NF-κB Activates Transcription of the RNA-Binding Factor HuR, via PI3K-AKT Signaling, to Promote Gastric Tumorigenesis

GASTROENTEROLOGY 2008;135:2030 –2042

NF-␬B Activates Transcription of the RNA-Binding Factor HuR, via PI3K-AKT Signaling, to Promote Gastric Tumorigenesis MIN–JU KANG,* BYUNG–KYU RYU,* MIN–GOO LEE,* JIKHYON HAN,* JIN–HEE LEE,* TAE–KYU HA,* DO–SUN BYUN,* KWON–SEOK CHAE,‡ BONG–HEE LEE,* HYANG SOOK CHUN,§ KIL YEON LEE,储 HYO–JONG KIM,¶ and SUNG–GIL CHI* *School of Life Sciences and Biotechnology, Korea University, Seoul; ‡Department of Biology Education, Teacher’s College, Kyungpook National University, Daegu; § Food Safety Research Division, Korea Food Research Institute, Kyonggi-do; and Departments of 储General Surgery and ¶Internal Medicine, School of Medicine, Kyung Hee University, Seoul, Korea

BASIC– ALIMENTARY TRACT

Background & Aims: HuR is a RNA-binding factor whose expression is commonly upregulated in some human tumor types. We explored the molecular mechanism underlying HuR elevation and its role in gastric cancer tumorigenesis. Methods: HuR expression and subcellular localization were determined by polymerase chain reaction, immunoblot, and immunohistochemical analyses. Its effect on tumor growth was characterized using flow cytometry, terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling, and soft agar analyses. Luciferase reporter, chromatin immunoprecipitation, and electrophoretic mobility shift assays were used to measure transcriptional activation by nuclear factor ␬B (NF-␬B) signaling. Results: Compared with normal gastric tissues, HuR was expressed at higher levels in gastric tumors, particularly in advanced versus early tumors; this increase was associated with enhanced cytoplasmic translocation of HuR. HuR overexpression increased proliferation of tumor cells, activating the G1 to S transition of the cell cycle, DNA synthesis, and anchorage-independent growth. Small interfering RNA–mediated knockdown of HuR expression reduced tumor cell proliferation and response to apoptotic stimuli. No genetic or epigenetic alterations of HuR were observed in gastric tumor cell lines or primary tumors; overexpression depended on phosphatidylinositol 3-kinase/AKT signaling and NF-␬B activity. AKT activation increased p65/RelA binding to a putative NF-␬B binding site in the HuR promoter, the stability of HuR target transcripts, and the cytoplasmic import of HuR. Conclusions: HuR is a direct transcription target of NF-␬B; its activation in gastric cancer cell lines depends on phosphatidylinositol 3-kinase/AKT signaling. HuR activation by this pathway has proliferative and antiapoptotic effects on gastric cancer cells.

P

osttranscriptional regulation is emerging as an important aspect of differential gene expression in neoplastic versus normal tissues. In particular, modulation of messenger RNA (mRNA) stability can have a significant impact on mRNA abundance and subsequent pro-

tein expression. Many growth factors and cytokines that potentially confer a growth advantage to the neoplastic cells are regulated at the level of mRNA stability. This regulation involves adenylate uridylate–rich elements (AREs) located in the 3= untranslated region of many short-lived mRNAs, and there has been growing interest in a particular pathway that regulates ARE-mediated mRNA stability.1 While the precise mechanisms that underlie mRNA turnover remain to be elucidated, it has been well documented that an interaction of a particular cis-acting ARE within the 3= untranslated region of transcripts with specific trans-acting RNA-binding factors either enhances or reduces mRNA stability.1 AREs, which most often reside in 3= untranslated region and usually possess single or multiple copies of the AUUUA pentamer or UUAUUUA(U/A)(U/A) nonamer, act as a repressor of gene expression by directing deadenylation followed by rapid degradation of mRNA.1,2 A number of trans-acting RNA-binding factors have been described, but only a few of them, including the Elav/Hu and hnRNP/AUF1 protein family, have been shown to definitively regulate mRNA stability.3 The Elav/Hu family stabilizes target mRNAs and enhances their translation, whereas AUF1 appears to be associated with enhanced mRNA turnover.3–5 These proteins undergo bidirectional nuclear-cytoplasmic shuttling, with mRNA stabilizing or destabilizing activity presumed to occur in the cytoplasm, possibly in association with translational machinery.6 HuR, a member of the Elav/Hu family, has been identified to bind target mRNA subsets bearing AREs through its RNA recognition motifs and directly implicated in regulating various cellular responses, including cell proliferation, differentiation, inflammation, replicative senescence, immune cell activation, and oncogenesis.7–10 Abbreviations used in this paper: ARE, adenylate uridylate–rich elements; FBS, fetal bovine serum; 5-FU, 5-fluorouracil; IGF, insulin-like growth factor; NF-␬B, nuclear factor-␬B; PI3K, phosphatidylinositol 3-kinase; PCR, polymerase chain reaction; RT, reverse transcription; siRNA, short interfering RNA. © 2008 by the AGA Institute 0016-5085/08/$34.00 doi:10.1053/j.gastro.2008.08.009

Accumulating evidence suggests that mRNA-binding proteins of the Elav/Hu and AUF1 family are important effectors in the pathophysiology of human cancers.11 HuR expression is aberrantly elevated in some human malignancies, including colon, brain, and ovary tumors, and correlates with advancing stages of malignancy, supporting a role for HuR in tumorigenesis.12–14 HuR is strongly expressed in highly proliferative tumors, including glioblastoma multiforme, in contrast to weak expression in low-grade brain tumors.13 It was shown that mRNA stability of Cyclin A and B1 is regulated by HuR in a cell cycle– dependent manner and that HuR interacts with AREs of many angiogenic factors and immunomodulating cytokines, including COX-2, c-Fos, c-Myc, IL-8, IL-6, TGF-␤, and TNF-␣.8,12,13 Accordingly, misregulated association of HuR with ARE-containing transcripts could allow for enhanced expression of growth-related genes that can influence the neoplastic and angiogenic potential of cancer cells. To understand the molecular mechanisms underlying elevated expression of HuR in human cancers, we characterized the genetic and epigenetic status of HuR and transcriptional regulators implicated in its overexpression in gastric cancers. We show that HuR is aberrantly up-regulated at the transcriptional level in a substantial fraction of gastric cancers and its elevation is associated with tumor progression. Our data also show that HuR is a direct transcription target of nuclear factor ␬B (NF-␬B) and that oncogenic phosphatidylinositol 3-kinase (PI3K)AKT signaling plays a crucial role in both expression and cytoplasmic import of HuR. Together, these findings indicate that activation of HuR via PI3K-AKT signaling contributes to the malignant progression of human gastric tumors.

