Survival advantages conferred to colon cancer cells by E-selectin-induced activation of the PI3K-NFκB survival axis downstream of Death receptor-3

Recombinant human E-selectin/Fc (rhE-selectin/Fc) was obtained from R&D Systems (Minneapolis, MN). Phenylethylisothiocyanate (PEITC) and LY294002 were purchased from Sigma (St Louis, MO). Calcein-AM was obtained from Invitrogen-Molecular Probes (Burlington, ON, Canada). Dimethylsulfoxyde was purchased from Fisher (Montreal, QC, Canada). Protein G-sepharose was purchased from GE Healthcare (Mississauga, ON, Canada). PP2 and PD098059 were purchased from Calbiochem (Mississauga, ON, Canada). Rabbit anti-DR3 clone H300 was obtained from Santa Cruz biotechnology, mouse anti-DR3 extracellular domain (DR3ecd), mouse anti-vinculin (hVIN-1), rabbit anti-active caspase 3, and irrelevant mouse IgG1κ (MOPC21) were purchased from Sigma (St Louis, MO). Mouse anti-DR3 clone B65 was obtained from Millipore (Nepean ON, Canada). Mouse anti-DR3 (hDR3) was purchased from R&D Systems (Minneapolis, MN). Rabbit anti-phospho Akt (Ser 473), rabbit anti-Akt, rabbit anti-NFκB p65 and mouse anti-caspase 8 (1C12) were all obtained from Cell Signaling Technology, (Beverly, MA). Mouse anti-TATA Binding Protein (TBP) antibody was purchased from AbCam (Cambridge, MA). Goat anti-mouse IgG (H+L) and goat anti-rabbit IgG (H+L) conjugated with horseradish peroxidase were from Jackson Immunoresearch (West Grove, PA).

HT29 colorectal adenocarcinoma cells were cultivated in McCoy 5A medium supplemented with 10% foetal bovine serum (FBS) and antibiotics. HT29LMM (highly metastatic HT29 cells) and Jurkat T cells were cultivated in RPMI medium containing 10% FBS. Caco2 colorectal adenocarcinoma cells were grown in DMEM high glucose medium supplemented with 10% FBS and Glutamax 1X. SW480 and SW620 are colorectal adenocarcinoma cells isolated from the primary site and lymph node secondary site from the same patient. They were cultivated in Leibovitz medium L15 containing 10% FBS. LoVo colorectal adenocarcinoma cells grade IV were grown in Ham F12K medium supplemented with 10% FBS. HIEC cells are normal human intestinal epithelial cells that were cultivated in OptiMEM containing 5% FBS and 5 ng/ml EGF [33]. HEK293, HeLa, MDA MB231 and MCF7 cells were cultivated in DMEM containing 10% foetal calf serum. All these cell lines were obtained from ATCC.

Human umbilical vein endothelial cells (HUVEC) were isolated by collagenase digestion of umbilical veins from undamaged sections of fresh cords, as described [34]. The cells used at passages ≤ 5 were grown to confluence in gelatin-coated tissue culture flasks in medium 199 containing 20% heat-inactivated FBS, endothelial cell growth supplement (60 μg/ml), glutamine (2 mM), heparin (25000 IU). Human micro-capillary endothelial cells (HMEC) were cultivated in MCDB medium containing 10% FBS, 1 μg/ml hydrocortisone and 10 ng/ml EGF. All cells lines were cultivated in the presence of antibiotics and maintained at 37°C in a 5% CO₂ humidified atmosphere.

