Targeting truncated RXRα for cancer therapy

Abstract

Retinoid X receptor-alpha (RXRα), a unique member of the nuclear receptor superfamily, is a well-established drug target, representing one of the most important targets for pharmacologic interventions and therapeutic applications for cancer. However, how RXRα regulates cancer cell growth and how RXRα modulators suppress tumorigenesis are poorly understood. Altered expression and aberrant function of RXRα are implicated in the development of cancer. Previously, several studies had demonstrated the presence of N-terminally truncated RXRα (tRXRα) proteins resulted from limited proteolysis of RXRα in tumor cells. Recently, we discovered that overexpression of tRXRα can promote tumor growth by interacting with tumor necrosis factor-alpha-induced phosphoinositide 3-kinase and NF-κB signal transduction pathways. We also identified nonsteroidal anti-inflammatory drug Sulindac and analogs as effective inhibitors of tRXRα activities via a unique binding mechanism. This review discusses the emerging roles of tRXRα and modulators in the regulation of cancer cell survival and death as well as inflammation and our recent understanding of tRXRα regulation by targeting the alternate binding sites on its surface.

Introduction

Retinoid X receptor-alpha (RXRα) is a unique and important member of the nuclear receptor superfamily (Fig. 1A). As a master regulator, RXRα acts through homodimerization with itself or via serving as obligatory partner for many other nuclear receptors, including retinoic acid receptor (RAR), thyroid hormone receptor (T3R), vitamin D receptor (VDR), Nur77, peroxisome proliferator-activated receptors (PPARs), liver X receptor (LXR), and farnesoid X receptor [1–9]. Many naturally occurring small molecules have been shown to bind to RXRα and modulate its activities [2–4,10–12]. 9-cis-Retinoic acid (9-cis-RA) was the first one that was identified as a natural RXRα ligand (Fig. 1B). Subsequently, several dietary fatty acids were found to bind to RXRα and act as natural RXRα ligands. These include docosahexaenoic acid (DHA), oleic acid, and phytanic acid. However, none of these molecules have been proved to be the bona fide endogenous ligands of RXRα [13,14].

Figure 1.

The scheme of RXRα/tRXRα and its ligands (A) Schematic representation of RXRα and tRXRα. DBD, DNA-binding domain; LBD, ligand-binding domain; AF-1, activation function 1; AF-2, activation function 2. (B) Chemical structures of 9-cis-RA, Targretin, UAB30 and peretinoin.

RXRα possesses a common structural organization that is shared by the nuclear receptor family: a disordered N-terminal A/B region containing activation function 1 (AF-1), a DNA-binding domain (DBD) containing two zinc fingers, and a C-terminal ligand-binding domain (LBD) composed of 12 α-helices and a short β-turn (Fig. 1A). The LBD consists of a canonical ligand-binding pocket (LBP), an activation function 2 (AF-2), a co-regulator-binding surface groove, and a dimerization surface (Fig. 1A). A well-described mechanism of RXRα action is that RXRα and its partners act as ligand-dependent transcription factors through binding to specific DNA-response elements of the target genes [1–9]. Ligand binding induces a conformational change that triggers a cascade of events such as co-regulator exchanging or binding, leading to positive or negative gene transcription and subsequent biological activities [1–9].

Genetic analysis demonstrates that RXRα is involved in a plethora of developmental and physiological pathways. Knockout of RXRα resulted in embryonic lethality [15]. Tissue-specific inactivation of RXRα has demonstrated a major role of RXRα in hepatocytes [16], skin [17], prostate [18], or adipose tissue [19]. Strong phenotypes observed in most RXRα mutant mice may be related to alternations in pathways regulated by its heterodimerization partners. The role of RXRα homodimer in vivo was unraveled recently. Ligand-activated RXRα homodimers up-regulate p21 expression through the direct binding of RXRα homodimers to the p21 promoter [20]. Characterization of mice lacking RXRα in myeloid cells reveals an important role of RXRα homodimers in the innate immune response to inflammatory stimuli [21].

Aside from its role in DNA binding and transactivation, RXRα also exerts extranuclear actions through transcription-independent mechanisms [22–26]. RXRα resides in the cytoplasm at different stages during development [27]. In response to differentiation [24], survival [28,29], apoptosis [22], and inflammation [25,26,28,29], RXRα migrates from the nucleus to the cytoplasm. For example, RXRα is translocated from the nucleus to the cytoplasm in response to endotoxin and other inflammatory mediators to inhibit its transactivation function [25,30], while an altered localization of RXRα to the splicing factor compartments occurs in highly malignant human breast cancer cells [31]. These observations revealed the intricacy of RXRα functions and underscored the importance of the RXRα nongenomic signaling.