Materials and Methods Human Tissues and Tumor Cell Lines A total of 190 gastric tissues, including 98 primary adenocarcinomas, 43 benign tumors (9 adenomas, 15 hamartomas, and 19 hyperplastic polyps), and 49 normal tissues, were obtained from 98 patients with cancer and 92 patients without cancer by surgical resection at the Kyung Hee University Medical Center (Seoul, Korea). Signed informed consent was obtained from each patient. Tumor specimens composed of at least 70% carcinoma cells and adjacent tissues found not to contain tumor cells were chosen for molecular analysis. Fifteen human gastric cancer cell lines were obtained from Korea Cell Line Bank (Seoul National University, Seoul, Korea) or American Type Culture Collection (Rockville, MD). The cells were maintained in Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum (FBS; Gibco BRL, Gaithersburg, MD) at 37°C in a humidified atmosphere with 5% CO2.

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2031

Semiquantitative Polymerase Chain Reaction Analysis Polymerase chain reaction (PCR) was initially performed over a range of cycles, and 2 ␮L of 1:4 diluted complementary DNA (12.5 ng/50 ␮L PCR reaction) undergoing 28 –36 cycles was observed to be within the logarithmic phase of amplification with primers HuR1 (5=-ATGACCCAGGATGAGTTACG-3=) and HuR2 (5=GGGTTGGCAAACTTCAC-3=) for HuR. PCR was performed for 30 –34 cycles at 95°C (1 minute), 58°C– 62°C (0.5 minute), and 72°C (1 minute). For DNA PCR analysis, 100 ng of genomic DNA was used for amplification of HuR with intron-specific primers HuRI-2 (5=-AAACACTGCCAGGCATCCTG-3=) and HuR7 (5=-TACGTCCTTCTGGGTCATGG-3=). Quantitation was achieved by densitometric scanning of the ethidium bromide–stained gels. Integration and analysis were performed using the Molecular Analyst software program (Bio-Rad, Hercules, CA). Quantitative PCR was repeated at least 3 times for each specimen, and the mean was obtained.

Immunoblot and Immunohistochemistry Assay Western analyses were performed using antibodies specific for HuR (Santa Cruz Biotechnology, Santa Cruz, CA), caspase-3 (Asp175; Cell Signaling, Beverly, MA), p-AKTser473 (Cell Signaling), p-GSK3-␤S9 (Cell Signaling), total Akt (Santa Cruz Biotechnology), and actin (Santa Cruz Biotechnology). Antibody binding was detected by enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ). Immunohistochemistry was performed using the Bond Polymer Intense Detection System (VisionBioSystems, Victoria, Australia). Briefly, 4-␮m sections of paraffin-embedded tissues were incubated for 15 minutes with monoclonal HuR antibody (3A2; Zymed, Carlsbad, CA) using the biotin-free polymeric horseradish peroxidase–linked antibody conjugate system. For the immunoreactive score, we established a 1- to 12-point system by multiplying the percentage of positive cells by the intensity of the staining score. Immunoreactive scores of 0 –5 were classified as negative, and scores of 6 –12 were regarded as positive.

Construction of Expression Plasmids, Small Interfering RNA, and Transfection HuR expression vector was constructed with a PCR-based approach using primers HuR11 (5=-ATGGTCTAATGGTTATGAAGAC-3=) and HuR3 (5=-TTTGTGGACTTGTTGGTTTTG-3=) and pcDNA3.1-His6-V5 vector (Invitrogen Corp, Carlsbad, CA). Constitutively active AKT (pUSEamp-CA-AKT) and dominant negative AKT (pUSEamp-DN-AKT) plasmids were purchased from Upstate Biotechnology (Lake Placid, NY). The phosphorylation-defective I␬B␣⌬N was kindly provided by Ballard (Vanderbilt University, Nashville, TN). HuR small interfering RNA (siRNA) (5=-TGTGAAAGTGATCCGCGAC-3=) was synthesized by Dharmacon Research (Lafayette, CO). AKT

BASIC– ALIMENTARY TRACT

December 2008

2032

KANG ET AL

siRNA (AKT1 stealth RNAi) was purchased from Invitrogen Corp.

Cell Cycle, DNA Synthesis, AnchorageIndependent Growth, and Apoptosis Analysis

BASIC– ALIMENTARY TRACT

For cell cycle analysis using flow cytometry, cells were fixed with 70% ethanol and resuspended in 1 mL of phosphate-buffered saline containing 100 ␮g/mL ribonuclease and 50 mg/mL propidium iodide. The assay was performed on a FACScan flow cytometer (Becton Dickinson, San Jose, CA), and the cell cycle profile was analyzed using MultiCycle software (Phoenix Flow Systems, San Diego, CA). For anchorage-independent growth assay, 2 ⫻ 104 cells were suspended in complete medium containing 0.3% agar and seeded above a layer of solidified 0.6% agar in Dulbecco’s modified Eagle medium. Complete medium was added every 3 days, and colonies were counted after 3 weeks. For apoptosis analysis, cells exposed to apoptotic stimuli were fixed with 4% paraformaldehyde and then incubated with 0.1% sodium citrate and 0.1% Triton X-100 solution at 4°C. Slides were incubated with terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling solution (Roche, Basel, Switzerland) in a humid chamber in the dark.

Reporter Constructs and Luciferase Assay The HuR promoter regions were cloned into the pGL3-basic vector (Promega, Madison, WI). The putative NF-␬B binding element was mutated (5=-CGTTTGTCAAA-3=) using site-directed mutagenesis. Cells were transfected with 500 ng of the promoter constructs using Lipofectamine 2000 (Invitrogen Corp). After normalization of each extract for protein content, luciferase activity was measured by using the Luciferase Assay System (Promega).

Electrophoretic Mobility Shift and Chromatin Immunoprecipitation Assay Synthetic oligonucleotides containing the putative NF-␬B binding site (5=-GCGCCGACCGGAAGTCCCGCCTCTC-3= for wild-type and 5=-GCGCCGACCGTTT GTCAAACCTCTC-3= for mutant type) were [␥-32p]adenosine triphosphate labeled at their 5= end. The annealed probes were incubated for 40 minutes with nuclear extract in a gel shift binding buffer (Amersham Pharmacia, Pittsburgh, PA). The reaction products were analyzed by electrophoresis on a 6% nondenaturing polyacrylamide gel. For chromatin immunoprecipitation, cells were incubated in 1% formaldehyde solution for 20 minutes. The cells were lysed and the pellet was resuspended in nuclei lysis buffer and sonicated. Immunoprecipitation was performed with p65/RelA antibody (Santa Cruz Biotechnology). PCR was performed using the fol-

GASTROENTEROLOGY Vol. 135, No. 6

lowing primer pairs: HL2-5 (5=-CGAGGGCCGGAACCCAGTTCGC-3=) and HL-6 (5=-GCGGCTCCTCCTCAGCGCGCA-3=).

