| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Tumor Biology |
at Serine 260 Impairs Its Metabolism and Function in Human Hepatocellular Carcinoma1
First Department of Internal Medicine, Gifu University School of Medicine, Gifu 500-8705, Japan [R. M-N., M. O., S. A., T. S., K. A., H. M.]; Division of Liver Diseases, Mount Sinai Medical Center, New York, New York 10029-6574 [S. L. F.]; and Laboratory of Molecular Cell Sciences, Tsukuba Institute, RIKEN, Tsukuba 305-0074, Japan [S. K.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(RXR), although the underlying mechanisms and biological implications have remained unclear. The present studies were based on the finding that the accumulated full-length RXR was phosphorylated at serine/threonine residues both in all human HCC tissues examined and in human HCC-derived HuH7 cells. Phosphorylation at serine 260 of RXR, a consensus site of mitogen-activated protein kinase, was closely linked to its retarded degradation, low transactivating activity, and the promotion of cancer cell growth. There was no genomic mutation in the RXR gene, and abrogation of phosphorylation by mitogen-activated protein kinase-specific inhibitors restored the degradation of RXR in an RXR ligand-dependent manner. These results suggest that phosphorylation of RXR may interfere with its metabolism and signaling in human HCC, which could lead to growth promotion of these tumors. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Retinoids transduce their signals primarily through two families of nuclear receptors, the RARs and RXRs (8)
. These receptors are ligand-dependent transcription factors that bind to cis-acting DNA sequences, called RAREs and RXREs, located in the promoter region of their target genes (9)
. RARs bind to the RARE in response to both all-trans-RA and 9cRA, whereas RXRs bind and activate transcription in response to only 9cRA (8
, 9)
. Both RARs and RXRs consist of three subtypes, , ß, and 
, characterized by a modular domain structure (9)
.
We reported previously that hepatocarcinogenesis is also accompanied by aberrant metabolism of RXR (10)
, which is the most abundant retinoid receptor in the liver and highly expressed in both human HCC as well as in human HCC-derived HuH7 cells (10
, 11)
. We found accumulation of full-length RXR (Mr 54,000) in the HCC tumors compared with nontumorous surrounding tissues in which degraded RXR fragment (Mr 44,000) was predominant (10)
. We next suggested that the nonlysosomal, calcium-dependent cysteine protease, m-calpain, might participate in the proteolysis of RXR via cutting at the NH2-terminal A/B domain (12)
. On the other hand, the cleavage at NH2-terminal A/B domain of RXR by cathepsin L-type protease has also been reported in primary cultures of hepatocytes (13)
. However, the mechanisms underlying RXR degradation and its biological implications have remained unclear. Because no mutations in the RXR gene were identified either in surgically resected human HCC tissues or in HuH7 cells,4
we hypothesized that posttranscriptional modification of RXR could account for its delayed degradation in HCC cells.
Recently, phosphorylation of RXR by MAPK has been reported (14, 15, 16)
. Phosphorylation of RXR at serine 260, a consensus MAPK site, results in attenuation of ligand-dependent transactivation by the vitamin D3 receptor/RXR complex in human keratinocytes, where it may play an important role in their malignant transformation by releasing cells from vitamin D3-dependent growth suppression (14)
. Similarly, stress-induced phosphorylation of RXR by MAPK kinase-4 (MKK4/SEK1) results in suppression of retinoid signaling in African green monkey kidney (COS-7) cells (15)
. On the other hand, mouse RXR is phosphorylated by JNK1 and JNK2 when overexpressed in COS-1 cells (16)
. Furthermore, some substrates of m-calpain, such as c-Fos, become resistant to proteolysis after being phosphorylated (17)
. These observations led us to investigate the phosphorylation state of RXR and its potential relationship to proteolytic degradation and function of RXR in HCC cells.
We found that the Mr 54,000 full-length RXR protein was phosphorylated, at least in part, at serine 260 by MAPK and became resistant to degradation in surgically resected HCC as well as in HuH7 cells. Moreover, phosphorylated RXR lost its transactivation activity, which appears to promote cancer cell growth. These results suggest that malfunction of RXR because of phosphorylation may contribute to the unrestrained growth of HCC cells, probably by impairing its ability to induce genes required for the normal control of cell growth.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(
N 197), anti-RXRß (C-20), anti-RXR (Y-20), and anti-p34 cdc2 kinase (H-297) antibodies and polyclonal anti-phospho p44/p42 MAPK (ERK1/ERK2) antibody were from Santa Cruz Biotechnology, Inc. (Santa Cruz Biotechnology, CA) and Cell Signaling Technology (Beverly, MA), respectively. Rhodamine-labeled antirabbit IgG polyclonal antibody and fluorescence isothiocyanate-labeled anti-BrdUrd monoclonal antibody were from ICN Pharmaceuticals, Inc. (Aurora, OH) and Progen Biotechnik (Heidelberg, Germany), respectively. Ro25-7386 (RXR-selective agonist), Ch55 (pan-RAR-selective agonist), and constitutively active MEK1 cDNA were generous gifts from Drs. M. Klaus (F. Hoffmann-La Roche, Basel, Switzerland), K. Shudo (The University of Tokyo, Tokyo, Japan), and Dr. N. G. Ahn (Howard Hughes Medical Institute, University of Colorado, Boulder, CO; Ref. 18
), respectively.
