Molecular mechanism for growth suppression of human hepatocellular carcinoma cells by acyclic retinoid

We have reported previously that acyclic retinoid, a synthetic retinoid X receptor a (RXRa)-ligand, suppresses the development of hepatocellular carcinoma (HCC) in patients with chronic liver disease. On the other hand, HCCs become refractory to physiological concentrations of the natural RXRa-ligand, 9-cis retinoic acid (9cRA), due to extracellular signal-regulated kinase (Erk) 1/2-mediated phosphorylation and inactivation of RXRa. Here, we show that acyclic retinoid restores the function of RXRa in human HCC-derived HuH7 cells by inactivating the RasErk 1/2 signaling system, thereby dephosphorylating RXRa. In contrast, 9cRA failed to suppress phosphoErk 1/2 levels and subsequent RXRa phosphorylation. Although 9cRA also suppressed Ras activity, it simultaneously down-regulated mitogen-activated protein kinase phosphatase-1, an enzyme that inactivates Erk, thereby leaving the phosphorylation status of Erk unchanged. A combination of 9cRA (a potent ligand) and acyclic retinoid (a weak ligand preventing phosphorylation) resulted in a marked cooperation in transactivation via the RXRresponse element and in inhibiting the proliferation of HuH7 cells. These events provide a novel molecular basis for the antitumor activity of acyclic retinoid against HCC.


Introduction
Retinoids are prime candidates for cancer chemoprevention, as they have profound effects on cell growth, apoptosis and differentiation (1).In fact, experimental studies have revealed potent roles of retinoids in reversing the carcinogenesis process and suppressing malignant transformation (2).Loss of retinoid activity or responsiveness is closely linked to carcinogenesis in several organs including hepatocellular carcinoma (HCC) (3).We documented previously a local depletion of retinoid in human HCC tissues due to its rapid conversion to an inactive metabolite (3) and further showed in experimental and clinical studies that supplementation with a synthetic retinoid analogue, acyclic retinoid, suppresses the development of HCC (4,5).However, the molecular mechanism(s) underlying this effect has been unclear.
Retinoids exert their biological activities primarily through retinoic acid receptor (RAR) and retinoid X receptor (RXR), two members of the nuclear receptor superfamily that act as ligand-dependent transcriptional regulators (6).RAR interacts with both all-trans retinoic acid (atRA) and 9-cis RA (9cRA), whereas RXR binds only to 9cRA.Acyclic retinoid is a ligand of both RAR and RXR (7).Both RAR and RXR consist of three subtypes, -a, -b and -g, characterized by a modular domain structure.RXR forms a homodimer as well as heterodimers with RAR and other nuclear receptors, including vitamin D receptor (VDR), thyroid hormone receptor and peroxisome proliferator-activated receptor (PPAR), and thereby participates in the regulation of respective downstream genes.Among three subtypes of RXR, RXRa is highly expressed in human HCC and HCC-derived HuH7 cells (8).These dimers bind to their respective response elements and subsequently activate or inhibit the expression of target genes.Transcriptional activity of retinoid receptors is regulated by several factors that modulate the composition of their receptor complexes (7).In the absence of RA, unliganded RAR/RXR heterodimer binds to co-repressor complexes that provide a link between the heterodimer and histone deacetylases, resulting in chromatin condensation and gene silencing by removing acetyl groups from nucleosomal histones.Ligand binding to the retinoid receptors induces their conformational change, which allows the interaction between RAR/RXR and coactivators.Co-activators form multiprotein complexes that have histone acetyltransferase activity to acetylate histone N-terminal tails.Acetylation of histones leads to nucleosomal repulsion and chromatin decondensation, which is thought to be indispensable for the transcriptional activation by retinoid receptor.
to accumulation of non-functional phosphoRXRa in the cancer cells, which may interfere with the function of remaining normal RXRa in a dominant-negative manner, thereby rendering the cancer cells much less sensitive to 9cRA.Hence, HCC cells evade 9cRA-induced apoptosis (12).In contrast, HCC cells are sensitive to acyclic retinoid and undergo apoptosis (12).Because over-expression of phosphorylation-site mutated unphosphoRXRa restored its activity and suppressed tumor cell growth (9), we have examined whether acyclic retinoid can suppress RXRa phosphorylation and thereby restore its function in HCC cells.
We have pursued this possibility using HuH7 cells, which reveal that acyclic retinoid dephosphorylates RXRa by inactivating the Ras-Erk system in HCC cells, potentiating the responses of RXRa to physiological concentration of 9cRA.Our findings identify these events as a key molecular mechanism underlying the anti-tumor activity of acyclic retinoid.
Cell culture and treatment HuH7 cells were obtained from the Japanese Cancer Research Resources Bank (Tokyo, Japan) and maintained in RPMI1640 medium supplemented with 1% FCS and 0.2% lactoalbumin hydrolysate as described previously (9,11).In an in vitro culture model, FCS present in the medium at concentration 41% interferes with the growth suppression by acyclic retinoid (13).HuH7 cells were plated on 100 mm-plastic dishes (8 Â 10 5 cells/dish) or 96-well plates (1.75 Â 10 3 cells/well) at the cell density of 30±40% confluency and were treated for 24 h in the same medium with the indicated concentrations of retinoids dissolved in ethanol at a final concentration of 0.05%.Cell numbers were counted using Trypan Blue dye exclusion method.

