¹Department of Obstetrics and Gynecology, National Taiwan University Hospital, No. 7 Chung-Shan South Road, Taipei, Taiwan

²Laboratory of Molecular & Cellular Toxicology, Institute of Toxicology, College of Medicine, National Taiwan University, Taipei, Taiwan

Correspondence to: C-Y Hsieh, Department of Obstetrics and Gynecology, National Taiwan University Hospital, No. 7 Chung-Shan South Road, Taipei, Taiwan. E-mail: cyhsieh@ha.mc.ntu.edu.tw

^aL-H Wei and M-L Kuo contributed equally to the article

Interleukin-6 (IL-6), a multifunctional cytokine, has recently been implicated in human cervical cancer, though the mechanism remains elusive. This study demonstrates that the anti-apoptotic protein Mcl-1 and IL-6 was concomitantly expressed in human cervical cancer tissues and cell lines, but not in normal cervix tissues. Upon IL-6 treatment, Mcl-1, but not other Bcl-2 family members, was rapidly up-regulated peaking at 4-8 h in human cervical cancer C33A cells. Supporting this observation, using anti-IL-6 or anti-IL-6 receptor antibody to interrupt the IL-6 autocrine loop in SiHa cells significantly reduced cellular level of Mcl-1. This study hypothesizes that the expression of Mcl-1 in cervical cancer cells is regulated by IL-6. The matter of which signaling pathways transduced by IL-6 is responsible for the Mcl-1 up-regulation is further investigated herein. Blocking the STAT3 or MAPK pathway with dominant-negative mutant STAT3F or the MEK inhibitor PD98059 failed to inhibit IL-6-mediated Mcl-1 expression. Meanwhile, the IL-6-induced Mcl-1 up-regulation was effectively abolished by treatment with PI 3-K inhibitors, LY294002. Additionally, overexpression of dominant-negative (dn) Akt in C33A cells could inhibit the IL-6-induced increase of Mcl-1. Finally, overexpression of IL-6 in C33A cells caused a markable resistance to apoptosis induced by doxorubicin or cisplatin. Transient transfection of IL-6-overexpressed cells with a mcl-1 antisense vector, leading to the attenuation of their apoptosis-resistant activity. In conclusion, the data herein suggest that IL-6 regulated the mcl-1 expression via a PI 3-K/Akt-dependent pathway that may facilitate the oncogenesis of human cervical cancer by modulating the apoptosis threshold. Oncogene (2001) 20, 5799-5809.

cervical cancer; interleukin-6; apoptosis; Mcl-1; PI 3-K/Akt

Introduction

Normal tissue homeostasis requires constant regulation of cell growth and cell death. Recent evidence implies that the failure of cells to undergo apoptotic cell death may influence the pathogenesis of various human diseases, including cancer, autoimmune diseases, and viral infections (Thompson, 1995). Bcl-2 and its related cytoplasmic proteins are key regulators of apoptosis. At least 15 Bcl-2 family members have been identified in mammalian cells, and either inhibit (e.g., Bcl-2, Bfl-1, Bcl-x_L, and Mcl-1) or promote cell death (e.g., Bax, Bik, and Bad) (Adams and Cory, 1998). All pro-survival bcl-2-like genes are oncogenic because they provide a growth advantage that may eventually lead to neoplastic transformation. Indeed, overexpression of Bcl-2, Bfl-1, Bcl-x_L, and Mcl-1 proteins have been noted in lymphoma, mast cell leukemia, multiple myeloma, gastric cancer, neuroblastoma, and prostate cancers (Schlaifer et al., 1995; Cervero et al., 1999; Choi et al., 1995; Dole et al., 1995; Krajewska et al., 1996). These findings reflect the important concept that apoptosis regulation by altering the level of Bcl-2 family proteins may contribute to neoplastic development (Chao and Korsmeyer, 1998). No doubt exists that clarifying the cellular factor(s) responsible for regulating Bcl-2 family proteins is an urgent task.

Interleukin-6 (IL-6), originally identified as a B-cell differentiated factor, is a multifunctional cytokine on various tissues and cells (Kishimoto, 1989). IL-6 regulates immune and inflammatory responses, hepatic acute-phase protein synthesis, hematopoiesis, and bone metabolism through its specific binding receptor (IL-6-Rp80) and signal transducer gp130 (Kishimoto et al., 1995). IL-6 is also important in regulating tumor cell growth, and can be produced by various types of cancer cells, including multiple myeloma, renal cell carcinoma, prostate carcinoma, ovarian carcinoma, and cervical carcinoma by autocrine and/or paracrine mechanisms (Kawano et al., 1988; Miki et al., 1989; Okamoto et al., 1997; Wu et al., 1992; Eustace et al., 1993).

IL-6 can act as an apoptosis inducer or inhibitor depending upon the types of cells (Kawano et al., 1988; Nakashima et al., 1999). The anti-apoptotic genes bcl-x_L and mcl-1, which were correlated with the status of the multiple myeloma, were found to be up-regulated by IL-6 in clinical multiple myeloma tissues and cell lines (Puthier et al., 1999b). Furthermore, the activation of JAK/STAT3 pathway was preferentially required for IL-6-induced bcl-x_L and mcl-1 gene expression in human multiple myeloma cells (Catlett-Falcone et al., 1999; Puthier et al., 1999a). However, current knowledge of the regulation mechanism of survival signaling elicited by IL-6 is largely restricted to cells of hematopoietic origin.