Results Aberrant Elevation of HuR Expression in Gastric Carcinomas Expression study of HuR using semiquantitative reverse-transcription (RT)-PCR initially revealed that normal and benign gastric tissues express similar levels of HuR transcript whereas primary carcinoma tissues and cell lines have variable levels (Figure 1A and B and Supplementary Figure 1A; see Supplementary material online at www.gastrojournal.org). Compared with noncancerous tissues, primary tumors and cell lines exhibited significantly higher levels of HuR (0.95–1.79 vs 1.15–3.01; P ⬍ .05; Figure 1C). Moreover, tumor-specific elevation of HuR levels was identified in 59 of 80 (73.8%) matched sets (Figure 1A). When HuR levels more than 2-fold (⬎2.62) of normal means (1.32) were arbitrarily set as abnormally high, 20 of 98 primary tumors (20.4%) and 7 of 15 cell lines (46.7%) were classified as abnormally high expressors (Figure 1C). HuR levels were significantly high in advanced tumors (16/59; 27.1%) compared with early tumors (4/39; 1.3%) (P ⬍ .05) but showed no association with grades (well and moderately differentiated tumors [10/47; 21.3%] vs poorly differentiated tumors [10/51; 19.6%]), histologic types (intestinal [12/59; 20.3%] vs diffused [8/39; 20.5%]) of tumors, and age and sex of the patients (Figure 1D and data not shown). HuR transcript levels are fairly consistent with its protein levels in both cell lines and tissues, suggesting that HuR expression is primarily controlled at the transcript level (Figure 1A and B). In addition, a fractionation assay using serum stimulation showed that HuR transcript level correlates with cytoplasmic translocation of its protein (Figure 1E and Supplementary Figure 1B; see Supplementary material online at www.gastrojournal.org). Moreover, an immunohistochemistry analysis showed that 8 of 30 primary tumors (26.7%) have detectable levels of cytoplasmic HuR, and cytoplasmic immunopositivity of HuR is more common in high expressors (4/6; 66.7%) compared with normal expressors (4/24; 16.7%) (Figure 1F). Although only 30 tumors were tested, the cytoplasmic immunopositivity showed a correlation with the stage (advanced [5/13; 38.5%] vs early [3/17; 17.6%]; P ⬍ .05) and the grade (poorly differentiated [5/15; 33.3%] vs well and moderately differentiated [3/15; 20%]; P ⬍ .05) of tumors. All normal mucous cells showed only nuclear positivity, but some of the parietal cells at the base of the fundus and body glands showed weak cytoplasmic positivity (Figure 1F and data not shown). Similarly, tumor-specific elevation of HuR levels was also found in 10 of 40 (25%) matched tissue sets obtained from patients with colorectal cancer, and its level was significantly higher in colorectal

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2033

Figure 1. HuR expression in gastric tissues and cell lines. (A) Tumor-specific elevation of HuR. HuR expression in cancer and adjacent noncancerous tissues was compared using matched tissue sets. P, patient; N, normal tissue; T, primary carcinomas. (B) HuR expression in cell lines. (C) Expression levels of HuR in tissues and cell lines. Bar indicates the mean level of each specimen group. (D) Correlation of HuR expression with tumor stage (E, early; A, advanced), differentiation status (WD, well-differentiated; MD, moderately differentiated; PD, poorly differentiated), and tumor types (I, intestinal; D, diffused). (E) Association of HuR transcript level with cytoplasmic translocation of its protein. Cells were stimulated by 10% serum for 12 hours, and subcellular localization of HuR was determined using immunoblot assay. U1 SnRNP70 and actin were used as a nuclear and cytoplasmic marker protein, respectively. (F) Immunohistochemistry analysis of HuR in tissues. Normal mucosa (a) adjacent to tumor showed only nuclear staining of HuR, whereas variable cytoplasmic immunopositivity was observed in primary tumors (b and c, poorly differentiated stage 2 tumors; d, well-differentiated stage 2 tumor; e, well-differentiated stage 3 tumor, and f, poorly differentiated stage 3 tumor).

cell lines and primary tumors compared with normal tissues. This indicates that overexpression of HuR transcript is not unique to gastric cancer (Supplementary Figure 1C–E; see Supplementary material online at www.gastrojournal.org).

Increased Tumor Cell Growth by HuR Elevation Ectopic overexpression of HuR in AGS cells (low expressor) led to a significant increase in cell number, whereas siRNA-mediated knockdown of HuR in MKN74 cells (high expressor) decreased cell growth (Figure 2A). Colony-forming ability of the cells was also up-regulated and down-regulated by transfection of WT-HuR and siHuR, respectively (Supplementary Figure 2A; see Supplementary material online at www.gastrojournal.org). Moreover, soft agar assay utilizing 2 AGS sublines expressing different levels of HuR showed that the number and size of the colony is significantly increased by HuR, support-

ing that HuR exerts a proliferative effect (Figure 2B). Flow cytometric analysis also showed that the G1 to S phase transition of the cell cycle is facilitated by HuR (Figure 2C). In addition, [3H]thymidine incorporation analysis revealed that HuR expression increases DNA synthesis (Supplementary Figure 2B; see Supplementary material online at www.gastrojournal.org). Moreover, the number of apoptotic cells and caspase-3 cleavage were markedly increased by HuR knockdown in MKN74 cells exposed to 5-fluorouracil (5-FU), while HuR knockdown itself in the absence of 5-FU treatment did not affect cell death (Figure 2D and E). Consistently, when HuR expression was inhibited, tumor cell response to various apoptotic stresses (etoposide, doxorubicin, taxol, ␥-irradiation, and hypoxia) was substantially increased (Supplementary Figure 2C and D; see Supplementary material online at www. gastrojournal.org). These results indicate that elevated

BASIC– ALIMENTARY TRACT

December 2008

2034

KANG ET AL

GASTROENTEROLOGY Vol. 135, No. 6

BASIC– ALIMENTARY TRACT

Figure 2. HuR regulation of tumor cell growth. (A) HuR stimulation of in vitro cellular growth. AGS and MKN74 cells were transfected with WT-HuR-V5 and siHuR, respectively. Data represent means of triplicate assays (bars, SD) (*P ⬍ .05; **P ⬍ .01). (B) HuR effect on anchorageindependent growth. Vector-transfected control and HuR-transfected AGS sublines (HuR-1 and HuR-2) were grown in soft agar for 3 weeks. Visible colonies (150 ␮m in diameter) were counted after crystal violet staining (**P ⬍ .01, ***P ⬍ .001). (C) HuR stimulation of a G1 to S transition of the cell cycle. (D) HuR effect on apoptosis. MKN74 cells transfected with siHuR were exposed to 5-FU (50 ␮mol/L) for 24 hours. Apoptosis was analyzed by flow cytometric measurement of the sub-G1 fraction. (E) Effect of HuR knockdown on caspase-3 activation. MKN74 cells were transfected with siControl or siHuR, and its effect on 5-FU–induced caspase-3 cleavage was examined by immunoblot assay.

