In this study, we have used WI-38 non-transformed fibroblasts as a platform to determine the effects of hyper-activated Ras and down-regulated p53 on the cluster of cancer-related chemokines. To this end, the cells were infected to express the constitutively active Ras^G12V mutant (Ras^High = Ras^H) along with dysfunctional modalities in p53 that are prevalent in human malignant diseases: (1) p53 knock down by specific shRNA (p53^Low = p53^L) that recapitulates the allelic loss of p53 or its abnormal degradation; (2) Over-expression of vectors expressing mutated p53, namely p53^R175H or p53^R248Q, representing frequent tumor-promoting mutations of p53 in human cancers [19,20,21,22].

The results of Figure 1 show that the combination of Ras hyper-activation + p53 dysfunction has induced a significant elevation in secretion of the members of the pro-malignancy chemokine network, consisting of CCL2, CCL5 and CXCL8, as well as of CXCL10 which is the chemokine showing more complex roles in malignancy (Figure 1A–D, respectively). Of the three modalities used for deregulation of p53, the p53^L (shRNA) and p53^R248Q forms were very effective in inducing the secretion of all chemokines, while the p53^R175H modality was much less efficient in most analyses and at times not active (see below). Overall, these results indicate that in cells that carry a normal phenotype, combined expression of the Ras and p53 oncogenic events shifts the balance in favor of a malignancy phenotype which is manifested by the elevated release of pro-cancerous chemokines.

After determination of the effects induced by hyper-activated Ras and dysfunctional p53 on chemokine secretion, we wished to know if either oncogenic modification could act alone to increase the release of the chemokines, or whether cooperation between them is required. Because of its high effectiveness in promoting chemokine release when combined with Ras hyper-activation (Figure 1), in this part of the study we focused on the p53^L mode of p53 inactivation (by p53 shRNA).

To this end, we have expressed in the non-transformed cells the hyper-activated form of Ras and p53^L, each alone or together. The results of Figure 2 show minimal effect of hyper-activated Ras alone (Ras^H), or of knocked-down p53 alone (p53^L) on chemokine secretion. However, when combined, Ras^H synergized with p53^L, together leading to a prominent release of CCL2, CCL5, CXCL8 and CXCL10 by the cells.

These results indicate that cooperation between oncogenic modifications in Ras and p53 is required for induction of the cancer-related chemokine cluster. This finding shows that this chemokine-cluster readout obeys the same regulatory patterns observed for other tumor-promoting characteristics, in that its induction requires synergism between at least two oncogenic modifications [30,31,32,33,34,35,36,37,38]. The above findings also indicate that the chemokine cluster serves as an appropriate manifestation of the malignancy phenotype of the cells, and more specifically they point to the important roles played by hyper-activated Ras and dysfunctional p53 in regulation of the inflammatory setup of the tumor microenvironment, as indicated by the chemokine readout. Thus, acting together, hyper-activated Ras and dysfunctional p53 may give rise to exacerbated release of inflammatory chemokines with tumor-supporting functions, that can promote pro-cancerous activities: in tumor cells (originating due to genetic/signaling modifications), and in microenvironmental cells at their vicinity.

Also, our findings suggest that in patients exhibiting hyper-activation of the Ras pathway—due to Ras mutations and/or over-expression of RTKs—and also deregulated p53, chemokine release by the tumor cells would be promoted and the tumor microenvironment would be exposed to their tumor-supporting effects. This mode of regulation may be a part of a more complex and reciprocal net of interactions that exists between chemokines and oncogenic modifications, since recent findings have shown that over-expression of the chemokine receptor CXCR4—that is associated with tumor metastasis [39,40]—has led to up-regulation of RTKs and to deregulation of the p53-MDM2 axis [41].

Published studies indicate that combination between oncogenic events can lead to complete transformation of cells, to the degree that they can establish full-blown tumors [30,31,32,33,34,35,36]. Along these lines, the Ras^H + p53^L cells that were used in our study, to be now termed “Ras^H/p53^L-in-culture” cells, were found to form tumors when inoculated to mice [3,4]. The gain of tumorigenic potential by these cells has motivated us to ask what would be the effect of exposure to host systems on the expression of the cancer-related chemokine cluster, in cells expressing oncogenic modifications.

