The MicroRNA let-7a Modulates Interleukin-6-dependent STAT-3 Survival Signaling in Malignant Human Cholangiocytes*

The inflammation-associated cytokine interleukin-6 (IL-6) can contribute to tumor growth and resistance to therapy by the activation of survival mechanisms. In several human cancers, IL-6-activated survival signaling involves the signal transducers and activators of transcription (Stat) factors or protein kinase cascades. microRNAs (miRNAs) are endogenous regulators of gene expression that are altered in expression in many cancers. However, the effect of inflammatory cytokines on miRNA expression and the role of miRNA in modulating IL-6-mediated cell survival are unknown. We investigated the involvement of miRNA in malignant cholangiocytes stably transfected to overexpress IL-6, which enhances tumor growth in vivo by inhibition of apoptosis. We provide evidence that (i) miRNA expression both in vitro and in vivo is altered by overexpression of IL-6; (ii) selective miRNAs including let-7a are up-regulated and contribute to the survival effects of enforced IL-6 activity; and (iii) let-7a contributes to the constitutively increased phosphorylation of Stat-3 by a mechanism involving the neurofibromatosis 2 (NF2) gene. These findings reveal a novel mechanism by which IL-6 mediates tumor cell survival that may be therapeutically targeted and emphasize the presence of complex interrelationships between deregulated expression of miRNA and transcription factors in human cancers.

Increased expression of the inflammation-associated cytokine interleukin-6 (IL-6) 2 occurs in chronic inflammatory conditions and in several human cancers such as multiple myeloma, prostate cancer, and cholangiocarcinoma. IL-6 has been implicated in tumor growth in many of these tumors, and elevated IL-6 expression has been associated with poor out-comes and resistance to chemotherapy (1). Experimentally, growth of prostate cancer and cholangiocarcinoma xenografts in athymic mice has been shown to be increased by enforced expression of IL-6 by activation of cell survival signaling (2,3). The mechanisms by which IL-6 promotes cell survival in cancers are of considerable interest because they may be therapeutically targeted.
IL-6-activated survival signaling has been shown to involve the signal transducers and activators of transcription (Stat) factors or various protein kinase cascades (4,5). Although constitutive activation of Stat has been described in many cancers, the precise mechanisms involved are incompletely understood. Cholangiocarcinomas are highly resistant to chemotherapy. However, inhibition of IL-6-dependent pathways such as the Jak-Stat pathway, phosphatidylinositol 3-kinase, or the p38 MAPK pathways can enhance chemotherapy-induced cell death. Thus, aberrant IL-6-dependent survival signaling may contribute to the refractoriness of cholangiocarcinoma to most chemotherapeutic agents.
We sought to understand the role of microRNAs (miRNAs) in IL-6-mediated tumor cell survival. miRNAs are endogenous regulators of gene expression that are altered in expression in many cancers (6)(7)(8). The expression of several miRNAs in cholangiocarcinoma xenografts in athymic mice is altered during in vivo treatment with gemcitabine (9). miRNAs are small endogenous molecules that can regulate gene expression in a sequence complementary manner. Several hundred miRNAs have been identified, and details of the mechanisms by which they regulate gene expression are being unraveled (10,11). Less is known about the mechanism by which miRNAs contribute to cellular behavior and function. Alterations in miRNA expression occur in many different cancers (8). Thus, individual miRNAs may play contributory or regulatory roles in tumor cell pathogenesis or behavior. Potential downstream targets of miRNA include oncogenes or tumor suppressor genes, but few miRNA-regulated targets relevant to tumor biology have been described such as the Ras oncogene (12). We postulated that genetic reprogramming resulting from altered miRNA regulatory networks may contribute to tumor cell response and resistance to chemotherapy.

EXPERIMENTAL PROCEDURES
Cell Lines and Cultures-Mz-1 and KMCH-1 human malignant cholangiocytes and their respective IL-6-overexpressing stable transfectants, Mz-IL-6 and KM-IL-6, were obtained and * This study was supported by grant DK069370 from the National Institutes of Health and the Scott and White Hospital Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains a supplemental figure. 1  cultured as described previously (2). Basal IL-6 expression was increased by ϳ1.5-fold in KM-IL-6 and ϳ3-fold in Mz-IL-6 cells relative to their respective controls. Transfections-20 l of 100 nM microRNA precursor, antisense inhibitor, or controls were added to 1 ϫ 10 6 cells suspended in 80 l of Nucleofector solution (Amaxa Biosystems, Koln, Germany) at room temperature. Electroporation was performed using the Nucleofector system (Amaxa Biosystems). Transfected cells were then resuspended in regular culture medium containing 10% serum for 48 -72 h prior to study.