HuR levels in tumor cells exert proliferative and antiapoptotic functions.

Absence of Genetic or Epigenetic Alterations of HuR in Cancer Cells We examined whether HuR overexpression is associated with gene amplification. Genomic levels of HuR in cell lines and tumor tissues were comparable to those in normal cells, whereas an elevated level of PIK3CA was observed in 9 of 15 cell lines (60%) and 17 of 80 primary tumors (21.3%) (Supplementary Figure 3A; see Supplementary material online at www.gastrojournal.org). Allelotyping analysis using an intragenic polymorphic marker failed to detect allelic imbalance, indicating the amplification or rearrangement of the gene in tumor specimens (data not shown). We next analyzed methylation status of 160 CpG sites located in the HuR promoter and the first

exon and intron regions using bisulfite DNA sequencing (Supplementary Figure 3B; see Supplementary material online at www.gastrojournal.org). However, all of the 6 cell lines, 3 matched sets, and 3 normal tissues tested exhibited unmethylation or rare methylation at the 160 CpG sites (Supplementary Figure 3C; see Supplementary material online at www.gastrojournal.org). All 15 cell lines showed no detectable change in HuR transcript level following treatment with 5-azadC (the demethylating agent) or TSA (the histone deacetylation inhibitor) or combined treatment with both (data not shown).

Activation of HuR Expression via PI3K-AKT Signaling We next examined whether elevation of HuR levels is associated with transcriptional activation of the gene. The high level of HuR in MKN28 was markedly reduced

December 2008

2035

mentary material online at www.gastrojournal.org). HuR transcript was reduced by LY294002 in a time- and dosedependent manner (Figure 3B). The inhibitory effect of LY294002 was also observed in other HuR-overexpressing cells (Supplementary Figure 4B; see Supplementary material online at www.gastrojournal.org). Additionally,

BASIC– ALIMENTARY TRACT

by the PI3K inhibitor LY294002 and the NF-␬B inhibitor BAY11-7082, while no or slight reduction was detected in cells treated with the p38 mitogen-activated protein kinase inhibitor SB203580, the extracellular signal–regulated kinase inhibitor UO126, and the Src inhibitor PP2 (Figure 3A and Supplementary Figure 4A; see Supple-

AKT ACTIVATION OF HuR IN GASTRIC CANCER

Figure 3. AKT regulation of HuR expression. (A) Effect of pharmacologic inhibitors on HuR expression. MKN28 was treated with each inhibitor (LY294002 [20 ␮mol/L], BAY11-7082 [4 ␮mol/L], SB203580 [10 ␮mol/L], UO126 [10 ␮mol/L], and PP2 [10 ␮mol/L]) for 12 hours. (B) A time- and dose-dependent inhibition of HuR expression by LY294002. (C) HuR activation by AKT. SNU216 cells were transfected with a constitutively active AKT (Myc-CA-AKT), and HuR level was determined at 48 hours after transfection. (D) Effect of AKT knockdown on HuR expression. (E) An association of AKT status with HuR expression. (F) Correlation of HuR expression with phospho-AKT level in cancer cells.

2036

KANG ET AL

mRNA decay assay revealed that HuR mRNA stability is not affected by LY294002 (Supplementary Figure 4C; see Supplementary material online at www.gastrojournal. org). Ectopic expression of an active AKT (CA-AKT) increased HuR levels in SNU216 cells (low expressor), while

GASTROENTEROLOGY Vol. 135, No. 6

siAKT transfection decreased HuR levels in MKN28 cells (high expressor) (Figure 3C and D). We also found that HuR level is associated with phospho-AKT levels (Figure 3E and F). Meanwhile, HuR knockdown did not affect AKT expression at both mRNA and protein levels, indi-

BASIC– ALIMENTARY TRACT

Figure 4. Involvement of NF-␬B in HuR overexpression. (A) Inhibition of HuR expression by BAY11-7082. (B) Effect of NF-␬B blockade on HuR expression in MKN28 cells. (C) Inhibition of AKT-mediated HuR activation by NF-␬B blockade in SNU216 cells. (D) HuR induction by AKT-activating stimuli. SNU216 cells maintained under serum-free condition were exposed to IGF-I, fibroblast growth factor (FGF), epidermal growth factor (EGF), or 10% FBS, and its effect on HuR expression was determined. (E) Role of AKT and NF-␬B in HuR induction by IGF-I and serum. SNU216 cells transfected with siAKT or DN-I␬B were serum starved for 48 hours and stimulated by 10% FBS or IGF-I for 12 hours.

cating that AKT is not regulated by overexpressed HuR in tumor cells (Supplementary Figure 4D; see Supplementary material online at www.gastrojournal.org).

Involvement of NF-␬B in AKT-Mediated HuR Induction HuR expression was also inhibited by the NF-␬B inhibitor BAY11-7082 in a dose- and time-dependent manner (Figure 4A and Supplementary Figure 5A; see Supplementary material online at www.gastrojournal. org). HuR level was decreased by transfection of DN-I␬B, which inhibits degradation of I␬B␣, thus rendering it a constitutively active molecule (Figure 4B). Moreover, activation of HuR expression by CA-AKT transfection was greatly attenuated by cotransfection of DN-I␬B (Figure 4C). In addition, HuR induction by various AKT-activating stimuli, such as insulin-like growth factor (IGF)-I, fibroblast growth factor, epidermal growth factor, or 10% FBS, was disrupted when the cells were pretransfected with DN-I␬B or siAKT (Figure 4D and E). These results show that NF-␬B is involved in HuR induction through PI3K-AKT signaling.

