Insulin-like Growth Factor-I Protects Colon Cancer Cells from Death Factor-induced Apoptosis by Potentiating Tumor Necrosis Factor α-induced Mitogen-activated Protein Kinase and Nuclear Factor κB Signaling Pathways¹

Instructive apoptosis is a kind of apoptosis in which death factors play a central role. It is essential in maintaining tissue homeostasis and eliminating deleterious cells (1, 2, 3). When this system under- or overfunctions, it contributes to the pathogenesis of a number of human diseases (4). In the intestinal mucosa, recent evidence shows that an excess of cell death is associated with inflammatory bowel diseases, whereas increased cell survival contributes to the outgrowth of colon cancer cells (5, 6).

The best characterized death factors Fas ligand and TNF⁴bind to ubiquitously expressed members of the TNFR superfamily. Fas ligand binds to Fas/CD95/APO-1 receptor, and TNF binds to two receptors, p55 (TNFR1) and p75 (TNFR2), that do not share any homology within their cytoplasmic domain. Both Fas and TNFR1 can activate apoptotic signaling pathways through a similar mechanism, recruiting directly or indirectly Fas-associated death domain protein and pro-caspase 8 (1, 2, 3). However, both TNFR1 and TNFR2 also associate with molecules that do not interact with Fas, especially the TRAF family of adaptor proteins and receptor-interacting protein. These molecules activate additional signaling pathways including NF-κB and the MAPK cascades (especially, JNK and p38), these latter being involved in the stimulation of AP-1 activity (7, 8, 9). TNF recruitment of both NF-κB and AP-1 transcription factors is pivotal to regulate many genes, especially those involved in expression of inflammatory cytokines and cell survival (10, 11, 12, 13, 14). Thus,TNF transmits one signal eliciting cell death and another that protects against cell death, this latter being closely linked to the proinflammatory signaling. In contrast, Fas-mediated signal appears to be simpler and does not lead to direct and efficient NF-κB and AP-1 activation. This may explain why Fas activation generally results in a more efficient apoptotic response (15).

The phenomenon of resistance by tumor cells to death factor-induced apoptosis is of major concern in cancer therapy. It contributes in a great part to the limited effectiveness of naturally occurring as well as peptide/cytokine-driven antitumor immune response generally observed in cancer patients. Moreover, this resistance may also antagonize the efficiency of chemotherapeutic drugs because many of them induce apoptosis of tumor cells by activating death factor/receptor systems,particularly the Fas/Fas ligand system (16, 17).

Several growth factors have been identified as regulators of cell survival (18), and among them IGF-I, IGF-II, and insulin have been reported to have a potent ability to protect a broad range of cells from a variety of proapoptotic challenge (19). The biological functions of the IGFs and insulin are pleiotropic and mediated by specific membrane receptors designated IGF-IR and IR,respectively. These receptors are heterotetrameric proteins with a highly homologous intracellular tyrosine kinase domain. An earliest step in signal transduction by both IGF-IR and IR is the extensive tyrosine phosphorylation of IRS-1, which initiates several distinct signaling pathways such as PI-3′K and MAPK cascades. In addition, the actions of IGFs, but not of insulin, are regulated by interactions with IGF binding proteins that modulate the IGF bioavailability to cell surface IGF-IR (reviewed in Refs. 20 and 21).

The gastrointestinal tract is one of the most responsive target tissues for IGFs (22), and several studies have shown that the IGF system contributes to homeostasis and functional integrity of the intestinal epithelium by regulating several basic cellular functions,such as proliferation and differentiation (23). Moreover,alterations of the IGF-I signaling have been reported to be associated with colorectal carcinoma both in vivo and in vitro, suggesting a role for the IGF axis in the pathogenesis of this disease (24, 25, 26). In this way, we have shown previously that two key processes, enterocyte-like differentiation(27, 28) and cell migration (29), were under the control of the IGF system in the human colon carcinoma cell line HT29-D4.