Cervical cancer is a leading cause of cancer-related deaths among women worldwide (Ponten et al., 1995). Epidemiologic and laboratory data suggest that cervical cancer typically arises from a series of causal steps including, HPV infections, oncogenes expression (e.g., c-MYC, Ha-RAS, and ERB-2) and/or up-regulation of pro-inflammatory cytokines (Pinion et al., 1991; Tartour et al., 1994, 1999). IL-6 has been reported to influence the pathogenesis of cervical cancer, because IL-6 is a central pro-inflammatory cytokine involved in female genital infection and because cervical cancer frequently develops in close association with chronic inflammation caused by infection with various sexually transmitted agents (Richter et al., 1999). Additionally, IL-6 was found to be highly expressed in invasive cervical carcinoma, but either not expressed or only barely expressed in normal cervixes or in pre-neoplastic lesions (Tartour et al., 1994). This study attempts to elucidate the role of IL-6 in the development of cervical cancers, with emphasizing its regulation of apoptosis. Our data indicate that IL-6 and Mcl-1 are co-expressed in human cervical cancer tissues and cell lines. Furthermore, IL-6 treatment effectively inhibits apoptosis of cervical cancer cells through induction of Mcl-1 via a PI 3K/Akt-dependent pathway.

Results

Concomitant expression of Mcl-1 and IL-6 in cervical cancer tissues and cell lines

Tumor specimens randomly selected from patients with invasive cervical cancer were immunohistochemically stained with various antibodies specific to anti-Bcl-2 family proteins, including Bcl-2, Bcl-x_L, and Mcl-1. In normal uterine cervix, Mcl-1 immunostaining was weakly present in mature superficial squamous epithelial cells, but not in other cell types (Figure 1a). Meanwhile, in invasive squamous cell carcinoma, diffuse and strong immunostaining was observed in the neoplastic cells for Mcl-1 proteins (Figure 1b). In contrast, no apparent immunoactivity of Bcl-2 was detected in either the neoplastic cells or the normal cervical epithium, but positive immunostaining of Bcl-2 occurred in the lymphocytes of stroma (Figure 1c). For Bcl-x_L, a mild but homogenous immunostaining was distributed in the basal cells of normal squamous epithelium (Figure 1d). Meanwhile, in squamous carcinoma, Bcl-x_L immunostaining also displayed a weak but heterogenous pattern (Figure 1e). The above observations suggest that Mcl-1, but not Bcl-2 or Bcl-x_L, appeared to have higher levels in human cervical cancer tissues than in normal cervix tissues.

Our previous investigation also showed that IL-6 levels were significantly increased in cervical cancer tissues (Wei et al., 2001), prompting us to investigate the possible relationship between the expression of IL-6 and Mcl-1. To address this issue, six randomly selected biopsies from invasive cervical cancer patients, and five independent cervical cancer cell lines, were employed to analyse the level of IL-6 and Bcl-2 family proteins using Western blotting. According to Figure 2a, a strict correlation exists between the levels of IL-6 and Mcl-1 in these six cervical cancer tissues, whereas the expression of other Bcl-2 family proteins, such as Bcl-x_L and Bax, is not correlated with the IL-6 level. Bcl-2 proteins could not be detected in the tumor specimen. Notably, the level of IL-6 and Mcl-1 proteins was extremely low and could not be detected in normal epithelial counterparts (data not shown).

Subsequently, whether the close relationship between the expression of IL-6 and Mcl-1 could also be observed in the human cervical cancer cell model was determined. Two cell lines, C33A and MS751, which produced little or no IL-6 (<50 pg/ml), also displayed an extremely low level of Mcl-1 (Figure 2b). In contrast, cell lines including SiHa, HeLa and HT3, which secreted large amounts of IL-6 (750-3000 pg/ml), displayed higher levels of Mcl-1 (Figure 2b). Interestingly, the expression of other Bcl-2 family proteins did not differ in these cancer cell lines.

These experimental results indicate that a highly coordinated expression of IL-6 and Mcl-1 was observed in both the clinical tumor specimen and the in vitro cell systems.

IL-6 regulates Mcl-1 expression in human cervical cancer cells

Whether IL-6 directly regulates the expression of Mcl-1 in human cervical cancer cells was explored next. Initially, C33A cells, a cell line which lacks IL-6 secretion, were serum starved for 48 h, then treated with 50 ng/ml of IL-6 for varying time periods. Western blotting revealed that upon IL-6 treatment, the level of Mcl-1 protein rapidly increased at 2 h, peaked at 4-8 h (approximately 5-6-fold), and then declined after 10 h (Figure 3a). Meanwhile, the expression level of the other Bcl-2 family protein, Bax, was not affected by IL-6. Whether the level of Mcl-1 changed with varying concentrations of IL-6 was further investigated. Upon stimulation with 1 ng/ml of IL-6, the level of Mcl-1 increased significantly, to around 2.5 times the level in the control cells (Figure 3b). Meanwhile, the expression of Mcl-1 was maximized at a concentration of 10 ng/ml (Figure 3b). To address the same question, another experiment was designed to interrupt the autocrine loop of IL-6 in the SiHa cells and examine its affect on the level of Mcl-1 protein. To interrupt IL-6 autocrine machinery, SiHa cells were exposed to different antibodies that could either specifically neutralize cytokine IL-6 or target the chain of IL-6 receptor. As illustrated in Figure 3c when SiHa cells were incubated with either anti-IL-6 antibody or anti-IL-6 receptor -chain antibody, it caused a reduction of approximately 50-60% in endogenous Mcl-1 level. However, treatment with control antibody IgG did not influence the level of Mcl-1 in the SiHa cells. In addition to the SiHa cells, the endogenous Mcl-1 level of the HeLa and HT-3 cells could also be partially attenuated by treatment with anti-IL-6 antibody (data not shown).

Based on the above results, we hypothesize that IL-6 functions as an autocrine factor that regulates the cellular level of Mcl-1 in human cervical cancers.

