prometastatic Matrix metalloproteinase-9 (MMP9) expression [129]. A similar study implying in vitro approaches using breast cancer cell lines demonstrated decreased migration and invasion upon PPARβ/δ activation with GW501516. PPARβ/δ mediated inhibition of breast cancer cell migration and invasion was proposed to be regulated via thrombospondin-1 (TSP-1) and its degrading protease, a disintegrin and metalloprotease domains with thrombospondin motifs 1 (ADAMTS1), as knockdown of ADAMTS1 reduced the effects of PPARβ/δ activation; and ADAMTS1 promoter activity was increased by GW501516 [130].

Interestingly, yeast-two hybrid screening identified the metastasis suppressor NDP Kinase alpha (NM23-H2) as a binding protein of PPARβ/δ [131]. NM23 genes have been shown to suppress metastasis development [132]. Overexpression of NM23-H2 in cholangiocarcinoma cells downregulated PPARβ/δ expression, impedes PPARβ/δ promoter activity and diminishes GW501516 induced cholangiocarcinoma cell proliferation. Reactivation of NM23-H2 was suggested as a therapeutic approach in cholangiocarcinoma metastasis [131].

Zuo and collaborators further demonstrated the importance of PPARβ/δ in metastatic cancer. Using an experimental mouse model of metastasis formation by tail vein injection of syngenic tumor cells (B16 melanoma and LLC1 Lewis lung carcinoma cells), the authors showed that PPARβ/δ knockdown in the respective cancer cells inhibited metastasis formation. Additionally, the potential of colon cancer cells (HCT116) to form metastasis in vivo was abolished completely upon genetic deletion of PPARβ/δ. Treatment of mice with the PPARβ/δ agonist GW0742 enhanced metastasis formation. The metastatic potential of PPARβ/δ in cancer cells was confirmed in orthotopic tumor models, confirming that also spontaneous metastasis formation was dramatically reduced upon knockdown of PPARβ/δ. Using heterozygous PPARβ/δ mice for syngenic tumor cell vein injection the authors further demonstrated that high expression of PPARβ/δ in cancer cells is the most important factor for metastasis formation as heterozygous PPARβ/δ mice developed fewer metastasis than their wildtype littermates, but exhibited the most important reduction of metastasis formation when injected with PPARβ/δ knockdown cancer cells. Transcriptome profiling of HCT116 wildtype and PPARβ/δ knockout cells identified gap junction protein alpha 1 (GJA1), vimentin (VIM), secreted protein acidic rich in cysteine (SPARC), neuregulin-1 (NRG1), CXCL8 (IL-8), stanniocalcin-1 (STC1), and synuclein gamma (breast cancer-specific protein 1; SNCG) as pro-metastatic PPARβ/δ targets. Finally, the authors further confirmed the correlation of high PPARβ/δ expression and significantly reduced metastasis-free survival in various cancer patient (colorectal, lung, breast) cohorts, including the largest reported cohort of 1609 breast cancer patients [98].

In profound contrast to the extensive in vivo study of Zuo, Lim and coworkers reported increased melanoma cell migration and invasion upon treatment with the PPARβ/δ antagonist 10 h as well as increased metastasis formation in PPARβ/δ knockout mice [133]. This antagonist had so far not been used in other studies and results were not confirmed employing well established antagonists as GSK0660 or GSK3787. Conversely, Ham and colleagues demonstrated that activation of PPARβ/δ in highly metastatic melanoma cell lines provoked an upregulation of Snail, a decrease of E-cadherin, and a stimulation of migration and invasion, which could be reversed by knockdown of PPARβ/δ. PPARβ/δ therefore seems to promote the high metastatic potential of aggressive melanoma [134].

Our group confirmed pro-metastatic effects of PPARβ/δ activation. Syngenic subcutaneous LLC1 tumor cell implantation resulted in significantly increased lung and liver metastasis when animals received the PPARβ/δ agonist GW0742. Interestingly, we also observed increased spontaneous metastatic spreading in a model with inducible conditional vascular-specific overexpression of PPARβ/δ, indicating that the proangiogenic function of PPARβ/δ importantly contributes to metastatic tumor progression [14].