HUVEC were trypsinized and grown for 24 hrs on gelatin-coated slides. These endothelial cells were treated with 20 ng/ml IL-1β for 4 h to induce the expression of E-selectin. The cultures were then placed in the laminar flow chamber GlycoTech (Gaithersburg, MD, USA) under a shear stress of 1 dyne/cm². In certain experiments, anti-human DR3 monoclonal Ab clone B65 or MOPC21 irrelevant antibody were added in the culture medium of HT29 cells, 30 min before their injection in the chamber. In other experiments, a knockdown of DR3 was performed by small interfering RNA, as previously described [9, 23]. Briefly, HT29 cells were transfected by electroporation with human DR3 siRNA (siRNA; sense, 5'-CCGUCCAGUUGGUGGGUAA-3', and antisense, 5'-UUACCCACCAACUGGACGG-3') or control siRNA purchased from Qiagen (Mississauga, ON, Canada). Tumor cells in suspension (2 × 10⁶ per assay) were labeled for 30 min with Calcein AM and washed twice with M199 medium before being added into the flow chamber. Videos were taken directly using a camera mounted on a TE2000 fluorescence microscope at ×20 magnification (Nikon, Melville, NY, USA).

Twenty-four hours after being plated, HT29 cells were left to grow for 96 hours with or without E-selectin or with the apoptosis inducer curcumin (75 μM) [35]. At the end of the treatments, the cell survival was evaluated with the Quick Cell Proliferation Assay Kit from BioVision (Mountain View, CA). The test evaluates the ability of viable cells to convert tetrazolium salt into formazan (WST-1 assay), which can be monitored at 450 nm.

Cells were washed twice and incubated in serum-free medium for 2 hours in the presence or not of the inhibitors (LY294002 or PP2). Thereafter, rhE-selectin was added for different periods of time. Cell extracts were prepared and PI3K and NFκB activation were assayed in western blotting by determining the phosphorylation of Akt at Ser 473 and nuclear translocation of p65NFκB, respectively.

The protocol was adapted from Andrews and Faller [36]. Cells were washed 3 times in PBS and were re-suspended in 1.6 ml of PBS. The cell suspension was briefly vortexed and 100 μl of total extract were collected and mixed in 20 μl of extraction buffer. The rest of the cell suspension was centrifuged (16,000 × g) for 10 seconds at 4°C, and the pellet was resuspended in 400 μl of buffer A (10 mM HEPES-NaOH pH 7.9, 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM DDT, 0.2 mM PMSF). The extract was left on ice for 10 min, vortexed for 10 seconds and centrifuged for 10 seconds at 4°C. The supernatant was removed and discarded, and the pellet was resuspended in 70 μl of buffer C (20 mM HEPES-NaOH pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DDT, 0.2 mM PMSF). The samples were incubated on ice for 20 minutes and centrifuged for 2 min at 4°C. Extraction buffer was added in each extract before heating. The amount of proteins was quantified by the Lowry method.

Total RNA was extracted from (1 × 10⁶) cells using Qiagen RNeasy kit (Mississauga, ON, Canada). All RNA samples were stored at -80°C until assay. The mRNA was reverse-transcribed with Qiagen Sensiscript reverse transcription kit using random hexamers. Nested PCRs were used to amplify a fragment of the tnfrsf25 gene using specific pairs of primers and the Qiagen Hotstart taq DNA polymerase kit according to the manufacturer protocol (PCR1-Forward primers: 5'-CGTCGGAGGGCTATGGAGCAGC-Reverse primers: 5'-GGCCGGCTGGTGCTGCTACGC. PCR2-Forward primers: 5'-GAGGATCCATGCAGGGCGGCACTCGTAGC-Reverse primers: 5'-ACCTCGAGTCACGGGCCGCGCTGCAG). PCR products were cloned in pcDNA3 (Invitrogen, Burlington ON, Canada) vector and were sequenced by CRCHUQ/CHUL sequencing platform (Québec Qc, Canada). The DR3 sequences were compared with those found in the BLAST database and analyzed with the Human Genome Browser Gateway http://​genome.​ucsc.​edu/​cgi-bin/​hgGateway and the AceView genes http://​www.​ncbi.​nlm.​nih.​gov/​IEB/​Research/​Acembly/​ databases.

