Cell–cell contacts induce STAT3 activity in colon carcinoma cells through an autocrine stimulation loop

Abstract

Purpose

Signal transducer and activator of transcription 3 (STAT3) is persistently activated in colorectal carcinoma (CRC) cells in the tumor context, but inactive in cultivated CRC cell lines. We approached mechanisms leading to STAT3 activation in CRC by mimicking “tumor-like” growth conditions and forcing cell–cell contacts of HT-29 CRC cells in culture. Further aims were to analyze in how far HT-29 cells growing in a tumorous manner spread STAT3 activity by secretion of soluble factors.

Methods

Non-adhesive growth and aggregation of HT-29 cells were achieved by cultivation on non-coated plastic surfaces. STAT3 activity was assessed by Western blot employing a phospho-STAT3-specific antibody as well as by electrophoretic mobility shift assay (EMSA). Expression changes of STAT3 target gene mmp-1 were quantified by real-time RT–PCR, Cytokine/chemokine patterns in conditioned media were characterized by cytokine arrays.

Results

Forced aggregation and non-adhesive growth of HT-29 CRC cells resulted in enhanced tyrosine phosphorylation of STAT3 and elevated expression of matrix metalloproteinase MMP-1. Furthermore, an activity was secreted into the medium that evoked STAT3 phosphorylation in adhesively growing HT-29 cells. The degree of activity in the conditioned medium was enhanced when wild-type STAT3 was overexpressed and reduced when a dominant variant of STAT3 was expressed in HT-29 cells. A characteristic panel of chemokines appeared in STAT3-activating conditioned medium.

Conclusions

Changing cultivation conditions of the CRC cell line HT-29 toward detachment and aggregation, thus toward the situation in tumors, induces STAT3 activity and evokes an autocrine STAT3 activation loop.

Electronic supplementary material

The online version of this article (doi:10.1007/s00432-010-0943-3) contains supplementary material, which is available to authorized users.

Introduction

Colorectal carcinoma (CRC) is a malignancy with high incidence and mortality. Roughly 50% of patients diagnosed with CRC eventually die of the disease. Various aspects of dysregulated signal transduction processes and oncogenic mutations have been implicated in the development of CRC, among them malfunction of the Wnt pathway, aberrant activity of k-Ras, mutations in tumor suppressors p53 and Rb, components of transforming growth factor- (TGF-) β-signaling and DNA repair genes (Oving and Clevers 2002).

In recent years, the involvement of STATs (signal transducers and activators of transcription) in different hematopoietic and solid cancers has attracted much attention, and it became evident that aberrant STAT signaling also plays an important role in CRC (Klampfer 2008).

Among the family of STATs, STAT3 mediates a particularly wide spectrum of cellular responses such as proliferation, differentiation or apoptosis, depending on the tissue context. In some cancers, inappropriate STAT3 activity was found significantly associated with poor prognosis and metastasis (Horiguchi et al. 2002; Masuda et al. 2002). We have recently shown that STAT3 is constitutively activated in the vast majority of tumor biopsies obtained from patients with CRC (Corvinus et al. 2005). We could also demonstrate that hyperactivity of STAT3 in CRC cell lines enhances malignancy parameters such as cell proliferation, anchorage-independent growth, invasiveness and expression of matrix metalloproteinases (Tsareva et al. 2007). Contributions of aberrant STAT3 activity to cancerous cell behavior probably involve upregulation of genes promoting cell cycle progression (cyclin D1, c-myc), preventing apoptosis (bcl-xL, mcl-1, survivin) and degrading extracellular matrix (Wilson et al. 1997; Nikkola et al. 2002; Tsareva et al. 2007). We were able to demonstrate that specific inhibition of STAT3 signaling can reduce growth of CRC xenograft tumors in mice (Corvinus et al. 2005).

Notably, most permanent cell lines derived from CRC are devoid of constitutive STAT3 activation, but show tyrosine phosphorylation of STAT3 in response to stimulation with cytokines or upon formation of xenograft tumors in mice (Corvinus et al. 2005 and unpublished results). Various upstream mediators were shown to evoke STAT3 phosphorylation including cytokine and growth factor receptors and several viral or cellular oncogene products such as Src, Fps, polyoma virus middle T-antigen and Sis (Bowman et al. 2000). Our previous results left us with the finding that within colorectal carcinoma tissue, cancer cells are subject to stimulation events that result in STAT3 activation.

The purpose of this study was to define processes that underlie this phenomenon. In particular, we were interested in whether CRC cells themselves are able to induce STAT3 activity under growth conditions resembling the situation in vivo.