Recently, an elegant study demonstrated the implication of PPARβ/δ in the pro-metastatic effects of dietary fats in colorectal cancer. Activation of PPARβ/δ with GW501516 induces cancer stem-like cell (CSC) expansion and accelerates liver metastasis in vivo. Analysis of promoters of self-renewal regulatory factors such as Oct4, Nanog, Sox2, and KLF4 identified a PPAR responsive element in the Nanog promoter. Activation of PPARβ/δ with GW501516 increased whereas knockout of PPARβ/δ decreased Nanog expression. Colonic CSC expansion was shown to be induced by PPARβ/δ through direct induction of Nanog expression via binding to its promoter. Furthermore, knockdown of Nanog abolished PPARβ/δ stimulation of hepatic metastasis formation. Similar to the exposure to GW501516, a high fat diet induced expression of Nanog, accelerated tumor growth and liver metastasis formation and knockout of PPARβ/δ completely inhibited these effects. This identifies a novel PPARβ/δ-mediated mechanism responsible for the contribution of dietary fat to colorectal cancer initiation and metastasis [135].

In conclusion, overwhelming evidence suggests that PPARβ/δ promotes metastasis.

#### **7. PPAR**β/δ **and Replicative Immortality**

Activation of PPARβ/δ with GW501516 was shown to inhibit angiotensin (Ang) II induced premature senescence of human vascular smooth muscle cells (hVSMCs). Ang II treatment of hVSMCs provoked an increase of senescence associated beta galactosidase activity (SA β-gal), which was inhibited by GW501515, an effect that could be reversed by hVSMCs knockdown. A significant reduction of SA β-gal activity was also observed upon pretreatment with N-acetyl-l-cysteine (NAC), a thiol antioxidant, suggesting that reactive oxygen species (ROS) mediate Ang II-induced premature senescence of hVSMCs. Activation of hVSMCs significantly reduced ROS accumulation as well as DNA damage in hVSMCs treated with Ang II. PPARβ/δ mediated transcriptional up-regulation of antioxidant genes (glutathione peroxidase (GPx)-1, manganese superoxide dismutase (Mn-SOD), heme oxygenase (HO)-1, and Thioredoxin (Trx)-1) had been identified as the major mechanism in the inhibition of premature senescence of hVSMCs [136]. In a following study, the authors identified upregulation of phosphatase and tensin homolog deleted on chromosome 10 (PTEN), leading to suppression of phosphatidylinositol 3-kinase (PI3K)/Akt pathway, by PPARβ/δ as a second mechanism of senescence inhibition in hVSMCs [137]. Increase of PTEN and suppression of PI3K/Akt by PPARβ/δ activation was also the main pathway identified for senescence inhibition of UV-induced keratinocytes by the agonist GW501516 [138]. In human coronary artery endothelial cells, inhibition of Ang II induced senescence by PPARβ/δ was found to be dependent of transcriptional activation of Sirtuin (SIRT) 1. Downregulation or inhibition of SIRT1 abolished the effects of PPARβ/δ on Ang II induced ROS production and premature senescence, and resveratrol, a SIRT1 activator, mimicked PPARβ/δ agonist effects [139]. PPARβ/δ activation has also been shown to prevent doxorubicin induced cardiomyocyte senescence. The PPARβ/δ agonist L165041 prevented telomeric repeat factor (TRF) 2 downregulation, partially rescued cell proliferation blockage, significantly attenuated cytoskeletal remodeling and the early loss of plasma membrane integrity and significantly reduced SA-β-gal activity. Senescence

inhibition was in this case shown to be dependent of B-cell lymphoma 6 protein (Bcl6) as a potent inhibitor of senescence, rendering cells unresponsive to antiproliferative signals from the p19ARF–p53 pathway. L1650141 increased the expression of Bcl6, which upon ligand binding, was released from PPARβ/δ and repressed its target genes, involved in DNA damage sensing and proliferation of checkpoint control [140]. In this context, it might be interesting to mention that our group observed an increase of TRF2, a protein that has a key role in the protective activity of telomeres [141], in tumor sorted endothelial cells from mice with vascular specific overexpression of PPARβ/δ (Wagner et al., unpublished results).