Post-translational modifications also play a critical role in the regulation of RXRα activities. Phosphorylation of the N-terminal domain of RXRα by mitogen-activated protein kinases occurs in response to several stress agents such as UV radiation, oxidative damage, or ribotoxic agents [32–37], leading to the inhibition of the transcriptional activity of RXRα heterodimers. RXRα could also be phosphorylated at Ser260 in its LBD [33,38]. In human hepatocellular carcinoma (HCC) cells, RXRα is heavily phosphorylated at Ser260, making it resistant to ubiquitination and proteasome-mediated degradation [38,39]. Nuclear export of RXRα in response to inflammatory signaling also involves c-Jun N-terminal kinase (JNK) phosphorylation of Ser260 [26]. RXRα is also a substrate for modification with small ubiquitin like modifier (SUMO) [40]. SUMOylation of RXRα at Lys108 in its AF-1 domain inhibits its transcriptional activity [40]. Interestingly, a recent study showed that inflammatory mediators increase SUMOylation of RXRα in a JNK-dependent manner [41]. Unlike SUMOylation, acetylation of RXRα by p300 promotes its DNA binding, thereby increasing its transcriptional activity [42].

We recently reported that an N-terminally truncated form of RXRα (tRXRα) (Fig. 1A) produced in cancer cells resides in the cytoplasm to promote the growth of tumor cells [29]. Our investigation of tRXRα action in the cytoplasm revealed an extensive interaction between tRXRα and tumor necrosis factor-α (TNFα) signaling. In this review, we will briefly summarize the role of RXRα in cancer and the cancer therapeutic effect of RXRα ligands (rexinoids). We will focus our discussion on the identification of tRXRα in cancer cells and its implication in the regulation of cancer survival and death as well as inflammation through its nongenomic interaction with TNFα-induced phosphoinositide 3-kinase (PI3K)/AKT and NF-κB signal transduction pathways. Finally, we will summarize recent advances in the discovery of tRXRα modulators with new RXRα-binding mechanisms.

RXRα and Cancer

Altered expression and function of RXRα are implicated in the development of a number of cancers and diseases. Although RXRα-knockout fetus dies at embryonic days [15], targeted disruption of RXRα gene leads to preneoplastic lesions in prostate [18], alopecia, epidermal interfollicular hyperplasia, keratinocyte hyperproliferation, and aberrant terminal differentiation in skin [17], the development of cervical malignant lesions [43], alteration of fatty acid oxidation and hepatocyte lifespan in liver [16], and the resistance to diet-induced obesity due to impaired adipocyte differentiation in adipose tissue [19]. Consistently, diminished RXRα expression is associated with the development of certain malignancies, such as thyroid carcinoma [44,45] and liver cancer [46], and levels of RXRα protein are often reduced in cancer cells and tumor tissues [45,47–51]. In addition to reduced levels of RXRα protein, altered RXRα function by phosphorylation is associated with the development of human HCC [38] and colon cancer [52]. Furthermore, several studies have demonstrated that alteration of the subcellular localization of RXRα is linked to the development of cancer [31]. Recent studies showed that RXRα binding to PML/RARα is absolutely required for the development of acute promeylocytic leukemia (APL) in transgenic mice [53–55], demonstrating the oncogenic potential of this protein when it acts inappropriately. Several groups have demonstrated that RXRα is proteolytically cleaved in cancer cells [23,45,51,56–60], and our illustration that tRXRα could enhance TNFα activation of PI3K/AKT and NF-κB pathways revealed that aberration in RXRα signaling by limited proteolysis plays an active role in cancer development [28,29,61–65].