Materials and methods

Cell lines and cell culture

Colon carcinoma cell line HT-29 was purchased from the ATCC. HT-29 derivatives over-expressing STAT3 (“HT-29 STAT3”) or a dominant-negative mutant of STAT3 (“HT-29 STAT3 d.n.”) were generated by retroviral infection as described (Tsareva et al. 2007). Cells were grown in RPMI 1640 medium containing 10% fetal calf serum, 200 mM L-Glutamine, 100 mM sodium pyruvate and 1 mg/ml gentamycin. Cultivation was performed in either coated tissue culture plasticware (Greiner Labortechnik) or non-coated Petri dishes for microbiology (Greiner) as indicated in the Sect. “Results”. Cells were harvested at 80% confluency after three PBS washes at 4°C. Whole-cell extracts were prepared from cell pellets as described above.

Conditioned medium from cells cultivated on non-coated plastic surface was obtained by seeding 1.8 × 10⁶ cells from a confluent cell culture flask into 8 ml RPMI 1640 and incubation of this cell suspension in a microbiological Petri dish for 4 days. After this period, medium was cleared from cells by centrifugation (10 min at 1,000 g) and passed through a sterile filter with a pore size of 0.2 μm (Greiner). For stimulation experiments, cells in tissue culture flasks grown to 70 to 80% confluency were incubated with undiluted conditioned medium for 30 min and further treated as described for cytokine stimulation assays. Generation of HT-29 derivatives expressing wild-type STAT3 (HT-29/STAT3) or dominant-negative STAT3 has been described (Corvinus et al. 2005).

Electrophoretic mobility shift assay (EMSA)

Whole-cell extracts were prepared as described above and subjected to Bradford protein determination. Protein concentrations were adapted to equal levels. For analysis of STAT3 complexes, the SIEm67 STAT binding site from the human c-fos promoter was used as a probe. Double stranded blunt ended oligonucleotides were annealed (5′-CATTTCCCGTAAATC-3′) and end-labelled using ³²P-γ-ATP and T4 polynucleotide kinase to a specific activity of 10,000 cpm/fmol as described (Corvinus et al. 2005). Binding reactions were performed by incubating 10,000 cpm of radiolabelled probe with 20 μg of cell lysate for 30 min at room temperature. For supershift reactions of STAT containing complexes, 2 μg of antibody to STAT3 (C-20; Santa Cruz) was added to the binding reactions before EMSA was performed. Samples were separated by electrophoresis through 6% native polyacrylamide gels and analyzed by autoradiography using BioMax sensitive films (Kodak).

Western blot

Western blotting was performed as described previously (Corvinus et al. 2005). Briefly, 20 μg of whole-cell extract was solublized in gel loading buffer (62.5 mM Tris/HCl pH 6.8; 2% SDS; 25% Glycerol; 1‰ phenol blue; 5% β-mercaptoethanol), boiled for 10 min and separated on a 7.5 or 10% acrylamide SDS gel. After protein transfer, nitrocellulose membranes were blocked in NET-G buffer for 1-h. STAT3 P-Tyr705 (Cell Signaling Technology, Beverly MA) and STAT3-α (c-20, Santa Cruz) were incubated for 36 h in a 1:1,000 dilution. Detection was performed with peroxidase-conjugated anti-rabbit IgG (Roth) at a dilution of 1:10,000 for 1 h. Visualization was performed using an enhanced chemiluminescence detection kit (Amersham).

RNA preparation and real-time PCR

For preparation of total RNA, cells were washed in PBS and homogenized in 1 ml TRIzol solution (Life Technologies). The RNA phase was purified by chloroform extraction and isopropanol precipitation. cDNA was synthesized in 20-μl reactions containing 5 μg total RNA, 4 μl 5 × M-MLV buffer (Promega), 20 mM DTT, 0.5 μg oligo(dT)15 (Promega), 0.5 mM of each dNTP, 40 U RNasin (Promega) and 200 U M-MLV reverse transcriptase (Promega). Mixtures were incubated according to the manufacturer’s recommendations. The reactions were terminated by addition of EDTA to a final concentration of 7 mM. Quantitative real-time PCR analysis was performed by using the iCycler Apparatus (Bio-Rad, Hercules, CA, USA) with the primer pair MMP-1 forward (5′-AGGGTCAAGCAGACATCATG-3′) and MMP-1 reverse (5′-AGCATCGATATGCTTCACAGT-3′). 25-μl reaction mixtures contained 1 μl of cDNA preparation, 1 × iQ SYBR Green supermix (Bio-Rad) and 0.4 μM of each respective primer. PCR comprised initial heating of samples to 95°C for 3 min followed by 35 repetitions of the following cycle: 40 s 58°C, 60 s 72°C, 20 s 95°C. The MMP-1 gene copy number in samples to be analyzed was quantified by simultaneously generating a standard curve for both MMP-1 and β-actin mRNA as an endogenous control from serial dilutions of cDNA (equivalent to cDNA amounts from 100 ng to 100 pg of initial total RNA). The target amounts of unknown samples were divided by the endogenous reference amount to obtain a normalized target value. Final results were expressed as n-fold differences in MMP-1 relative to β-actin mRNA. Data acquisition and analysis was carried out using the iCycler iQTM software (Bio-Rad).