Total RNA was extracted from (1 × 10⁶) cells using Qiagen RNeasy kit and one μg was used for a reverse transcription using Omniscript reverse transcriptase (Qiagen). Then, the full length DR3 was amplified by PCR using Qiagen Hotstart polymerase and the following primers: 5'-CGTCGGAGGGCTATGGAGCAGC and 5'-GGCCGGCTGGTGCTGCTACGC following the manufacturer's instructions. Thereafter, the region from exon 5 to exon 7 of DR3 was amplified by PCR, as previously described, using DR3 full length PCR product as a template and the following primers: 5'-CCCGCAGAGATACTGACTGTGGGAC and 5'-GTAGCCAGGGGTCCAGCTGTTACC. The resulting products were separated by agarose gel electrophoresis.

For more precise quantification, targeted PCR reactions were carried out, and the amplified products were analyzed by automated chip-based microcapillary electrophoresis on an Agilent 2100 Bioanalyzer instrument (Agilent Technologies, Santa Clara, CA) as previously described [37]. Amplicon sizing and relative quantification was performed by the manufacturer's software. The primers used were Forward: 5'-TTCCCGCAGAGA TACTGACTG and Reverse: 5'-AGCACCTGG ACCCAGAACA.

1) by DNA fragmentation . HT29 cells were treated with rhE-Selectin/Fc at 10 μg/ml for 4 hours or 24 hours, or were treated with phenethyl isothiocyanate at 50 μM for 24 hours. Cells were washed twice with PBS, fixed with 3,7% formaldehyde and stained with Hoechst for 60 min at room temperature in the dark. The cells were examined with a Nikon Eclipse 800 equipped with a 40 × objective lens. 2) by caspase activation . Caspase 8 and 3 activities were evaluated by western blotting using anti-caspase 8 and anti active-caspase-3 antibodies. The assays were performed on pools of cells containing both floating and adhering cells.

We previously reported that the adhesion of HT29 colon cancer cells to endothelial cells under static conditions is mediated by the binding interaction between DR3 expressed by cancer cells and E-selectin expressed by endothelial cells [23, 38]. Considering that the adhesion of cancer cells to the endothelium in vivo occurs under flow and shear stress conditions, we ascertained the role of DR3 in mediating adhesion of colon cancer cells to E-selectin under flow conditions using a laminar flow chamber.

HUVEC forming a tight monolayer on gelatin-coated glass slides were treated or not for 4 hours with IL-1β to induce the expression of E-selectin. Then, the cultures were placed in a laminar flow chamber in which medium circulated under a flow that gave a physiological shear stress of 1 dyne/cm² [9]. Live HT29 cells (2 × 10⁶ per assay) stained with Calcein AM and pre-treated or not with anti-DR3 antibody or an siRNA that knocks down the expression of DR3 were injected in the flow system and video sequences were taken at 25 minute intervals. The cells attached to the endothelium were counted in more than 5 fields per condition. Results showed that, after the first 25 min, no HT29 cancer cell adhered to endothelial cells that did not express E-selectin. However, they adhered in a time-dependent manner to HUVEC expressing E-selectin and the adhesion was blocked by treating the endothelial layer with an anti-Eselectin antibody (data not shown and [9]). These findings clearly indicated that the adhesion of HT29 cells to endothelial cells was E-selectin-dependent. As shown in Figure 1A-F (and additional files 1, 2, 3 and 4), the adhesion was also DR3-dependent given that inhibiting DR3 with the anti-DR3 antibody or knocking down its expression with siRNA led to a 7-fold reduction of the adhesion of HT29 cells to HUVEC expressing E-selectin.

These results suggest that the adhesion of colon cancer cells in blood circulation relies mainly on DR3/E-selectin interaction. In a previous study, we described three distinct mechanisms by which circulating cancer cells interact with E-selectin to initiate transendothelial migration: formation of a mosaic between cancer cells and endothelial cells, paracellular diapedesis at the junction of three endothelial cells, and transcellular diapedesis [9]. The results of the present study now suggest that DR3 expressed by colon cancer cells is a major partner of E-selectin in inducing these mechanisms of diapedesis in vivo. In particular, it is possible that DR3 binding to E-selectin is the initial event that activates E-selectin oligomerization and thereby ERK-mediated disruption of the adherent junctions and diapedesis [38]. Another possibility is that the DR3/E-selectin binding triggers the release of chemokines or cytokines, such as VEGF, by endothelial cells or cancer cells, which later triggers diapedesis [39, 40].

