Original Paper

Oncogene (2004) 23, 2138–2145. doi:10.1038/sj.onc.1207332 Published online 15 December 2003

Secretion of cytokines and growth factors as a general cause of constitutive NFkappaB activation in cancer

Tao Lu1, Swati S Sathe1, Shannon M Swiatkowski1, Chetan V Hampole1 and George R Stark1

1Department of Molecular Biology, Lerner Research Institute, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA

Correspondence: GR Stark, Lerner Research Institute/NB21, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA. E-mail: starkg@ccf.org

Received 1 May 2003; Revised 5 September 2003; Accepted 10 November 2003.



The constitutive activation of nuclear factor kappaB (NFkappaB) helps a variety of tumors to resist apoptosis and desensitizes them to chemotherapy, but the causes are still largely unknown. We have analysed this phenomenon in eight mutant cell lines derived from human 293 cells, selected for NFkappaB-dependent expression of a marker gene, and also in seven tumor-derived cell lines. Conditioned media from all of these cells stimulated the activation of NFkappaB (up to 30-fold) in indicator cells carrying an NFkappaB-responsive reporter. Therefore, secretion of extracellular factors as the cause of constitutive activation seems to be general. The mRNAs encoding several different cytokines and growth factors were greatly overexpressed in the tumor and mutant cells. The pattern of overexpression was distinct in each cell line, indicating that the phenomenon is complex. Two secreted factors whose roles in the constitutive activation of NFkappaB are not well defined were investigated further as pure proteins: transforming growth factor beta2 (TGFbeta2) and fibroblast growth factor 5 (FGF5) were both highly expressed in some mutant clones and tumor cell lines, each activated NFkappaB alone, and the combination was synergistic. Our data indicate that a group of different factors, expressed at abnormally high levels, can contribute singly and synergistically to the constitutive activation of NFkappaB in all of the mutant and tumor cell lines we studied. Since several NFkappaB target genes encode secreted proteins that induce NFkappaB, autocrine loops are likely to be ubiquitously important in the constitutive activation of NFkappaB in cancer. We provide the first evidence of the general, complex, and synergistic activation of NFkappaB in tumor and mutant cell lines through the action of secreted factors and suggest that the same explanation is likely for the constitutive activation of NFkappaB in cancers.


TGFbeta2, FGF5, microarrays, mutant 293 cells, ELISA



Nuclear factor kappaB (NFkappaB) is the generic name for a family of dimeric factors that binds to many promoters to initiate transcription. The most common form of NFkappaB is a p65/p50 heterodimer, while other forms occur less frequently (Verma et al., 1995). NFkappaB is kept quiescent in the cytoplasm of cells by its repressor, IkappaB. In response to a stimulus, IkappaB is phosphorylated and degraded, and the released NFkappaB translocates to the nucleus, where it initiates transcription (Baeuerle and Henkel, 1994; Siebenlist et al., 1994). In addition to its role in inflammation, NFkappaB plays a major role in activating genes involved in oncogenesis. Constitutively high levels of NFkappaB have been detected in many tumors (Baldwin, 2001). Loss of the regulation of NFkappaB may contribute to the deregulated growth and resistance to apoptosis often observed in cancer, including tumors of the prostate, breast, and ovary (Deng et al., 2002; Romieu-Mourez et al., 2002). Therefore, inhibition of constitutive NFkappaB is a promising approach to inhibit tumor growth and metastasis.

It is well known that NFkappaB can be activated by many different stimuli, including but not limited to interleukin-1 (IL1), tumor necrosis factor alpha (TNFalpha), UV or gamma radiation, double-stranded RNA and other toll-like receptor ligands, phorbol esters, and reactive oxygen species (Pahl, 1999; O'Neill, 2002). In several types of tumor, NFkappaB is induced in an autocrine manner through the release of the cytokines TNFalpha, IL1alpha, and IL1beta (Wolf et al., 2001; Arlt et al., 2002; Coward et al., 2002). However, the generality of this phenomenon and the basis of constitutive secretion are not known. To investigate the constitutive activation of NFkappaB further, we selected a group of eight mutant cell lines in which NFkappaB is constitutively activated, from chemically mutagenized human HEK293 cells (Sathe et al., manuscript submitted). We show here that these mutants overexpress multiple cytokines that can activate NFkappaB, and similar findings were made in all of the several different tumor cells that we studied. Our data indicate that the secretion of extracellular factors is likely to be a general explanation for the constitutive activation of NFkappaB in cancer, that the patterns of secretion are complex, that some of these factors may act synergistically, and that secretion is a consequence of mRNA overexpression.