Detection of cytokines/chemokines

Cytokines and chemokines were analyzed with a commercial RayBio Human Cytokine Array V. Conditioned medium (1 ml) was incubated with blocked array membranes over-night according to the manufacturer’s protocol. Membranes were then probed with the supplied mix of biotin-conjugated antibodies for 2 h at RT. Upon incubation with HRP-conjugated streptavidin, signals were detected by exposing the membranes to X-ray films (Kodak).

Results

Non-adhesive growth of colon carcinoma cells induces STAT3 activity

To better understand by which mechanisms colon carcinoma cells within tumors gain STAT3 activity, we sought to provide growth conditions that would more closely resemble the tissue environment than standard cell culture. We observed that HT-29 cells cultivated on the non-coated plastic surface of Petri dishes (compare “Material and methods”) displayed a more dedifferentiated appearance compared to cells grown in regular cell culture flasks, like rounding-up, detachment and formation of clusters in suspension (Fig. 1a). We tested whether these morphological changes coincided with altered STAT3 activation. Notably, growing cells on non-coated surfaces evoked strong STAT3 activity in non-stimulated HT-29 cells and in HT-29 cells overexpressing STAT3 both with regard to specific DNA binding and tyrosine phoshorylation (Fig. 1b, c). The effect was not evident in cells transfected with dominant-negative STAT3. This finding was in clear contrast to the situation observed with cells grown in regular tissue culture flasks.

Fig. 1.

Comparison of growth behavior and STAT3 activity in HT-29 cells cultivated on coated and non-coated surfaces. a Growth characteristics of HT-29 cells cultivated on regular tissue culture plasticware (“t.c.”, top) and on non-coated surfaces (“non-coated”, bottom). b Western blot analysis of tyrosine phosphorylation (top) and expression (bottom) of endogenous, constitutively active (c.a.) and dominant-negative (d.n.) STAT3 variants in HT-29-derived cell lines. Cells were grown for 4 days on tissue culture (t.c.) or non-coated plastic material (non-coated) as indicated. They were lysed and probed with antibody to phosphorylated STAT3 (top) or STAT3 (bottom), respectively. c EMSA analysis for STAT3 DNA-binding activity of HT-29-derived cell lines grown on tissue culture (“t.c.”) or on non-coated plasticware (“non-coated”). HT-29 parental cells and cell lines stably expressing constitutively active (c.a.) or dominant-negative (d.n.) mutants were treated as in a and subjected to EMSA using the SIE m67 element as a probe. The identity of the STAT3-containing complexes was verified by supershift employing a specific antibody to STAT3. Positions of STAT3 complexes are indicated by arrows. Results shown are representative for three independent experiments

Activation of STAT3 in non-adhesively growing CRC cells involves an autocrine loop

We next tested whether growth in contact with non-coated surfaces induced the secretion of a STAT3-activating factor. HT-29 cells as well as HT-29 cells overexpressing STAT3 were grown in Petri dishes for 24 h. Subsequently, the conditioned medium was used to cultivate a new batch of cells in regular culture flasks. The STAT3 activity status was determined after 30 min (Fig. 2). Interestingly, exposure of cells to conditioned medium evoked strong STAT3 tyrosine phosphorylation compared to non-conditioned medium both in parental cells and in the derivative overexpressing STAT3. Similar results were obtained by assessing specific DNA binding of STAT3 (data not shown).

Fig. 2.

Analysis for STAT3-activating properties of culture medium conditioned by HT-29 cells grown on non-coated surfaces. a Left-hand side: HT-29 cells were grown for 4 days on tissue culture (t.c.) or non-coated plastic material (non-coated) as indicated. Right-hand side: Cells were grown for 30 min on tissue culture plastic surface in medium conditioned by the indicated cell lines grown on non-coated surface. Conditioned medium was obtained by centrifugation of cells and subsequent sterile filtration of supernatant. Cells were harvested after treatment, lysed and subjected to Western blotting with antibody to phosphorylated STAT3. Blots were reprobed with anti-STAT3. b Analysis as in a with HT-29 cells overexpressing STAT3. Results are representative for three independent experiments

Conditioned medium from HT-29-STAT3 cells had a stronger effect than medium from parental cells, indicating that STAT3 activity is directly related to the production of the activating factor. This notion is supported by the observation that expression of STAT3 d.n. blocked the secretion of stimulating activity from cells grown on non-coated surface. These results argue for an autocrine loop operative in HT-29 cells under the conditions that promote cellular dedifferentiation.