The PI3-K/Akt pathway is required for IL-6-mediated Mcl-1 expression

Unraveling which signaling pathways are involved in IL-6-mediated Mcl-1 up-regulation in human cervical cancer cells is the next topic of interest. Extensive studies of hematipoietic cells have demonstrated that binding IL-6 to the subunit of its receptor triggers the recruitment of gp130, subsequently activating downstream signaling pathways, including JAK/STAT3, MAPK, and PI 3-K (Catlett-Falcone et al., 1999; Chen et al., 1999; Ogata et al., 1997). First, possible involvement of the STAT3 pathway in IL-6-mediated Mcl-1 up-regulation was examined. To examine this issue, control vector and STAT3 dominant-negative mutant (STAT3F), in which Tyr-705, a phosphoacceptor site of STAT3, was mutated to phenylalanine (Nakajima et al., 1996), were introduced into C33A cells. Stable expression of STAT3F resulted in inhibition of the endogenous STAT3 function, as evidenced by the decreasing of phosphorylated STAT3 with a specific anti-phospho-STAT3 antibody (Figure 4a). However, the IL-6-mediated up-regulation of Mcl-1 was not affected in neo control or STAT3F cells (Figure 4b), suggesting that STAT3 is not involved in IL-6-stimulated Mcl-1 up-regulation in cervical cancer C33A cells. Whether the remaining two signaling pathways, MAPK and PI 3-K, were required for IL-6-induced Mcl-1 expression was also examined. To address this, two chemical inhibitors PD98059 and LY294002, which specifically inhibit MEK1 and PI 3-K respectively, were used. As Figure 4c, upper blot, verifies, the IL-6-induced increase of phosphorylated ERK, as expected, was reduced by treatment with 25 or 50 M PD98059. Whereas PD98059 did not affect IL-6 mediated up-regulation of Mcl-1 (Figure 4c, lower blot).

As Figure 4d reveals, either 25 or 50 M LY294002 significantly decreased the level of Mcl-1 induced by IL-6. Under the identical circumstances, IL-6-stimulated PI 3-K activity was completely blocked by LY294002, as detected by the decreased phosphorylated form of Akt proteins using a specific antibody (Figure 4e). Total intracellular Akt proteins were not altered during drug treatment (Figure 4e). Because Akt could be phosphorylated by IL-6 (Figure 4e), a dominant-negative (dn) Akt vector was transfected into C33A cells herein to examine its influence on IL-6-mediated Mcl-1 up-regulation. A stable dn-Akt-overexpressed cell line exhibited a marked reduction of Mcl-1 expression induced by IL-6 as compared to neo control or parental C33A cells (Figure 4f). The above findings indicate that the PI 3-K and its downstream Akt are activated by IL-6 and mediate the signaling to up-regulate the level of Mcl-1.

IL-6-overexpressing cells confers resistance to apoptosis

To determine the exact role of IL-6 in human cervical cancer cells, the IL-6 expressing vector (pCMV-IL-6) and control vector (pCMV) were transfected into human C33A cells. Following transfection, cells were cultured in a medium containing 600 g/ml of G418. Each colony that grew after G418 selection was picked and expanded. The C33A cells transfected with the IL-6 expression vectors were designated as C33A/IL-6 cells (C33A/IL-6-cl-5), with a vector control of C33A/neo. ELISA analysis confirmed that these IL-6-overexpressing clones produced substantially high levels of IL-6 (>2000 pg/ml) than neo control cells (<50 pg/ml) (Figure 5a). Furthermore, the amounts of IL-6 receptor were quantified by using immunoblotting with an antibody, which recognized the ligand binding subunit of the IL-6 receptor. However, neither the IL-6-overexpressing cells nor the neo control cells caused any change in the level of IL-6 -chain receptor (Figure 5b). Compatible with IL-6 levels, immunoblotting showed that these IL-6 transfectants did indeed display a higher level of Mcl-1 protein than the neo control cells (Figure 5c). The data from the IL-6 transfectants is consistent with the results from exogenous IL-6 treatment (as shown in Figure 3a,b) revealing that IL-6 can directly regulate the expression of Mcl-1. These IL-6-overexpressing clones have been examined their growth properties by determining MTT assay. No significant difference exists in growth rate between IL-6 transfectants and neo control cells (data not shown).

To explore the anti-apoptotic effect of IL-6, C33A/IL-6-c5, C33A/neo, or parental C33A cells were treated with 3.5 M of doxorubicin for 24 h. The characteristic morphology of apoptosis was subsequently determined by staining with Hoechst 33258. As illustrated in Figure 6a I-VI, the typical apoptotic morphologies including, chromatin condensation and nuclear fragmentation, were clearly observed in neo control cells after treatment with doxorubicin, but not in IL-6-overexpressing cells (C33A/IL-6-c5). Furthermore, MTT colormetric analysis was also employed herein to determine the general cell survival of IL-6-overexpressing C33A clones (c1 and c5) and control cells. IL-6-overexpressing cells consistently exhibited much more resistance to various concentrations of cisplatin treatment than neo control cells (Figure 6b). The above experimental results suggest that IL-6-overexpressing C33A cells exhibited a marked resistance to drugs, which may mediate through the inhibition of apoptosis.

Finally, it is interesting to elucidate the contribution of Mcl-1 in the anti-apoptotic effect of IL-6. To clarify the contribution of Mcl-1, this study attempts to establish a stable clone that could constitutively express the antisense mcl-1 in C33A/IL-6-c5 cells. However, no stable clones that survived and could express antisense mcl-1 were obtained, and thus a transient-transfection death assay was conducted. C33A/IL-6-c5 cells were cotransfected with GFP expression plasmid (pCMV-GFP) plus either empty vectors (pCDNA3) or antisense mcl-1 vectors. Forty-eight hours after transfection, transfected cells were transferred to a serum-free medium containing doxorubicin for a further 24 h. After treatment, the extent of cell death was determined based on the number of surviving cells expressing GFP fluorescence. As Figure 7a displays, transfection of the antisense mcl-1 vector, but not the control vector, increased the sensitivity of C33A/IL-6-c5 cells to doxorubicin. Supporting this observation, the antisense mcl-1 transfection also effectively reduced the endogenous level of Mcl-1 in C33A/IL-6-c5 cells (Figure 7b). These findings suggest that IL-6 regulated the anti-apoptotic mechanism in human cervical cancer cells primarily through up-regulation of Mcl-1.