In contrast to these studies reporting senescence inhibition upon PPARβ/δ activation, Zhu and coworkers observed stimulation of Harvey sarcoma ras virus gene (Hras)-induced senescence by PPARβ/δ. 7,12-dimethylbenz[a]anthracene (DMBA)-initiation led to a higher percentage of malignant squamous cell carcinomas and a lower percentage of benign papillomas in PPARβ/δ knockout compared to wildtype animals. In vitro, Hras expressing PPARβ/δ knockout keratinocytes displayed less senescence as investigated by SA β-gal staining. The authors identified as the molecular mechanisms of this senescence induction by PPARβ/δ a potentiation of the RAF/MEK/ERK pathway and an inhibition of the PI3K/AKT pathway [142]. In a very similar study appearing in the same year, the authors showed that increased endoplasmatic reticulum (ER) stress attenuated senescence in part by up-regulating phosphorylated protein kinase B (p-AKT) and decreasing phosphorylated extracellular signal-regulated kinase (p-ERK), which was repressed by PPARβ/δ [143].

Cellular senescence has been linked to the development of endothelial cell dysfunction in atherosclerosis. Especially oxidative stress induced by ROS from lipid loaded macrophage foam cells has been linked to premature senescence of the vasculature. Riahi and coworkers exposed endothelial cells to the secretome of such foam cells and observed an increase of endothelial SA β-gal activity, p16 and p21 expression as well as a decrease of phosphorylated retinoblastoma protein. They found that senescence was induced by 4-hydroxnonenal (4-HNE) through stimulation of pro-oxidant thioredoxin-interacting protein (TXNIP). The lipid peroxidation product 4-HNE activated PPARβ/δ promoter activity. The PPARβ/δ agonist GW501516 enhanced TXNIP expression, whereas the antagonist GSK0660 reduced TXNIP promoter activity and inhibited 4-HNE induced senescence [144].

In contrast to the prosenescent effects of PPARβ/δ in endothelial cells, Bernal and colleagues reported that PPARβ/δ maintains the proliferative undifferentiated phenotype of adult neuronal precursor cells, probably through activation of SOX2, one self-renewal regulatory factor [145]. This is in line with the findings from Wang and colleagues showing that colonic cancer stem cell expansion was induced by PPARβ/δ through direct transcriptional activation of Nanog [135].

It has been described that PPARβ/δ amplifies Wnt signaling activity through direct interaction with β-catenin and direct transcriptional activation of the Wnt coreceptor low-density lipoprotein receptor-related protein (LRP) 5 [146]. Senescence associated reprogramming has been shown to upregulate an adult tissue stem-cell signature in lymphoma cells, activate Wnt signaling and distinct stem-cell markers. Former senescent lymphoma cells had a higher in vivo tumor initiation potential than their non-senescent counterparts [147]. Given these highly interesting findings, it will be extremely exciting to further clarify the role of PPARβ/δ in cancer related senescence, replicative immortality and cancer stemness.

#### **8. PPAR**β/δ **and Metabolism**

It is well established that the high tumor cell growth rate due to proliferation is connected to profound metabolic changes [12]. As early as 1927, Otto Warburg described an anomaly in cancer cell metabolism compared to normal cells—cancer cells largely depend on aerobic glycolysis for energy production [148–150]. Cancer metabolism is not only linked to proliferation, but also to tumor angiogenesis as rapidly growing tumor cells will turn on the "angiogenic switch" for increased oxygen supply in the tissue. Lack of oxygen results in hypoxia in the tissue, which results in stabilization of hypoxia- induced factors (Hif) [151–153] and subsequent activation/inhibition of downstream target genes, e.g., VEGF [154], erythropoietin [155], WT1 [156], PPARα [157], glucose transporters (Glut-1 and Glut-3) and many other target genes involved in cancer metabolism (for a recent review see [158]). In contrast to PPARα, PPARβ/δ seems not to be directly regulated by Hif-1; but Hif-1 expression is stimulated by calcineurin A [159] and PPARβ/δ activates calcineurin [85]. Consequently, we observed an increase in calcineurin and Hif-1 expression in the hearts of mice treated with the PPARβ/δ agonists GW0742 and GW501516 [85]. Whether this signaling cascade is relevant for PPARβ/δ-dependent cancer progression remains to be established. Hypoxic stress has been shown to induce transcriptional activation of PPARβ/δ in HCT116 colon cancer cells. PPARβ/δ associated with p300 upon hypoxic stress in these cells. The p300 and the PI3K/Akt pathways seem to play a role in the regulation of PPARβ/δ transactivation as PI3K inhibitors or siRNA knockdown of Akt suppressed the PPARβ/δ transactivation in response to hypoxia [160]. Interestingly, hypoxia-induced IL-8 and VEGF expression was significantly attenuated in PPARβ/δ-deficient colon cancer cells linking expression of PPARβ/δ in cancer cells to tumor angiogenesis and immune response [160]. The in vivo relevance of these findings for tumor growth remains to be determined. In addition, prostacyclin synthase, which catalyzes the conversion of prostaglandin H2 (PGH2) to prostaglandin I2 (PGI2) is upregulated in fibroblasts and cancer cells in response to hypoxia. PGI2 in turn stimulates PPARβ/δ and subsequent VEGF expression [161], which provides an additional link between hypoxia, metabolism, PPARβ/δ in the tumor stroma and angiogenesis. PPARβ/δ also protects chronic lymphocytic leukemia and breast cancer cells from harsh environmental conditions, i.e., hypoxia and low glucose concentrations, which was related to increased antioxidant expression, substrate utilization and mitochondrial performance providing additional evidence for PPARβ/δ as a positive regulator of cancer growth [28,35].