Rexinoids and Cancer Therapy

The pleiotropic actions of RXRα under both physiological and pathophysiological conditions suggest RXRα as an important drug target for pharmacologic interventions and therapeutic applications. This is highlighted by the FDA approval of the RXR-based drug Targretin (bexarotene) for treating cutaneous T-cell lymphoma (CTCL) patients who are refractory to at least one prior systemic therapy [66]. Targretin selectively binds to RXRs and does not have significant RAR binding and transactivation [5,67,68]. Side effects have been reported to be associated with the Targretin treatment, including hyperlipidemia and hypothyroidism [66,69]. Hyperlipidemia is thought to be associated with the modulation of RXR heterodimers with PPARs and LXRs [70] and hypothyroidism may be due to the inhibition of the TSH production by Targretin through the thyrotrope-restricted RXRγ isoform [71]. To overcome these side effects, efforts to develop novel RXR-based drug for CTCL treatment are ongoing. Recently, it was reported that certain analogs of Targretin possess improved biological properties [72]. The mechanism of Targretin action has yet to be fully elucidated. It has been shown that the drug can induce apoptosis, differentiation and cellular senescence, inhibit metastasis and angiogenesis, and block cell cycle progression, which was reviewed recently [73–76].

Rexinoids also show significant effect in non-small-cell lung cancer (NSCLC) [77], in which altered RXR signaling has been observed [78]. Preclinical data showed that Targretin can prevent and overcome acquired paclitaxel resistance in NSCLC [79]. Furthermore, Targretin could act synergistically with standard first-line platinium-based chemotherapy [77,78]. In a randomized Phase III trial comparing Targretin in combination with cisplatin/vinorelbine to cisplatin/vinorelbine alone in a total of 623 patients for over survival as the primary efficacy endpoint, a subgroup (32%) of Targretin-treated patients who developed NCI Grade 3/4 hypertriglyceridemia had longer median survival compared with control patients [80]. In another Phase III trial to determine the effects of addition of Targretin to standard first-line carboplatin and paclitaxel therapy, similar results were obtained that increased survival was correlated with the occurrence of Grade 3/4 hypertriglyceridemia in patients treated with Targretin [81].

Rexinoids also induce differentiation in AML patient samples and in various AML cell lines [82,83]. As a differentiation agent, Targretin is being tested for AML treatment [84]. A Phase I dose escalation study in elderly and relapsed AML patients was conducted to investigate whether Targretin could be used in combination with decitabine [84]. It was found that the combination was well tolerated, but produced only modest responses. However, greater AML blast differentiation was observed in patients with clinical response, suggesting a potential of rexinoids in AML therapy. Recent studies suggested that Targretin mediates the RXR/LXR-regulated gene expression that was deregulated in AML cells [85].

Other studies have also revealed an emerging role of RXRα in APL [86]. RXRα has been demonstrated to be a binding partner of PML/RAR. Genome-wide epigenetic studies suggested that PML/RARα/RXR complex acted as a local chromatin modulator [55]. Recruitment of RXRα by the APL fusion protein is crucial for oncogenic transformation and is required for the development of APL in transgenic mice [53–55]. Therefore, rexinoids have great potentials for treating APL.

Malfunction of RXRα due to phosphorylation by the Ras-mitogen-activated protein kinase signaling pathway is profoundly associated with the development of HCC and thus may be a critical target for HCC chemoprevention. Acyclic retinoid (also known as peretinoin), a synthetic retinoid that binds to both RXR and RAR, can prevent phosphorylation of RXRα by inhibiting the activities of Ras–Raf–Erk system through an undefined mechanism [87]. Clinical studies have shown that it is effective in suppressing HCC recurrence and improving patient survival rates following curative therapy [87]. A recent gene expression profiling study using liver biopsy from HCC patients underwent therapy revealed that peretinoin not only enhances the expression of retinoid target genes but also regulates various signal transduction pathways involved in hepatocarcinogenesis [23].

Proteolytic Cleavage of RXRα in Cancer Cells

Numerous studies have shown limited proteolytic cleavage of RXRα protein in tumor cells [23,45,51,56–60]. Matsushima-Nishiwaki et al. [56] showed a 44 kDa tRXRα in liver cells and found that m-calpain could cleave RXRα into the tRXRα lacking N-terminal A/B region in HuH7 HCC cells (Fig. 1A). A recent study showed that tRXRα accumulates more in several HCC cell lines than in normal immortal human hepatocytes [88]. Cathepsin L-type protease could also cleave RXRα at its N-terminal region, producing a 44 kDa tRXRα in rapid growing hepatocytes, which alters thyroid hormone responsiveness [58], and in HCC cells [60]. In osteosarcoma cells, the production of an aberrant 45 kDa tRXRα is implicated in the resistance to the antiproliferative effects of calcitriol and retinoids [59]. Two tRXRα proteins with 47 and 44 kDa were detected in all seven prostate cancer cell lines, which may have altered subcellular localization [51]. Interestingly, a 44 kDa tRXRα was found in mitochondria [23], suggesting that tRXRα may play a nongenomic role. In our study, we found that tRXRα is produced in many different types of cancer cells [29]. Moreover, tRXRα was detected in primary tumors but not in tumor surrounding tissues or distant normal tissues from the same cancer patients [29], demonstrating a close association of tRXRα production with cancer. We also identified calpain II as a protease that could cleave RXRα protein in vitro and in vivo. Activation of calpain II by ionomycin enhances the production of tRXRα in cancer cells, which is regulated in a glycogen synthase kinase 3 beta-dependent manner [63].