To analyze this question, following the inoculation of the Ras^H/p53^L-in-culture cells to mice, cancer cells were excised from the tumors that have developed, and were brought back to culture [4]. The tumor cells, termed herein TUMOR-Ras^H/p53^L cells, were found to exert a powerful elevation in the release of the cancer-related chemokines when compared to control non-modified cells (Figure 3). To assess the specific contribution of the tumor microenvironment to the expression of the chemokines, we have compared the TUMOR-Ras^H/p53^L cells to their in-culture counterparts. This comparison revealed that exposure to the host microenvironment has further induced prominent elevation in the release of the chemokines (Figure 4A,B). In both cell types, CXCL8 was the most prominent chemokine released by the cells. Of note, the chemokine CXCL10 obeyed the regulatory pattern observed for the typical tumor-promoting chemokines CCL2, CCL5 and CXCL8, however in general it was released in lower levels than the other chemokines. The lower efficiency obtained for CXCL10 induction by the Ras and p53 modifications, as well as by the in vivo constrains, may reflect the more complex roles of this chemokine in cancer.

Taken together, these results indicate that malignant transformation due to co-expression of oncogenic modifications induces the expression of the cancer-related chemokines; however, the expression of the chemokines can be further promoted by exposure of the cells to elements residing at the tumor microenvironment. These findings emphasize the importance of host systems in selecting cells that have preferential expression of tumor-promoting traits, including gene products that affect angiogenesis and additional pro-cancerous activities, such as those attributed to the cancer-related chemokines [4,42,43,44,45].

The results shown in Figure 3 and Figure 4A–B actually show that TUMOR-Ras^H/p53^L cells retained an amplified pattern of chemokine release when they were brought back to culture, indicating that they carry permanent modification/s which enable them, in vitro, to keep releasing exacerbated levels of the chemokines. Specifically, the high basal release of the cancer-related chemokine cluster by these cells may reflect the constitutive activation of transcription factors known to regulate such chemokines. To address this possibility, we have determined the activation level of AP-1, a key inducer of inflammatory chemokines in the context of immune activities [46,47,48], in TUMOR-Ras^H/p53^L cells. Interestingly, we found that while the total levels of the c-Jun subunit of AP-1 has been consistently reduced in these cells, the phosphorylation level of the protein remained high, thus giving rise to elevated basal activation of AP-1 in these cells (Figure 4C). In comparison, no constant activation of c-Jun was detected in Ras^H/p53^L-in-culture cells (Figure 4C). Overall, the AP-1 transcription factor had a higher basal activation in the TUMOR-Ras^H/p53^L cells that have emerged from the in vivo passage, than control cells in which the expression of Ras and p53 was not modified.

The elevated basal activation of AP-1 in TUMOR-Ras^H/p53^L cells (Figure 4C) and the high chemokine amounts released by these cells (Figures 4A–B), point to the possibility that inflammatory mediators play key roles in regulating the inflammatory chemokines in Ras^H/p53^L cells that were exposed to host systems in vivo. Inflammatory cytokines, such as Tumor Necrosis Factor α (TNFα) and Interleukin 1β (IL-1β) are known to elevate the release of inflammatory chemokines in the course of immune activities, and to activate the transcription factors AP-1 and NF-κB under such conditions [46,47,48,49,50,51,52,53,54,55,56]. Furthermore, these two cytokines were recently identified as tumor-promoting factors in many malignant diseases [24,25,26,27,28,29]. These published findings suggested that TNFα or IL-1β may be potential candidates whose activities in vivo may have led to the extremely high levels of the chemokine cluster in TUMOR-Ras^H/p53^L cells.

At the tumor site, the sources for TNFα or IL-1β may be in host cells that are located in the tumor microenvironment, or in the tumor cells themselves. Therefore, we have determined the possibility that TUMOR-Ras^H/p53^L cells have acquired the ability to permanently release TNFα and/or IL-1β, that then act in autocrine manners on the cells and induce the release of the inflammatory chemokines. Our data indicated that this was not the case, because TNFα or IL-1β were not secreted by the TUMOR-Ras^H/p53^L cells (data not shown).

In view of the above, we raised an alternative possibility hypothesizing that the elevated levels of activated AP-1 in TUMOR-Ras^H/p53^L cells actually attest for repeated selective pressures that were applied by an inflammatory tumor microenvironment in the host, leading eventually to selection of cells that carry high basal activation of transcription factor/s that promote the expression of the chemokines. In such a case, it is possible that the TUMOR-Ras^H/p53^L cells manifest high responsiveness to inflammatory cytokines, which has led to the activation of AP-1.