Discussion

Using an immunohistochemical approach, the current results show that Mcl-1 is the most abundant anti-apoptotic proteins among Bcl-2 family members in human cervical cancer tissues. Mcl-1 and Bcl-2 are expressed at different stages of differentiation in numerous normal tissues. These two proteins commonly appeared in gradients with opposing directions, such that the expression of Bcl-2 tended to be higher in the less differentiated cells lining the basement membrane whereas Mcl-1 was more intense in the differentiated cells (Krajewski et al., 1995). These findings imply that the expressions of the Mcl-1 and Bcl-2 proteins are mutually exclusive. Thus, it is not surprising that Mcl-1 expression is more widespread in cervical cancer than Bcl-2 since most cervical cancers are, histologically, well-differentiated, large cell type squamous cell carcinomas (Robert and Fu, 1990). Although immunohistochemical staining does not provide actual quantitative data for target proteins, it can be utilized to distinguish which type of cells are positive for such an expression. The quantitative data from Western blotting (see Figure 2a) clearly revealed that the level of Mcl-1 protein is much higher than that of other Bcl-2 family members in cervical cancer specimens. Other investigators obtained a similar finding (Crawford et al., 1998), namely that Mcl-1 expression was positive in 57% of primary and 75% of recurrent cervical cancers.

Mcl-1 was originally identified as an early gene induced during differentiation of ML-1 human myeloblastic leukemia cells (Kozopas et al., 1993). Unlike Bcl-2, Mcl-1 contains a PEST sequence, which probably presents itself as a labile protein with an estimated half-life of 1-3 h (Yang et al., 1995). The intracellular distribution and pattern of expression of Mcl-1 overlap with those of Bcl-2 with a prominent mitochondrial localization, indicating they share the similar function of delayed apoptotic cell death (Yang et al., 1995). Indeed, the role of Mcl-1 in preventing cell death was first demonstrated in Chinese hamster ovary cells undergoing apoptosis in response to c-myc over-expression (Reynolds et al., 1994). Similarly, Mcl-1 can interacting with Bax in hematopoietic FDC-P1 cells and can prolong cell viability under various cytotoxic conditions (Zhou et al., 1997). Additionally, it has been proven that mcl-1 is tightly regulated by the GM-CSF signaling pathway and is one component of the GM-CSF viability response (Chao et al., 1998; Moulding et al., 1998). However, the above studies almost all focused on hematepoietic cell systems. The present investigation is the first to demonstrate that Mcl-1 is also critical in overriding apoptotic cell death in human cervical cancer cells.

Recently, many investigators have considered the possible role of IL-6 in the pathogenesis of human cervical cancers (Castrilli et al., 1997; Tjiong et al., 1999), but they made little progress. The lack of progress most likely results from the downstream effector of IL-6 remaining unidentified. The present study provides a novel finding, namely that inflammatory cytokine IL-6 can up-regulate the expression of Mcl-1, as evidenced by the coexpression of IL-6 and Mcl-1 in the tumor specimen and in in-vitro cell model. Our previous investigations (Wei et al., 2001) have demonstrated that IL-6 and its -chain receptor are over-expressed in human cervical cancer tissues but not in their normal counterparts, suggesting that IL-6 is an autocrine factor in human cervical cancer. Besides, tumor size is closely correlated with the level of IL-6. Results of the present study at least partially support previous clinical observations, namely that overexpression of IL-6 rendered cervical cancer cells is more refractory to apoptosis (Figure 5a). The increase of the apoptosis threshold in cervical cancer cells with IL-6 armed them against attack by immune cells and allowed pre-neoplastic or neoplastic cells to maintain their growth. Supporting the above findings, IL-6 has been found to be crucial to the oncogenesis of human multiple myeloma (Kawano et al., 1988). The anti-apoptotic Mcl-1 protein was also tightly regulated by IL-6 in human multiple myeloma cells (Puthier et al., 1999b).

Binding of IL-6 and its receptor triggers the recruitment of the signal-transducing protein gp130, and subsequently activates JAK/STAT or Ras/MAP kinase pathways, which have been implicated in the survival and/or proliferation of multiple myeloma (Catlett-Falcone et al., 1999; Ogata et al., 1997). Further study showed that IL-6 mediated Mcl-1 up-regulation in human multiple myeloma cells through the JAK/STAT3 rather than the MAPK pathway (Puthier et al., 1999a). However, the results herein have confirmed that a PI 3-K/ Akt-, but not STAT3 or MAPK, dependent pathway resulted mainly from IL-6-induced Mcl-1 up-regulation in human cervical cancer cells. This finding implies that the signaling pathways leading to the Mcl-1 induction may vary according to the type of cell models or stimuli employed. Our earlier studies and others (Chen et al., 1999; Qiu et al., 1998) have obtained similar results, showing that the PI 3-K/ Akt pathway is important for IL-6-induced Mcl-1 expression in human hepatoma and prostatic cancer cells. The PI 3-K/ Akt signaling pathway has recently attracted extensive attention because it may affect multiple cellular processes, including increasing cellular proliferation, inhibiting apoptosis and eventually facilitating tumorigenesis (Klippel et al., 1998; Tlsty, 1999). Recently, PIK3CA has been shown to be an oncogene in cervical cancer (Ma et al., 2000). Amplification of PIK3CA may activate PI 3-K/ Akt signaling and contributes to cervical tumorigenesis.