Limited proteolytic cleavages of RXRα mainly occur at its N-terminus [23,45,51,56–60]. Comparing with its DBD and LBD, the function and regulation of the N-terminal A/B domains of RXRα have not been well studied. The fact that the N-terminal A/B domains of nuclear receptor family members are highly variable suggests that they may mediate specific functions. Phenotypic analysis of mice expressing RXRα with its N-terminal A/B region deleted indicated that the RXRα AF-1 domain is functionally important for efficiently transducing the retinoid signal during embryonic development [89]. An interesting feature of the N-terminal A/B region is that it contains many consensus phosphorylation sites and is therefore the target of multiple kinases, such as JNK [32,34,36,37,90–92]. Hyperphosphorylation of A/B region (Ser61, Ser75, and Thr87) can induce apoptosis [93], while phosphorylation of RXR at Ser260 has been correlated with the unrestrained growth of certain HCC [38]. However, little is known about the mechanisms through which the N-terminal A/B region and its phosphorylation site participate in the regulation of RXRα activity.

Regulated proteolysis is a key step in a number of different signaling pathways that respond to developmental cues or external stimuli [94–100]. Caspase-mediated cleavage of the BH3-only protein Bid generates a truncated protein (tBid), and the subsequent translocation of tBid to mitochondria is implicated in death receptor signaling [99]. Similarly, caspase-3 mediates retinoic acid-induced degradation of the APL PML/RARα fusion protein [101], and the cleavage product of PML/RARα contributes to ATRA-mediated differentiation in APL [102]. Proteolytic processing of Notch and nuclear translocation of truncated product is a crucial step in the transduction of the Notch signaling [96]. Cleavage of the androgen receptor by calpain produces a truncated receptor protein that may play a role in the development of androgen-independent prostate cancer [103]. Similarly, cleavage of MET, a membrane-bound receptor tyrosine kinase, results in a truncated nMET, which is localized in the nuclei of malignant cells to promote the growth of castration-resistant prostate cancer cells through its activation of both SOX9 and β-catenin [104]. Thus, proteolytic cleavage likely represents an important mechanism that regulates the biological function of RXRα.

Nongenomic Action of tRXRα in Cancer

tRXRα and PI3K/AKT survival signaling

PI3K is a heterodimeric protein composed of a catalytic subunit (p110α/β/γ/δ) and a regulatory subunit (p85α/β) that participate in multiple cellular processes, including cell growth, transformation, differentiation, and survival in a number of cell types and human cancers [105–108]. The p85α/p110α heterodimer is the major form of PI3Ks, in which the p85α regulatory subunit binds to the p110α catalytic subunit to integrate signals from various cellular proteins, providing an integration point for activation of p110α and downstream molecules such as AKT.

We found that tRXRα but not RXRα could act to mediate TNFα activation of PI3K/AKT in a number of cancer cell lines [29] (Fig. 2). Unlike RXRα that resides in the nucleus, tRXRα is cytoplasmic in response to TNFα treatment, and interacts with the p85α regulatory subunit, leading to an enhanced activation of the PI3K/AKT survival pathway and anchorage-independent cell growth in vitro and cancer cell growth in animals (Fig. 2). Abnormal activation of the PI3K/AKT pathway is often observed in cancer cells, contributing to their growth and survival properties and drug resistance [105–108]. Knocking down of tRXRα could reduce basal AKT activation in some cancer cells, demonstrating that tRXRα may play a critical role in the aberrant activation of PI3K/AKT signaling in cancer cells. Interestingly, the interaction of tRXRα with p85α and its activation of PI3K/AKT signaling were induced by inflammatory cytokine TNFα but not by some growth factors such as epidermal growth factor, implying that tRXRα may specifically act in inflammatory environment. These results provided another example that a nuclear receptor can act outside of the nucleus to regulate an important biological process and identified tRXRα as a key molecule involved in the aberrant activation of PI3K/AKT pathway in cancer cells. However, many important questions regarding the nongenomic regulation of the PI3K/AKT pathway by tRXRα and ligands remain to be answered. It is unclear whether the cytoplasmic localization of tRXRα results from its nuclear export or cytoplasmic retention due to its interaction with cytoplasmic proteins such as p85α. As the wild-type RXRα fails to interact with p85α, the N-terminal region deleted from RXRα is expected to play a critical role in regulating RXRα activities, which remains to be determined. As RXRα is often phosphorylated in cancer cells, it will be interesting to examine whether phosphorylation or other modifications of RXRα are involved in the regulation of RXRα cytoplasmic localization and its interaction with p85α. How tRXRα interacts with p85α is still unknown at present. However, it is worth noting that numerous nuclear receptors including estrogen receptor, androgen receptor, glucocorticoid receptor, and RAR have been shown to interact with p85α [109–112], implying the existence of a more general mechanism for their interaction with p85α.