DR3 is a member of the TNF receptor family whose activation is typically associated with apoptosis [24]. Along these lines, the ectopic expression of DR3 in HEK293 or HeLa cells induced marked apoptosis [30]. Accordingly, we next investigated whether the activation of DR3 by E-selectin triggers apoptosis. We found that chimeric rhE-selectin/Fc taken as ligand did not induce apoptosis in HT29 cells, even at concentrations twice as those required to induce DR3-mediated activation of p38 [23]. This is illustrated in Figure 2A-C which shows that rhE-selectin/Fc at a concentration of 10 μg/ml did not induce nuclear fragmentation even after 24 h exposure. In contrast, phenylethyl isothiocyanate, a death receptor-independent inducer of apoptosis in these cells exerted a strong apoptotic response [41, 42] (Figure 2D). Consistent with these findings, we found that E-selectin, in contrast to curcumin, did not reduce cell survival even after 96 h of exposure, as determined by the WST-1 assay (Figure 2E). In the in vivo context, these results suggest that the DR3-mediated adhesion of colon cancer cells to endothelial cell E-selectin may trigger activation of survival pathways in cancer cells that impair apoptosis.

Inhibition of ERK is associated with a weak increase in the activation of caspase-3 in LoVo colon cancer cells treated by rhE-selectin/Fc [23]. This suggests that another pathway is involved in conferring resistance to apoptosis to colon cancer cells adhering to E-selectin. We thus evaluated the contribution of the PI3K pathway given that it is a major pro-survival pathway. By measuring the phosphorylation of AKT at Ser473 (Figure 3A), we found that exposure of HT29 cells to rhE-selectin/Fc induced a time-dependent activation of PI3K which peaked at 15 min. The activation of PI3K by E-selectin is dependent on DR3 activation given that it was abolished by two DR3 neutralizing antibodies (Figure 3B). Interestingly, the E-selectin-induced phosphorylation of Akt at Ser473 was sensitive to LY294002, a well-known inhibitor of PI3K activity (Figure 3C). In line with the findings that showed that PI3K activation was downstream of Src in response to different cytokines including TNFα, we found that the phosphorylation of Akt at Ser473 was also sensitive to Src inhibition by the pan Src inhibitor PP2 (Figure 3D)[43, 44]. Interestingly, DR3 contains an ITAM motif within its death domain that harbors two tyrosine residues (Y376 and Y394) that have been suggested to be phosphorylated via Src activation [45]. In light of our results, it is thus possible that Src-dependent activation of the PI3K pathway may originate from an Src-mediated phosphorylation of one of these tyrosines. Hence, these findings suggest that E-selectin-mediated activation of Src may trigger phosphorylation of DR3 which would converge on the activation of the PI3K pathway, a major regulator of cell survival [46]. Accordingly, we next investigated the signaling events by which the activation of PI3K downstream of DR3 may mediate the survival of colon cancer cells.

Earlier findings have highlighted the point that, depending on cell types and cellular context, DR3 activation was associated either with apoptosis following the recruitment of the apoptotic cascade on the death domain, or survival following activation of the pro-survival factor NFκB [31]. Hence, we next investigated the status of NFκB following activation of DR3 by E-selectin. As shown in Figure 4, we found that E-selectin induced a LY294002-sensitive and thereby PI3K-dependent activation of NFκB, as evaluated by the translocation of NFκB-p65 subunit into the nucleus (Figure 4). Previous studies have reported that NFκB was activated by DR3 and other TNFR following the activation of NFκB-inducing kinase downstream of the recruitment of TRAF2 to the receptor death domain [47, 48]. In turn, this leads to increased survival [29, 31, 49]. Here our findings suggest that the activation of NFκB downstream of DR3 may be independent of the TRAF2 pathway and would depend on the activation of the PI3K/Akt pathway, presumably downstream of a Src-dependent tyrosine phosphorylation of DR3 within the ITAM motif [50, 51]. This possibility is in line with the finding that cell survival downstream of CD95/Fas is associated with its tyrosine phosphorylation, upstream of the activation of the PI3K/AKT pathway [26]. Consistent with a role of PI3K/NFκB pathways in protecting HT29 cells from apoptosis in response to E-selectin, we further found that the inhibition of PI3K by LY294002 increased the cleavage of caspase 8 in response to E-selectin (Figure 5A). We previously reported that ERK contributes to protect colon cancer cells from apoptosis following activation of DR3 by E-selectin [23]. Accordingly, the co-inhibition of both ERK and PI3K, respectively by PD098059 and LY294002, was associated with a synergistic activation of both caspase-8 and caspase-3 in response to E-selectin (Figure 5B). This result indicates that both ERK and the PI3K/Akt//NFκB axis contribute to confer apoptosis resistance to colon cancer cells in response to E-selectin. In addition, it confirms the pro-survival function of the ERK pathway downstream of DR3, as we previously reported [23].