Supporting this hypothesis were the findings of Figure 5, showing analyses of TUMOR-Ras^H/p53^L cells following stimulation by TNFα or IL-1β. Measuring the activation of the AP-1 subunit c-Jun, we found that IL-1β and more potently TNFα activated the AP-1 signaling pathway in TUMOR-Ras^H/p53^L cells and in control cells (Figure 5A). However, the results of Figure 5B show that the overall phosphorylation levels of c-Jun in TUMOR-Ras^H/p53^L cells was higher than in control cells. Because we could not obtain sufficient reduction of c-Jun expression by siRNA (data not shown), we have determined the direct ability of IL-1β and TNFα to activate AP-1 by using a dual luciferase assay with AP-1 reporter. These assays were performed only on the TUMOR-Ras^H/p53^L cells, because they were showing higher responsiveness to the cytokines than control cells, and have indicated that the two cytokines, and primarily TNFα induced the activation of this pathway in the TUMOR-Ras^H/p53^L cells )Figure 5C).

In parallel, we have determined the activation of the NF-κB pathway, first asking if TNFα and IL-1β reduce the levels of the IκBα inhibitor of NF-κB in control and TUMOR-Ras^H/p53^L cells. Using the IκBα readout, we found that TNFα and IL-1β acted primarily on the TUMOR-Ras^H/p53^L cells to reduce IκBα levels, with TNFα showing more prominent effects compared to IL-1β (Figure 6A). The results of Figure 6B show that the effects of the cytokines on IκBα levels were more prominent in TUMOR-Ras^H/p53^L cells than in control cells. These results suggest that the two cytokines activate NF-κB in TUMOR-Ras^H/p53^L cells. Support for this possibility was provided by dual luciferase assays with NF-κB reporter, showing the activation of this pathway by both cytokines, with higher activation achieved by the TNFα stimulation (Figure 6C).

Overall, the TUMOR-Ras^H/p53^L cells, which were exposed to host-microenvironmental factors in vivo have shown preferential response to TNFα or IL-1β stimulation, possibly by activating both NF-κB and AP-1 signaling pathways.

The above results suggest that the TUMOR-Ras^H/p53^L cells attest for processes that may take place in vivo, in which the activity of the inflammatory cytokines induces in the cells a specific setup and transcriptional up-regulation that eventually lead to elevated release of the inflammatory chemokines at the tumor site. To recapitulate this in vivo process, we exposed the cells that were modified to express oncogenic Ras and p53 and were kept in-culture, to the inflammatory mediators TNFα and IL-1β. Here, of the four chemokines we chose to focus on CCL2 and CCL5 because together their effects represent a broad spectrum of cancer-related activities that are exerted by the chemokines included in the cluster studied in this research.

Specifically, we used cells that expressed Ras^H together with the different modifications of p53: p53^R175H, p53^R248Q or p53^L. In all of these cells, the stimulation by the inflammatory cytokines has led to increased CCL2 and CCL5 secretion which was way beyond the one obtained by the combined oncogenic modifications of hyper-activated Ras and deregulated p53 together (Figure 7A,B, respectively). Moreover, the effect of the cytokines was obtained regardless of the p53 modification used, because potent induction of the chemokines by TNFα and IL-1β was obtained in all forms of deregulated p53, namely p53^R175H, p53^R248Q and p53^L. Of note, the chemokine amounts obtained by stimulation of the different cell types kept in culture, expressing the oncogenic modifications, with the cytokines, were in general similar to those of TUMOR-Ras^H/p53^L cells, or even higher (depending on the chemokine type).

An important point that was revealed by these analyses is that the inflammatory cytokines induced the release of the cancer-related chemokine cluster to greater extent than the combination of deregulated Ras and p53 together, and acted powerfully also without the oncogenic modification, as indicated by their ability to potently induce the release of CCL2 and CCL5 by control cells that were not modified with Ras hyper-activation and p53 down-regulation. Based on all the above findings, our conclusion is that when the chemokine cluster is used as readout for a malignancy phenotype, the tumor microenvironment has a primary role in dictating the malignancy trait of the cells. Nevertheless, when in vivo settings are concerned, it is possible that interactions between the inflammatory milieu and the oncogenic modifications would lead to a vicious cycle that potentiates the release the cancer-related chemokines.

In this study, WI-38 human lung fibroblasts were used, following immortalization by hTERT as previously described [3,4]. These cells originally express wild type (WT) Ras and WT p53, and they were infected with the following constructs, as previously described [3,4]: Ras^High = Ras^H—A constitutively active human H-Ras^G12V mutant expressed by a hygromycin-expressing vector; p53 shRNA (p53^Low = p53^L), p53^R175H, p53^R248Q expressed by puromycin-expressing vectors. Controls included WI-38 cells infected with the empty vector of hygromycin and with a puromycin-expressing vector (shRNA against the mouse but not the human NOXA gene was used as control for shRNA of p53).