Figure 2.

The nongenomic tRXRα actions The cytoplasmic tRXRα acts through its interaction with TNFα-signaling proteins to regulate cell survival, inflammation, and apoptosis. tRXRα effect is negatively regulated by its modulator Sulindac (S). In addition, tRXRα may target mitochondria through heterodimerization with Nur77 to modulate mitochondria-dependent apoptosis.

Because of its role in oncogenesis and drug resistance, the PI3K/AKT pathway has therefore been targeted extensively to develop therapeutics against cancer and related diseases, and to overcome drug resistance. However, current targeting strategies that rely on direct inhibition of PI3K/AKT activities have posed profound adverse effects and are thus far confined to the preclinical and clinical evaluation due to toxicity and lack of selectivity. Thus, the identification of tRXRα-mediated activation of PI3K/AKT signaling pathway in cancer cells may provide new strategies to inhibit the activation of PI3K/AKT in cancer cells by targeting tRXRα. Such tRXRα-based PI3K/AKT inhibitors are likely to be more specific and tumor selective than conventional PI3K/AKT inhibitors.

tRXRα and apoptosis

Apoptosis, programed cell death, is abnormally regulated in cancer cells and the efficacy of chemotherapeutic drugs depends largely on their ability to induce apoptosis [113,114]. Apoptosis often occurs following either triggering of cell surface death receptors (the extrinsic pathway) or perturbation of mitochondria (the intrinsic pathway) [115]. The intrinsic pathway is initiated by the release of apoptogenic factors such as cytochrome c from mitochondria, while the extrinsic pathway often involves the activation of the initiator caspase-8 through stimulation of death receptors of the TNF-receptor superfamily. It is likely that tRXRα is involved in the regulation of both intrinsic and extrinsic apoptotic pathways.

Previous studies showed that RXRs and ligands are implicated in the regulation of mitochondria-dependent apoptosis through modulating the Nur77-Bcl-2 apoptotic pathway due to its heterodimerization with Nur77 [116] (Fig. 2). Nur77, also known as TR3 or NGFI-B, is an immediate-early response gene and an orphan member of the nuclear receptor superfamily [117–119]. It mediates apoptosis of numerous types of cancer cells by a variety of apoptotic stimuli including retinoid-related apoptotic molecules, TPA, calcium ionophore, and many chemotherapeutic agents, perhaps being the most potent pro-apoptotic member in the nuclear receptor superfamily [22,120–130]. We discovered that Nur77 could migrate from the nucleus to the cytoplasm where it targets mitochondria through its interaction with Bcl-2, leading to cytochrome c release and apoptosis [131,132]. The unique property of the Nur77-Bcl-2 apoptotic pathway is the conversion of Bcl-2 from an anti-apoptotic molecule to a pro-apoptotic one upon binding by Nur77 [132,133]. RXRα cotranslocates with Nur77 from the nucleus to the cytoplasm in response to NGF [24], apoptotic stimuli [22,134], and IGFBP-3 [135]. Interestingly, Casas et al. [23] detected a 44 kDa tRXRα in mitochondria. Whether and how cytoplasmic tRXRα modulates the Nur77-Bcl-2 pathway remain to be investigated.