Long chain fatty acids (LCFA) represent energy sources, components of cell membranes and are further processed into signaling molecules. Dietary fatty acids are linked to cancer risk especially colon cancer. Saturated fatty acids were positively associated with colon cancer risk, while polyunsaturated fatty acids showed inverse associations [162]. Experimental studies, however suggested that saturated long chain fatty acids (SLCFA) inhibit while unsaturated long chain fatty acids (ULCFA) might increase proliferation of different cancer cell lines [163,164]. A recent report provided novel mechanistic insights into this problem linking long chain fatty acid metabolism and cancer [165]. Saturated fatty acids bind to fatty acid binding protein 5 (FABP5) and displace endogenous ligands and retinoic acid (RA) from this transport protein. Thus, these ligands are not delivered to PPARβ/δ and its transcriptional activity is reduced while RA is diverted to the retinoic acid receptor (RAR), which becomes activated. In contrast, binding of unsaturated long-chain fatty acids to FABP5 has similar consequences for the displacement of RA and its subsequent binding to RARs, but results in nuclear import of the ULCFA/FABP5 complex and subsequent activation of PPARβ/δ, which in turn results in increased cancer cell proliferation [165]. Although these results identify a central role for FABP5 for cancer cell proliferation and might explain the differences observed regarding PPARβ/δ and cancer cell proliferation dependent on the presence/absence of FABP5 and amounts of RA and endogenous PPAR ligands, the situation for in vivo experimental and clinical studies might be even more complex due to the interplay of the different hallmark capabilities.

PPARβ/δ is, however, not only activated by fatty acids presented by FABP5 in tumorigenesis. In mammary epithelium, overexpression of PDK1 resulted in increased phosphorylation of Akt and GSK3β and augmented expression of PPARβ/δ protein. Treatment with GW501516 increased the number of mammary tumors and reduced survival, which was even more pronounced in animals with PDK1 overexpression. This dramatic effect correlated with an increase in a specific metabolic gene signature indicative of glycolysis and greater levels of fatty acid and phospholipid metabolites in PDK1 overexpressing mice treated with GW501516 compared to treated wild-type control mice [38]. As these metabolic changes are common also in human tumors [166] and enable high tumor cell proliferation [167], it is possible that this mechanism plays a common role in tumor types with PPARβ/δ overexpression. In addition, GW501516 increases expression of glucose transporter 1 (Glut-1) and solute carrier family 1 member 5 (SLC1A5), which results in an increased influx of glucose and