MicroRNA Isolation and Expression Profiling-miRNA was isolated by PAGE purification of total RNA, and expression profiling was performed using a custom-generated microarray as described previously (9). Microarrays were scanned using a GenePix 4000A array scanner (Axon Instruments, Union City, CA). Normalization was performed by expressing each miRNA replicate relative to a control miRNA (Ambion, Austin, TX) added to each sample, thus allowing comparisons between chips. Data were analyzed using GeneSpring 7.0 Software (Silicon Genetics, Redwood City, CA), and an average value of the median intensity of each replicate in four groups was generated. MicroRNA expression levels were clustered using a self-organizing tree algorithm using the MultiExperiment Viewer Version 3.1 from The Institute for Genomic Research (13).
Quantitative Real-time PCR-RNA was isolated using the ToTALLY RNA isolation kit (Ambion), and cDNA was generated by reverse transcription using 1 g of total RNA and the reverse transcription kit (Invitrogen). Mature let-7a miRNA expression was assessed using a TaqMan human microRNA assay kit (Applied Biosystems, Foster City, CA). Real-time PCR was performed using a MX 3000P TM PCR instrument (Stratagene, San Diego, CA).
Cytotoxicity Assay-Cell viability was assessed using a commercially available tetrazolium bio-reduction assay as described previously (14). 10,000 viable cells/well were seeded into 96-well plates and incubated with gemcitabine, 5-fluorouracil, camptothecin, or appropriate diluent controls in a final volume of 200 l of medium containing 0.5% fetal bovine serum.
Caspase Assay-Cells were plated in 96-well plates (20,000 cells/well) and incubated with different chemotherapeutic agents or diluent control. Caspase 3/7 activity was assayed using the fluorometric Apo-ONE homogenous caspase 3/7 assay (Promega, Madison, WI) and a Cytofluor microplate fluorescence plate reader.
Stat-3 Kinase Assay-Stat-3 activity was assessed in cell lysates after immunoprecipitation using monoclonal P-Stat-3-Tyr 705 antibody (Cell Signaling Technology) and using a tyrosine kinase activity assay kit (Chemicon, Temecula, CA).
Luciferase Reporter Vectors-The pMIR-NF2-luc and pMIR-NF2-MUT-luc firefly luciferase reporter vectors, which contain the intact or mutated putative let-7a recognition sequence from the 3Ј-UTR of NF2, respectively, cloned downstream of the firefly luciferase gene were constructed as follows. Synthetic oligonucleotides encompassing the intact or mutated let-7a recognition sequence, sticky ends for HindIII and SpeI to eliminate digestion of the insert, and a unique BlpI site to test for positive clones were synthesized and annealed. The oligonucleotides used were 5Ј-CTAGTGCTCAGCTACAAGAGAT-TCTCCTGCCTCAA-3Ј (sense) and 5Ј-AGCTTTGAGG-CAGGAGAATCTCTTGTAGCTGAGCA-3Ј (antisense) for pMIR-NF2-luc, or 5Ј-CTAGTGCTCAGCTATAGGCGCTT-TGCTCGATGGAA-3Ј(sense) and 5Ј-AGCTTTCCATCGAG-CAAAGCGCCTATAGCTGAGCA-3Ј (antisense) for pMIR-NF2-MUT-luc. The recognition sequence is underlined in the former sequence, and the random mutations introduced are italicized in the latter sequence. To anneal the oligonucleotides, 2 g of each strand was added to 46 l of DNA annealing buffer (30 mM HEPES, pH 7.4, 100 nM potassium acetate, 2 mM magnesium acetate) for a final volume of 50 l and incubated at 90°C for 3 min and then at 37°C for 1 h. The annealed insert was then directly ligated into the HindIII and SpeI cloning sites of the pMIR-REPORT luciferase expression vector (Ambion). tumor cell xenografts (n ϭ 4 each) were treated with gemcitabine (120 mg/kg intraperitoneally) every 3 days for a total of five doses. Tumor volume was assessed at baseline and at the end of treatment. The starting size of tumors was similar in both groups, and the studies were performed with tumors ranging from 200 to 230 mm 3 . The data represent the mean and 95% confidence intervals from four tumors. p Ͻ 0.05 for differences between change in volume from baseline for Mz-IL-6 when compared with Mz-1 xenografts. C, xenograft sections were obtained, and TUNEL staining for apoptotic cells was performed. The data represent mean and standard deviation of the number of TUNEL-positive cells from eight random high power fields (HPF). There was a decrease in chemosensitivity in vivo and in the numbers of TUNEL-positive cells in Mz-IL6 xenografts. Thus, enhanced IL-6 expression decreases chemotherapy-induced apoptosis in vivo.