Clustered growth of HT-29 colon carcinoma cells enhances expression of the MMP-1 gene

We have recently shown that activated STAT3 transcriptionally upregulates of the matrix metalloproteinase MMP-1 (Tsareva et al. 2007). In order to examine the effect of the non-adhesive growth of HT-29 cells on MMP-1, the level of MMP-1 mRNA was measured by real-time quantitative PCR. HT-29 cells were incubated on tissue cell culture flasks or on microbiological Petri dishes for 4 days in RPMI 1640 before isolation of total RNA. The MMP-1 mRNA copy number in HT-29 and HT-29/STAT3 cells from the Petri dishes was significantly higher than that from cells growing in regular flasks (fourfold and threefold, respectively, Fig. 3). These results corroborate previous findings, i.e. MMP-1 expression is in correlation with STAT3 activity in HT-29 cells.

Fig. 3.

Influence of growth conditions of HT-29 derivatives on MMP-1 expression. HT-29- and HT-29/STAT3 cells were grown for 4 days on regular tissue culture plasticware (“t.c.”) or on non-coated surfaces. The total RNA was isolated and MMP-1 mRNA levels were measured by quantitative real-time RT–PCR. Three independent experiments, each run in triplicate, were performed. Results were normalized and expressed as the means ± S.E. High statistical significance of differences (P < 0.01) is indicated by double asterisks

Forced cell–cell contacts induce secretion of chemokines by HT-29 cells

To identify potential autocrine/paracrine factor(s) responsible for STAT3 stimulation in HT-29 cells, cytokines and growth factors were screened for specific appearance in the STAT3-activating conditioned medium. Concentrations of 79 different factors were compared in conditioned media from HT-29 cells grown in regular cell culture flasks and on the non-coated plastic surface of Petri dishes using a commercial human cytokine array (RayBio). Membranes immobilized with 79 capture antibodies specific for cytokine and growth factors were exposed in parallel to the two media. Binding of specific antigens was subsequently detected by a mixture of detection antibodies.

In the growth medium of the colon carcinoma cells cultivated on non-coated plastic surface, elevated abundance of six proteins was identified in comparison with control medium: monocyte chemotactic protein (MCP)-1, angiogenin (Ang), interferon-gamma inducible protein (IP)–10, IL-1α, macrophage inflammatory proteins (MIP)-1δ and -3α (Fig. 4).

Fig. 4.

Comparative analysis of the abundance of cytokines and growth factors in conditioned medium from cells grown on coated and non-coated surfaces. Simultaneous detection of multiple cytokines and growth factors using a protein array. Conditioned media issued from HT-29 cells growth in regular cell culture flasks (left) and on non-coated plastic surface of Petri dishes (right) were incubated with cytokine array membranes. The membranes were then incubated with a combination of multiple biotin-conjugated antibodies to cytokines. Signals were visualized by enhanced chemiluminescence

Discussion

STAT3 is an important player in the pathogenesis of various human cancers. Like in a number of other solid tumors, it is constitutively active in the large majority of human colorectal cancer tissues (Corvinus et al. 2005).

Obviously, STAT3 may contribute to oncogenesis in the colon epithelium by exerting a promoting effect on the cell cycle by increasing motility and invasiveness and/or by suppressing apoptosis. For colon carcinoma, all these notions are supported by recent results from our laboratory which showed that heterologous overexpression of STAT3 or cytokine-mediated activation of endogenous STAT3 in colon carcinoma cell lines accelerates cell proliferation, enhances invasiveness and elevates the abundance of anti-apoptotic proteins Cyclin D1 and Bcl-2 (Corvinus et al. 2005; Tsareva et al. 2007 and unpublished results). In situ techniques, however, are required to verify whether such changes coincide spatially with STAT3 activity within the heterogeneous tumor tissue.

Although aberrant STAT3 activity is now considered as functionally important for the occurrence and/or maintenance of various malignant tumors, it is poorly understood by which mechanisms excessive STAT3 activity in cancer is triggered. Unlike for other signaling molecules, no naturally occurring genetic mutations or amplifications of STAT3 associated with oncogenesis have been identified so far, indicating that persistent STAT3 activity is most probably due to dysregulated events upstream of STAT3 within the signaling pathways. Various recent findings, including our own, point to an important role of cytokines and growth factors secreted by tumor cells.