Activation of NFkappaB in indicator cells by conditioned media from constitutive mutants and tumor cells

We tested 16 different tumor cell lines for constitutive NFkappaB activity by electrophoretic mobility shift assays (EMSAs): fibrosarcoma HT1080, glioma T98G, lung cancer A549, colon cancer HT29, HCT116, RKO, and SW480, ovarian cancer SKOV-3, breast cancer MCF–7, cervical cancer Hela, prostate cancer PC3 and DU145, melanoma Mel-8 and Mel-29, and lymphoma IM9 and Di cells. Normal fibroblast WI38, epithelial hTERT-HME1, and melanocyte CMN cells were used as controls. Most of the tumor cell lines showed a relative increase in NFkappaB activity compared to the normal controls (Figure 1). The cell lines HT1080, T98G, RKO, Hela, DU145, Mel-29, and IM9, which showed substantial increases, were studied further.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

EMSA assay for NFkappaB in tumor cell lines. WI38, HTERT-HME1, and CMN cells were used as normal controls. Cells cultured to near-confluency were analysed by EMSA. Cell lines shown with an asterisk were chosen for further study

Full figure and legend (54K)

Conditioned media from all eight mutant cells (Sathe et al., manuscript submitted) activates NFkappaB, up to 17-fold compared to control parental 293C6 cells (Figure 2a). Similar effects were seen when the same mutant cells were cocultivated with 293IL1R indicator cells (data not shown). To test whether the secretion of activators is also observed in tumor cells with constitutively active NFkappaB, the same indicator cells were treated with conditioned media from seven of the tumor cell lines, selected for their high constitutive NFkappaB activity. The conditioned media from all the tumor cells activated the NFkappaB-dependent promoter, whereas the media from three control normal cells did not (Figure 2b). The media from HT1080 and Mel-29 cells gave the strongest effects. Therefore, with no exception, all of the constitutive mutants and tumor cells investigated secrete extracellular factors that activate NFkappaB.

Figure 2.
Figure 2 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Activation of NFkappaB in indicator cells by conditioned media from constitutive mutants and tumor cells. Each data point represents mean + s.d. from three independent experiments. (a) Conditioned media from mutant cells. 293IL1R indicator cells in 24-well plates at approx90% confluency were treated with media from 293C6 cells and the constitutive mutants. Cells were lysed and samples were assayed for luciferase activity after 24 h. NFkappaB activity is represented as relative luciferase activity, normalized to the amount of protein. (b) Conditioned media from tumor cells. The experiments were carried out as in (a)

Full figure and legend (94K)

Expression of cytokines and growth factors in cells with constitutively active NFkappaB

We used an array containing immobilized cDNAs for 96 different cytokines and growth factors (for list, see http://www.superarray.com). Total RNA from control 293C6 cells and the eight mutant clones was reverse-transcribed to cDNAs, which were used to probe the array. In the constitutive mutants, most cytokine genes were expressed at levels similar to the levels in parental 293C6 cells. However, a few genes were markedly overexpressed (Table 1). Among these, IL1beta, TNFalpha, receptor activator of NFkappaB ligand (RANKL), OX40 ligand (OX40L), bone morphogenetic protein 2 (BMP-2), and fibroblast growth factor 5 (FGF5) were also highly induced by IL1beta treatment of 293C6 cells, indicating that they are likely to be targets of NFkappaB. The others were either unaffected or only slightly induced by IL1beta, suggesting that thay may not be direct targets.

Northern analysis confirmed the overexpression of mRNAs for these cytokines in the mutant lines (Figure 3a) and revealed a generally similar pattern of overexpression in the tumor cells (Figure 3b). Interestingly, many of the induced genes are angiogenic factors, which are highly correlated with transformation, tumor invasion, and metastasis (Coussens and Werb, 2001; Alitalo and Carmeliet, 2002; Kascinski, 2002; Karkkainen et al., 2002; Klomp et al., 2002; Mentlein and Held-Feindt, 2002). The expression of OX40L, CD27 ligand (CD27L), and RANKL, normally expressed primarily in B and T cells, was found to be dramatically increased in most of the mutant cells and in some of the tumor cell lines. Transforming growth factor beta2 (TGFbeta2) was overexpressed in several tumor lines, including HT1080, T98G, DU145, and Mel-29, and FGF5 was overexpressed only in DU145 cells (Figure 3b).

Figure 3.
Figure 3 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Northern analysis of cytokine mRNAs overexpressed in the constitutive mutants and tumor cells. (a) Analyses of constitutive mutants. RNA (15 mug) for each sample was loaded in each lane. RNAs from 293C6 cells, with or without IL1beta treatment for 4 h, were also assayed. The 28 S and 18 S RNAs were used as loading controls. (b) Analyses of tumor cells. The assays were carried out as in (a)

Full figure and legend (251K)