Identification of HuR as a Direct Transcription Target of NF-␬B We found a putative NF-␬B binding site (5=-CGGAAGTCCCG-3=) in the 5= upstream region of the HuR promoter (nucleotides ⫺133/⫺123 relative to the transcription start site). To determine whether this site could confer NF-␬B responsiveness to a heterologous reporter, promoter regions (nucleotides ⫺600/⫺61) were cloned (Pro/600-Luc, Pro/401-Luc, Pro/321-Luc, and Pro/224-Luc) (Figure 5A). While the empty luciferase vector was not affected by IGF-I treatment, inclusion of HuR promoter sequence increased luciferase activity (Figure 5B). A reporter comprising nucleotides ⫺61 to ⫺224 (Pro/224-Luc) exhibited the highest activity, while a reporter construct without this region (⌬224-Luc) or without the putative NF-␬B site (⌬␬B/224-Luc) failed to respond to IGF-I. Inclusion of nucleotides ⫺600 to ⫺225 attenuated the response, suggesting the possible negative regulatory role of this upstream region. The MT-␬B-Luc, which contains a mutated NF-␬B site (5=-CGTTTGTCAAA-3=), showed the basal level of activity. Pro/224-Luc reporter activity was strongly increased by serum addition or CA-AKT transfection, and this effect was diminished by LY294002 or BAY11-7082 and cotransfection of DN-I␬B (Figure 5C and Supplementary Figure 6A; see Supplementary material online at www.gastrojournal. org). To define whether p65/RelA could physically bind to the HuR promoter, we performed an electrophoretic mobility shift analysis using a 24 – base pair synthetic oligonucleotide (WT-␬B-BS) containing the putative NF-␬B site. One prominently shifted band was detected when the probe was incubated with nuclear extract purified from 10% serum- or IGF-I–treated AGS cells, while

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2037

the band intensity was clearly diminished by siAKT or DN-I␬B transfection (Figure 5D). There was no band shift seen when the nuclear extract was incubated with MT-␬B-BS, a mutated form of the ␬B-BS. The specificity of this binding was confirmed by efficient competition with an excess of unlabeled WT-␬B-BS but not MT-␬B-BS (Figure 5E). To further determine whether p65/RelA occupies the NF-␬B site in living cells, we performed chromatin immunoprecipitation experiments. A clear association was observed between p65/RelA and HuR chromatin when the cells were treated with IGF-I (Supplementary Figure 6B; see Supplementary material online at www.gastrojournal.org). The p65/RelA binding to the HuR chromatin was detectable at 6 hours and remained up to 24 hours following treatment but was disrupted when AKT was blocked (Figure 5F). These results thus show that HuR is a direct transcription target of NF-␬B.

Role of HuR in AKT-Induced Tumor Resistance to Apoptosis Stresses WT-HuR transfection in SNU216 cells increased transcript levels of IL-8, COX-2, and VEGF, whereas siHuR transfection in MKN74 cells decreased its levels (Figure 6A and Supplementary Figure 7; see Supplementary material online at www.gastrojournal.org). In SNU216 cells, the times required for 50% loss of the transcripts were increased by CA-AKT transfection (7.1 to 13.1 hours for IL-8, 6.7 to 11.5 hours for COX-2, and 5.5 to 8.8 hours for VEGF), but this effect was diminished by DN-I␬B, supporting that AKT enhances the mRNA half-lives of HuR targets via NF-␬B signaling (Figure 6B). CA-AKT expression reduced apoptotic response of MKN74 cells to 5-FU, but this effect was diminished by HuR knockdown (Figure 6C). Antiapoptotic effect of CA-AKT was further enhanced by HuR overexpression (Figure 6D). These results thus indicate that PI3K-AKT signaling exerts its oncogenic effects partially through HuR activation.

AKT Regulation of Cytoplasmic Relocalization of HuR Protein To explore whether PI3K-AKT signaling is also involved in the cytoplasmic relocalization of HuR, the role of AKT in serum-induced HuR translocation was assessed. An elevation of cytosolic HuR following serum stimulation was substantially attenuated by siAKT transfection but not influenced by DN-I␬B transfection (Figure 7A). Absence of DN-I␬B effect and no detectable HuR gene induction until 6 hours after serum stimulation suggest that the role of AKT in HuR translocation is independent of its HuR transactivating function (Figure 7B). Moreover, cytosolic HuR was apparently increased at 30 minutes after serum addition even in the presence of cycloheximide, indicating that PI3K-AKT signaling can facilitate the nuclear export of preexisting HuR (Figure 7C). An immunofluorescence assay revealed that the cytoplasmic import of HuR was inhibited by siAKT trans-

BASIC– ALIMENTARY TRACT

December 2008

2038

KANG ET AL

GASTROENTEROLOGY Vol. 135, No. 6

BASIC– ALIMENTARY TRACT

fection (Figure 7D). We also observed that HuR translocation stimulated by 5-FU treatment is blocked by AKT knockdown (Figure 7E). These results thus indicate that PI3K-AKT signaling might play an important role in the cytoplasmic import of HuR proteins as well as the transcriptional activation of the gene.

Discussion Elevated HuR expression has been found in a wide variety of human cancers, including breast, colon, lung, and ovarian cancers, and correlated with advancing stages of malignancy.15–19 Many ARE-containing genes, which confer a growth advantage to the neoplastic cells, have been identified as targets of HuR.4,8 –10 It is therefore conceivable that increased expression of HuR facilitates tumor growth by preventing the turnover of the mRNAs integral to tumor development and progression. Although cytoplasmic immunopositivity of HuR was previously described to associate with high cyclooxygenase-2 expression and reduced survival, its expression status, regulation mechanisms, and roles in tumor cell growth have not been defined in human gastric cancer.15 In the present study, we found that HuR mRNA level is significantly elevated in primary carcinomas and cell lines compared with normal or benign tumor tissues. Intriguingly, 74% (59/80) of matched sets were found to have tumor-specific HuR elevation and HuR overexpression is significantly high in advanced tumors versus early-stage tumors, suggesting that HuR elevation may provide tumor cells with a selective growth advantage. Consistently, we observed that HuR enhanced cellular growth, the G1 to S transition of the cell cycle, and the colony-forming ability of tumor cells and decreased apoptotic sensitivity to etoposide, doxorubicin, taxol, ␥-irradiation, and hypoxic stress, showing that HuR has proliferative and antiapoptotic activities. Therefore, our study shows that HuR is commonly up-regulated in gastric carcinomas and its activation facilitates tumor progression. Increased copy number of the gene is one of the genetic mechanisms leading to high mRNA levels of protooncogenes in tumor cells. However, our allelotyping and methylation studies show that HuR is not a target of genetic or epigenetic alterations in gastric cancer. Given these findings, we explored the possibility that HuR expression might be under the direct control of specific