Next, we verified whether a mutation in DR3 could further contribute to the lack of apoptosis induced by E-selectin. By using PCR approach, we cloned and sequenced DR3 cDNA, and we found major variations in the expression profile of DR3. From 36 different clones, we discovered that, in addition to the full-length version of DR3, HT29 cells expressed splice variants of DR3. One of them is characterized by a loss of exon 6 (DR3Δ6) (Additional file 5). The joint between the last two nucleotides of exon 5 and the first two nucleotides of exon 7 leads to a shift in the reading frame introducing a premature stop codon, located at the beginning of exon 8 (Figure 6A). This variant codes for a new protein whose last 37 amino acids are not found in any of the known variants of DR3. This protein has no trans-membrane and death domain (Figure 6B) and thus is unable to trigger apoptosis. Interestingly, by PCR amplification of the region around exon 6, we found that the relative proportion of DR3Δ6 was higher in metastatic colon cancer cells (HT29, HT29LMM, SW620) in comparison to normal colon epithelial cells (HIEC) and endothelial cells (HUVEC, HMEC), as well as in metastatic cancer cells that are not of colon origin (Jurkat, Hela, MCF7, MDAMB231) (Figure 6C). Notably, it is particularly clear that the relative level of of DR3Δ6 to full length DR3 is higher in metastatic SW620 cells relative to non-metastatic SW480 cells taken from the primary tumor site of the same patient. In fact, more precise quantification by targeted PCR reactions and analysed of the amplified products by chip-based microcapillary electrophoresis indicated that the ratio of DR3Δ6 to full length DR3 doubled in SW620 cells relative to SW480 (Figure 6D). These findings strongly suggest that the expression of DR3Δ6 is associated with a metastatic phenotype in colon cancer. In turn, this raises the possibility that, during the acquisition and progression of malignancy, colon cancer cells evolved to develop alternative splicing mechanisms favoring the shifting of a death-receptor toward a survival receptor. Along these lines, it was shown that a variant of DR3, (DR3β), differs from the described DR3 isoform 2 by the inclusion of a 28 amino-acid stretch in the extracellular domain. Whereas DR3 was expressed in all the cell lines and lymphoma samples tested, DR3β expression was restricted to lymphoid T-cell and immature B-cell lines and to some cases of follicular lymphoma. This is consistent with our finding that different isoforms of DR3 can contribute to cancer [28]. It is difficult at present to fully understand the mechanism of alternative splicing regulation acting on DR3. One possibility relies on the phosphorylation of serine-arginine rich proteins (SRPs) known to be major regulators of alternative splicing in colon cancer cells [52]. This is further supported by the fact that PI3K which is activated by E-selectin-mediated stimulation of DR3 also regulates the phosphorylation of SRPs [53]. Interestingly, death decoy receptor-3 (DcR3), another member of the TNF receptor superfamily, is a soluble receptor that is highly expressed in various tumors including colon cancer and that act as a negative regulator of DR3[54]. Although, DR3Δ6 differs in sequence from DcR3, it is possible that it also acts as a decoy receptor for the activation of DR3 by E-selectin.