We report in this study that engagement of IGF-IR and IR induces a full resistance against IFN/TNF-induced apoptosis in HT29-D4 cells. The findings further indicate that the antiapoptotic function of IGF-I is mediated via enhancement of the inflammatory/survival signaling pathways generated by the TNF activation itself and involves activation of MAPK/ERK and MAPK/p38 in combination with NF-κB. This model of resistance to cell death is further supported by the observation that IGF-I, which is unable to prevent Fas- and IFN/Fas-induced apoptosis,became able to prevent this apoptosis when TNF is simultaneously added.

Tissue culture flasks and multiwell plates were purchased from Falcon (Lincoln Park, NJ). DMEM, RPMI-HEPES, FCS, and other cell culture reagents were purchased from Life Technologies (Grand Island,NY). PAGE reagents were purchased from Bio-Rad (Hercules, CA). IGF-I and IGF-II were purchased from Bachem (Bubendorf, Switzerland). Des-(1, 2, 3)-IGF-I and des-(1, 2, 3, 4, 5, 6)-IGF-II were from GroPep (Adelaide,Australia). IFN was purchased from Genzyme (Cambridge, MA). TNF and neutralizing mAbs against TNFR1 (clone 16803.1) and TNFR2 (clone 22221.311) were purchased from R & D Systems (Minneapolis, MN). Bovine insulin, BSA, WMN, CPHC, CHX, and PI were purchased from Sigma (L’Isle d’Abeau, France). BIM was from Boehringer-Manheim (Meylan, France). PD098059, LY294002, SB203580, and rapamycin were purchased from Alexis Biochemicals (San Diego, CA). BAY 11-7082 was from Biomol (Plymouth Meeting, PA). The antihuman IR mAb (B6), the agonistic antihuman Fas IgM mAb (CH-11), FITC-conjugated goat antimouse IgG + IgM Ab,and FITC-conjugated annexin V were purchased from Immunotech(Marseille, France). The antihuman IGF-IR mAb (α-IR3) was purchased from Oncogene Science (Uniondale, NY), and the anti-phosphotyrosine mAb(PY20) was from ICN Biomedical (Aurora, OH). The antihuman FAK Ab(A-17) was from Santa Cruz Biotechnology (Santa Cruz, CA), and the mAb raised against β-catenin was from Transduction Laboratories(Lexington, KY). Nitrocellulose sheets (Hybond-C extra), horseradish peroxidase-coupled antimouse secondary Ab, and enhanced chemiluminescence detection reagents were purchased from Amersham(Aylesbury, United Kingdom).

The HT29-D4 human colon adenocarcinoma cell line was cultured routinely in DMEM supplemented with 10% FCS as reported elsewhere(27, 28). For each experiment, HT29-D4 cells were seeded at a density of 2.5 × 10⁵cells/cm² in six-well tissue culture dishes. After 24 h, the cells were washed with HBSS and further incubated in FCS-free DMEM containing 0.1% BSA (serum-free DMEM) for 24 h at 37°C. After washing, cells were incubated with or without IFN (40 ng/ml) for 5 min at 37°C and then washed again, and apoptosis was induced by adding TNF (4 ng/ml) or CH-11 anti-Fas Ab (500 ng/ml) in the presence or absence of various concentrations of IGF family peptides. Whenever used, metabolic and kinase inhibitors were added to cells at the concentrations indicated under the Figure 6, Figure 7, Figure 8, Figure 9, 60 min prior to the addition of des-(1, 2, 3)-IGF-I. At the end of incubation,nonadherent and adherent cells (recovered by 0.53 mmEDTA/0.05% trypsin) were combined and assayed for apoptosis as described below. All of the experiments were made in duplicate and repeated at least three times.

Adherent and nonadherent cells were separately counted on a Coulter Counter ZM (Coultronics France, Margency, France). The percentage of dead cells was calculated as a ratio of detached cells to the total number of cells/well.

To quantify cells with advanced DNA fragmentation, we used the technique described by Nicoletti et al. (30). Briefly, cells (1 × 10⁶/ml) were treated by an hypotonic fluorochrome solution containing 0.1% Triton X-100, 0.1% sodium citrate, and 50 μg/ml PI. Flow cytometric analysis was done on a FacSort (Becton Dickinson, San Jose, CA) for quantifying the proportion of hypodiploid nuclei(pre-G₁ peak). Ten thousand events were examined for each determination, and the data were analyzed using the Cell Quest software package (Becton Dickinson).