Activation of NFkappaB by TGFbeta2 and FGF5

Two factors secreted by several of the mutant cell lines and tumor cells whose roles as activators of NFkappaB have not been obvious to date were selected for more careful study. Enzyme-linked immunosorbent assays (ELISA) revealed that TGFbeta2 was secreted by mutant C6P1Z12, at about 8000 pg/106 cells in 24 h, and at a lower level by several of the tumor cell lines (Table 2). Titration with a neutralizing antibody directed against TGFbeta2 confirmed the specificity of the ELISA result (Table 2). The array data showed that the mRNA for FGF5 was highly expressed in mutant C6P1Z12 and tumor DU145 cells (Table 1 and Figure 3a). We treated indicator cells with different concentrations of TGFbeta2, FGF5 or both. Each induces NFkappaB activity singly, in a dose dependent manner, with FGF5 being less potent than TGFbeta2 (Figure 4a). In addition, the effect of the two proteins was clearly synergistic when they were applied together (Figure 4a). An EMSA assay to assess the effects of TGFbeta2 and FGF5 on NFkappaB activation over time in parental 293C6 cells (Figure 4b) confirmed the activation by TGFbeta2 and FGF5 alone, and again revealed synergistic activation of NFkappaB. Northern analysis shows that TGFbeta2 and FGF5 also induce expression of the endogenous NFkappaB target gene IL8, singly and synergistically, in 293C6 cells (Figure 4c).

Figure 4.
Figure 4 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Individual and synergistic effects of TGFbeta 2 and FGF5 on NFkappaB activation. (a) Luciferase reporter assays of 293IL1R indicator cells, treated with TGFbeta2, FGF5 or both for 24 h. The luciferase activities shown are normalized to the activity found in untreated cells. TB2 represents TGFbeta2, F5 represents FGF5, numbers in parentheses represent the concentrations. Each data point represents mean + s.d. from three independent experiments. (b) EMSA for NFkappaB in TGFbeta2- or FGF5-treated 293C6 cells. Cells in 60-mm dishes were treated with TGFbeta2 (0.1 mug/ml), FGF5 (2 mug/ml) or both. Samples were collected and analysed by EMSA. To show specificity, IL1 was used as a control. The binding of the labeled probe was competed by a 100-fold excess of unlabeled probe. (c) Induction of IL8 expression by TGFbeta2 and FGF5 in 293C6 cells. Cells plated in 100-mm dishes were treated with TGFbeta2 (0.1 mug/ml), FGF5 (2 mug/ml) or both for 4 or 10 h. RNA (15 mug) was loaded in each lane

Full figure and legend (204K)



To investigate the mechanism behind the frequent constitutive activation of NFkappaB in tumors, we screened 16 different tumor cell lines and found that most had constitutive NFkappaB activity. Seven were chosen for further study. We also studied a group of eight mutant cell lines with constitutive NFkappaB (Sathe et al., manuscript submitted). NFkappaB can be activated via multiple pathways (Pahl, 1999) and, recently, it has been shown that constitutive NFkappaB activation can result from the autocrine action of secreted cytokines (Wolf et al., 2001; Arlt et al., 2002; Coward et al., 2002). However, the generality of this phenomenon is not known. To test this point, we assayed conditioned media from all of the cells noted above and found that NFkappaB was activated in every case. We identified a group of mRNAs encoding several important cytokines that are active in metastasis (Karkkainen et al., 2002), angiogenesis (Karkkainen and Petrova, 2000), transformation, and invasion (Bhat-Nakshatri et al., 2002). These mRNAs were overexpressed in distinct and different patterns in different constitutive mutants. Some of the mRNAs encode well-known NFkappaB-activating cytokines, including IL1beta and TNFalpha, which are overexpressed in some tumor cells, as reported by other groups. For example, IL1beta is overexpressed in the human pancreatic carcinoma cell lines A818-4 and PancTu–1 (Arlt et al., 2002) and TNFalpha is overexpressed and functions in an autocrine loop to activate NFkappaB in human cutaneous T-cell lymphoma HuT-78 cells (O'Connell et al., 1995; Giri and Aggarwal, 1998). In our analysis, IL1beta and TNFalpha were overexpressed in many constitutive mutants and in some of the cancer cells, but not in all of them, indicating they are just a part of the picture.

Some of the overexpressed cytokines belong to the TNF superfamily. For example, OX40L, also known as gp34, is expressed by activated B, T, dendritic and endothelial cells (Koubek et al., 1999). We show that OX40L is overexpressed in several tumor cell lines of different tissue origins, including human glioma T98G and prostate cancer DU145 cells.

RANKL is expressed primarily in T cells, in T-cell-rich organs, such as thymus and lymph nodes, and in stromal cells (Xing et al., 2002). RANKL is highly expressed and involved in the tumorigenesis and metastasis of multiple myelomas (Sezer et al., 2002), giant cell tumors of the bone (Roux et al., 2002), prostate cancers (Brown et al., 2001), and neuroblastomas (Michigami et al., 2001). We show that RANKL is overexpressed in most constitutive mutants, and in colon cancer RKO, cervix cancer Hela, and prostate cancer DU145 cells. In addition, we show that RANKL can be induced by IL1beta treatment (Figure 3a), indicating that it is a target of NFkappaB, a result that has not been reported previously.