Carcinogenesis involves an imbalance between the regulation of cell proliferation and apoptotic cell death. One related study has demonstrated that transformation of colorectal epithelium to adenomas and carcinomas is associated with progressive inhibition of apoptosis (Elder et al., 1996). Supportive evidence from our previous study showed that suppression of apoptosis by overexpression of Bcl-2 promoted cellular mutagenesis and inhibited DNA repair (Kuo et al., 1999). Therefore the IL-6-mediated Mcl-1-dependent anti-apoptotic effect should be an important mechanism for human cervical carcinogenesis. To the best of our knowledge, this is the first work to demonstrate that IL-6 up-regulates the apoptotic threshold through a PI 3-K/ Akt-dependent up-regulation of Mcl-1, which may potentially influence the oncogenesis of human cervical cancer.

Materials and methods

Tissues and cell culture

Human cervical cancer samples were obtained from surgical specimens of FIGO stage Ib-IIa uterine cervical cancer patients. The tumors were primary site cancers that had not been treated with chemotherapy or irradiation.

Tumor cell lines were obtained from the American Type Culture Collection (Rockville, MD, USA). The following were maintained in DMEM: C33A (carcinoma, cervix, human), SiHa (squamous carcinoma, cervix, human), Hela (epitheloid carcinoma, cervix, human), and MS-751 (epidermoid carcinoma, cervix, metastasis to lymph node, human). Meanwhile, HT-3 (carcinoma, cervix, metastasis to lymph node, human) was maintained in McCoy's 5a medium (Life Technologies, NY, USA). All of the culture mediums were supplemented with 10% (vol/vol) fetal bovine serum (FBS). The cell lines were cultured in a humidified atmosphere of 5% CO₂ and 95% air at 37°C.

Antibodies and reagents

Recombinant human IL-6, antibody to human IL-6, antibody to human bcl-2, and antibody to human Bcl-x_L were obtained from R&D systems. Meanwhile, antibodies to human Mcl-1, human Bax, human IL-6 receptor chain, and phospho-Erk, were from Santa Cruz Biotechnology. Antibodies to STAT, phospho-STAT, Akt and phospho-Akt were from Upstate Biotechnology. Furthermore, the kinase inhibitors, such as LY294002, and PD98059, were from Sigma. Meanwhile, mammalian expression plasmid for the dominant-negative mutant of STAT3, STAT3F, was kindly provided by Dr T Hirano (Nakajima et al., 1996), while dominant-negative mutant of Akt vector was kindly donated by Dr RH Chen, at the Institute of Molecular Medicine, National Taiwan University, Taipei, Taiwan. Finally, the plasmid containing the antisense Mcl-1 was constructed as described previously (Jee et al., 2001).

Immunohistochemical assay

Immunohistochemical analysis of Mcl-1, Bcl-2, and Bcl-x_L proteins in tissue sections was performed as described previously (Krajewski et al., 1994). Briefly, the slides were re-hydrated in PBS for 15 min and the endogenous peroxydase was inhibited by 3% H₂O₂/methanol for 10 min at room temperature. For blocking, 5% non-fat milk/PBS was used for 30 min at room temperature. Slides were incubated with either monoclonal anti-human Bcl-2 antibody, Bcl-x_L antibody, or rabbit polyclonal anti-human Mcl-1 antibody at a 1 : 50 dilution for 16 h at 4°C. The peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology) were used in a 1 : 200 dilution and incubated for 1 h at room temperature. For all antibodies, staining was developed by immersing slides in 0.06% 3,3'-diaminobenzidine tetrahydrochloride (DAKO), followed by counterstaining with Gill's Hematoxylin V.

Enzyme immunoassay

The IL-6 levels of the cell culture supernatant were determined via enzyme immunoassay with commercially available kits (R&D Systems) according to the manufacturer's instructions. IL-6 measurements in the cell culture supernatant were taken twice, and the average value of IL-6 was recorded as pg/ml.

Immunoblotting

Cells were lysed in a lysis buffer (1% Triton X-100, 150 mM NaCl, 1 mM EGTA, 1% NP-40, 1 mM NaF, 1 mM Na₃VO₄, 2 mM phenylmethyl- sulfonylfluoride (PMSF), 1 g/ml aprotinin and leupeptin in PBS) and the tissue samples were homogenized in a lysis buffer. The lysates were centrifuged at 12 000 r.p.m. for 25 min at 4°C. Protein concentration was measured with a Bio-Rad protein assay (Hercules, CA, USA). A 50 g protein sample was separated by 12.5% SDS-PAGE, transferred onto polyvinylidene difluoride (PVDF) membrane, and immunoblotted with various antibodies. Bound antibodies were detected using appropriate peroxidase-coupled secondary antibodies, followed by an enhanced chemiluminescent detection system (ECL, Boehringer Mannheim).

Cell transfection and generation of stable cell lines

The pCMV-IL-6, a constitutive expression vector, carries 0.64 kb full-length human IL-6 cDNA under the control by the CMV promoter/enhancer sequence. The pCMV-IL-6 or pCMV was transfected into the C33A cell line with the TransFast^TM transfection reagent (Promega). After 24 h, cells were replated in DMEM with 10% FBS and 600 g/ml G418. Clones resistant to G418 were selected and expanded. The level of IL-6 and IL-6 receptor chain of each clone was determined using enzyme immunoassay and Western blotting, respectively.

For transfection of dominant-negative Akt and dominant-negative STAT3, C33A cells (1 ´ 10⁶) were transfected with TransFast^TM transfection reagent (Promega) according to manufacturer's instructions. For selecting stable clones, G418 was added to cultured medium 72 h after transfection.