Clones were selected after screening by restriction digestion with BlpI. Mz-IL-6 cells were co-transfected with 1 g of pMIR-NF2-UTR or pMIR-NF2-MUT-UTR construct and 1 g of pRL-TK Renilla luciferase expression construct without (empty vector control) or with anti-let-7a inhibitor using Tran-sIT-siQUEST transfection reagent (Mirus, Madison, WI). Luciferase assays were performed 48 h after transfection using the Dual-Luciferase reporter assay system (Promega). For validation of the effect of anti-let-7a, we used the pRL-TK let-7a firefly luciferase expression construct in which two let-7a sites are inserted into the XbaI site in the 3Ј-UTR. Cells were cotransfected with pRL-TK, and luciferase assays were performed 48 h after transfection. Firefly luciferase activity was normalized to Renilla luciferase activity for each sample.
Western Blotting-For immunoblot analysis of cells in culture, cell lysates were obtained from cells grown in 100-mm dishes, whereas for analysis of xenograft tissue, lysates were obtained after tissue homogenization. Equivalent amounts of protein were resolved by electrophoresis in a 4 -20% Tris-HCl gel (Bio-Rad) and then transferred to nitrocellulose membranes. After blocking, membranes were incubated with primary antibodies and infrared dye-labeled secondary antibodies. The protein of interest was then detected using the LI-COR Odyssey infrared imaging system (LI-COR Bioscience, Lincoln, NE). Nuclear and cytoplasmic fractions were obtained using the NE-PER extraction kit (Pierce) according to the manufacturer's instructions.
Xenograft Model-Studies were performed under an Institutional Animal Care and Use Committee (IACUC) approved protocol. Eight-week-old male athymic nu/nu mice (Charles River Laboratories, Wilmington, MA) were maintained in accordance with IACUC procedures and guidelines. 5 ϫ 10 6 Mz-1 or Mz-IL-6 cells were suspended in 0.25 ml of extracellular matrix gel, and the mixture was injected subcutaneously into the right and left flanks. Serial measurements of xenograft growth were performed, and tumor volume was estimated using the formula 4/3 (L*W*H/8). Once tumor volume was 200 -230 mm 3 , xenografts were excised, and the tissue was homogenized for miRNA isolation or immunoblot studies or used for miRNA inhibition studies. For the latter, mice with Mz-IL-6 xenografts were injected intratumorally with 4 ng/mm 3 tumor volume of either anti-let-7a or diluent. The following day, gemcitabine (150 mg/kg) was administered intraperitoneally every 3 days for three doses. The change in tumor size was assessed, and tumors were excised after 10 days. Homogenates were obtained for Western blot analysis. Sections of tumors were obtained for immunofluorescence studies using mouse anti-P-Stat-3 (Tyr 705 ) (1:75 dilution) or rabbit
Statistical Analysis-In vivo responses were compared using the t test. Data are expressed as the mean Ϯ 95% confidence interval unless otherwise noted. The difference between two groups was analyzed using a double-sided Student's t test, and the null hypothesis was rejected at the 0.05 level.