In parallel to the cell lines kept in culture, cells expressing H-Ras^G12V and p53^L (termed “Ras^H/p53^L-in-culture”) were inoculated to nude mice [4]. The cells gave 100% incidence of local tumor uptake (without metastases), and cells obtained from such a tumor have been brought back to culture [4]. In contrast to primary murine host cells that cannot survive the continuous growth in culture (such as macrophages or fibroblasts), the cells excised from the tumor have given rise to a cell line that could be continuously grown in vitro, termed herein “TUMOR-Ras^H/p53^L”. Pathologic examination determined the sarcoma type of the resulting cells [4], having high mitotic rate.

Up-regulation of Ras expression and down-regulation of p53 expression were verified in all cell lines by real time qPCR (Supplementary Figure 1; Verification for some of the cell lines by Western blotting was also provided in [4]). All cell lines, including TUMOR-Ras^H/p53^L cells, had similar growth rates in culture (data not shown).

The different cell types, as appropriate, were cultured in growth medium. Then, the cells were washed twice in PBS and incubated overnight in Low Protein Medium (LPM) containing TNFα (50 ng/mL; 300-01A) or IL-1β (500 pg/mL; 200-01B) (PeproTech, Rocky Hill, NJ, USA). Cytokine concentrations were selected based on titration analyses performed in other studies in the laboratory. Control cells were grown under similar conditions, with 0.1% BSA, the solubilizer of TNFα and IL-1β.

Chemokine expression in the supernatants of the cells was determined by sandwich ELISA, at the linear range of absorbance, using recombinant human proteins for generation of standard curves. As expected of a biological system, there were fluctuations in the levels of chemokines released by the same cell line in the various assays. However, it is important to note that the general ratios between the secretion levels of the different cell types used in this study, as well as the ratios between the different chemokines in the same cell line, were kept proportionate in the different assays along the whole study.

The following antibodies were used: Coating antibodies—mouse monoclonal antibodies against human CCL2 (500-M71, PeproTech), CCL5 (500-M75, PeproTech), CXCL8 (508402; BioLegend, San Diego, CA, USA) and CXCL10 (500-P93, PeproTech); Detecting antibodies—biotinylated goat antibodies against human CCL2 (500-P34Bt, PeproTech), CCL5 (BAF278, R&D Systems), CXCL8 (BAF208; R&D Systems, Minneapolis, MN, USA) and CXCL10 (500-P93Bt, PeproTech). After the addition of streptavidin-horseradish peroxidase (HRP, Jackson ImmunoResearch Laboratories, West Grove, PA, USA), the substrate TMB/E solution (Chemicon, Temecula, CA, USA) was added. The reaction was stopped by the addition of 0.18 M H2SO4 and optical density (OD) was measured at 450 nm. p values were calculated by Student’s t test.

The different cell types, as appropriate, were lysed in RIPA lysis buffer. Cell lysis was followed by conventional Western blot procedures. The following proteins were analyzed by the appropriate antibodies: phosphorylated c-Jun (sc-822; Santa Cruz Biotechnology, Santa Cruz, CAF), c-Jun (610326; BD Transduction Laboratories, San Jose, CA), IκBα (4814; Cell Signaling Technology) and GAPDH (MAB374; Chemicon).

After washings, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies, as appropriate: sheep anti-mouse-HRP (NA931; Amersham Pharmacia Biotech, Buckinghamshire, UK) or goat anti-rabbit-HRP (111-035-003; Jackson ImmunoResearch Laboratories). The membranes were subjected to enhanced chemilluminescence (Amersham), and bands on immunoblots were quantified by densitometry. Because of consistent reduction in c-Jun expression levels (total c-Jun) in TUMOR-Ras^H/p53^L cells, “Relative c-Jun phosphorylation levels” were calculated by dividing the densitometry values obtained for phosphorylated-c-Jun, by the values of total c-Jun, and then by GAPDH which served as a loading control (phosphorylated c-Jun/total Jun/GAPDH).

The cells were transfected with vectors containing a luciferase gene under the control of an AP-1 binding sites (5 × coll-TRE-tata; kindly provided by Tsaffrir Zor, Tel Aviv University), or NF-κB binding sites (3 × KBL; kindly provided by Stefan Wiemann, DKFZ Heidelberg, Germany). A construct coding for renilla luciferase was used for normalization of the results according to transfection yields (kindly provided by Tsaffrir Zor, Tel Aviv University). 48 h after transfection, the cells were stimulated by TNFα (8 h) or IL-1β (6–8 h) (time points were selected following a kinetic analysis), and processed with the reagents provided in Dual-Luciferase Assay System Kit (Promega, Madison, WI, USA). Luciferase activity was determined using the same kit, according to manufacturer’s instructions.