Internucleosomal DNA fragmentation and nuclear condensation

Morphological evaluation of apoptosis was performed by Hoechst staining. Briefly, the medium was gently removed at the indicated time to prevent the detachment of cells. Cells were fixed with 4% formaldehyde in PBS for 5 min at room temperature, incubated with Hoechst dye 33258 at 5 g/ml in PBS for 5 min, washed and finally mounted in PBS : glycerol (3 : 1). Fluorescent H33258-stained nuclei were observed with a fluorescence microscope (Olympus, Japan).

Cell viability determined by MTT assay

Cells were plated into 96-well microplates at a density of 5 ´ 10³ cells/well for cell viability assay. Briefly, the cells were cultured at 37°C for the indicated time, and 30 l of MTT solution (5 mg/ml) was added into each well, then incubated for 4 h in darkness. The formazan grain was then dissolved in DMSO, and the absorbance at 570 nm was read using an ELISA plate reader.

Transient transfection and detection of cell viability

C33A/IL-6-c5 cells were plated 24 h before transfection at a density of 1 ´ 10⁵ cells/well in a 6-well plate. The cells were co-transfected with 2 g green fluorescent protein (GFP) expression vector and plasmid containing the antisense Mcl-1 (8 g) or pCDNA3 control vector using the TransFast^TM transfection reagent (Promega). Transfection was performed in duplicate. Twenty-four hours after transfection, the cell medium was replaced with a fresh serum-free medium for 24 h, and then treated with Doxorubicin (3.5 M). Following 24 h treatment, cells were washed and fixed in PBS containing 4% formaldehyde. These cells were washed twice with PBS and observed under a fluorescence microscope. The GFP-positive cells in each well were then counted.

Acknowledgements

The authors would like to thank the National Science Council of the Republic of China for financially supporting this research under Contract No. NSC89-2314-B-002-261.

Adams JM, Cory S. (1998). Science 281, 1322-1326. Article MEDLINE

Castrilli G, Tatone D, Diodoro MG, Rosini S, Piantelli M, Musiani P. (1997). Br. J. Cancer 75, 855-859. MEDLINE

Catlett-Falcone R, Landowski TH, Oshiro MM, Turkson J, Levitzki A, Savino R, Ciliberto G, Moscinski L, Fernandez-Luna JL, Nunez G, Dalton WS, Jove R. (1999). Immunity 10, 105-115. MEDLINE

Cervero C, Escribano L, San Miguel JF, Diaz-Agustin B, Bravo P, Villarrubia J, Garcia-Sanz R, Velasco JL, Herrera P, Vargas M, Gonzalez M, Navarro JL, Orfao A. (1999). Am. J. Hematol. 60, 191-195. MEDLINE

Chao DT, Korsmeyer SJ. (1998). Annu. Rev. Immunol. 16, 395-419. MEDLINE

Chao JR, Wang JM, Lee SF, Peng HW, Lin YH, Chou CH, Li JC, Huang HM, Chou CK, Kuo ML, Yen JJ, Yang-Yen HF. (1998). Mol. Cell. Biol. 18, 4883-4898. MEDLINE

Chen RH, Chang MC, Su YH, Tsai YT, Kuo ML. (1999). J. Biol. Chem. 274, 23013-23019. Article MEDLINE

Choi SS, Park IC, Yun JW, Sung YC, Hong SI, Shin HS. (1995). Oncogene 11, 1693-1698. MEDLINE

Crawford RA, Caldwell C, Iles RK, Lowe D, Shepherd JH, Chard T. (1998). Br. J. Cancer 78, 210-214. MEDLINE

Dole MG, Jasty R, Cooper MJ, Thompson CB, Nunez G, Castle VP. (1995). Cancer Res. 55, 2576-2582. MEDLINE

Elder DJ, Hague A, Hicks DJ, Paraskeva C. (1996). Cancer Res. 56, 2273-2276. MEDLINE

Eustace D, Han X, Gooding R, Rowbottom A, Riches P, Heyderman E. (1993). Gynecol. Oncol. 50, 15-19. MEDLINE

Jee SH, Shen SC, Chiu HC, Tsai WL, Kuo ML. (2001). Oncogene 20, 198-208. MEDLINE

Kawano M, Hirano T, Matsuda T, Taga T, Horii Y, Iwato K, Asaoku H, Tang B, Tanabe O, Tanaka H, et al. (1988). Nature 332, 83-85. MEDLINE

Kishimoto T. (1989). Blood 74, 1-10. MEDLINE

Kishimoto T, Akira S, Narazaki M, Taga T. (1995). Blood 86, 1243-1254. MEDLINE

Klippel A, Escobedo MA, Wachowicz MS, Apell G, Brown TW, Giedlin MA, Kavanaugh WM, Williams LT. (1998). Mol. Cell. Biol. 18, 5699-5711. MEDLINE

Kozopas KM, Yang T, Buchan HL, Zhou P, Craig RW. (1993). Proc. Natl. Acad. Sci. USA 90, 3516-3520. MEDLINE

Krajewski S, Bodrug S, Gascoyne R, Berean K, Krajewska M, Reed JC. (1994). Am. J. Pathol. 145, 515-525. MEDLINE

Krajewski S, Bodrug S, Krajewska M, Shabaik A, Gascoyne R, Berean K, Reed JC. (1995). Am. J. Pathol. 146, 1309-1319. MEDLINE

Krajewska M, Krajewski S, Epstein JI, Shabaik A, Sauvageot J, Song K, Kitada S, Reed JC. (1996). Am. J. Pathol. 148, 1567-1576. MEDLINE

Kuo M, Shiah SG, Wang CJ, Chuang SE. (1999). Mol. Pharmacol. 55, 894-901. MEDLINE

Ma YY, Wei SJ, Lin YC, Lung JC, Chang TC, Whang-Peng J, Liu JM, Yang DM, Yang WK, Shen CY. (2000). Oncogene 19, 2739-2744. MEDLINE

Miki S, Iwano M, Miki Y, Yamamoto M, Tang B, Yokokawa K, Sonoda T, Hirano T, Kishimoto T. (1989). FEBS Lett. 250, 607-610. MEDLINE