Another TNF superfamily member, CD27L, mainly mediates the interaction between B- and T lymphocytes (Hishima et al., 2000). Recently, CD27L was reported to be overexpressed in human meningioma and glioma cells (Held-Feindt and Mentlein, 2002) and in thymic carcinoma (Hishima et al., 2000). Thus, CD27L is suggested to be a new marker in brain tumors of nonhematopoietic origin (Held-Feindt and Mentlein, 2002). In our study, CD27L was strongly induced in most of the mutants, and in human tumors of different origin, such as fibrosarcoma HT1080, cervix cancer Hela, prostate cancer DU145, and melanoma Mel-29 cells. Both RANKL (Galibert et al., 1998; Mizukami et al., 2002) and CD27L (Yamamoto et al., 1998) were reported previously to activate NFkappaB in different cells. However, no previous report suggests the involvement of OX40L in NFkappaB activation.

Some of the other cytokines overexpressed in the cells we have studied are angiogenic and involved in tumor progression and invasion. For example, pleiotrophin (PTN) is a proto-oncoprotein that is strongly expressed in different human tumor cell lines. Expression of PTN in tumors accelerates their growth and stimulates angiogenesis (Zhang et al., 1999). In our study, PTN was one of the most strongly expressed cytokines in the mutants, and was also overexpressed in glioma T98G, cervix cancer Hela, and melanoma Mel-29 cells.

Much evidence suggests that colony-stimulating factor 1 (CSF-1) plays an autocrine or paracrine role in cancer (Kacinski, 1995; Sapi and Kacinski, 1999; Hermouet et al., 2000). Circulating CSF-1 was suggested to be a marker of tumor progression (McDermott et al., 2002). CSF-1 was reported to induce its own expression through the activation of NFkappaB (Brach et al., 1991). Thus, it is possible that CSF-1 functions to activate NFkappaB and further induce its own expression, forming autocrine and paracrine loops to promote constitutive NFkappaB activity in tumor cells. CSF-1 was highly overexpressed in most of the mutants and in several tumor cell lines, including fibrosarcoma HT1080, glioma T98G, prostate cancer DU145, and melanoma Mel-29 cells.

Vascular endothelial growth factor C (VEGF-C), a member of the vascular endothelial growth factor superfamily, is a mitogen for both vascular and lymphatic endothelial cells (Clauss, 2000). Evidence suggests a strong correlation between VEGF-C expression and lymph node metastasis formation (Yonemura et al., 1999). Most recent reports suggest that VEGF-C plays an extremely important role in invasion in breast (Mattila et al., 2002), cervical (Ueda et al., 2002), colon (Furudoi et al., 2002), and gastric cancer (Takahashi et al., 2002). In our study, VEGF-C was highly overexpressed in most of the mutants, and in fibrosarcoma HT1080, glioma T98G, and prostate cancer DU145 cells. So far, very few studies have suggested that TGFbetas can activate NFkappaB (Li et al., 1998), and the relationship between different forms of TGFbetas, NFkappaB activation, and tumorigenesis is largely unknown. Interestingly, we found that both TGFbeta2 and FGF5 activate NFkappaB, singly and synergistically. TGFbeta2 is a member of a large superfamily of factors that regulate many cellular processes, including cell proliferation, differentiation, and death. It is not surprising that the TGFbetas are important in the genesis of cancers of various origins, including gastric (Iacopetta et al., 1999), colorectal (Miyaki et al., 1999), hepatocellular (Furuta et al., 1999), cervical (Chen et al., 1999), lymphoma (Schiemann et al., 1999), and breast cancers (Chen et al., 1998). Recently, we have found that TGFbeta2 can prevent TNFalpha-triggered apoptosis through the activation of NFkappaB (Lu et al., manuscript submitted). In the present study, we found TGFbeta2 to be overexpressed in mutant C6P1Z12 cells, as well as in HT1080, T98G, DU145, and Mel-29 tumor cells.

FGF5, reported to be overexpressed in human pancreatic cancer tissues and COLO-357 pancreatic cancer cells (Kornmann et al., 1997), functions in autocrine and paracrine loops. In our study, FGF5 was overexpressed in several constitutive mutants, especially in C6P1Z12 cells. TGFbeta2 and FGF5 were also overexpressed together in prostate cancer DU145 cells, confirming that the coexpression found in a mutant cell line also exists in a tumor cell line. It has been suggested that another FGF family member, FGF1, can activate and increase NFkappaB nuclear translocation in Jurkat T cells (Byrd et al., 1999), but the correlation between FGF5 and NFkappaB activation has not been reported. We also find that FGF5 and TGFbeta2 can activate NFkappaB synergistically.