tRXRα has also been shown to modulate the extrinsic apoptotic pathway upon ligand binding [29]. TNFα is a multifunctional cytokine that controls diverse cellular events such as cell survival and death that control the destiny of cancer cells [136–141]. The diverse cellular effects of TNFα are mediated by TNFα receptor-1 (TNFR1). Upon binding to TNFα, TNFR1 forms a complex that consists of TNF-receptor-associated death domain (TRADD) protein, receptor interacting protein (RIP), and Fas-Associated protein with Death Domain (FADD). When TNFR1 signals apoptosis, FADD binds to pro-caspase-8, resulting in its activation and apoptosis. Alternatively, when TNFR1 signals survival, TNF-receptorassociated factor 2 (TRAF2) is recruited to the complex, leading to activation of AP-1 and NF-κB pathways [136–141]. TNFα binding to TNFR1 could also result in activation of the PI3K/AKT survival pathway [142]. Although TNFα is capable of inducing apoptosis of cancer cells through death receptor-dependent mechanism, such an effect is often antagonized by its own survival function through its activation of NF-κB and PI3K/AKT pathways [143–148]. The balance of TNFα-induced survival- and death-signaling is therefore pivotal in determining the fate of TNFα-responding cells [136–142,144,146,149–152]. Since TNFα is produced by malignant or host cells in the tumor microenvironment but not in normal cells, there has been tremendous interest in developing strategies to shift TNFα signaling from survival to death. Although tRXRα is required for TNFα activation of PI3K/AKT to promote the survival of cancer cells, we found that tRXRα binding to nonsteroidal anti-inflammatory drug (NSAID) Sulindac (Fig. 3) or analogs resulted in caspase-8 activation and apoptosis, demonstrating that tRXRα and ligands could regulate the extrinsic apoptotic pathway [29]. The activation of caspase-8-dependent apoptosis by Sulindac and analogs was attributed to their inhibition of tRXRα-dependent activation of the PI3K/AKT survival signaling. This is supported by the observation that Sulindac and analogs inhibit the binding of tRXRα to p85α [29]. Similarly, inhibition of tRXRα-mediated activation of NF-κB survival pathway also leads to apoptosis [65]. These results demonstrate that inhibition of tRXRα-mediated activation of PI3K/AKT and NF-κB pathways could effectively shift TNFα signaling from survival to death. However, currently it cannot be excluded that tRXRα may be directly involved in the regulation of the death-inducing signaling complex comprised of FADD and pro-caspase-8.

Figure 3.

Chemical structures of molecules that act via binding to novel binding regions in RXRα

tRXRα and inflammation

Recent reports suggest a strong link between chronic inflammation and the development of cancer [153,154]. Chronic inflammation due to infection, autoimmune disease, malignant and benign tumors, or other pathologies has become a recognized risk factor for epithelial-derived malignancies [155]. The transcription factor NF-κB is known as a master inflammatory transcriptional regulator, which is highly active in cancer cells. In quiescent cells, NF-κB proteins are sequestered in the cytoplasm by inhibitory molecules, the inhibitor of NF-κB (IκB) proteins [136–141]. NF-κB requires a signaling pathway for activation. Such NF-κB-activating pathways are triggered by a variety of extracellular stimuli and lead to the phosphorylation and subsequent proteasome-mediated degradation of IκB proteins [136–141]. Activated NF-κB migrates into the nucleus to regulate the expression of multiple target genes associated with cancer cell migration, proliferation, anti-apoptosis, angiogenesis, and metastasis [136–141]. In cancer cells, NF-κB is often constitutively activated as a result of undergoing inflammation or the consequence of formation of an inflammatory microenvironment during malignant progression.

The role of RXRα in regulating inflammation was suggested by findings that several anti-inflammatory agents act as RXR ligands. Dietary omega-3 fatty acids, such as DHA, exert their beneficial effects primarily through their anti-inflammatory effects. DHA induces growth inhibition and apoptosis by inhibiting NF-κB activity [156] and suppressing cytokine production in macrophages [157], whereas R-etodolac which is known to bind to RXRα [158] decreases constitutive and RANKL-stimulated NF-κB activation in macrophages and suppresses TNFα-induced IκB-kinase (IKK) phosphorylation and subsequent NF-κB activation in human multiple myeloma cells [159]. LGD1069 down-regulates COX-2 expression in breast cancer cells [160] and inhibits angiogenesis and metastasis in solid tumors [161]. Oral administration of Targretin reduces inflammation in a group of patients with plaque-type psoriasis [162]. Thus, the anti-inflammatory effects of RXRα and its ligands in various cell types underscore its function in the prevention and treatment of cancer and diseases.