Moulding DA, Quayle JA, Hart CA, Edwards SW. (1998). Blood 92, 2495-2502. MEDLINE

Nakajima K, Yamanaka Y, Nakae K, Kojima H, Ichiba M, Kiuchi N, Kitaoka T, Fukada T, Hibi M, Hirano T. (1996). EMBO J. 15, 3651-3658. MEDLINE

Nakashima K, Wiese S, Yanagisawa M, Arakawa H, Kimura N, Hisatsune T, Yoshida K, Kishimoto T, Sendtner M, Taga T. (1999). J. Neurosci. 19, 5429-5434. MEDLINE

Ogata A, Chauhan D, Teoh G, Treon SP, Urashima M, Schlossman RL, Anderson KC. (1997). J. Immunol. 159, 2212-2221. MEDLINE

Okamoto M, Lee C, Oyasu R. (1997). Cancer Res. 57, 141-146. MEDLINE

Pinion SB, Kennedy JH, Miller RW, MacLean AB. (1991). Lancet 337, 819-820. MEDLINE

Ponten J, Adami HO, Bergstrom R, Dillner J, Friberg LG, Gustafsson L, Miller AB, Parkin DM, Sparen P, Trichopoulos D. (1995). Int. J. Cancer 60, 1-26. MEDLINE

Puthier D, Bataille R, Amiot M. (1999a). Eur. J. Immunol. 29, 3945-3950. Article MEDLINE

Puthier D, Derenne S, Barille S, Moreau P, Harousseau JL, Bataille R, Amiot M. (1999b). Br. J. Haematol. 107, 392-395. MEDLINE

Qiu Y, Robinson D, Pretlow TG, Kung HJ. (1998). Proc. Natl. Acad. Sci. USA 95, 3644-3649. Article MEDLINE

Reynolds JE, Yang T, Qian L, Jenkinson JD, Zhou P, Eastman A, Craig RW. (1994). Cancer Res. 54, 6348-6352. MEDLINE

Richter HE, Holley RL, Andrews WW, Owen J, Miller KB. (1999). Am. J. Obstet. Gynecol. 181, 940-944. MEDLINE

Robert ME, Fu YS. (1990). Semin. Diagn. Pathol. 7, 173-189. MEDLINE

Schlaifer D, March M, Krajewski S, Laurent G, Pris J, Delsol G, Reed JC, Brousset P. (1995). Blood 85, 2671-2674. MEDLINE

Tartour E, Fossiez F, Joyeux I, Galinha A, Gey A, Claret E, Sastre-Garau X, Couturier J, Mosseri V, Vives V, Banchereau J, Fridman WH, Wijdenes J, Lebecque S, Sautes-Fridman C. (1999). Cancer Res. 59, 3698-3704. MEDLINE

Tartour E, Gey A, Sastre-Garau X, Pannetier C, Mosseri V, Kourilsky P, Fridman WH. (1994). Cancer Res. 54, 6243-6248. MEDLINE

Thompson CB. (1995). Science 267, 1456-1462. MEDLINE

Tjiong MY, van der Vange N, ten Kate FJ, Tjong AHSP, ter Schegget J, Burger MP, Out TA. (1999). Gynecol. Oncol. 73, 285-291. MEDLINE

Tlsty TD. (1999). Nat. Genet. 21, 64-65. Article MEDLINE

Wei LH, Kuo ML, Chen CA, Cheng WF, Cheng SP, Hsieh FJ, Hsieh CY. (2001). Gynecol. Oncol. 82, 49-56. MEDLINE

Wu S, Rodabaugh K, Martinez-Maza O, Watson JM, Silberstein DS, Boyer CM, Peters WP, Weinberg JB, Berek JS, Bast Jr RC. (1992). Am. J. Obstet. Gynecol. 166, 997-1007. MEDLINE

Yang T, Kozopas KM, Craig RW. (1995). J. Cell Biol. 128, 1173-1184. MEDLINE

Zhou P, Qian L, Kozopas KM, Craig RW. (1997). Blood 89, 630-643. MEDLINE

Figure 1 Immunohistochemical staining of Bcl-2, Mcl-1, and Bcl-x_L proteins in human cervical cancer tissues. (a) In normal squamous epithelium of exocervix, Mcl-1 is weakly expressed in mature and superficial cells. (b) Diffuse and strong immunostaining of Mcl-1 detected in squamous cell carcinoma. (c) Bcl-2-positive staining was found on the lymphocytes of the stroma, but not in the dysplastic squamous epithelium. (d) The basal cells of normal squamous epithelium are positive for Bcl-x_L. (e) A weak and heterogenous immunostaining of Bcl-x_L was in squamous cell carcinoma. Arrow, the typical immunohistochemical positive cells for the cervical cancer tissue section. (Hematoxylin counterstaining, peroxidase-diaminobenzidine; Bars, 50 m)

Figure 2 The co-expression of IL-6 and Mcl-1 in human cervical cancer and cell lines. (a) Western blot analysis of IL-6, Mcl-1, Bcl-x_L, and Bax proteins in biopsies derived from six invasive cervical carcinomas (I-VI). Each tumor tissue was homogenized and extracted, the equal amount of protein extracts were subjected to immunoblotting as described in the Materials and methods. -tubulin was used as an internal control. (b) Five cervical cancer cell lines: C33A, MS-751, SiHa, Hela, and HT3 were plated at a density of 10⁶ cells/ml. Then, the cells were cultured for further 24 h, and IL-6 levels of the culture supernatant were determined by EIA (upper panel) as described in Materials and methods. For determination of Mcl-1 expression, each cell line was cultured at a density of 1 ´ 10⁶ cells/ml in serum-free medium for 48 h. Cells were lysed and protein extracts were obtained to determine the level of Mcl-1, Bax and Bcl-x_L proteins by immunoblotting (down blot)