Site-directed Mutagenesis in RXR.
Human RXR expression vector, pRShRXR (19)
, was kindly provided by Dr. R. M. Evans (The Salk Institute, La Jolla, CA). Human RXR mutants [threonine 81 and 82
alanine (T82A), serine 259 and 260alanine (S260A), and their combination (T82A/S260A) as well as threonine 81 and 82aspartate (T82D), serine 259 and 260aspartate (S260D), and their combination (T82D/S260D)] were constructed using the Mutan-Super Express Km site-directed mutagenesis kit and protocols supplied by TAKARA Biomedicals (Tokyo, Japan). Briefly, RXR cDNA was subcloned into the EcoRI site of pKF18k plasmid vector containing dual amber mutations on the kanamycin-resistant gene. PCR was performed using a combination of sense primers containing the desired mutations and an antisense primer to revert the amber mutations. Sense primers used were as follows (the alanine or aspartate mutations are underlined): T82A, 5'-CCACTCCATGTCGGTGCCCGCCGCACCCACCCTGGGCTTC-3'; S260A, 5'-GAACCCCGCCGCGCCGAACGACCCTGTCACCAACATTTGC-3'; T82D, 5'-CCACTCCATGTCGGTGCCCGACGACCCCACCCTGGGCTTC-3'; and S260D, 5'-GAACCCCGACGACCCGAACGACCCTGTCACCAACATTTGC-3'. With this PCR, amplified mutagenic-selection DNA functioned as a PCR primer, yielding nicked double-stranded plasmid. When this nicked DNA was transformed into Escherichia coli MV1184, the nick was repaired, and only transformants containing a desired site-specific mutation were grown in the presence of kanamycin. T82A/S260A and T82D/S260D double mutants were constructed through exchange of the ScaI-BssHII fragment of S260A or S260D mutant with that of T82A or T82D mutant, respectively. The A/B mutant lacking the A/B domain was constructed by PCR, using sense primer 5'-AAAGAATTCGACATGACCAAGCACATCTGCGCCATC-3' and antisense primer 5'-AAAGAATTCGCAGACATGGACACCAAACAT-3'. Sequences of mutants were verified by sequencing. The mutant hRXR cDNAs were cloned into the EcoRI-XhoI site of pcDNA3.1/Myc-His(+) mammalian expression vector (Invitrogen Corp., Carlsbad, CA), which contains an oligonucleotide encoding a polyhistidine (His6) metal-binding peptide at the COOH terminus of the desired protein.
Tissue Specimens.
HCC and its surrounding noncancerous tissues were obtained by surgical resection from 10 patients infected with hepatitis viruses B (3 cases) or C (7 cases). In all of the cases, tumor and its surrounding tissues were classified histologically as moderately differentiated HCC and liver cirrhosis, respectively. A histologically normal liver specimen was obtained from a patient with hypercitrullinemia who underwent liver transplantation. The study was approved by Gifu University School of Medicine Ethics Committee, and all of the patients gave informed consent.
Cell Culture and Treatment.
HuH7 cells were obtained from the Japanese Cancer Research Resources Bank (Tokyo, Japan) and maintained in RPMI 1640 (Life Technologies, Inc., Gaithersburg, MD) supplemented with 1% FCS (Life Technologies, Inc.) and 0.2% lactoalbumin hydrolysate (Sigma Chemical Co.). Hc cells, normal human hepatocytes, were purchased from Applied Cell Biology Research Institute (Kirkland, WA) and maintained in the attached CS-C complete medium. Hc cells were synchronized in G0-G1 phase by incubating with isoleucine-free RPMI 1640 containing 3% dialyzed FCS for 48 h prior to treatments or transfections. HuH7 and Hc cells were treated for 24 h in 1% FCS and 0.2% lactoalbumin-containing RPMI 1640 and 3% FCS-containing isoleucine-free RPMI 1640, respectively, with the indicated concentrations of chemicals dissolved in ethanol at the final concentration of 0.05%.
Preparation of Nuclear Extracts and Western Blot Analysis.