In summary, our data strongly suggest that a group of cytokines is overexpressed at the mRNA level in mutant and tumor cells with constitutive NFkappaB activity. Except for TGFbeta2, we have not investigated their expression at the protein level. However, the observation that many cytokines are overexpressed in different patterns in the mutant and tumor cell is closely linked to NFkappaB activation. Our data for IL1beta-treated 293C6 cells (Table 1) indicates that some of the mRNAs expressed at high levels can be induced by NFkappaB, for example, IL1beta, TNFalpha, FGF5, and RANKL. It is difficult at this point to sort out cause and effect, that is which cytokine causes NFkappaB activation and which is overexpressed as a consequence of this activation. However, as is well known for IL1beta and TNFalpha, many cytokines can activate NFkappaB and further induce their own expression. Furthermore, in addition to the synergistic effect of TGFbeta2 plus FGF5, it is possible that other cytokines can collaborate to produce stronger activation of NFkappaB. In summary, for the first time, our data reveal that complex patterns of cytokine secretion are general in cancer cells with constitutive NFkappaB and that certain combinations of secreted factors may function in synergy. The basic cause of cytokine overexpression and secretion is an interesting and important topic for future investigation. Since the eight mutant cell lines were obtained under conditions that were found previously to cause ablation of single proteins in signaling pathways, and since most of these mutants are recessive (Sathe et al., manuscript submitted), we hypothesize that the loss of expression of a single negative regulator may trigger the aberrant coordinate expression of a group of different cytokines that activate NFkappaB and that further regulate their own expression through complex autocrine or paracrine loops. To understand the causes of the secretion of multiple cytokines will be of fundamental importance in understanding an important aspect of cancer.


Materials and methods

Cell culture

293C6 (293ILIR cells transfected with E-selectin-driven zeocin-resistance and thymidine kinase genes) and derived mutant cell lines (C6P1Z1, C6P2Z1, C6P2Z2, C8P1Z3, C8P4Z5, C6P1Z12, C18P3Z13, C8P1Z23) (Sathe et al., manuscript submitted), human normal fibroblast WI38, human fibrosarcoma HT1080, human glioma T98G, human lung carcinoma A549, human colon cancer HT29, HCT116, RKO, and SW480, human ovarian cancer SKOV-3, human breast cancer MCF-7, human cervix cancer Hela, human prostate cancer PC3, and DU145 cells (American Type Culture Collection, Manassas, VA, USA), normal human melanocyte CMN cells, and melanoma Mel-8 and Mel-29 cells (kind gifts from the Department of Surgical Oncology of the University of Illinois at Chicago) were cultured in Dulbecco's modified Eagle's media supplemented with 100 U/ml penicillin, 100 mug/ml streptomycin, and 10% fetal calf serum. The media for 293C6 and derived mutants were also supplemented with G418 (400 mug/ml). 293IL1R indicator cells were cultured in the same media as 293C6 cells, plus puromycin (1 mug/ml). Human lymphoma IM9 and Di cells (American Type Culture Collection) were cultivated in RPMI-1640 media with 100 U/ml penicillin and 100 mug/ml streptomycin, and 10% fetal calf serum. Normal human epithelial htert-HME cells (Clonetics Corporation, San Diego, CA, USA) were cultivated in special mammary epithelium basal media (Clonetics Corporation).

To collect conditioned media, cells were cultivated to approx90% confluency under the conditions described above, the media was replaced, and the cells were grown for another 24 h. The conditioned media were collected, floating cells were pelleted at 3000 g at 4°C, for 10 min, and the supernatant solution was aliquotted into sterile tubes and stored at -80°C.

Construction of stable 293IL1 indicator cells

The kappaB-luciferase plasmid p5XIP10 kappaB (contains five tandem copies of the NFkappaB site from the IP10 gene, Sizemore et al., 2002) and the pBabe puromycin-resistance plasmid were stably cotransfected into 293IL1R cells (which carry transfected IL1 receptors) by using the FUGENE 6 reagent, following the protocol provided by Roche Diagnostics Corporation (Indianapolis, IN, USA). Puromycin-resistant clones were selected in 1 mug/ml puromycin.

NFkappaB luciferase reporter assay

293IL1R indicator cells were grown in 24-well plates to approx90% confluency, then treated with either conditioned media, TGFbeta2, or FGF5 at 37°C for 24 h. The cells were then washed with cold phosphate-buffered saline, and lysed in 80 mul of 5 times lysis buffer (Promega Corporation, Madison, WI, USA). After incubation on ice for 15–20 min, cell debris was pelleted at 15 000 g, for 4 min at 4°C. A measure of 80 mul of luciferase assay substrate (Promega Corporation) was added to 20 mul of supernatant solution before assay in a luminometer. The relative luminescence was normalized to total protein (Bio-Rad Laboratories, Hercules, CA, USA).

Cytokine gene expression profile by microarray analysis

293C6 cells and constitutive NFkappaB mutants were cultured to approx90% confluency and total RNA was extracted by using the TRIzol reagent, following the protocol provided by Invitrogen Life Technologies (Carlsbad, CA, USA). By reverse transcription, cDNA probes were made from total RNA and labeled with [alpha-32P]dCTP (NEN, Boston, MA, USA), following the protocol from Promega Corporation (Madison, WI, USA). The labeled cDNA probes were hybridized to a membrane containing 96 specific cDNA fragments of human cytokine genes (Human Common Cytokine GEArray Q Series) under conditions described by SuperArray Inc. (Bethesda, MD, USA). Membranes were exposed to phosphorimaging screens for 2 days. Signal intensities for individual genes were quantified by using ImageQuant software (Molecular Dynamics). The mean signals were calculated from quadruplicate spots. Data were analysed with the software provided by the supplier.