glutamine in different cancer cell types and subsequently augments cancer cell proliferation [15]. Furthermore, animals with direct specific overexpression of PPARβ/δ in the mammary epithelium were prone to the development of mammary tumors [36]. Infiltrating mammary ductal carcinomas developed after a latency of 12 months; GW501516 reduced tumor latency to 5 months. Histologically, PPARβ/δ overexpression was confirmed in the mammary epithelium. In agreement with the study by Pollock et al. [38], increased Akt phosphorylation was detected, but also mTOR was activated. Inhibition of mTOR by everolimus reduced cell proliferation and the malignant phenotype indicating the importance of this signaling pathway for PPARβ/δ-dependent mammary tumorigenesis. Microarray and metabolomic analyses revealed a marked increase in the levels of phosphatidylcholine metabolites, lysophosphatidylcholine, lysophosphatidic acid and arachidonic acid metabolites, which correspond to PPARβ/δ-dependent gene regulation involved in prostaglandin biosynthesis. Lysophosphatidic acid stimulated mTOR activation through Akt, and phosphatidic acid directly mediates activation of mTOR [36]. These results provided robust evidence for PPARβ/δ induced metabolic changes resulting in mTOR activation in mammary tumorigenesis. Taken together, several metabolites increase PPARβ/δ activity and PPARβ/δ stimulation induces complex metabolic alterations, which are mostly protumorigenic.

#### **9. PPAR**β/δ **and Immune Function**

PPARβ/δ agonists have been reported to inhibit the tumor necrosis factor (TNF) α induced up-regulation of monocyte chemoattractant protein (MCP)-1 and vascular cell adhesion protein (VCAM)-1 in endothelial cells, to inhibit cytokine induced nuclear translocation of NF-kappaB and to reduce monocyte binding to activated vascular cells [74]. They modulate acute inflammation by targeting the neutrophil-endothelial cell interaction and reducing tumor necrosis factor alpha induced endothelial chemokine ligand (CXCL) 1 release and VCAM-1, E-selectin and ICAM-1 expression [77]. Another study described potent inhibitory effects of the PPARβ/δ agonist GW0742 on lipopolysaccharide target genes as cyclooxygenase (COX)-2 and inducible nitric oxide synthase (iNOS) in macrophages. Lipopolysaccharide (LPS) is the most abundant component within the cell wall of Gram-negative bacteria. It can stimulate the release of inflammatory cytokines in various cell types, leading to an acute inflammatory response towards pathogens. It has been suggested that PPARβ/δ functions in modulating the program of macrophages during inflammatory responses [168]. PPARβ/δ modulation has been proposed to attenuate inflammation in atherosclerosis. A comparison between wildtype and PPARβ/δ knockout macrophages revealed that proinflammatory genes such as MMP9, IL-1β and MCP-1 were down-regulated in PPARβ/δ knockout macrophages. However, activation of PPARβ/δ with GW501516 suppressed the expression of MCP-1 and IL-1β, indicating that activation of PPARβ/δ is anti-inflammatory. As an explanation for this seemingly discrepancy of PPARβ/δ function in inflammation, a ligand-dependent interaction of PPARβ/δ with the anti-inflammatory transcriptional repressor BCL-6 had been suggested. Without ligand, PPARβ/δ binds BCL-6. When activated with a PPARβ/δ ligand, BCL-6 is released and suppresses proinflammatory pathways [75]. Monocytes can be differentiated in either a proinflammatory (M1 or classically activated macrophage, induced by TNFα, bacterial LPS or interferon gamma) or an anti-inflammatory (M2 or alternatively activated macrophage, induced by interleukins). PPARβ/δ has an important role in the development of the M2 phenotype, as PPARβ/δ knockout cells were unable to acquire this alternatively activated macrophage phenotype upon interleukin-4 or-10 stimulation [169]. In contrast, Thulin and colleagues demonstrated that PPARβ/δ is regulated by the microRNA miR-9 in monocytes and that activation of PPARβ/δ might be of importance in M1 proinflammatory, but not in M2 anti-inflammatory macrophages, as the PPARβ/δ agonist GW501516 induced expression of PPARβ/δ target genes in proinflammatory M1, but not in M2 macrophages [170]. Further studies confirmed the implication of PPARβ/δ in the modification of macrophage functions and the reprogramming of their activation status. Treatment of macrophages with modified low-density lipoproteins (LDLs) induced arginase I expression, which was abolished by the PPARβ/δ antagonist GW9662. In contrast, the PPARβ/δ agonist GW0742 strongly induced