Preparation of cell extracts and western blot analysis
Preparation of cell extracts and western blot analyses of total Erk 1/2, phos-phoErk 1/2, MKP-1 and GAPDH were performed as described previously (9).For detection of phosphoRXRa, RXRa was affinity-purified from the cell extracts using anti-RXRa antibody-immobilized Sepharose beads, and then subjected to western blot analyses employing anti-phosphoserine antibody (9).As a control, proteins precipitated with non-immunized rabbit antibody-linked beads were analyzed.Protein concentrations in the samples were determined by Bio-Rad protein assay kit (Hercules, CA).The signals were detected with an Amersham Bioscience ECL system (Piscataway, NJ).Densitometric analysis was performed using the NIH image version 1.61 software.

Determination of Ras activity
Ras activities were determined using a Ras activation assay kit (Upstate Biotechnology, Lake Placid, NY) according to the manufacturer's instruction.This assay detects the activated Ras bound to Raf-1.

Luciferase reporter assay
A reporter plasmid, tk-CRBPII-RXRE-Luc, containing multiple copies of the DR1 type RXRE sequences within rat cellular retinol-binding protein type II (CRBPII) promoter up-stream of luciferase cDNA, was kindly provided by the late Dr K.Umesono (Kyoto University, Kyoto, Japan).Transfection with tk-CRBP-II-Luc reporter and measurement of luciferase activity of cell lysates were performed using a dual-luciferase reporter assay system (Promega, Madison, WI) as described previously (9).Cells were cotransfected with pRL-CMV (Renilla luciferase) as an internal standard to normalize transfection efficiency (14).Furthermore, to see clear differences in the transcriptional activation via RXRE, wild-type RXRa was transfected to HuH7 cells.

Transfection of mutant RXRa
Construction of a human RXRa mutant cDNA, in which serines 259 and 260 were mutated to alanine (S260A), an unphosphorylated mimic of RXRa, and its transfection to HuH7 cells were performed as described previously (9).

Statistical analysis
Data are expressed as the mean and SD.Statistical significance was assessed with one-way analysis of variance followed by Sheffe's t-test.