Northern analyses

Human I.M.A.G.E. Consortium cDNA clones of the following cytokine genes were from ResGen invitrogen Corporation (Huntsville, AL, USA): TGFbeta2, FGF5, VEGF-C, BMP-2, PTN, CSF-1, OX40L (tumor necrosis factor (ligand) superfamily member 4 (TNFSF4), CD27L (TNFSF7), RANKL (TNFSF11), TNFalpha, and IL1beta. Plasmids were amplified and purified by using the QIAGEN purification kit following the protocol provided by QIAGEN Incorporation (Valencia, CA, USA). The sequence of each I.M.A.G.E. clone was confirmed after purification. Inserts of 250–500 base pairs were cut out by using different restriction sites. The cDNAs were labeled with [alpha-32P]dCTP, using the Megaprime DNA labeling system, following the protocol provided by Amersham Biosciences (Buckinghamshire, England, UK).

Cells cultured to approx90% confluency were washed with cold phosphate-buffered saline. Total RNA was extracted with the TRIzol reagent at room temperature following the protocol provided by Invitrogen Life Technologies (Carlsbad, CA, USA). A measure of 15 mug of total RNA was loaded onto each lane, electrophoresed in an agarose–formaldehyde gel and transferred onto a Hybond-N+ membrane (Amersham Pharmacia Biotech, Piscataway, NJ, USA). After UV crosslinking, the transfers were hybridized with [alpha-32P]dCTP-labeled probes and analysed by autoradiography at -80°C. For Northern analysis of IL8 expression, 293C6 cells were treated with either TGFbeta2 (0.1 mug/ml) or FGF5 (2 mug/ml) for the indicated times. Total RNAs were extracted and Northern analysis was carried out as described above.


The oligomer used for an NFkappaB binding site was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA): 5'-AGTTGAGGGGACTTTCCCAGGC-3', labeled with [italic gamma-32P]ATP by the polynucleotide kinase method, following the protocol provided by Promega Corporation (Madison, WI, USA). 293C6 cells were treated with either TGFbeta2 (0.1 mug/ml), FGF5 (2 mug/ml) or both for 1 or 4 h. The cells were washed and collected in phosphate-buffered saline and pelleted at 3000 g at 4°C for 4 min. The cells were lysed in a hypotonic buffer (20 mM HEPES, pH 7.9, 20 mM NaF, 10 mM Na3VO4, 2 mM Na4P2O7, 10 mM EDTA, 10 mM EGTA, 20 mM DTT, 100 mM NaCl, 10% (v/v) glycerol, 1 mug leupeptin, 1 mug/ml pepstatin, 1 mug/ml aprotonin, and 0.5 mM phenylmethanesulfonyl fluoride. The mixture was vortexed and kept on ice for 15–20 min. Samples were centrifuged at 15 000 g at 4°C for 4 min. Equal amounts of supernatant solution (normalized for total protein) were incubated with 5 mug/mul of poly(dI-dC) in binding buffer (20 mM HEPES, pH 7.9, 60 mM KCl, 4 mM MgCl2, 0.2 mM EDTA, 1 mM DTT, 10% (v/v) glycerol, and 2% (v/v) polyvinyl alcohol) for 10 min, and then incubated with 1 mul of italic gamma-32P-labeled kappaB probe for another 20 min at room temperature. Samples were loaded onto 6% polyacrylamide gels (acrylamide : N,N'-methylene bisacrylamide, 30 : 1) in 0.25 times Tris borate buffer, pH 8.0. After electrophoresis, the gels were dried and analysed by autoradiography at -80°C.

ELISA for TGFbeta2 quantitation

Conditioned media were collected as described above. ELISA (Quantikine-Human TGFbeta2 Immunoassay) was carried out following the protocol from R&D Systems (Minneapolis, MN, USA). The amount of TGFbeta2 was normalized to the cell number, determined in parallel. Neutralizing anti-TGFbeta2 antibody (R&D Systems, Minneapolis, MN, USA) was used to test the specificity of the ELISA assay.