Preparation of nuclear extracts and whole cell extracts and Western blot analyses of RXR, RXRß, RXR, and ERK as well as p34 cdc2 kinase as an internal standard of nuclear proteins (20)
were performed as described previously (10)
. For detection of phosphorylated RXR, RXR was affinity purified from tissue nuclear extracts using anti-RXR antibody-immobilized Hi-Trap NHS-activated Sepharose beads and then subjected to Western blot analyses using anti-phosphoserine or anti-phosphothreonine antibodies. As the control, proteins precipitated with nonimmunized rabbit antibody-cross-linked beads were analyzed. For detection of His6-tagged mutant RXRs overexpressed in HuH7 and Hc cells, mutant proteins were purified from whole cell extracts using the Ni-NTA agarose column (Qiagen, Hilden, Germany) and subjected to the Western blot analyses. Protein concentrations in the samples were determined by Bio-Rad (Hercules, CA) protein assay kit. The signals were detected with an Amersham-Pharmacia ECL system. Densitometric analysis was performed using the NIH image version 1.61 software.
Metabolic Labeling of RXR and Its Detection.
After phosphate starvation for 24 h, cells were incubated for another 24 h with 32Pi (1 mCi/ml of the medium per 75-cm2 flask). Nuclear extracts were prepared, and RXR protein in each nuclear extract was immunoprecipitated as described previously (10)
. Immunoprecipitated RXR was electrophoresed in 8% polyacrylamide gels in the presence of SDS, and the radioactive bands were detected by an autoradiography performed using a Fujix BAS 2,500 Bio-imaging analyzer (Fuji Photo-Film, Tokyo, Japan).
Transfection with RXR and MEK1 cDNAs.
Transfection was performed by either electroporation with Gene-Pulser (Bio-Rad) or Lipofectamine Plus reagent (Life Technologies, Inc.) as described in previous reports (10
, 21)
.
Luciferase Reporter Assay.
A reporter plasmid, tk-CRBPII-Luc, which contains multiple copies of the DR1 type RXRE sequences within rat CRBPII promoter upstream of luciferase cDNA (19)
, was kindly provided by the late Dr. K. Umesono (Kyoto University, Kyoto, Japan). Cells were cotransfected with a combination of wild-type or a mutant hRXR-expressing plasmid (300 ng/35-mm dish) and tk-CRBPII-Luc reporter (750 ng/35-mm dish), along with pRL-CMV (Renilla luciferase, 100 ng/35-mm dish; Promega Corp., Madison, WI) as an internal standard to normalize transfection efficiency (21)
. Twenty-four h after transfection, the cells were washed with 137 mM NaCl, 2.7 mM KCl, 9 mM PO4-, pH 7.4 (PBS) and treated with 1 µM 9cRA for an additional 24 h in either 1% FCS and 0.2% lactoalbumin-containing RPMI 1640 (HuH7 cells) or 3% FCS-containing isoleucine-free RPMI 1640 (Hc cells). Thereafter, cell lysates were prepared, and luciferase activity of each cell lysate was measured using the LumiCount microplate luminometer (Packard Instrument Co., Meriden, CT). Changes in firefly luciferase activity were calculated and plotted after normalization with changes in Renilla luciferase activity in the same sample.
Immunocytochemistry.
After 1-h incubation with 20 µM BrdUrd, cells were fixed with 10% formalin in PBS for 10 min at room temperature, washed twice with PBS, and permeabilized with 0.3% Triton X-100 in PBS for 10 min at room temperature. Cells were treated with 1 N HCl for 1 h at room temperature and neutralized with 0.1 M borate buffer (pH 8.5). After blocking with nonimmune rabbit IgG (final, 100 µg/ml) for 1 h, samples were incubated with anti-RXR antibodies (N 197; final, 1:50 dilution) overnight at 4°C. After washing with PBS-0.1% Tween 20, the cells were incubated at room temperature with rhodamine-labeled second antibody (1:100) for 90 min and successively with fluorescence isothiocyanate-labeled anti-BrdUrd antibody (1:20) for 30 min. Fluorescent images were obtained by a Nikon Eclipse E800 laser microscope and analyzed using Fuji-S Fuji photofolder and Adobe Photoshop version 5.0.
Statistical Analysis.