Figure 3 IL-6 regulates the expression of Mcl-1 in human cervical cancer cells. IL-6 induces Mcl-1 expression in C33A with (a) time-dependent and (b) concentration-dependent manner. Serum-starved C33A cells (5 ´ 10⁵/ml) were treated with 50 ng/ml IL-6 for different periods of time (0, 2, 4, 6, 8, and 10 h) or treated with various concentrations of IL-6 (0, 1, 10, 50, and 100 ng/ml) for 6 h. After treatment, total proteins were extracted and the levels of Mcl-1 were determined by Western blotting. Relative intensities were quantitated by densitometer. (c) Interruption of IL-6 autocrine loop in SiHa cells lead to an attenuation of Mcl-1 level. Serum-starved SiHa cells (5 ´ 10⁵/ml) were treated with anti-IL-6 (50 and 100 ng/ml), anti-IL-6 receptor chain (100 ng/ml), or control IgG (100 ng/ml) antibody for 24 h. The expression of Mcl-1 and Bax protein was determined by immunoblotting and the relative intensities were quantitated by densitometer

Figure 4 IL-6 up-regulates Mcl-1 through PI3K but not MAPK or STAT3 pathway. (a) STAT3-related pathway is not involved in IL-6 up-regulates Mcl-1. Upper blot, Control vector- and dominant-negative STAT3 (STAT3F)-overexpressed C33A cells were checked for their functions by immunoblotting with anti-phospho-STAT3 antibody, which specifically recognizes the phosphotyrosine at Tyr 705 of STAT3 protein. (b) STAT3F and Neo stable transfectants were treated with 50 ng/ml IL-6 or none for 4 h. The level of Mcl-1 protein was determined by Western blot analysis. (c) PD98059 inhibits IL-6-induced ERK1/2 phosphorylation but not Mcl-1 expression. Cell lysates were prepared and used for Western blotting with antibody specific to the phosphorylated ERK1/2 (upper blot) or anti-Mcl-1 antibody (lower blot). (d) PI 3-K inhibitor blocks IL-6-mediated Mcl-1 expression. Serum-starved C33A cells (5 ´ 10⁵/ml) were treated with various agents as indicated. Total cellular proteins were extracted and the expressions of Mcl-1 (upper) or Bax (lower) were determined by Western blot. (e) PI 3-K inhibitors reduce IL-6-induced serine phosphorylation of AKT. Serum-starved C33A (5 ´ 10⁵/ml) were pre-treated with LY294002 (25 or 50 M) for 1 h before IL-6 (50 ng/ml) was added. After further 60 min incubation, total proteins were extracted and the expressions of phospho-Akt and Akt were determined by Western blotting. (f) Dominant-negative Akt (dn-Akt) inhibits IL-6-induced Mcl-1 up-regulation. C33A cells were transfected with dnAkt or control vectors. After G418 selection, cells stably expressed dnAkt or neo control were treated with IL-6 (50 ng/ml) or none for 6 h. Cell lysates were subjected to SDS-PAGE and analysed by Western blot using anti-Mcl-1 and anti-Bax antibodies

Figure 5 Overexpression of IL-6 in human cervical C33A cells resulted in up-regulation of Mcl-1. (a) ELISA analysis of the amount of secreted IL-6 for different IL-6-transfected clones such as C33A/IL-6-cl-5 and control cells (C33A, C33A/neo). For this assay, each clones was incubated in serum free DMEM for 24 h, and then aliquots of medium from each culture was collected to performed ELISA according to the manufacture procedure of R&D. The means (bars, s.d.) were calculated from at least three independent experiments for each sample. (b) Determination of the level of IL-6 receptor -chain for each transfectant was determined by Western blotting. (c) Elevated level of Mcl-1 in IL-6-overexpressing C33A cells. Cell extracts were prepared and immunoblotting preformed as described in Materials and methods

Figure 6 Expression of IL-6 protects C33A cells against apoptosis. (a) Morphological examination of IL-6-transfected C33A and control cells after treatment with doxorubicin. C33A cells were treated with none (I) or 3.5 M doxorubicin (II) for 24 h; C33A/neo cells were treated with none (III) or 3.5 M doxorubicin (IV) for 24 h; C33A/IL-6-c5 cells were treated with none (V) or 3.5 M doxorubicin (VI) for 24 h. After treatment, each cells were determined their morphological changes by staining Hoechst 33258 fluorescent dye. (b) Overexpression of IL-6 in C33A against cisplatin cytotoxicity. C33A/IL-6-cl, c5, neo, and parental C33A cells were exposed to various concentrations of cisplatin for 24 h. Then, the survived cells were fixed and subjected for MTT colorimetric analysis. The means (bars, s.d.) were calculated from at least three independent experiments for each sample

Figure 7 Anti-sense mcl-1 abolishes IL-6-mediated anti-cell death activity. (a) C33A/IL-6-c5 cells were transiently transfected with pCMV-GFP (1 g/well) and either an empty expression control vector (pCDNA3), or a vector encoding antisense mcl-1 (4 g/well). After 48 h of transfection, the medium was replaced with serum-free medium containing 3.5 M doxorubicin for a further 24 h. Subsequently, cells were fixed with 70% ethanol and GFP-positive cells in each field (magnified ´40) were quantitated under fluorescence microscope. The results shown here represent the mean±s.d. of three independent experiments performed in duplicate. (b) The endogenous Mcl-1 level in C33A/IL-6-c5 cells was effectively reduced by anti-sense mcl-1. C33A/IL-6-c5 cells were transfected with anti-sense mcl-1 vector or control vector pCDNA3 for 48 h. After transfection, cell lysates were prepared and analysed by Western blotting using anti-Mcl-1 and anti-tubulin antibodies