IL-6 Survival Signaling Involves
Alterations in miRNA Expression-To identify miRNA that may contribute to survival signaling and chemoresistance, we first assessed the effect of IL-6 on miRNA expression. Mz-ChA-1 human cholangiocarcinoma cells were stably transfected to overexpress IL-6 (Mz-IL-6 cells) and implanted as xenografts in athymic nude mice. When compared with Mz-1 control cell xenografts, the growth rate of Mz-IL-6 xenografts was increased (Fig.  1A). Moreover, there was a loss of sensitivity of Mz-IL-6 tumor xenografts to the chemotherapeutic agent gemcitabine (Fig. 1B), in conjunction with a decrease in the number of TUNEL-positive (apoptotic) cells when compared with controls (Fig. 1C). We used a miRNA microarray to assess the expression of human miRNAs in tumor cell xenografts and in two different cholangiocarcinoma cell lines overexpressing IL-6. The pattern of miRNA expression in IL-6overexpressing Mz-IL-6 and KM-IL-6 cells differed from their controls (supplemental Fig. 1). A cluster of miRNAs that were increased with enforced IL-6 expression both in vivo as well as in vitro was identified and included several members of the let-7 family and miRNA that have been implicated in oncogenesis such as miR-21 ( Fig. 2A) (17). These data showing altered miRNA expression profiles in vivo suggest that some effects of IL-6 on tumor cell growth and apoptosis may be mediated by miRNA-dependent regulation of gene expression. The relative expression of several members of the let-7 family was altered in vivo (Fig. 2B). Of these, let-7a was chosen for further study based on consistency and level of expression. The expression of mature let-7a miRNA was confirmed to be increased by realtime PCR in IL-6-overexpressing cells in vitro, as well as in tumor cell xenografts in vivo when compared with controls (Fig. 2C).
let-7a Contributes to Survival Signaling by IL-6-To determine the relevance of enhanced let-7a expression to survival signaling, we next evaluated the effect of let-7a inhibition on the response to chemotherapy in vitro. The effect of the anti-let-7a inhibitor was assessed by co-transfecting with the pRL-Tk let-7a firefly luciferase construct that contains two let-7a sites in the 3Ј-UTR of the firefly luciferase reporter. An increase in luciferase activity confirmed the efficacy of anti-let-7a under C, cells were transfected with either control or anti-let-7a inhibitors for 48 h as above and then incubated with the indicated chemotherapeutic agent. Caspase-3/7 activation was assessed using a fluorometric assay after 24 h. Data represent mean Ϯ 95% confidence interval from three separate studies done in triplicate. D, transfected cells were incubated with 100 M gemcitabine for 24 h, and lysates were obtained for Western blot analysis for activated caspase-3 or PARP expression. Inhibition of let-7a increased caspase activity and PARP cleavage in response to gemcitabine. *, p Ͻ 0.05 relative to respective controls.

let-7a Modulates IL-6 Survival Signaling
the conditions used for our studies (Fig. 3A). Cytotoxicity in response to diverse agents was enhanced by preincubation with antisense inhibitors to let-7a (Fig. 3B). Moreover, there was an increase in caspase-3/7 activity in response to chemotherapy in cells preincubated with anti-let-7a when compared with a control inhibitor (Fig. 3C). Similarly, an increase in PARP cleavage and activated caspase-3 was noted in response to anti-let-7a in Western blots from gemcitabine-treated cells (Fig. 3D). Thus, inhibition of let-7a increases chemotherapy-induced apoptosis. Taken together, these data are consistent with an effect of let-7a on IL-6-mediated anti-apoptotic survival pathways.
let-7a Regulates Stat-3 Phosphorylation-Survival signaling by IL-6 can involve activation of the Stat family of transcription factors or protein kinases such as p38 MAPK and phosphatidylinositol 3-kinase. Constitutive phosphorylation and activation of Stat-3 in cholangiocarcinoma cells have been shown to be IL-6-dependent (18). Consistent with these observations, there was an increase in Tyr 705 -phosphorylated Stat-3 (p-Stat-3) in Mz-IL-6 cells when compared with controls. An increase of Ser 473 phosphorylated Akt was also noted in these cells. We next evaluated the effect of antisense inhibition of miR-21 and let-7a on these pathways (Fig. 4A). Constitutive Stat-3 phosphorylation in Mz-IL-6 cells was markedly decreased by antilet-7a but not by anti-miR-21. Furthermore, incubation of Mz-IL-6 cells with anti-let-7a decreased basal Stat-3 kinase activity to 76.9 Ϯ 2.5% of controls after 48 h (n ϭ 8, p ϭ 0.002). Stat-3 has several well characterized targets that inhibit apoptosis and can modulate survival, such as survivin, Bcl-X L , and Mcl-1 (19). Of these, survivin expression can predict prognosis in human cholangiocarcinoma, and Mcl-1 expression can modulate