Data are expressed as the means ± SD. Statistical significance was assessed with one-way ANOVA, followed by Sheffe’s t test.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
in Human HCC Tissues.
shows the result of Western blot analyses of nuclear extracts prepared from HCC, its noncancerous surrounding tissue, and normal liver with an anti-RXR (N 197) antibody that recognizes a ligand binding sequence in the E-domain and thus reacts with both full-length Mr 54,000 and fragmented Mr 47,000 and Mr 44,000 RXRs (10)
. Although N 197 could recognize all three subtypes of RXR, Western blotting of the same sample with specific antibody to RXRß and RXR did not reveal any bands (data not shown), suggesting that the visualized bands represented RXR. A major species at Mr 54,000 plus minor Mr 47,000 and Mr 44,000 species were detected in HCC tissues (Fig. 1A
, Lane 1). In contrast, in noncancerous surrounding and normal liver tissues, Mr 54,000 bands were faintly (Fig. 1A
, Lane 2) and barely (Fig. 1A
, Lane 3) detected, respectively, whereas fragmented Mr 44,000 bands were predominant in these tissues. A control experiment using anti-p34 cdc2 kinase antibody confirmed a constant amount of nuclear proteins among samples (data not shown). Similar observations were made in all 10 cases examined (Fig.1D)
, suggesting that the accumulation of full-length RXR was limited to cancerous liver tissues.
|
was identified with both anti-phosphoserine (Fig. 1B)
and anti-phosphothreonine (Fig. 1C)
antibodies after immunoprecipitation with N 197-immobilized beads. Strong bands were expressed in tumor tissues only at Mr 54,000 but not at Mr 47,000 or Mr 44,000 (Fig. 1
, B and C, Lanes 1), indicating that full-length RXR was exclusively phosphorylated at both serine and threonine residues in human HCC. No band was detected when proteins precipitated with nonimmunized rabbit antibody-immobilized beads were analyzed (data not shown). Much weaker expression was seen in noncancerous adjacent tissues (Fig. 1
, B and C, Lanes 2), and the protein was barely detectable in a histologically normal liver sample (Fig. 1
, B and C, Lanes 3). Phosphorylated bands were not detected with anti-phosphotyrosine antibody (PY-20; Wako Pure Chemicals; data not shown). Similar observations were made in 10 cases examined (Fig. 1
, E and F), suggesting a potential link between retarded degradation and phosphorylation of RXR in human HCC.
Accumulation of Phosphorylated Full-length RXR in Human HCC Cells and Its Degradation by RXR-selective Ligands.
To study the relationship between phosphorylation and degradation of RXR as well as its biological implications, we used cultures of a human HCC cell line, HuH7 cells, and normal human hepatocytes, Hc cells. We first tested whether in vivo results were reproduced in cultured cells. By Western blot (Fig. 2A)
, HuH7 expressed RXR, but not RXRß or RXR (Fig. 2
A, Lanes 1–3), most of which was the Mr 54,000 full-length form (Fig. 2A
, Lane 1), whereas the majority of RXR in Hc cells was the Mr 44,000 truncated form (Fig. 2A
, Lane 5), establishing the utility of HuH7 and Hc cells as representatives of clinical HCC and normal liver tissues, respectively. HuH7 cells transfected with plasmid encoding a full-length RXR cDNA expressed an increased amount of Mr 54,000 RXR with reduced Mr 47,000 fragment (Fig. 2A
, Lane 4), whereas Hc cells expressed only an increased amount of Mr 44,000 fragmented RXR (Fig. 2A
, Lane 6), suggesting that RXR mRNA was alternatively spliced in Hc cells and/or that RXR protein translated from the same mRNA was differentially metabolized between the two cell types. The latter idea was supported by the in vivo data that only phosphorylated RXR escaped degradation. We did not detect any mutations in the full-length RXRcDNA derived from either human HCC tissues or HuH7 cells (data not shown).
|
was constitutively phosphorylated, we performed metabolic labeling experiments. After treatment of the cells under the indicated conditions, half of each culture was radiolabeled, and phosphorylated RXR was immunoprecipitated and detected by SDS-PAGE/autoradiography (Fig. 2B)
, and the other half was used to monitor the levels of both Mr 54,000 and Mr 44,000 RXRs by Western blot analyses (Fig. 2C)
. As seen in Fig. 2B
, Lane 1 (inset), Mr 54,000 full-length RXR was phosphorylated. Radioactive bands were detected only at Mr 54,000 but not at Mr 44,000, suggesting that the Mr 44,000 truncated RXR peptide was minimally or not phosphorylated. Next, a correlation between phosphorylation and degradation of RXR was assessed by inhibiting phosphorylation with a broad-spectrum protein kinase inhibitor, STS, and/or a ligand for RXR, 9cRA, which has been reported to attenuate phosphorylation via induction of phosphatase activity (22)
. The amounts of both Mr 54,000 and Mr 44,000 RXR were almost unchanged when phosphorylation was inhibited by 30% with STS alone (Fig. 2
, B and C, Column 2 and Lane 2, each inset). 9cRA alone did not affect either phosphorylation or degradation (Fig. 2
, B and C, Column 3 and Lane 3, each inset). However, a combination of these reduced phosphorylation by 70% and reduced the amount of Mr 54,000 RXR by 50%, with emergence of the Mr 44,000 fragment (Fig. 2
, B and C, compare Column 4 and Lane 4, each inset), suggesting that a combination of dephosphorylation and ligand-binding may induce degradation of RXR. In contrast, Hc cells did not contain any detectable levels of phosphorylated RXR (Fig. 2B
, Column 5 and Lane 5, inset) and expressed mainly Mr 44,000 RXR (Fig. 2C
, Column 5 and Lane 5, inset).