In contrast to tumor cell lines originating from other tumors, all colon carcinoma cell lines employed in our previous study were negative for STAT3 activity in the absence of cytokine stimulation. Interestingly, they showed profound activity upon implantation and xenograft tumor formation in immunodeficient mice. Attempts to promote cancerous cell behavior by adapted culture conditions for the colon carcinoma model cell line HT-29 (and, thus, to mimic aspects of the tumor tissue situation) yielded evidence that STAT3 activation is induced via autocrine stimulation in the tumor. Autocrine loops involving factors that trigger JAK/STAT pathways have been identified as crucial for malignant cell behavior in several tumor models. In breast carcinoma cells, a mechanism of autocrine-mediated STAT3 activation has been identified which correlates with cell proliferation (Li and Shaw 2002). In brain tumors, STAT3 is considered to play a central role in autocrine activation of the vascular endothelial growth factor (VEGF) system (Schaefer et al. 2002). Different examples for IL-6-triggered autocrine stimulation of tumor cells have been published recently such as acute myeloid leukemia cells (Schuringa et al. 2000) as well as carcinoma cells of the kidney (Angelo et al. 2002) and the prostate (Giri et al. 2001). Stem cell factor (SCF) and its receptor c-Kit have been suggested to be involved in an autocrine loop during CRC progression, since both proteins were found jointly upregulated in many CRC biopsies, but not in healthy and pre-malignant tissue (Bellone et al. 2006). Our assay, however, did not reveal significant SCF secretion by HT-29 cells, excluding the possibility that SCF/c-Kit function underlies autocrine STAT3 activation. A very recently reported example for an involvement of STAT3 in autocrine cancer cell signaling is endometrial carcinoma, where STAT3 mediates the oncogenic effects of autocrine growth hormone (Tang et al. 2010).

Our data suggest that STAT3 in the non-neoplastic environment of transformed cells may become stimulated by diffusible molecules originating from the tumor tissue. We observed the specific appearance of a characteristic chemokine panel in conditioned medium with autocrine STAT3-stimulatory activity. Interestingly, the identified molecules (monocyte chemotactic protein-1, angiogenin, interferon-γ inducible protein 10, IL-1α, macrophage inflammatory proteins 1δ and 3α) all belong to groups of chemokines taking part in inflammatory processes. Two of them were described as important factors in tumor formation (angiogenin and MCP-1), for one of these proteins (MCP-1), an inducibility by STAT3 has been reported (Kim et al. 2002). The cytokine arrays also showed that the abundance of IL-10, oncostatin M (OSM), growth regulated oncogene (GRO), macrophage colony stimulating factor (M-CSF) and MIP-1ß was reduced in the medium from the HT-29 cells grown on non-coated surface (data not shown). We have as yet failed to demonstrate a STAT3-stimulatory effect by application of pure individual factors from the above-described panel of chemokines (data not shown). At present, this leaves us with the interpretation that indirect or combinatorial effects under the observed phenomenon. Very recently, macrophage migration inhibitory factor (MIF) was identified as a chemokine promoting invasion of drug-resistant colon carcinoma cells in an autocrine fashion (Dessein et al. 2010). We have not observed enhanced MIF expression in our assay, suggesting that activation of MIF-triggered signaling is a specific feature of drug-resistant CRC tumor cells.

STAT3 has recently emerged as an attractive target for tumor treatment. In cultured tumor cells, blockage of STAT3 was found to downmodulate proliferation and to induce programmed cell death. Transfection of a dominant-negative form of STAT3 led to growth inhibition and apoptosis of breast, brain and prostate cancer cells (Garcia et al. 2001). RNA interference and inhibitor approaches have been pursued successfully to block STAT3-associated malignancy parameters of various cancer cell types (Weerasinghe et al. 2007; Tadlaoui Hbibi et al. 2009).

Most interestingly, growth of experimental melanomas could be suppressed efficiently in mice by the introduction of a functionally deficient STAT3 variant (Niu et al. 1999). Interruption of the probable IL-6 autocrine loop in prostate cancer xenografts by administration of a neutralizing anti-IL-6 antibody-induced tumor regression (Smith and Keller 2001). Our findings suggest that targeting STAT3 may also be an interesting option to fight CRC and other cancers in which STAT3 activation is linked with tumor progression.

Electronic supplementary material

Below is the link to the electronic supplementary material.