responses to chemotherapy (20). Consistent with a role for Stat-3 in mediating the survival effects of IL-6   NF2 is a target of let-7a. A, the location of the putative let-7a target site in the NF2 3Ј-UTR is shown. A comparison of base pairs between mature human let-7a (hsa-let-7a), human NF2, rat rno-let-7a, and rat NF2 shows sequence conservation between species. The sequence of the mutated target site with mutations to disrupt base pairing between let-7a binding sites and NF2 is also shown. B, Mz-IL-6 cells were transfected with the Renilla luciferase expression construct pRL-tk and either the luciferase construct pMIR-NF2-luc or pMIR-NF2-MUT-luc (in which mutations were introduced in the let-7a target site) with either anti-let-7a or control inhibitor. After 48 h, Dual-Luciferase assays were performed. An increase in relative firefly luciferase with pMIR-NF2-luc (black bars) but not with the pMIR-NF2-MUT-luc construct (gray bars) confirms that the let-7a complementary sequence in the 3Ј-UTR of NF2 is a target of modulation by let-7a. The data represent the mean and standard deviations from six determinations from three independent transfections. C, Mz-1 cells were co-transfected with either let-7a or control precursor, pRL-TK, and either pMIR-NF2-luc or pMIR-NF2-MUT-luc constructs. After 48 h, Dual-Luciferase assays were performed. A decrease in relative firefly luciferase was observed with pMIR-NF2-luc but not with p-MIR-NF2-MUT-luc. The data represent the mean and standard errors from six determinations from three independent transfections. *, p Ͻ 0.05 relative to respective controls.
NF2 Is a Target for let-7a-To elucidate potential mediators of let-7a modulation of Stat-3 phosphorylation, we performed a bioinformatics analysis. The let-7a sequence was compared with proposed regulators of Stat-3, using the criteria of Doench and Sharp (15) but modified to include 85% sequence complementarity at positions 2-9 of the miRNA. The tumor suppressor gene NF2, a known modulator of Stat-3 activation, was identified as a putative target for let-7a. Moreover, interrogation of various target prediction databases such as TargetScan, miRScan, miRanda, and PicTar did not identify any other known regulators of Stat-3 as potential targets for let-7a (21). The location of the let-7a complementary site in the 3Ј-UTR of NF2 is shown in Fig. 5. The site is conserved in human and rat homologs of NF2.
NF2 has been previously shown to regulate Stat-3 phosphorylation by a mechanism involving the hepatocyte growth factor tyrosine kinase substrate HRS (22). We verified that NF2 was a target for let-7a using luciferase reporter constructs containing the let-7a recognition sequence from the 3Ј-UTR of NF2 inserted downstream of the luciferase gene (pMIR-NF2-luc), along with a similar construct in which random mutations were introduced at sites involved in base-pairing (pMIR-NF2-MUT-luc) (Fig. 5A). Transfection with anti-let-7a increased reporter activity in Mz-IL-6 cells, whereas let-7a precursor decreased reporter activity in Mz-1 cells However, these effects were ameliorated when the mutated reporter construct pMIR-NF2-MUT was used in place of pMIR-NF2-luc (Fig. 5, B and C). Constitutive expression of NF2 was decreased and p-Stat-3 increased in Mz-IL-6 cells when compared with Mz-1 controls. Moreover, inhibition of let-7a increased NF2 expression and concomitantly decreased p-Stat-3 (Fig. 6A). Conversely, p-Stat-3 was increased during incubation of Mz-1 cells with NF2 siRNA (Fig.  6B). Incubation with the let-7a precursor miRNA increased nuclear Stat-3 expression, suggesting that let-7a enhances activation of Stat-3 and nuclear translocation. Similar effects were also observed with NF2 siRNA (Fig. 6C). These studies support a mechanism by which enhanced IL-6 production enhances constitutive Stat-3 phosphorylation and activation via a mechanism involving let-7a-mediated inhibition of NF2.
let-7a Can Mediate Downstream Effects of IL-6 Overexpression on Stat-3 Phosphorylation-We evaluated whether an increase in let-7a expression alone could recapitulate the effects of IL-6 on NF2 and Stat-3. Mz-1 cells were transfected with either let-7a precursor or control precursors. Interestingly, NF2 expression was decreased, and Stat-3 phosphorylation was increased in Mz-1 cells transfected with let-7a precursor. Furthermore, the effect of let-7a expression on Stat-3 phosphorylation was blocked by enforced expression of NF2 (Fig. 7A). To assess the involvement of NF2 on IL-6-dependent signaling, we measured phospho-Stat-3 and NF2 levels in Mz-1 and Mz-IL-6 cells after transfection with NF2 or Lac Z. Aberrant expression of NF2 successfully blocked IL-6-dependent Stat-3 activation (Fig. 7B). Thus, enforced expression of NF2 is sufficient to overcome the effects of IL-6 on constitutive Stat-3 expression in malignant cholangiocytes.