  1. Alitalo K and Carmeliet P. (2002). Cancer Cell, 1, 219–227. | Article | PubMed | ISI | ChemPort |
  2. Arlt A, Vorndamm J, Muerkoster S, Yu H, Schmidt WE, Folsch UR and Schafer H. (2002). Cancer Res., 62, 910–916. | PubMed | ISI | ChemPort |
  3. Baeuerle PA and Henkel T. (1994). Annu. Rev. Immunol., 12, 141–179. | Article | PubMed | ISI | ChemPort |
  4. Baldwin AS. (2001). J. Clin. Invest., 107, 241–246. | PubMed | ISI | ChemPort |
  5. Bhat-Nakshatri P, Sweeney CJ and Nakshatri H. (2002). Oncogene, 21, 2066–2078. | Article | PubMed | ISI | ChemPort |
  6. Brach MA, Henschler R, Mertelsmann RH and Herrmann F. (1991). Pathobiology, 59, 284–288. | PubMed |
  7. Brown JM, Corey E, Lee ZD, True LD, Yun TJ, Tondravi M and Vessella RL. (2001). Urology, 57, 611–616. | Article | PubMed | ISI | ChemPort |
  8. Byrd VM, Ballard DW, Miller GG and Thomas JW. (1999). J. Immunol., 162, 5853–5859. | PubMed | ChemPort |
  9. Chen T, Carter D, Garrigue-Antar L and Reiss M. (1998). Cancer Res., 58, 4805–4810. | PubMed | ISI | ChemPort |
  10. Chen T, de Vries EG, Hollema H, Yegen HA, Vellucci VF, Strickler HD, Hildesheim A and Reiss M. (1999). Int. J. Cancer, 82, 43–51. | Article | PubMed | ISI | ChemPort |
  11. Clauss M. (2000). Semin. Thromb. Hemost., 26, 561–569. | Article | PubMed | ISI | ChemPort |
  12. Coussens LM and Werb Z. (2001). J. Exp. Med., 193, 727–740. | Article | PubMed | ISI | ChemPort |
  13. Coward WR, Okayama Y, Sagara H, Wilson SJ, Holgate ST and Church MK. (2002). J. Immunol., 169, 5287–5293. | PubMed | ISI |
  14. Deng J, Miller SA, Wang HY, Xia W, Wen Y, Zhou BP, Li Y, Lin SY and Hung MC. (2002). Cancer Cell, 2, 323–334. | Article | PubMed | ISI | ChemPort |
  15. Furudoi A, Tanaka S, Haruma K, Kitadai Y, Yoshihara M, Chayama K and Shimamoto F. (2002). Oncology, 62, 157–166. | Article | PubMed | ChemPort |
  16. Furuta K, Misao S, Takahashi K, Tagaya T, Fukuzawa Y, Ishikawa T, Yoshioka K and Kakumu S. (1999). Int. J. Cancer, 81, 851–853. | Article | PubMed | ISI | ChemPort |
  17. Galibert L, Tometsko ME, Anderson DM, Cosman D and Dougall WC. (1998). J. Biol. Chem., 273, 34120–34127. | Article | PubMed | ISI | ChemPort |
  18. Giri DK and Aggarwal BB. (1998). J. Biol. Chem., 273, 14008–14014. | Article | PubMed | ISI | ChemPort |
  19. Held-Feindt J and Mentlein R. (2002). Int. J. Cancer, 98, 352–356. | Article | PubMed | ChemPort |
  20. Hermouet S, Corre I and Lippert E. (2000). Leuk. Lymphoma, 38, 39–48. | PubMed |
  21. Hishima T, Fukayama M, Hayashi Y, Fujii T, Ooba T, Funata N and Koike M. (2000). Am. J. Surg. Pathol., 24, 742–746. | Article | PubMed | ChemPort |
  22. Iacopetta BJ, Soong R, House AK and Hamelin R. (1999). J. Pathol., 187, 428–432. | Article | PubMed | ISI | ChemPort |
  23. Kacinski BM. (1995). Ann. Med., 27, 79–85. | PubMed | ISI | ChemPort |
  24. Karkkainen MJ, Makinen T and Alitalo K. (2002). Nat. Cell Biol., 4, E2–E5. | Article | PubMed | ISI | ChemPort |
  25. Karkkainen MJ and Petrova TV. (2000). Oncogene, 19, 5598–5605. | Article | PubMed | ISI | ChemPort |
  26. Kascinski B. (2002). Cancer Treat. Res., 107, 285–292. | PubMed |
  27. Klomp HJ, Zernial O, Flachmann S, Wellstein A and Juhl H. (2002). Clin. Cancer Res., 8, 823–827. | PubMed | ISI | ChemPort |
  28. Kornmann M, Ishiwata T, Beger HG and Korc M. (1997). Oncogene, 15, 1417–1424. | Article | PubMed |
  29. Koubek K, Stary J, Kumberova A, Klamova H and Filipec M. (1999). Eur. J. Haematol., 63, 1–10. | PubMed | ChemPort |
  30. Li JM, Shen X, Hu PP and Wang XF. (1998). Mol. Cell. Biol., 18, 110–121. | PubMed | ISI | ChemPort |