Acyclic retinoid reduced phosphoserine RXRa levels and Erk activity
In our previous study, we validated human HCC-derived HuH7 cells as a faithful representative of clinical HCC tissues in terms of MAP kinase-mediated phosphorylation of RXRa (9).Therefore, we have used HuH7 cells in the present study to examine if acyclic retinoid might prevent the phosphorylation of RXRa and restore its function.Because malfunction of RXRa is due to phosphorylation at a serine residue (9), we first examined if acyclic retinoid might dephosphorylate a phosphoserine of RXRa.As depicted in Figure 1A, levels of phosphoserine RXRa (lane 1) band was reduced markedly following treatment with acyclic retinoid alone (lane 2) and Fig. 1.Effects of retinoids on the levels of phosphoserine RXRa (A) and phosphoErk 1/2 (B) in HuH7 cells.HuH7 cells were treated with vehicle (panels A and B, lanes and column 1), 1 mM acyclic retinoid (lanes and column 2), 1 mM 9cRA (lanes and column 3) or their combination (lanes and column 4) for 24 h.(A) RXRa was affinity-purified from the cell extract using anti-RXRa-antibody-immobilized Sepharose and then subjected to western blot analysis using anti-phosphoserine antibody.(B) Western blot analyses of phosphoErk 1/2 and total Erk 1/2 were performed with the respective specific antibodies using the same extract as in panel A. Relative intensity of the band was determined by densitometry (columns 1±4).Values represent the amounts of phosphoErk divided by those of total Erk (the mean AE SD, n 5).An asterisk represents significant difference (P 5 0.05) compared to columns 1 and 3.
We have shown previously that Erk 1/2, constitutively expressed in the HCC cells (11), plays a crucial role in RXRa phosphorylation (9).We thus examined if acyclic retinoid might suppress Erk activity.As depicted in Figure 1B, the amount of phosphoErk 1/2, active forms of Erk, was also reduced significantly (P 5 0.05) following treatment with acyclic retinoid alone (lanes and column 2) and with the combination of acyclic retinoid plus 9cRA (lanes and column 4), but not with 9cRA alone (lanes and column 3), as has been reported previously (11).The amount of total Erk 1/2 was not affected by either treatment (Figure 1B).These results suggest that HuH7 cells are refractory to 9cRA, but sensitive to acyclic retinoid, in regulating the activation of Erk 1/2 and subsequent RXRa phosphorylation.

Differential regulation of Ras and MKP-1 levels between acyclic retinoid and 9cRA
Our previous work suggested that the down-regulation of transforming growth factor (TGF-a) interrupted autocrine growth signals via the epidermal growth factor (EGF) receptor, the receptor for TGF-a, providing a potential mechanism for apoptosis induction by acyclic retinoid in HuH7 cells (15).Therefore, we examined if acyclic retinoid might downregulate the activity of Ras, a small GTP-binding protein, which mediates activation of the MAP kinase pathway following stimulation by TGF-a (16).As shown in Figure 2A, Raf-1-bound Ras activities were inhibited by 70% in acyclic retinoid-treated HuH7 cells (lane and column 2).Similar inhibition was also observed following 9cRA-treatment (Figure 2A, lane and column 3).
To provide an explanation for the discrepancy between 9cRA's effects on Ras activity (suppressed; Figure 2A, lane and column 3) and on phosphoErk levels (unchanged; Figure 1B, lanes and column 3), we analyzed the effect of 9cRA on the levels of MKP-1, a phosphatase that inactivates Erk 1/2.As depicted in Figure 2B, 9cRA suppressed MKP-1 levels by 40% (lanes and column 3), whereas acyclic retinoid did not reduce the levels of MKP-1 either in the absence (lanes and column 2) or in the presence of 9cRA (lanes and column 4).These results suggest that because 9cRA suppresses simultaneously both activating protein (Ras) and inactivating enzyme (MKP-1), 9cRA leaves the levels of phosphoErk 1/2 unaffected.In contrast, as acyclic retinoid downregulates Ras activity without accompanying a decrease in MKP-1 levels, it suppresses phosphoErk 1/2 levels and prevents generation of phosphoRXRa.