let-7a Modulates NF2 and Stat-3 in Vivo-To explore the in vivo relevance of these observations, we assessed the expression of p-Stat-3, NF2, and the Stat-3-regulated anti-apoptotic proteins in homogenates from xenograft tumors. The results corresponded to those observed in vitro with an increase in basal expression of p-Stat-3 and decrease in NF2 in Mz-IL-6 xenografts when compared with control cell xenografts (Fig.  8A). Intratumoral administration of anti-let-7a increased NF2 and decreased p-Stat-3 expression in Mz-IL-6 xenografts in vivo (Fig. 8, B and C). Moreover, a decrease in tumor growth consistent with increased gemcitabine toxicity was observed in response to anti-let-7a when compared with tumors that were When compared with control siRNA transfected cells, the phosphorylation of Stat-3 was increased by NF2 siRNA. C, Mz-1 cells were incubated with either let-7a precursor or siRNA to NF2 for 48 h. Nuclear and cytoplasmic fractions were obtained, and immunoblots were performed for Stat-3 expression. TATA box-binding protein and ␤-actin were used as nuclear and cytoplasmic markers, respectively, and as loading controls. An increase in nuclear Stat-3 occurs with either let-7a precursor or with siRNA to NF2.

let-7a Modulates IL-6 Survival Signaling
untreated. Anti-let-7a (n ϭ 6) or diluent (n ϭ 4) was administered intratumorally into Mz-IL-6 xenografts. Animals subsequently received a course of three doses of gemcitabine 150 mg/kg intraperitoneally given every 3 days. The mean change in tumor growth at the end of treatment was 0.6 Ϯ 2.0% in tumors that received anti-let-7a when compared with 18.4 Ϯ 8.2% in controls (p ϭ 0.02). In combination, these findings identify a previously unrecognized mechanism that contributes to constitutive Stat-3 phosphorylation involving NF2, a target of regulation by the let-7a microRNA.

DISCUSSION
Although the association between chronic inflammation and malignancy has been recognized for many decades, the role of miRNA in cancer cell biology has only recently been appreciated. The role of cytokines as stimulators of miRNA expression have not been explored, and here, we demonstrate a role for persistent IL-6 stimulation on altered miRNA expression in a human cancer. The demonstration of an inflammation-associ-ated cytokine-regulated miRNA-mediated survival mechanism is highly relevant to both tumor biology and regulation of cytokine signaling. Moreover, these studies emphasize the emerging complexity of miRNA-mediated cellular responses.
There is compelling evidence of a critical role for activated Stat-3 in human cancers. Constitutively activated Stat-3 is observed in many cancers, and abrogation of Stat-3 activation results in the loss of the malignant phenotype (23). Moreover, cells expressing persistently activated Stat-3 are dependent on it for survival. Thus, Stat-3 can act as an oncogene and may contribute to tumor growth (24 -26). Although several cytokines including IL-6 can induce Stat-3 tyrosine phosphorylation, the mechanisms by which it is constitutively activated in cancers are unknown. Although activating Stat-3 mutations have not been described, aberrant expression of modulators of Stat expression or phosphorylation such as PIAS-3, modulators of upstream Stat-3 activation such as SOCS-1, or as yet uncharacterized Stat-3 tyrosine phosphatases may all contribute (5). None of these mechanisms have been shown to predominate in FIGURE 7. let-7a and NF2 expression are mediators of the effects of IL-6 on Stat-3 phosphorylation. A, Mz-1 cells were transfected with 30 nM specific let-7a or control precursors, along with 6 g of pcDNA3/5ЈF encoding NF2 or Lac Z control plasmids for 48 h. Representative immunoblots are shown along with quantitative data representing the means Ϯ S.E. from four separate blots. let-7a overexpression increases Stat-3 phosphorylation, which is blocked by overexpression of NF2. *, p Ͻ 0.05 relative to expression in control precursor group. B, Mz-1 or Mz-IL-6 cells were transfected with 6 g of NF2 or Lac Z control plasmids. Immunoblot analysis for NF2 and phosphorylated Stat-3 was performed 48 h after transfection. A decrease in NF2 expression along with an increase in Stat-3 phosphorylation is observed in Mz-IL-6 cells when compared with Mz-1 controls. Enforced expression of NF2 is sufficient to overcome the effect of IL-6 overexpression on constitutive Stat-3 phosphorylation. Representative immunoblots and quantitative data (mean Ϯ S.E.) from four separate blots are shown.