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2039

oncogenic signaling. We observed that HuR level is tightly associated with AKT phosphorylation level and HuR gene transcription is up-regulated via PI3K/AKT/NF-␬B signaling. Recent studies showed that AKT stabilizes ARE transcripts by phosphorylating mRNA decay-promoting factors, such as BRF1, KSRP, and TTP, thus resulting in binding to 14-3-3 and impairment of its mRNA decay activity, showing that PI3K-AKT signaling exerts an mRNA-stabilizing effect by inhibiting mRNA-destabilizing factors.16 –18 Our finding thus describes a novel mechanism by which AKT modulates the turnover of ARE transcripts, a transcriptional activation of the mRNAstabilizing factor HuR in an NF-␬B– dependent fashion. Interestingly, recent studies showed that gastrin and its precursor progastrin, which exert proliferative and antiapoptotic effects on intestinal epithelial and colon cancer cells, increase COX-2 and IL-8 expression via PI3K/NF-␬B signaling and enhance expression and nucleocytoplasmic translocation of HuR through the p38 signaling pathway.19,20 Our preliminary study has shown that mRNA stability of HuR targets, including COX-2, IL-8, and VEGF, are decreased by the p38 inhibitor SB203580 or SB202190, and cotreatment of cells with LY294002 and SB203580 leads to a synergistic reduction in HuR expression (data not shown). It is thus conceivable that p38 and PI3K signaling might have a cross talk in NF-␬B regulation of HuR, particularly in gastrin/progastrin-expressing cancers. Further studies will be required to dissect the collaborative action of PI3K and p38 signaling in HuR regulation in human cancer cells. HuR is predominantly localized in the nucleus, but it rapidly translocates to the cytoplasm in response to various stimuli, and HuR subcellular localization is intimately linked to its function.6,21 The nucleocytoplasmic shuttling of HuR is modulated by different signaling cascades involving AMP-activated protein kinase.22–24 AMP-activated protein kinase induces the acetylation and phosphorylation of importin ␣1, and dual modification of importin ␣1 is required for the nuclear import of HuR.23,24 Recently, Abdelmohsen et al25 showed that HuR selectively up-regulates and down-regulates the mRNA stability of SIRT1 depending on phosphorylation by the cell cycle checkpoint kinase Chk2, indicating that the posttranslational modification of HuR is a critical feature for its activities. Cytoplasmic HuR level is elevated

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™

Figure 5. Identification of HuR as a direct transcription target of NF-␬B. (A) A putative NF-␬B binding site in the HuR promoter and construction of reporter plasmids for luciferase assay. (B) Disruption of promoter responsiveness to IGF-I by mutation or deletion of the NF-␬B binding site. SNU216 cells transfected with promoter constructs were exposed to IGF-I (100 ng/mL) for 12 hours. Relative luciferase activity was normalized by the ␤-galactosidase activity (*P ⬍ .05; **P ⬍ .01). (C) Suppression of the promoter response to serum by AKT or NF-␬B inhibition. (D) Electrophoretic mobility shift assay of p65/RelA binding to the putative NF-␬B binding site. A synthetic 24 – base pair oligonucleotide (WT-␬B-BS) was incubated with nuclear extract purified from 10% serum-treated or IGF-I–treated AGS cells. (E) Specificity verification of the probe binding. The probe was incubated with nuclear extract in the presence of an excess of unlabeled WT-␬B-BS or MT-␬B-BS. (F) Disruption of p65/RelA binding to the HuR chromatin by AKT knockdown. SNU216 cells transfected with siAKT or siControl were exposed to IGF-I for 0 –24 hours. Cross-linked chromatin was immunoprecipitated with antibodies against p65/RelA or rabbit immunoglobulin G and analyzed by PCR using primers that flank the NF-␬B binding site.

BASIC– ALIMENTARY TRACT

December 2008

2040

KANG ET AL

GASTROENTEROLOGY Vol. 135, No. 6

BASIC– ALIMENTARY TRACT

Figure 6. HuR implication in oncogenic function of AKT. (A) HuR regulation of its target transcripts. (B) Regulation of half-lives of HuR target transcripts by AKT/NF-␬B signaling. The values indicate the percentages of initial mRNA signal remaining relative to the steady-state level at the time of addition of actinomycin D (Act D, 5 ␮g/mL). (C) Role for HuR in the antiapoptotic action of AKT. MKN74 cells transfected with CA-AKT were exposed to 5-FU (100 ␮mol/L), and the effect of HuR knockdown on AKT-mediated survival was evaluated using terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling assay. (D) Further increase of the antiapoptotic action of AKT by HuR. SNU216 cells transfected with CA-AKT were exposed to etoposide (100 ␮mol/L), and the effect of WT-HuR transfection on cell survival was evaluated. *P ⬍ .01; **P ⬍ .001.

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2041

Figure 7. The role of AKT in cytoplasmic relocalization of HuR. (A) Effect of AKT blockade on cytoplasmic import of HuR. AGS cells transfected with siAKT (200 nmol/L) or DN-I␬B (2 ␮g) were serum starved for 48 hours and then stimulated by adding 10% FBS for 2 hours. The subcellular distribution of HuR was assessed using fractionation and immunoblot assays. The numbers indicate relative values of protein level. (B) Induction kinetics of HuR mRNA by addition of serum. (C) Transactivation-independent role for AKT in HuR relocalization. AGS cells pretreated with cycloheximide were exposed to 10% FBS in the presence or absence of LY294002 (20 ␮mol/L), and cytosolic HuR level was determined. (D) An immunofluorescence assay for AKT role in HuR relocalization. AGS cells were stimulated by 10% FBS for 2 hours, and the subcellular distribution of HuR was examined. (E) The role of AKT in 5-FU–induced HuR translocation. SNU216 cells were cotransfected with GFP-HuR and siAKT. After 48 hours, the cells were exposed to 5-FU for 10 hours and HuR localization was examined.

in many types of human cancers, including gastric carcinoma, and associated with tumor progression and poor prognosis of the patients.15,18,21 Nevertheless, the molecular mechanism underlying the cytoplasmic abundance of HuR in cancer cells remains largely undefined. We show that AKT facilitates the nuclear export of HuR independently of its transactivating function for the HuR gene. Furthermore, exposure of tumor cells to the chemotherapeutic agent 5-FU leads to the cytoplasmic translocation of HuR, and this effect is attenuated by blocking AKT. Given that HuR has never been reported to be mutated and AKT is up-regulated in a vast majority of cancers, it is conceivable that the cytoplasmic abundance of HuR in human cancers might be associated with oncogenic activation of AKT signaling. Taken together, our data show that AKT plays a crucial role in the cytoplasmic relocalization of HuR protein as well as the transcriptional activation of the gene, and AKT activation of HuR contributes to AKT-mediated tumor cell survival.

In conclusion, HuR is frequently activated through the PI3K/AKT/NF-␬B pathway in gastric cancers and elevated HuR expression contributes to tumor progression by exerting proliferative and antiapoptotic effects. The direct control of HuR by oncogenic signaling and its implication in tumor growth suggest that HuR would be a valuable therapeutic target for treatment and also be a useful prognostic and diagnostic marker for gastric cancers.

Supplementary Data Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2008.08.009. References 1. Chen CY, Shyu AB. AU-rich elements: characterization and importance in mRNA degradation. Trends Biochem Sci 1995;20:465– 470.