This method was used to detect both apoptosis and necrosis from the same cell samples (31). Cells (1 × 10⁶/ml) were simultaneously stained with FITC-annexin V and PI as recommended by the supplier (Immunotech) and subjected to flow cytometric analyses to detect the percentage of apoptotic (FITC-stained) and necrotic (PI-stained) cells in a given population. A minimum of 10,000 cells was examined for each sample.

Cells were lysed by incubation in 50 mm HEPES (pH 7.5)containing 100 mm NaCl, 1% Triton X-100, 1 mmEDTA, 1 mm DTT, 1 mm phenylmethylsulfonyl fluoride, and a mixture of protease inhibitors (aprotinin, leupeptin,iodoacetamin, and pepstatin, 1.0 μg/ml each). Lysates were clarified by centrifugation (12,000 × g for 3 min),and equal amounts of proteins were subjected to SDS-PAGE on 7.5%polyacrylamide slab gels, then electrophoretically transferred onto nitrocellulose sheet for 1 h at 100 V, and analyzed by immunoblotting with anti-FAK and anti-β-catenin Abs. Immunoreactive proteins were identified by horseradish peroxidase-conjugated secondary antibody, followed by enhanced chemiluminescence reagents, with the technique recommended by the manufacturer.

Confluent HT29-D4 cells were starved in serum-free DMEM for 24 h. After washing, cells were incubated with or without TNF (4 ng/ml) or CH-11 (500 ng/ml) for 1 h at 37°C and then stimulated with IGF-I (50 ng/ml) for different times at 37°C. Cells were rapidly washed with cold PBS and then lysed with 50 mm HEPES (pH 7.5) containing 150 mm NaCl, 1% Triton X-100, 1 mg/ml bacitracin, 5 mm sodium orthovanadate, 100 mmNaF, 10 mm EDTA, 1 mm phenylmethylsulfonyl fluoride, and the same mixture of protease inhibitors as above. Cellular lysates were then submitted to Western Blotting procedure as described above with PY20 anti-phosphotyrosine Ab.

Confluent HT29-D4 cells were maintained in serum-free DMEM for 24 h and then gently recovered with 0.53 mm EDTA in PBS. The cells were then washed and resuspended in the same medium at 5.0 × 10⁶ cells/ml with Abs against TNFR1, TNFR2, and Fas and an isotype-matched control Ab at a concentration of 20 μg/ml for 90 min at 4°C. Cells were washed twice and incubated with FITC-conjugated goat antimouse IgG + IgM Ab at a dilution of 1:150 for 30 min at 4°C and then washed and fixed at 4°C in 2% paraformaldehyde. Cells were subjected to flow cytometry, and the relative fluorescence intensity of the cells was compared with the fluorescence intensity of the same cells stained with the control Ab. Results were presented as the number of cells(10,000/analysis) versus the log of fluorescence intensity.

IL-8 concentration in serum-free DMEM was determined using a commercially available ELISA kit, according to the manufacturer’s recommendations (Immunotech).

Adherent cells were fixed in situ, and floating cells were collected by centrifugation. Cells were fixed with 2.5%glutaraldehyde in 0.2 m sodium cacodylate buffer(pH 7.4) for 2 h, washed overnight in the same buffer containing 7.5% saccharose, postfixed in 1% osmium tetroxide, and then dehydrated in ethanol, embedded in Epon, and processed for examination with a Jeol 100C microscope.

The data were analyzed using the Statview software package(Abacus, Berkeley, CA). Results were expressed as the mean ± SD of triplicate determinations. Significant effects were determined using the nonparametric Mann-Whitney test. A statistically significant difference was considered to be present at P < 0.05.