let-7a Modulates IL-6 Survival Signaling
tumor cells. The contribution of miRNA modulation of NF2 expression warrants further investigation as an alternative mechanism contributing to constitutive Stat-3 activation. We speculate that the NF2-dependent mechanism may be more relevant to constitutively increased Stat-3 phosphorylation in the setting of chronic IL-6 stimulation, rather than transient, non-sustained activation of Stat-3 in response to acute stimulation of IL-6 signaling, which has been well characterized and involves, among others, Jak-Stat interactions.
NF2 is located on chromosome 22q12.2 and encodes for merlin, a putative tumor suppressor gene. Merlin has strong binding to HRS, a potent regulator of receptor tyrosine kinase trafficking, and the interaction of HRS and Merlin can result in inhibition of Stat activation (22). Merlin has been shown to act as growth regulator, and its decreased expression could partly contribute to the increased growth rate observed in IL-6-overexpressing tumor cell xenografts. In response to IL-6 stimulation, activation of Stat-3 is associated with the endocytotic pathway (27). Thus, a plausible mechanism by which decreased NF2 expression in response to persistent IL-6 stimulation results in activation of Stat-3 could involve facilitating its association with the endocytotic pathway through an HRS-mediated mechanism.
The miRNA family of let-7 and its homologs have been implicated as cancer-associated miRNAs in recent studies (28,29). Although we focused our studies on let-7a, we note that the relative expression of other members of the let-7 family such as let-7d and let-7f-2 were also increased in vivo (Fig. 2B). It is quite likely that these other members of the let-7 family that are differentially altered in response to increased IL-6 stimulation may also have cellular actions. NF2 is not a predicted target for either let-7d or let-7f-2. However, both these miRNAs could potentially target SOCS-1, an established inhibitor of the Jak-Stat-3 pathway, and thereby modulate IL-6-dependent Stat-3 activation.
In reported series of lung cancers, let-7 expression is downregulated in association with Ras expression in the setting of activating Ras mutations, and decreased expression of let-7a2 has been shown to correlate with a poor prognosis (12,30,31). However, these observations are likely to be cell type-specific since let-7 is only sporadically reduced in tumor types other than lung cancer. Our experimental model differs considerably from these studies in representing a state of persistent cytokine stimulation, and it is unknown whether let-7 expression can be modulated by IL-6 in a similar manner in lung cancer. Although augmenting let-7 expression is being touted as a potential ther-FIGURE 8. STAT-3 phosphorylation is increased by enforced expression of IL-6 in vivo. A, tumor cell xenografts in nude mice (three for each cell type) were excised once they grew to a volume of ϳ200 mm 3 and tissue-homogenized. The expression of Tyr 705 phosphorylated Stat-3, its negative upstream modulator NF2, and the downstream anti-apoptotic Stat-3 targets survivin and Mcl-1 were assessed by Western blot analysis. Quantitative data of mean and 95% confidence intervals from three separate blots are shown. B, Mz-IL-6 tumor xenografts were treated with anti-let-7a or control miRNA inhibitor, and homogenates were obtained for immunoblot analysis. Representative blots and quantitative data showing the average and 95% confidence interval of four separate blots are shown. C, immunohistochemistry for phosphorylated Stat-3 and NF2 was performed on xenograft sections, showing a decrease in phospho-Stat-3 expression and an increase in NF2 in xenografts injected with anti-let-7a when compared with controls. *, p Ͻ 0.05 relative to respective controls. MARCH 16, 2007 • VOLUME 282 • NUMBER 11 apeutic strategy, such approaches may be inappropriate for cancers that are associated with elevated IL-6 levels such as cholangiocarcinoma. In contrast, potential interventions to decrease survival signaling and Stat-3 activation by IL-6 may be a useful approach for these cancers.