Acyclic retinoid potentiates 9cRA-induced transactivation via RXRE
We have shown previously that compared with phosphorylated RXRa, an unphosphorylated form of RXRa exerts much higher transactivation activity in response to 9cRA (9).Acyclic retinoid at !5 mM can serve as a ligand for RXRa, but has much less potent transactivation activity than that of 9cRA (7,17,18).Together with the results obtained in the present study, the data suggest that 9cRA cannot prevent phosphorylation of RXRa, but is a potent ligand when RXRa is kept unphosphorylated, whereas acyclic retinoid can prevent the phosphorylation, but is a weak ligand.This led us to explore whether a combination of 9cRA (a potent ligand) and acyclic retinoid (a weak ligand preventing phosphorylation) might exert a cooperativity on transactivation via RXRE.This suggestion was borne out in the findings illustrated in Figure 3.
HuH7 cells, which had been transfected with an RXREcontaining gene promoter, were treated with incremental concentrations of 9cRA or acyclic retinoid alone, or with a combination of incremental concentrations of 9cRA and a fixed concentration (1 mM) of acyclic retinoid, and then transactivation activity was measured.9cRA induced transactivation in a dose-dependent manner at !100 nM (Figure 3, curve 2).These activities were further enhanced ~20-fold by simultaneous addition of acyclic retinoid (Figure 3, curve 3), which alone had no transactivating activity (curve 1).In the previous report we found that acyclic retinoid alone induced RXREmediated transactivation activity (17).Although we cannot explain the reason for this discrepancy, we speculate that the differences in the experimental condition (techniques employed for transfection, culture media conditions and cell densities) would be responsible for it.
This current result suggests that there might be a 20-fold molar excess of non-functioning phosphoRXRa compared with normal RXRa in the cancer cells.Importantly, even 9cRA at physiological concentrations as low as 1±10 nM (6,19), which alone was not transactivating, became active in conjunction with 1 mM acyclic retinoid, a pharmacological concentration in patients' plasma (4).These results suggest that prevention of RXRa phosphorylation by acyclic retinoid might be linked to the restoration of its function and transactivation activity in response to physiological concentration of 9cRA in vivo.

Acyclic retinoid and 9cRA cooperatively suppress proliferation of HuH7 cells
We have suggested previously that malfunction of RXRa due to phosphorylation appears to be associated with aberrant proliferation of HuH7 cells, and that introduction of an unphosphorylated form of RXRa suppresses cell growth (9).Therefore, we examined if the combination of acyclic retinoid plus 9cRA might cooperatively suppress the growth of HuH7 cells (Figure 4).To see clearer differences in the growth suppression between single treatment with each of two retinoids and their combination, we used the culture medium containing FCS, which was known to regress the effect of acyclic retinoid (13).Acyclic retinoid alone at concentrations 1 mM minimally suppressed cell growth ( 3%; Figure 4, curve 1) as we have reported previously (12).9cRA was not effective until 100 nM, and slightly suppressed the growth at 1 mM, culminating in a 23% inhibition (Figure 4, curve 2).In contrast, in conjunction with 1 mM acyclic retinoid, 9cRA suppressed cell growth from 0.1 nM and resulted in 25% inhibition at 10 nM (Figure 4, curve 4).Growth suppression by 9cRA was also restored when HuH7 cells were transfected with unphosphorylated mimic of RXRa (S260A mutant; Fig. 3. Cooperativity between 9cRA and acyclic retinoid in RXRE-mediated transactivation activity in HuH7 cells.Cells were cotransfected with CRBPII-RXRE-luciferase reporter gene, tk-CRBPII-Luc, and wild-type RXRaexpressing vector along with pRL-CMV (Renilla luciferase) as an internal standard using lipofection.Cells were then treated for 24 h with indicated concentrations of acyclic retinoid alone (curve 1), 9cRA alone (curve 2) or the combination of 1 mM acyclic retinoid plus indicated concentrations of 9cRA (curve 3).Cell lysates were prepared and luciferase activity in each cell lysate was measured and plotted as fold-induction compared with the activity of vehicle-treated cells after normalization with Renilla luciferase activity.Values are the mean AE SD (n 5).An asterisk and a dagger represent significant differences (P 5 0.05) from curves 1 and 2, respectively, at the indicated concentration.Fig. 4. Cooperativity between 9cRA and acyclic retinoid in growth suppression of HuH7 cells.After cells were cultured in 96-well plates and treated for 24 h with indicated concentrations of acyclic retinoid alone (curve 1), 9cRA alone (curve 2) or the combination of 1 mM acyclic retinoid plus indicated concentrations of 9cRA (curve 4), cell number was counted by Trypan Blue dye exclusion method.Some cells were transfected with a mutant (S260A) RXRa cDNA-expressing plasmid vector and then treated with indicated concentrations of 9cRA for 24 h (curve 3).Values are the mean AE SD (n 5).An asterisk represents significant differences (P 5 0.05) from curves 1 and 2 at the indicated concentration.
Figure 4, curve 3).These results suggest that prevention of phosphorylation of RXRa and thus restoration of its function by acyclic retinoid enhance the susceptibility of cancer cells to 9cRA, enabling growth suppression at physiological concentrations.This mechanism may provide a potential explanation for inhibition of the development of HCC by acyclic retinoid in vivo, although other possibilities remain.