  31. Lu T, Burdelya LG, Swiatkowski SM, Boiko AD, Howe PH, Stark GR and Gudkov AV, manuscript submitted.
  32. Mattila MM, Ruohola JK, Karpanen T, Jackson DG, Alitalo K and Harkonen PL. (2002). Int. J. Cancer, 98, 946–951. | Article | PubMed | ISI | ChemPort |
  33. McDermott RS, Deneux L, Mosseri V, Vedrenne J, Clough K, Fourquet A, Rodriguez J, Cosset JM, Sastre X, Beuzeboc P, Pouillart P and Scholl SM. (2002). Eur. Cytokine Network, 13, 121–127.
  34. Mentlein R and Held-Feindt J. (2002). J. Neurochem., 83, 747–753. | Article | PubMed | ChemPort |
  35. Michigami T, Ihara-Watanabe M, Yamazaki M and Ozono K. (2001). Cancer Res., 61, 1637–1644. | PubMed | ISI | ChemPort |
  36. Miyaki M, Iijima T, Konishi M, Sakai K, Ishii A, Yasuno M, Hishima T, Koike M, Shitara N, Iwama T, Utsunomiya J, Kuroki T and Mori T. (1999). Oncogene, 18, 3098–3103. | Article | PubMed | ISI | ChemPort |
  37. Mizukami J, Takaesu G, Akatsuka H, Sakurai H, Ninomiya-Tsuji J, Matsumoto K and Sakurai N. (2002). Mol. Cell. Biol., 22, 992–1000. | Article | PubMed | ISI | ChemPort |
  38. O'Connell MA, Cleere R, Long A, O'Neill LA and Kelleher D. (1995). J. Biol. Chem., 270, 7399–7404. | PubMed | ISI | ChemPort |
  39. O'Neill LA. (2002). Mol. Cell, 10, 969–971. | Article | PubMed | ISI |
  40. Pahl HL. (1999). Oncogene, 18, 6853–6866. | Article | PubMed | ISI | ChemPort |
  41. Romieu-Mourez R, Landesman-Bollag E, Seldin DC and Sonenshein GE. (2002). Cancer Res., 62, 6770–6778. | PubMed | ISI | ChemPort |
  42. Roux S, Amazit L, Meduri G, Guiochon-Mantel A, Milgrom E and Mariette X. (2002). Am. J. Clin. Pathol., 117, 210–216. | Article | PubMed | ISI | ChemPort |
  43. Sapi E and Kacinski BM. (1999). Proc. Soc. Exp. Biol. Med., 220, 1–8. | Article | PubMed |
  44. Sathe SS, Sizemore N, Li X, Vithalani K, Commane M, Swiatkowski SM and Stark GR, manuscript submitted.
  45. Schiemann WP, Pfeifer WM, Levi E, Kadin ME and Lodish HF. (1999). Blood, 94, 2854–2861. | PubMed | ISI | ChemPort |
  46. Sezer O, Heider U, Jakob C, Zavrski I, Eucker J, Possinger K, Sers C and Krenn V. (2002). Blood, 99, 4646–4647. | Article | PubMed | ISI | ChemPort |
  47. Siebenlist U, Franzoso G and Brown K. (1994). Annu. Rev. Cell Biol., 10, 405–455. | Article | PubMed | ISI | ChemPort |
  48. Sizemore N, Lerner N, Dombrowski N, Sakurai H and Stark G. (2002). J. Biol. Chem., 277, 3863–3869. | Article | PubMed | ISI | ChemPort |
  49. Takahashi A, Kono K, Itakura J, Amemiya H, Feng TR, Iizuka H, Fujii H and Matsumoto Y. (2002). Oncology, 62, 121–127. | Article | PubMed | ISI | ChemPort |
  50. Ueda M, Terai Y, Yamashita Y, Kumagai K, Ueki K, Yamaguchi H, Akise D, Hung YC and Ueki M. (2002). Int. J. Cancer, 98, 335–343. | Article | PubMed | ISI | ChemPort |
  51. Verma IM, Stevenson JK, Schwarz EM, Van Antwerp D and Miyamoto S. (1995). Genes Dev., 9, 2723–2735. | Article | PubMed | ISI | ChemPort |
  52. Wolf JS, Chen Z, Dong G, Sunwoo JB, Bancroft CC, Capo DE, Yeh NT, Mukaida N and Van Waes C. (2001). Clin. Cancer Res., 7, 1812–1820. | PubMed | ISI | ChemPort |
  53. Xing L, Bushnell TP, Carlson L, Tai Z, Tondravi M, Siebenlist U, Young F and Boyce BF. (2002). J. Bone Miner. Res., 17, 1200–1210. | Article | PubMed | ISI | ChemPort |
  54. Yamamoto H, Kishimoto T and Minamoto S. (1998). J. Immunol., 161, 4753–4759. | PubMed | ISI | ChemPort |
  55. Yonemura Y, Endo Y, Fujita H, Fushida S, Ninomiya I, Bandou E, Taniguchi K, Miwa K, Ohoyama S, Sugiyama K and Sasaki T. (1999). Clin. Cancer Res., 5, 1823–1829. | PubMed | ISI | ChemPort |
  56. Zhang N, Zhong R and Deuel TF. (1999). J. Biol. Chem., 274, 12959–12962. | Article | PubMed |