Both genomic and nongenomic actions of RXRα could account for the modulation of inflammation in macrophages and cancer cells by RXRα and ligands. For genomic action, interactions between RXRα and proinflammatory transcription factors, particularly NF-κB and AP-1, have been well described in several reviews [163,164]. RXRα may also employ nongenomic action to negatively regulate the NF-κB signaling pathway. The role of nongenomic action of RXRα in the regulation of inflammation is also suggested by mounting evidence that subcellular localization of RXRα is altered in response to inflammation [25,30]. Lipopolysaccharide treatment of animals altered the subcellular location of RXRα [25]. RXRα undergoes rapid nuclear export in response to signals initiated by the proinflammatory cytokine IL-1β in hepatoma cells [26]. Our recent discovery that TNFα could induce cytoplasmic localization of tRXRα underscores the significance of tRXRα cytoplasmic action in the regulation of inflammation (Fig. 2). As discussed above, TNFα is a multiple cytokine that induces not only the extrinsic apoptotic and PI3K/AKT pathways but also the NF-κB inflammatory pathway through its sequential recruitment of various adaptors including TRADD, RIP1, and TRAF2 to the cytoplasmic membrane [136–141]. This is followed by the recruitment and activation of the classical IKK complex [136–141]. Once activated, the IKK complex phosphorylates IκBα, which is subsequently ubiquitinated and degraded via the proteasome pathway. We recently reported that tRXRα could promote TNFα activation of NF-κB pathway through its interaction with TRAF2 and enhance TNFα-induced RIP1 ubiquitination [65]. A nitrostyrene derivative that binds to tRXRα could inhibit tRXRα interaction with TRAF2 and its induction of NF-κB activation. These results demonstrate that tRXRα is directly involved in the activation and regulation of the inflammasome. It remains to be clarified whether other RXRα modulators known to have anti-inflammatory effect also target tRXRα-mediated activation of the NF-κB pathway.

Ligands that Bind to a Novel Site to Regulate the Nongenomic Activities of tRXRα

We identified that NASID Sulindac (also called CLINORIL^®) currently used for treating pain and inflammation can inhibit the binding interaction between tRXRα and p85α, resulting in caspase-8-dependent apoptosis [29]. Furthermore, we synthesized several Sulindac analogs and identified several Sulindac derivatives including K-80003, K-8008, and K-8012 (Fig. 3), which can also bind to tRXRα to inhibit the TNFα-induced interaction of tRXRα with p85α and activate the TNFα-dependent apoptotic pathway [29]. K-80003, K-8008, and K-8012 are much more effective than Sulindac in inhibiting the growth of various cancer cells, including A549 lung cancer, PC3 prostate cancer, and ZR-75-1 and MB231 breast cancer cells. Both K-80003 and K-8008 inhibit HepG2 tumor growth in animals and do not show apparent toxic effects. Several natural products including CF31 [28] and nitrostyrene derivatives [65] could also activate this death pathway by directly binding to tRXRα.

Our identification of NSAID Sulindac and its analogs as tRXRα binders and modulators prompted us to study their binding mode. The crystal structure of RXRα-LBD in complex with K-8008 or K-8012 [62] demonstrates the existence of different binding site. The complex structures exist as noncrystallographic homotetramer similar to the reported apo-homotetramer [165,166], in which bottoms of two homodimers interface form a tetramer. In a tetramer, two K-8008 molecules were found to bind to one homotetramer in a hydrophobic region that is near the entry and the edge of the cognate LBP [62]. The K-8008 binding region is close to the dimer–dimer interface that does not overlap with the binding region of 9-cis-RA (Fig. 4A). With respect to the monomeric and the dimeric RXRα-LBD, the K-8008 binding region is located on the surface of the RXRα molecules. RXRα has been shown to form transcriptionally silent homotetramers in solution, which rapidly dissociate into active homodimers upon binding of agonists or antagonists [165–167]. Therefore, it is intriguing that binding of K-8008, an RXRα antagonist, to a new region does not induce tetramer dissociation, a similar phenomenon observed in the binding of danthron [168]. The structural basis of K-8008 binding suggests that RXRα tetramerization represents a key mechanism for the regulation of RXR nongenomic actions.

Figure 4.