arginase I expression. PPARβ/δ activity in macrophages therefore impacts the balance of Th1/Th2 responses through specific induction of arginase I expression and activity [171]. Myelin-derived phosphatidylserine was found to mediate PPARβ/δ activation in macrophages after myelin uptake, a pathway leading to suppression of the production of inflammatory mediators, ameliorating experimental autoimmune encephalomyelitis (EAE), an animal model of multiple sclerosis [172]. Mukundan et al. identified PPARβ/δ as a transcriptional sensor of apoptotic cells in macrophages. Apoptotic cell feeding stimulated PPARβ/δ expression in macrophages, which then induced expression of opsonins, enhanced apoptotic cell clearance by macrophages and increased anti-inflammatory cytokine production [173]. As another mechanism of PPARβ/δ function in macrophages, induction of the immunoreceptor CD300a has been postulated. The PPARβ/δ agonist GW501516 activated CD300a expression in macrophages. Mice lacking CD300a showed chronic intestinal inflammation upon high fat diet and an increase in proinflammatory cytokines, specific for the M1 macrophage type. The PPARβ/δ/CD300a pathway could therefore contribute to the anti-inflammatory action in macrophages [174]. Adhikary and colleagues investigated the global PPARβ/δ-regulated signaling network in human monocyte-derived macrophages. They found a robust induction of PPARβ/δ expression upon monocyte to macrophage differentiation. Using PPARβ/δ agonists and inverse agonists, they identified two mechanisms by which PPARβ/δ regulates immune-modulatory genes: 1) canonical regulation through DNA binding at PPARβ/δ–RXR sites (PPREs), induced by agonists and repressed by inverse agonists, and 2) repression by agonists in the absence of PPARβ/δ DNA binding (inverse regulation). Inverse regulation concerned NF-kappaB and the signal transducer and activator of transcription (STAT)1 target genes, resulting in the inhibition of multiple proinflammatory mediators consistent with anti-inflammatory effects of PPARβ/δ activation. Interestingly, they could also demonstrate specific immune stimulatory effects induced by PPARβ/δ agonists, a pro-survival effect on macrophages and inhibition of CD32B surface expression and stimulation of T cell activation. This confirms the strong anti-inflammatory function of PPARβ/δ, but also indicates context-dependent specific immune-stimulatory actions of PPARβ/δ activation [175]. The same group aimed at elucidating the role of PPARβ/δ in the pro-tumorigenic polarization of tumor associated macrophages (TAMs) in ovarian cancer. In vitro, PPARβ/δ target genes such as pyruvate dehydrogenase kinase (PDK) 4 and angiopoietin-like protein (ANGPTL) 4 were robustly induced in monocyte derived macrophages, but the ligand response in TAMs was impaired and most PPARβ/δ target genes were refractory to synthetic agonists. Next, the authors compared freshly isolated ascites-associated TAMs from ovarian cancer patients with monocyte-derived macrophages from healthy donors. Many PPARβ/δ target genes as PDK4, ANGPTL4, and carnitine palmitoyl transferase (CPT) 1A were found to be up-regulated in TAMs and were refractory to stimulation with the PPARβ/δ agonist L-165041. The deregulation and unresponsiveness of target genes in TAMs was found to be due to the presence of endogenous activators in malignancy associated ascites, as ascites caused an equal deregulation in normal macrophages. Lipidomic analysis of ascites samples revealed high levels of polyunsaturated fatty acids (PUFA) [176], known as PPARβ/δ activators [177]. The deregulation of PPARβ/δ target genes by PUFA ligands stimulates the pro-tumorigenic conversion of host-derived monocytic cells and might contribute to tumor progression [176]. Very little is known about the PPARβ/δ function in other key immune cell types except macrophages. In 2008, protein expression of PPARβ/δ in activated human T-cells was described. It has been shown that PPARβ/δ is a transcriptional target of human type I interferon (IFN), stimulates T-cell proliferation and inhibits IFN induced apoptosis, which is partially mediated through enhanced extracellular signal-regulated kinases (ERK) 1/2 signaling [178]. More recently, it has been demonstrated that PPARβ/δ overexpression/activation in vivo inhibits thymic T-cell development by decreasing proliferation of CD4-CD8- double-negative stage 4 (DN4) thymocytes [179]. PPARβ/δ has further been reported to drive maturation of monocyte–derived dendritic cells towards an atypical phenotype with reduced stimulatory effects on T-cells [180]. An interesting in vivo study using murine models of septic shock induction confirmed general anti-inflammatory effects of PPARβ/δ activation. PPARβ/δ deletion had detrimental effects on cardiac and renal function, liver injury, lung