Discussion
We here demonstrate a novel action of a synthetic retinoid, acyclic retinoid, to prevent phosphorylation of RXRa (Figure 1A) by inhibiting the activities of Ras-Raf-Erk 1/2 system (Figures 1B and 2A), and thereby to restore the function of RXRa in HCC cells (Figure 3) in addition to its previously established function as an RXR-ligand (7).9cRA, a natural RXRa-ligand, failed to suppress the RXRa phosphorylation (Figure 1A), probably because 9cRA inhibited both activated Ras (activating protein of Erk) and MKP-1 (inactivating enzyme of Erk) (Figure 2) at the same time, thereby leaving the Erk 1/2 activity unchanged (Figure 1B).Growth suppression by 9cRA was restored when HuH7 cells were either co-treated with acyclic retinoid or transfected with unphosphorylated mimic of RXRa (S260A RXRa; Figure 4).Thus, the dephosphorylating ability of acyclic retinoid may play a key role in its growth inhibitory effect on HCC (Figure 5).The concentration of the respective retinoid used in the present study (1 mM for acyclic retinoid and 10 nM for 9cRA) is within the physiological ranges (4,6,19), implying that acyclic retinoid and 9cRA may work cooperatively in vivo in the liver.
Retinoid receptors have anticancer properties (1,6) and loss of the receptors is largely involved in the carcinogenic process of various types of cancers (20±25).In particular, recent studies indicate that RXRa is required for apoptosis-induction and/or growth suppression of cancer cells via an autonomous RXR pathway (26,27) or by heterodimerization with other nuclear receptors (10,28).Thus, restoration of impaired RXRa activity may provide a key strategy to render HCC cells sensitive to RA and thus prevent cancer development (29).Solomon et al. originally suggested a similar strategy to use an Erk inhibitor (e.g.PD98059) to prevent phosphorylation of RXRa and restore its function as well as consequent vitamin D signaling via VDR/RXRa heterodimer, thereby reversing the resistance of cancer cells to 1,25-dihydroxyvitamin D 3 (30).The novel activities of acyclic retinoid as an Erk inhibitor may provide a clue in uncovering the mechanism of its specific antitumor activity against HCC.Although the ligand activity of acyclic retinoid is much weaker than that of 9cRA (Figure 3), the retinoid might work cooperatively with endogenous 9cRA in vivo and thereby suppress the growth of the tumor whose RXRa is phosphorylated.We have found recently that potential downstream genes inhibiting the cancer cell growth include STAT1 (31) and p21 (18).
The detailed molecular mechanism by which acyclic retinoid suppresses RXRa phosphorylation remains to be elucidated.Others have reported a similar effect of atRA in inhibiting Erk 1/2 activation induced by pro-angiogenic agents in the endothelial cells; however, the molecular mechanisms also remain obscure (32).Recently, Sah et al. (33) have reported that an RAR-selective retinoid inhibits EGF-induced Erk 1/2 activation and subsequent cell proliferation in a human Fig. 5. Schematic representation of the effect of acyclic retinoid on RXRa phosphorylation in HCC cells and its implication for specific antitumor activity of the retinoid.In normal hepatocytes, 9cRA binds to RXRa and transactivates downstream genes via RXRE that may regulate cell proliferation p.In HCC cells, Ras/MAP kinase pathway is highly activated q and phosphorylates RXRa at serine residue r, impairing functions of the receptor.Acyclic retinoid suppresses Ras/MAP kinase sequence, dephosphorylates RXRa, restores the function of the receptor, and subsequently inhibits the cancer cell growth s.Acyclic retinoid might inhibit Ras-Erk 1/2 system independently of RXRa s. uterine cervical cell line.In that study the retinoid seemed to act in an RAR-dependent manner because the effect is reversed by co-treatment with an RAR antagonist (33).In contrast, in the present study acyclic retinoid suppressed Erk activities