BASIC– ALIMENTARY TRACT

December 2008

2042

KANG ET AL

BASIC– ALIMENTARY TRACT

2. Lagnado CA, Brown CY, Goodall GJ. AUUUA is not sufficient to promote poly(A) shortening and degradation of an mRNA: the functional sequence within AU-rich elements may be UUAUUUA(U/A) (U/A). Mol Cell Biol 1994;14:7984 –7995. 3. Good PJ. The role of elav-like genes, a conserved family encoding RNA-binding proteins, in growth and development. Semin Cell Dev Biol 1997;8:577–584. 4. López de Silanes I, Zhan M, Lal A, et al. Identification of a target RNA motif for RNA-binding protein HuR. Proc Natl Acad Sci U S A 2004;101:2987–2992. 5. Wilson GM, Brewer G. Identification and characterization of proteins binding A⫹U-rich elements. Methods 1999;17:74 – 83. 6. Fan XC, Steitz JA. HNS, a nuclear-cytoplasmic shuttling sequence in HuR. Proc Natl Acad Sci U S A 1998;95:15293–15298. 7. Ma WJ, Cheng S, Campbell C, et al. Cloning and characterization of HuR, a ubiquitously expressed Elav-like protein. J Biol Chem 1996;271:8144 – 8151. 8. Wang W, Caldwell MC, Lin S, et al. HuR regulates cyclin A and cyclin B1 mRNA stability during cell proliferation. EMBO J 2000; 19;2340 –2350. 9. Benjamin D, Moroni C. mRNA stability and cancer: an emerging link? Expert Opin Biol Ther 2007;7:1515–1529. 10. Wang W, Yang X, Cristofalo VJ, et al. Loss of HuR is linked to reduced expression of proliferative genes during replicative senescence. Mol Cell Biol 2001;21:5889 –5898. 11. Lopez de Silanes I, Lal A, Gorospe M. HuR: post-translational paths to malignancy. RNA Biol 2005;2:11–13. 12. Dixon DA, Tolley ND, King PH, et al. Altered expression of the mRNA stability factor HuR promotes cyclooxygenase-2 expression in colon cancer cells. J Clin Invest 2001;108:1657–1665. 13. Nabors LB, Gillespie GY, Harkins L, et al. HuR, a RNA stability factor, is expressed in malignant brain tumors and binds to adenine- and uridine-rich elements within the 3= untranslated regions of cytokine and angiogenic factor mRNAs. Cancer Res 2001;61:2154 –2161. 14. Denkert C, Weichert W, Pest S, et al. Overexpression of the embryonic-lethal abnormal vision-like protein HuR in ovarian carcinoma is a prognostic factor and is associated with increased cyclooxygenase 2 expression. Cancer Res 2004;64:189 –195. 15. Mrena J, Wiksten JP, Thiel A, et al. Cyclooxygenase-2 is an independent prognostic factor in gastric cancer and its expression is regulated by the messenger RNA stability factor HuR. Clin Cancer Res 2005;11:7362–7368. 16. Schmidlin M, Lu M, Leuenberger SA, et al. The ARE-dependent mRNA-destabilizing activity of BRF1 is regulated by protein kinase B. EMBO J 2004;23:4760 – 4769.

GASTROENTEROLOGY Vol. 135, No. 6

17. Benjamin D, Schmidlin M, Min L, et al. BRF1 protein turnover and mRNA decay activity are regulated by protein kinase B at the same phosphorylation sites. Mol Cell Biol 2006;26:9497–9507. 18. Gherzi R, Trabucchi M, Ponassi M, et al. The RNA-binding protein KSRP promotes decay of beta-catenin mRNA and is inactivated by PI3K-AKT signaling. PLoS Biol 2007;5:82–95. 19. Subramaniam D, Ramalingam S, May R, et al. Gastrin-mediated interleukin-8 and cyclooxygenase-2 gene expression: differential transcriptional and posttranscriptional mechanisms. Gastroenterology 2008;134:1070 –1082. 20. Rengifo-Cam W, Umar S, Sarkar S, et al. Antiapoptotic effects of progastrin on pancreatic cancer cells are mediated by sustained activation of nuclear factor-␬B. Cancer Res 2007;67: 7266 –7274. 21. Tran H, Maurer F, Nagamine Y. Stabilization of urokinase and urokinase receptor mRNAs by HuR is linked to its cytoplasmic accumulation induced by activated mitogen-activated protein kinase-activated protein kinase 2. Mol Cell Biol 2003;23:7177– 7188. 22. Martínez-Chantar ML, Vázquez-Chantada M, Garnacho M, et al. S-adenosylmethionine regulates cytoplasmic HuR via AMP-activated kinase. Gastroenterology 2006;131:223–232. 23. Wang W, Yang X, López de Silanes I, et al. Increased AMP:ATP ratio and AMP-activated protein kinase activity during cellular senescence linked to reduced HuR function. J Biol Chem 2003; 278:27016 –27023. 24. Wang W, Yang X, Kawai T, et al. AMP-activated protein kinaseregulated phosphorylation and acetylation of importin ␣1: involvement in the nuclear import of RNA-binding protein HuR. J Biol Chem 2004;279:4837– 4888. 25. Abdelmohsen K, Pullmann R Jr, Lal A, et al. Phosphorylation of HuR by Chk2 regulates SIRT1 expression. Mol Cell 2007;25: 543–557.

Received February 5, 2008. Accepted August 18, 2008. Address requests for reprints to: Sung–Gil Chi, PhD, School of Life Sciences and Biotechnology, Korea University, 136-701 Seoul, Republic of Korea. e-mail: [email protected]; fax: (82) 2-927-5458. The authors disclose the following: Supported in part by grants from the Korea Science and Engineering Foundation (R01-2006-00010688-0 and Biofood R&D R053123), the Korea Research Foundation (2003-070-C00031), and the National Cancer Center, Korea (0420230), Republic of Korea.

December 2008

Supplementary Figure 1. (A) Semiquantitative RT-PCR analysis of HuR expression in normal and benign tumor tissues. N, normal gastric tissues; Ha, hamartomas; Hp, hyperplastic polyps; Ad, adenomas. (B) An association of HuR transcript levels with the cytoplasmic translocation of its protein products. AGS cells were transfected with increasing dose of HuR-V5 expression plasmids. The cells were serum starved for 48 hours and then stimulated by adding 10% FBS for 8 hours. The subcellular distribution of HuR proteins was assessed using fractionation and immunoblot assays. (C) Expression of HuR transcript in colorectal carcinoma cell lines. HuR transcript was amplified by RTPCR using exon-specific primers, and 10 ␮L of the PCR products was resolved on a 2% agarose gel. GAPDH was used as an endogenous expression control. (D) Tumor-specific elevation of HuR expression. HuR mRNA levels were compared between cancer and adjacent noncancerous tissues using matched tissue sets obtained from the same patients with colorectal cancer. P, patient; N, normal tissue; T, tumor tissue. (E) Expression levels of HuR (HuR/GAPDH) in colorectal tissue and cell line specimens. Quantitation was achieved by densitometric scanning of HuR RT-PCR products in ethidium bromide–stained gels, and absolute area integrations of the curves representing each specimen were compared after adjustment for GAPDH. Quantitative PCR was repeated at least 3 times for each specimen, and the means were obtained. Bar indicates the mean expression level of each specimen group.