Next, we examined the ligand specificity of RXR degradation using receptor-selective retinoids, because 9cRA can bind to both RXR and RAR. As shown in Fig. 2D
, both 9cRA (pan-RAR and RXR agonist) and Ro25-7386 (RXR-selective agonist) enhanced RXR degradation in the presence of STS (Fig. 2D
, Columns 3 and 5), but not in the presence of Ch55 (pan-RAR-selective agonist, Column 7). 9cRA induced the degradation of RXR in a dose-dependent manner (ED50, 60 nM) in the presence of a low concentration (50 nM) of STS, suggesting that this effect is physiologically relevant (data not shown).
Involvement of MAPK in the Reduced Degradation of RXR.
Because STS suppresses several kinase activities (23)
, we examined the effect of kinase-selective inhibitors to identify which protein kinase(s) might be involved. HuH7 cells were treated with a combination of 9cRA and one of four different protein kinase inhibitors, and the amount of Mr 54,000 RXR was assessed semiquantitatively by Western blotting. As shown in Fig. 3A
, degradation was induced to almost the same extent by STS (Column 3) as well as SB203580 (p38 MAPK inhibitor; Column 5) and PD98059 (MEK inhibitor; Column 7) in the presence of 9cRA. PD98059 also induced the degradation in the absence of 9cRA (Column 6). However, KN-93 (calmodulin kinase II inhibitor; Column 9) did not induce the degradation even at a concentration as high as 10 µM. These results suggest that the current system in HuH7 cells may be controlled, at least, by the MAPKs. In support of this conclusion, HuH7 cells expressed high levels of phosphorylated (activated) p42 MAPK (ERK2), whereas Hc normal hepatocytes did not (Fig. 3B
, compare Lanes 1 and 2). When Hc cells were transiently transfected with constitutively active MEK1 cDNA, they expressed high levels of ERK1/ERK2 (Fig. 3B
, Lane 3), and the degradation of RXR was attenuated (Fig. 3B,
compare Lanes 4 and 5).
|
Degradation by Phosphorylation at Serine 260. phosphorylation and its reduced degradation, we constructed plasmids encoding mutant RXRs in which threonine 82 and/or serine 260, the putative phosphorylation sites by MAPKs (14
, 16)
, were mutated to either alanine (unphosphorylated form) or aspartate (phosphomimic). HuH7 and Hc cells were transfected, and the amount of full-length Mr 54,000 RXR was semiquantitated by Western blotting after treatment with 9cRA, which was required for degradation. The antigenicities of these mutant RXRs were similar to the wild-type RXR (data not shown). As depicted in Fig. 4A
, whereas native and overexpressed wild-type RXRs were not degraded in HuH7 cells (Columns 2 and 4, respectively), the alanine 260-mutant (S260A) as well as alanine 82-mutant/alanine 260-mutant (T82A/S260A) mimicked the unphosphorylated type and restored the degradation after treatment with 9cRA (Columns 8 and 10, respectively). Mutation of threonine 82 to alanine (T82A) did not affect degradation (Fig. 4A
, Column 6). A similar result was obtained when protein synthesis was inhibited by treatment with cycloheximide during incubation with 9cRA (Fig. 4B)
, confirming the degradation of S260A mutant RXR. Because comparable expression was observed between wild-type and S260A mutant RXR (data not shown), possible interference with transcription and/or translation of the mutant genes was not likely. These results suggested that phosphorylation at serine 260 by MAPKs may render RXR resistant to degradation, whereas phosphorylation at threonine 82 may not be an important determinant of RXR degradation. Mutant RXRs with aspartate 82 and/or aspartate 260 (T82D, S260D, and T82D/S260D) mimicked the phosphorylated type and did not degrade into smaller peptides in HuH7 cells when overexpressed (data not shown).
|
s were overexpressed in Hc cells, these phosphorylated RXR mimics were found to be resistant to proteolysis (Fig. 4C
, Columns 4 and 5, respectively). On the other hand, both wild-type and T82D mutant RXR decayed (Columns 2 and 3, respectively), as in the cases of T82A, S260A, and T82A/S260A mutant RXRs (data not shown). A similar result was also obtained in the presence of cycloheximide (Fig. 4D)
.
Inactivation of RXR by Phosphorylation at Threonine 82 and Serine 260.