Novel binding regions in RXRα (A) K-8008 binds to a novel binding region: the K-8008 binding region is away from the 9-cis-RA binding area and located on the surface of monomeric RXRα. It shows the superposition of the monomer of RXRα-LBD/K-8008 complex structure (brown) and the apo protein structure (purple, from PDB entry 1G1U). K-8008 is shown as sticks (carbon in magenta and nitrogen in blue). The classic ligand-binding site is indicated by a VDW ball model of 9-cis-RA (in cyan/red) taken from a superimposed 1FBY of PDB. (B) The proposed binding region for compound 23. The compound 23-binding region overlaps with the coactivator-binding region. Here, compound 23 was docked to the structure 3FUG (in pink) of PDB and the docked conformation (in VDW balls) was displayed with the coactivator peptide (in green) in the structure of 3FUG.

Ligands Targeting the Co-regulator-binding Site of tRXRα

Drug discovery and development efforts targeting RXRα have been focused on identifying and optimizing rexinoids that bind to RXRα canonical LBP. However, as mentioned above, there are key limitations of using rexinoids that include rising of plasma triglyceride levels, suppression of the thyroid hormone axis, and induction of hepatomegaly. Therefore, targeting alternate sites on RXRα for regulating its activities could become a new strategy for RXRα-based drug discovery. Compounds that bind to alternate sites have been identified for other nuclear receptors [169–171], including estrogen receptor, androgen receptor, VDR, and T3R. Among the reported alternate sites on nuclear receptors, the co-regulator binding site is the most studied one. Recently, by employing a docking-based virtual screening approach, we identified some small molecules that bind to the co-regulator-binding surface of RXRα, a region where the binding sites of corepressor and the coactivator overlap (Fig. 4B). One of the identified binder, compound 23 (Fig. 3), could regulate the biological functions of tRXRα. Compound 23 could inhibit the TNFα-induced interaction between tRXRα and p85α, inhibit AKT activation in vitro and in animals, and induce apoptosis [61]. Compound 23 represents the first example of an RXRα modulator that acts via the co-regulator-binding site rather than the classical LBP. Furthermore, compound 23 does not bind to the LBP. Thus, targeting the alternate binding sites on the surface of RXRα for therapeutic intervention may become a new paradigm for nuclear receptor-based drug discovery.

Conclusion and Perspective

RXRα is a unique and important drug target as evidenced by the success of rexinoid Targretin in treating CTCL and the enormous favorable results from testing rexinoids in various cancer models. However, the mechanisms by which RXRα modulate carcinogenesis are complex and remain to be fully defined, which has hampered the exploitation of the therapeutic potential of RXRα. The identification of tRXRα in cancer cells and the illustration of its roles in the control of apoptosis, survival, and inflammation offer new strategies to develop improved therapeutics against cancer by targeting tRXRα.

RXRα-based drug development is also hampered by the side effects associated with targeting its cognate LBP that is highly conserved among many nuclear receptors. Therefore, one of the current challenges in developing RXRα-based drugs is to identify selective RXRα modulators that possess the desired pharmacological activities without unwanted side effects. Many new modulators are being developed. In addition, screening for natural and synthetic RXRα ligand is ongoing. Thus, the findings that RXRα is cleaved in tumor cells and that Sulindac-derived small molecules and others act at the alternate binding sites on the surface of tRXRα will provide new rational for drug design and screening approach. Such an approach may help to identify small molecules specific to tumor- or disease-selective RXRα (i.e. tRXRα or RXRα with abnormal modifications) and may also circumvent side effects associated with binding to the cognate RXRα LBP. However, many questions remain unanswered regarding tRXRα production and function, and the underlying mechanisms of tRXRα actions need to be determined. Binding of Sulindac analogs to the tetrameric form of RXRα LBD is interesting. However, little is known about the biological function of the RXRα tetramer with respect to the regulation of the nongenomic RXRα action. Characterizing the surface-binding sites in tRXRα and developing selective inhibitors targeting the surface-binding sites may support a transition from the traditional paradigm of drugs targeting the LBP to a novel rational approach targeting functionally important surface sites, which may lead to more effective and specific therapeutics.

Funding

This work was supported by the grants from the National Natural Science Foundation of China (Nos. 91129302, 91429306, and U1405229), the Xiamen Science and Technology Project (No. 3502Z20123015), the Xiamen Oceanic Administration (No. 14PYY051SF04), the U.S. Army Medical Research and Material Command (No. W81XWH-11-1-0677), the National Institutes of Health (No. CA179379), and the California Breast Cancer Research Program (No. 20IB-0138).