inflammation and survival, which could not be attenuated by administration of the specific agonist PPARβ/δ GW0742. In wildtype animals, selective activation of PPARβ/δ attenuated the multiple organ injury and dysfunction and improved survival when administered acutely in rodent models of endotoxemia and polymicrobial sepsis. PPARβ/δ activation was proposed as an anti-inflammatory therapeutic approach for the treatment of conditions involving local and systemic inflammation [181]. Using an experimental model for multiple sclerosis, it has been shown that PPARβ/δ limits the expansion of pathogenic T helper cells and production of Interleukin 12 and Interferon gamma, thereby limiting autoinflammation in the central nervous system [182]. Similar, in acute cerulein and taurocholate induced pancreatitis mouse models, treatment with the PPARβ/δ agonist GW0742 reduced expression of proinflammatory enzymes and cytokines, neutrophil invasion and tissue and inflammatory deterioration in the pancreas [183]. In contrast, PPARβ/δ has been shown to be a negative regulator of mesenchymal stem cell (MSC) immunosuppressive function, as PPARβ/δ inhibition or genetic deletion enhanced the immunosuppressive properties of MSCs, involving an increased NF-kappaB, ICAM-1 and VCAM-1 activity [184]. Interestingly, also in natural killer (NK) cells, inhibition of PPARβ/δ was beneficial to restore cytotoxic anti-tumor activity. Obesity induced a PPAR driven lipid accumulation in NK cells causing inhibition of their cellular metabolism and inhibiting their function. PPARβ/δ agonists mimicked obesity effects and inhibited trafficking of the cytotoxic machinery to the NK cell-tumor junction, disenabling NK cells to reduce tumor growth in obesity in vivo. Inhibition of PPARβ/δ restored NK cell cytotoxicity [185]. Finally, it may be concluded that most studies identified PPARβ/δ function as anti-inflammatory, mainly in the setting of atherosclerosis. However, only few cancer related investigations exist. In this context, PPARβ/δ has pro-inflammatory and pro-tumorigenic functions by converting host monocytes in macrophages favoring tumor progression [176,185] or impairing antitumor cytotoxicity of NK cells. Surely, more cancer related studies addressing the question how PPARβ/δ acts in different immune regulatory cells, tissues and conditions, are needed.

#### **10. Conclusions and Outlook**

PPARβ/δ functions have been studied extensively. We summarized here known PPARβ/δ effects on cell proliferation, induction of angiogenesis, cell death, function of tumor suppressors, replicative immortality and senescence, invasion and metastasis, tumor metabolism and immune function and mentioned underlying molecular mechanisms. Although not all cited manuscripts were directly related to cancer, one has to keep in mind that the different hallmark capabilities interplay during tumor progression [12,13]. Some controversies regarding the effects of PPARβ/δ activation for cancer progression still exist, which might relate to the different cellular or animal models used. The majority of reports, however, suggest that activation of PPARβ/δ might result in modifications of the hallmark capabilities in favor of a pro-tumorigenic profile. Thus, in contrast to the earlier notion of the therapeutic potential of PPARβ/δ agonists as "exercise mimetics" and potential treatments for metabolic syndrome [186–188], extreme caution should be applied when considering PPARβ/δ agonists for therapeutic purposes given their pro-tumorigenic properties.

For future approaches using PPARβ/δ modulation for potential cancer therapy, collaborations between different laboratories and pathologists are urgently needed to define exact expression patterns of PPARβ/δ in different types, stages and grades of cancer. Currently, already the antibody validation is a limiting factor. Reproducible immunostaining protocols established between different laboratories and precise annotation of cell types would be required to define, which patients might benefit from PPARβ/δ modulation according to expression pattern in cells of the different hallmark capabilities.

**Author Contributions:** Conceptualization, N.W. and K.-D.W.; writing—original draft preparation, N.W. and K.-D.W.; writing—review and editing, N.W. and K.-D.W.; funding acquisition, N.W. and K.-D.W. Both authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by "Fondation ARC pour la recherche sur le cancer", grant number n\_PJA 20161204650 (N.W.), Gemluc (N.W.), and Plan Cancer INSERM, Fondation pour la Recherche Médicale (K.-D.W.).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