We assessed the effect of phosphorylation as well as degradation of RXR on its transactivation activity via RXRE using the RXR-responsive reporter, CRBPII-RXRE-luciferase (Fig. 5)
. Hc cells expressed 
2-fold higher transactivation activity than HuH7 cells. 9cRA treatment enhanced the activity 2-fold in both cell types (Fig. 5, A
and B, Columns 2), suggesting that Hc cells possess 2-fold higher ligand-dependent transactivation activity than HuH7 cells. Overexpression of wild-type RXR increased the ligand-dependent activity in both cell types. However, the effect was more potent in Hc cells (7-fold increase; Fig. 5B
, Column 4) than in HuH7 cells (5-fold increase; Fig. 5A
, Column 4), implying that phosphorylation of RXR in HuH7 cells might interfere with its transactivation activity. Overexpression of T82A mutant RXR in HuH7 cells resulted in a further (6-fold) potentiation (Fig. 5A
, Column 6), whereas overexpression of T82D mutant RXR in Hc cells led to a significantly smaller (2.5-fold) increase (Fig. 5B
, Column 6), implying that phosphorylation of threonine 82 might reduce ligand-dependent transactivation activity. On the other hand, overexpression of S260A or T82A/S260A mutant RXR in HuH7 cells resulted in 3–3.5-fold increases (Fig. 5A
, Columns 8 and 10, respectively), whereas overexpression of S260D and T82D/S260D mutant RXRs in Hc cells suppressed ligand-dependent activities almost completely (Fig. 5B
, Columns 8 and 10, respectively). Overexpression of T82D, S260D, and T82D/S260D mutant RXRs in HuH7 cells did not enhance ligand-dependent transactivation activity, whereas overexpression of T82A, S260A, and T82A/S260A mutant RXRs in Hc cells resulted in almost similar levels of ligand-dependent transactivation activity, as detected with wild-type RXR (data not shown). Columns 11 and 12 in Fig. 5A
display the result when the A/B mutant lacking the A/B domain, a mimic of Mr 44,000 protein, was overexpressed in HuH7 cells. The truncated RXR did not display ligand-induced transactivation activity. Similar results were obtained using a luciferase reporter gene, DR-5 Luc (21)
, the transactivation of which is driven via RARE by the RAR/RXR heterodimer (data not shown).
|
. phosphorylation on cellular proliferation by determining the BrdUrd labeling index (Fig. 6)
. HuH7 and Hc cells were transfected with wild-type or various mutant RXR cDNAs and, after treatment with 9cRA, BrdUrd labeling indexes were determined. Because transfection efficiency was relatively low (approximately 5–7% for each cDNA in both types of cells), there was concern that the effect of overexpressed RXR mutants would be diluted by the influence of high levels of endogenous RXR expression in HuH7. To surmount this problem, we performed BrdUrd and RXR double staining and estimated the proliferation only of transfected RXR-overexpressing cells. We analyzed RXR-transfected cells, where nuclear RXR was more than twice as great as endogenous RXR in corresponding empty vector-transfected cells, and then determined the ratio between the number of BrdUrd double-positive cells against the total RXR-positive cell number. In HuH7 cells (Fig. 6A)
, overexpression of T82A (Column 2), S260A (Column 3), or T82A/S260A (Column 4) mutant RXRs resulted in a significant (22–44%) reduction in the BrdUrd-positive cell numbers compared with wild-type RXR transfected cells (Column 1). In contrast, in Hc cells (Fig. 6B)
, overexpression of biologically inactive T82D (Column 2), S260D (Column 3), and T82D/S260D (Column 4) mutant RXRs significantly increased the number of BrdUrd-positive cells.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
phosphorylation in human HCC. RXR was phosphorylated in human HCC tissue samples (Fig. 1)
as well as in a human HCC cell line (Fig. 2B)
. Phosphorylated RXR was resistant to proteolytic degradation (Figs. 1
, 2
, and 4
), lacked transactivation activity (Fig. 5)
, and exerted dominant-negative inhibition (Fig. 5)
, which was likely linked to the promotion of HCC cell growth (Fig. 6)
. These findings are summarized in Fig. 7
. These changes was triggered by MAPK induction in HCC cells (Fig. 3)
, as has been reported in vivo (24)
, and not attributable to a mutation in the RXR gene (data not shown). Only full-length RXR (Mr 54,000) but not the truncated receptor (Mr 44,000) was phosphorylated (Figs. 1
and 2
) at serine 260 (Figs. 3
and 4
). Dephosphorylation at this site by site-directed mutagenesis restored the degradation (Fig. 4)
as well as transactivation activities (Fig. 5)
and suppressed growth of the cancer cells (Fig. 6)
, particularly in the presence of the ligand. Suppression of cell growth was also observed when phosphorylation was modified. Accumulation of full-length RXR may represent the presence of the phosphorylated inactive receptor in HCC.