even in HCC cells in which RXRa function is impaired, suggesting the possibility of RXRa-independent mechanism.In fact, our preliminary data suggest that the inhibitory effect of acyclic retinoid on Erk does not require de novo protein synthesis (Matsushima-Nishiwaki et al., unpublished observation), suggesting a dispensability of retinoid receptor-mediated transactivation, which often accompanies autoinduction of the receptors (20,22,25).Thus, acyclic retinoid might exert its antitumor activity independently of nuclear receptors, as has been reported with other synthetic retinoids including 4hydroxyphenyl retinamide (4HPR) and CD437 (34±40).Our previous study demonstrated that acyclic retinoid downregulates TGF-a expression in HuH7 cells, and that inclusion with TGF-a prevents acyclic retinoid-induced apoptosis in the cells (15).Because TGF-a is a ligand for EGF receptor, acyclic retinoid may directly interfere with the signal transducing sequence of TGF-a-EGF receptor-Ras-Raf-Erk system.
Phosphorylation has recently been shown to modulate the function of many nuclear receptors.The effect of phosphorylation on the function of each receptor is distinct depending upon the type of the receptor and/or kinases involved as well as upon the sites of phosphorylation.Phosphorylation of RXRa at Tyr249 by MAP kinase kinase-4 and by its downstream mediator c-Jun N-terminal kinase (JNK) has been reported, which results in the reduced transactivation via RAR/RXR complex (41).In contrast, others have reported that RXRa phosphorylation by JNK does not affect the transactivating properties of either RXRa homodimer or RXRa/RARa heterodimer (42).On the other hand, phosphorylation of RAR seems to enhance its transcriptional activities, irrespective of RAR subtypes and types of kinases (43±45).Much like RXR/RAR, controversy persists regarding the phosphorylation of PPAR, depending upon the experimental conditions (46±48).It is intriguing to see if acyclic retinoids affect phosphorylation of other nuclear receptors and their biological activities.
HCC has become one of the most frequent cancers in the world (49).HCC arises in patients with chronic viral hepatitis and cirrhosis at a high rate (3±5% annually) and is a major cause of morbidity and mortality in patients with advanced liver disease.Thus, prevention of HCC by targeting such a high-risk group is of great significance.We have successfully prevented the development of HCC with the use of acyclic retinoid in a clinical trial (4,5).In that study, we proposed a concept of `clonal deletion' therapy, in which the retinoid eradicates occult malignant cell clones under the detection limit of diagnostic images by inducing apoptosis (50).The present study may suggest a specific molecular mechanism by which acyclic retinoid renders HCC cell clones susceptible to the retinoid therapy and thereby removes the invisible cancer cells.

Fig. 2 .
Fig. 2. Effects of retinoids on the levels of activated Ras (A) and MKP-1 (B) in HuH7 cells.HuH7 cells were treated with vehicle (panels A and B, lanes and columns 1), 1 mM acyclic retinoid (lanes and columns 2), 1 mM 9cRA (lanes and columns 3) or their combination (panel B, lanes and column 4) for 24 h.(A) Levels of Ras bound to Raf-1.Ras was precipitated in the cell extract using Raf-1/Ras-binding domain-immobilized agarose and then subjected to western blot analysis using anti-Ras antibody.(B) Western blot analyses of MKP-1 and GAPDH were performed with the respective specific antibodies using the same extract as in panel A. (A) Relative intensity of the band was determined by densitometry (columns 1±3).Values are the mean AE SD (n 5).Double asterisks represent significant difference (P 5 0.01) compared to column 1. (B) Values represent the amounts of MKP-1 divided by those of GAPDH (the mean AE SD, n 5).An asterisk represents significant difference (P 5 0.05) compared with columns 1, 2 and 4.