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2042.e1

Supplementary Figure 2. Proliferative and antiapoptotic effect of HuR. (A) Effect of HuR expression on colony-forming ability of tumor cells. AGS and MKN74 cells were transfected with WT-HuR and siHuR, respectively, and maintained in the presence of G418 (1600 ␮g/mL) for 4 weeks. Assays were performed in triplicate (**P ⬍ .01). (B) [3H]thymidine uptake assay of HuR effect on DNA synthesis. Cells were transfected with increasing doses of WT-HuR expression plasmid (AGS cells) or siHuR (MKN74 cells). After transfection, 2 ⫻ 104 cells/well were cultured on 24-well multiplates for 44 hours and then were pulse labeled for 4 hours with 1 ␮Ci/mL [3H]thymidine. The radioactivity incorporated into trichloroacetic acid–precipitable materials was counted by a liquid scintillation counter. Data represent means of triplicate assays (bars, SD) (*P ⬍ .05; **P ⬍ .01). (C) A representative example of HuR effect on tumor cell growth. MKN74 cells were transfected with siHuR and exposed to 5-FU for 24 hours. (D) Effect of HuR knockdown on tumor cell sensitivity to apoptotic stresses. Data represent means of triplicate assays (*P ⬍ .05).

2042.e2

KANG ET AL

Supplementary Figure 3. No genomic amplification and aberrant promoter methylation of HuR in gastric tumor specimens. (A) Genomic level of HuR and PIK3CA in gastric cancer cell lines and primary tumor tissues. Genomic DNA was isolated from the same cells used for mRNA expression analysis. For comparison and validation, the PIK3CA gene was included. Exon 11-2 region of the HuR gene and intron 11 region of the PIK3CA gene were amplified by PCR using intron-specific primer sets. For tissue analysis, HuR gene levels were compared between cancer and adjacent noncancer tissues obtained from the same patients. P, patient; N, normal tissue; T, tumor tissue. Ten microliters of the PCR products were resolved on a 2% agarose gel. GAPDH was used as an endogenous control. (B) A map of the CpG sites of the 5= upstream region of the HuR gene. One hundred sixty CpGs (within nucleotide ⫺663 to ⫹665) analyzed by bisulfite DNA sequencing are represented by bold vertical lines and numbered 1–160. The transcription start site is indicated by an arrow at ⫹1. (C) Representative examples of bisulfite genomic sequencing in cancer cell lines and tissues. One microgram of genomic DNA was incubated with 3 mol/L sodium bisulfite, and 50 ng of bisulfite-modified DNA was subjected to PCR amplification of the HuR promoter region using primer sets HMS9 (sense, 5=-AGGGTATTTTAGGTTAGCGTTG-3=) and HMS8 (antisense, 5=-TTATCCCGCACCTATCTATCCTACC-3) for nucleotides ⫺663 to ⫺305 (23 CpG sites), HMS5 (sense, 5=-GGTAGGATAGATAGGTGCGGGATAA-3=) and HMS2 (antisense; 5=-ACCCCCCAC CACCTCACCTAA-3=) for nucleotides ⫺305 to ⫹169 (80 CpG sites), and HMS11 (sense, 5=-ATTTAGGTAGGGCGGTGGGGGGTT-3=) and HMS10 (antisense, 5=-CAAAACTCCCGTATATCTACTAA-3=) for nucleotides ⫹169 to ⫹655 (57 CpG sites). The PCR products were cloned into pCRII vectors (Invitrogen Corp), and 5 clones of each specimen were sequenced by automated fluorescence-based DNA sequencing to determine the methylation status. The white circles indicate an unmethylated CpG (0 of 5 clones), and the gray squares indicate a partially methylated CpG (ⱕ3 of 5 clones).

GASTROENTEROLOGY Vol. 135, No. 6

Supplementary Figure 4. Suppression of HuR by the PI3K inhibitor. (A) Effect of pharmacologic inhibitors on HuR mRNA expression. MKN28 was treated with LY294002 (20 ␮ mol/L), BAY117082 (4 ␮mol/L), SB203580 (10 ␮mol/L), UO126 (10 ␮mol/L), and PP2 (10 ␮mol/L) for 12 hours. (B) HuR inhibition effect of LY294002 in various tumor cells. Four HuR-overexpressing tumor cells were treated with LY294002 for 12 hours, and expression level of HuR mRNA was determined using semiquantitative RT-PCR. (C) No influence of LY294002 on the mRNA stability of HuR determined by mRNA decay assay. (D) No detectable effect of HuR knockdown on AKT level. MKN28 cells were transfected with siControl or siHuR, and its effect on AKT level was examined using semiquantitative RT-PCR and immunoblot assay.

Supplementary Figure 5. (A) Suppression of HuR by the NF-␬B inhibitor. HuR inhibition effect of BAY11-7082 in gastric tumor cells. Four HuR-overexpressing tumor cells were treated with BAY11-7082 (4 ␮mol/L) for 12 hours, and expression level of HuR mRNA was determined using semiquantitative RT-PCR.

December 2008

Supplementary Figure 6. (A) A positive and negative regulation of the HuR promoter activity by CA-AKT and DN-I␬B, respectively. SNU216 cells were transfected with increasing doses of CA-AKT or stimulated with 10% FBS, and its effect on Pro/224-Luc reporter activity was analyzed. Implication of NF-␬B signaling was determined by analyzing the effect of DN-I␬B transfection on the serum-induced luciferase reporter activity. (B) In vivo binding of p65/RelA to the HuR promoter. SNU216 cells exposed to IGF-I and cross-linked chromatin were immunoprecipitated with antibodies against p65/RelA or rabbit immunoglobulin G. The released DNA was analyzed by PCR using primers that flank the NF-B binding site.

Supplementary Figure 7. Effect of HuR knockdown on its target expression in gastric tumor cells. HuR-overexpressing MKN74 cells were transfected with increasing doses of siHuR, and its effect on protein expression of HuR target genes was examined using immunoblot assay.

AKT ACTIVATION OF HuR IN GASTRIC CANCER

2042.e3