|
(Figs. 1
and 2B
, respectively), probably because normal hepatocytes did not have detectable levels of active MAPKs (Fig. 3B)
. When active MAPKs were induced in Hc cells by transfecting constitutively active MEK1 cDNA, RXR degradation was inhibited, as has been observed for HuH7 cells (Fig. 3B)
, suggesting that MAPKs play a role in phosphorylation and retarded degradation of RXR. Although Mr 54,000 RXR in noncancerous surrounding liver was slightly phosphorylated (Fig. 1)
, this might be explained by the presence of cirrhosis in these tissues, in which hepatocytes proliferate in a regenerative response to inflammation and necrosis. Similarly, normal Hc hepatocytes in growth phase also expressed a slight amount of phosphorylated Mr 54,000 RXR (data not shown). Full-length Mr 54,000 RXR overexpressed in Hc cells was detected in the experiment in Fig. 4C
, but not in Fig. 2A
, for unclear reasons. However, this might be related to the difference in the method of transfection [electroporation (Fig. 2A)
versus lipofection (Fig. 4C)
].
As shown in Fig. 5A
, if transactivation activity in HuH7 cells is solely regulated by phosphorylation state at both threonine 82 and serine 260, all T82A, S260A, and T82A/S260A mutants should have higher activity compared with wild-type receptor. However, only the T82A mutant showed higher activity than the wild-type RXR. The A/B mutant, which mimics the cleaved Mr 44,000 form of RXR (12
, 13)
, loses ligand-induced transactivation activity, consistent with the previous report (25)
. Accordingly, we speculate that loss of transactivation activity upon cleavage of the A/B domain explains why the S260A and T82A/S260A mutants did not enhance transactivation activity more than the wild-type. In supporting this hypothesis, the T82A mutant showed the highest activity, i.e., that the T82A mutant restored the transactivating activity due to dephosphorylation at threonine 82, which was sustained due to its relative resistance to proteolysis, whereas the S260A mutant restored the transactivating activity to a lesser extent, which decayed as a result of its degradation to the A/B domain truncated form. In contrast, Hc cells possess high transactivation activity of RXR, although most of it exists as a degraded form in this cell type. Therefore, it might be possible that phosphorylation and degradation of RXR and their effects on transactivation activity may be cell type context dependent.
Controversial results have been reported regarding biological consequences of phosphorylation of RXR. Treatment of COS and CV1 cells with a protein phosphatase inhibitor leads to an increase in transactivation of RXR-responsive reporter genes (26
, 27)
. On the other hand, hyperphosphorylation by JNKs does not affect the transactivation properties of either RXR homodimers or RXR/RAR heterodimers in COS-1 cells (16)
. In contrast, phosphorylation of human RXR at serine 260 can interfere with the vitamin D3 receptor/RXR-dependent as well as RXR/RXR-dependent signaling pathways and thereby confer a hormone-resistant growth advantage in ras-transformed human keratinocytes (14)
. Repression in the activity of phosphorylated RXR by MKK4/SEK1 and cAMP-dependent protein kinase has also been reported in COS cells (15
, 28)
. Thus, regulation of RXR function by phosphorylation appears to be cell context dependent. Most steroid and nuclear receptors are phosphoproteins (17
, 29)
. It is important to assess the effect of the phosphorylation of RXR on transactivation through other nuclear hormone receptors/RXR heterodimers. We found that overexpression of the S260D mutant in Hc cells inhibited transactivation through an RARE and peroxisome proliferator-activated receptor-responsive element (data not shown). Because RXR forms heterodimers with other nuclear receptors, alterations in the phosphorylation of RXR may affect other nuclear receptor pathways as well. Moreover, it will be intriguing to determine whether other members of nuclear receptors might be phosphorylated in HCC tissues.
Divergent biological consequences of RXR signaling might be also dependent on which amino acids are phosphorylated. We investigated here the influence of phosphorylation at threonine 82 and/or serine 260, the putative consensus sites of the MAPK family (14
, 16)
. We cannot rule out the possibility that phosphorylation at other sites might also affect transactivation activity and cell growth. It has been reported that in addition to threonine 82 (threonine 87 in mouse) and serine 260 (serine 265 in mouse), mouse RXR is phosphorylated by JNK1 and JNK2 at serine 61 (serine 56 in human) and serine 75 (serine 70 in human; Ref. 16
). Furthermore, serine 27 can be phosphorylated by a cAMP-dependent protein kinase (28)
. We are now investigating whether phosphorylation at these amino acid residues might affect degradation and function of RXR.
We have demonstrated that RXR ligands are required for dephosphorylation (Fig. 2B)
and/or degradation (Figs. 2
3
4
) of RXR as well as for its transactivation activity (Fig. 5)
. In normal bronchial epithelial cells, RA inhibits both JNK and ERK1/ERK2 activities by inducing the dual-specificity MAPK phosphatase 1 (22)
. Inhibition of phosphorylation with kinase inhibitors might restore the function of RXR in part and thereby, in the presence of the ligand, stimulate the expression of MAPK phosphatase 1, which would in turn inhibit phosphorylation of the receptor. The results in Fig. 4D
underscore the importance of ligand binding to RXR for its metabolism. On the basis of a previous study (30)
, ligand-binding to RXR promotes its degradation via the ubiquitin-proteasome pathway. Similarly, we found recently that S260A and T82A/S260A mutant RXRs undergo rapid ubiquitination and subsequent degradation by the proteasome in HuH7 cells upon 9cRA-treatment,5
implying that phosphorylation at serine 260 might interfere with ubiquitination and thereby prevent proteasomal degradation. Finally, ligand binding is essential for recruiting coactivators and exerting transactivation activity (8
, 9)
. Thus, RXR ligands appear to function indirectly to inhibit kinases, thereby promoting degradation and transactivation. It will also be important to assess the effect of phosphorylation of RXR on homo- and heterodimerization.
Despite recent advances in diagnosis and treatment of HCC, the 5-year survival rate barely reaches 50% because of the high incidence of second primary tumors (31)
. The impaired function of phosphorylated RXR and its improvement by RXR ligands may partly explain the close relationship between hepatocarcinogenesis and a reduction in hepatic retinoid content at an early phase of hepatocarcinogenesis (5)
. Restoration of RXR function may therefore be a potential approach to treating HCC. In this regard, we have demonstrated a chemopreventive effect of a retinoid analogue, a ligand to RXR, on the occurrence of second primary HCC (6
, 7)
. This compound, named acyclic retinoid, induced apoptosis in HuH7 cells (32)
via induction of an apoptosis-inducing enzyme, tissue transglutaminase (33, 34, 35, 36)
,6
the transcription of which is governed by RAR-RXR heterodimers (37)
. Recently, we have found that the acyclic retinoid restored the activity of RXR by inducing dephosphorylation of the receptor, even in the absence of kinase inhibitor as well as by functioning as a ligand in HuH7 cells.7
This might well be linked to its apoptotic effect via tissue transglutaminase in HCC cells.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported in part by Grants-in-Aid 12670472 (to M. O.), 13670579 (to S. K.), and 10557055 (to H. M.), and Special Coordination Funds for Promoting Science and Technology (to S. K.) from the Ministry of Education, Culture, Sports, Science and Technology; Haraguchi Memorial Cancer Research Fund (to R. M-N.); grants from the Foundation for Total Health Promotion (to M. O.); Grant RO137340 from the National Institute of Diabetes and Digestive and Kidney Diseases (to S. L. F.); and grants for the Multibioprobe Research Program from RIKEN (to S. K.). 
2 To whom requests for reprints should be addressed, at Laboratory of Molecular Cell Sciences, Tsukuba Institute, RIKEN, Koyadai, Tsukuba, Ibaraki 305-0074, Japan. Phone: 81-298-36-3522; Fax: 81-298-36-9050; E-mail: kojima{at}rtc.riken.go.jp
3 The abbreviations used are: HCC, hepatocellular carcinoma; RA, retinoic acid; RAR, RA receptor; RARE, RAR responsive element; RXR, retinoid X receptor; RXRE, RXR responsive element; 9cRA, 9-cis-RA; MAPK, mitogen-activated protein kinase; JNK, c-Jun-NH2-terminal kinase; STS, staurosporine; BrdUrd, 5-2'-bromodeoxyuridine; ERK, extracellular signal-regulated kinase; MEK1, mitogen-activated ERK activating kinase 1; CRBPII, cellular retinol-binding protein type II.
4 S. Adachi, unpublished data.
5 S. Adachi, unpublished observations.
6 T. Sano and S. Kojima, Induction of transglutaminase-induced apoptosis by acyclic retinoid in human hepatocellular carcinoma, manuscript in preparation.
7 R. Matsushima-Nishiwaki, unpublished observations.
Received 3/26/01. Accepted 8/ 9/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
by cathepsin L-type protease is a potential mechanism for modulating thyroid hormone action. J. Biol. Chem., 273: 33166-33173, 1998.. J. Clin. Investig., 103: 1729-1735, 1999.[Medline]
by activated c-Jun NH2-terminal kinases. J. Biol. Chem., 274: 18932-18941, 1999. residue, phosphorylated by protein kinase A is essential for cyclic AMP-mediated downregulation of RXR function. Biochem. Biophys. Res. Commun., 279: 853-857, 2000.[Medline]
) phosphorylates RXR
), which renders the receptor resistant to proteolysis (
). Phosphorylation also impairs its transactivation activity (
), which may be linked to the uncontrolled growth of the cancer cells (
).