The flow cytometric analysis shown in Fig. 1,A indicates that HT29-D4 cells constitutively expressed Fas,TNFR1 and TNFR2. However, as shown in Fig. 1,B, incubation of cells with TNF alone (up to 50 ng/ml) did not induce a significant increase in cell death above basal levels. In contrast, incubation of cells with anti-Fas Ab (CH-11) induced a 38% cell death at 24 h. Pretreatment of cells with IFN (40 ng/ml) resulted in induction of sensitivity to killing by TNF (4 ng/ml) and enhancement of cell death induced by anti-Fas Ab (500 ng/ml). Preliminary dose-response relationship with IFN, TNF, and anti-Fas Ab revealed that these concentrations were optimal for induction of cell death measured by counting floating dead cells (not shown). Using these concentrations,stimulation of TNFR and Fas resulted in 84 and 69% cell death,respectively, at 24 h (Fig. 1,B). It should also be noted that a 5-min pulse of IFN treatment was sufficient for an optimal sensitizing effect of cells to TNF- and anti-Fas Ab-mediated apoptosis. Longer exposures to IFN did not significantly improve death factor-mediated apoptosis but induced ∼15% cell death after 24 h in the absence of any additional proapoptotic stimuli (not shown). Fig. 1,C, left panel, indicates that TNFR1 Ab inhibited IFN/TNF-induced apoptosis by ∼45%. This incomplete inhibition was in fact attributable to the agonistic activity of this Ab (Fig. 1,C, right panel). In contrast, addition of TNFR2 Ab, alone or in combination with TNFR1 Ab, did not interfere with cell death (Fig. 1,C, both panels). Fig. 1 D shows a flow cytometric analysis of PI-stained nuclei, which was used as an alternative method to quantify apoptosis. About 60 and 38% of the cells contained <2N DNA content (pre-G₁ peak) after a 24-h treatment with IFN/TNF and IFN/anti-Fas Ab, respectively.

Fig. 1 (B and D) also shows that treatment of cells with des-(1, 2, 3)-IGF-I at 50 ng/ml induced a potent resistance(>85%) to IFN/TNF-induced apoptosis, whatever the assay used. In contrast, des-(1, 2, 3)-IGF-I did not significantly inhibit apoptosis driven by Fas cross-linking and provided only a poor protection(<25%) against apoptosis induced by IFN/anti-Fas Ab. Des-(1, 2, 3)-IGF-I was used instead of IGF-I because this NH₂terminally truncated IGF-I analogue binds to IGF-IR with the same affinity as native IGF-I but does not bind to any of the IGF binding proteins secreted by the HT29-D4 cells (27).

We next sought to confirm that the IGF-I-mediated cytoprotective effect was in apoptotic and not merely in necrotic cells. For this purpose, we used a simultaneous labeling of cells with annexin V-FITC and PI after a 8-h stimulation of the cells with IFN/TNF or IFN/anti-Fas Ab (Fig. 2,A). Annexin V bound to phosphatidylserine that translocated to the outer leaflet of the plasma membrane during the early phase of apoptosis, whereas these apoptotic cells with intact cell membranes did not stain with PI (Fig. 2,A, quadrant R2). This early stage of apoptosis was rapidly followed in vitro by the loss of membrane integrity and PI staining of the cells during a secondary necrotic stage (Fig. 2,A, quadrant R1). Because of the dynamic nature of this process, the percentage of trapped annexin-positive, PI-negative apoptotic cells was lower than the ones measured with assays based on cell death or DNA fragmentation, i.e., 34 and 22% for IFN/TNF- and IFN/anti-Fas Ab-induced apoptosis, respectively. Fig. 2,A also indicates that addition of des-(1, 2, 3)-IGF-I totally suppressed the appearance of cells that were actively undergoing apoptosis (quadrant R2) when IFN/TNF was the proapoptotic stimuli. In contrast, des-(1, 2, 3)-IGF-I could not do it for apoptotic cells driven by IFN/anti-Fas Ab (Fig. 2,A). To confirm these results by another approach, we evaluated the ability of des-(1, 2, 3)-IGF-I to prevent the caspase-dependent cleavage of FAK (32) and β-catenin(33). Fig. 2,B shows that des-(1, 2, 3)-IGF-I totally inhibited these proteolytic cleavages when apoptosis was induced by IFN/TNF but not IFN/anti-Fas Ab. Finally, it is important to note that all of these early- and late-stage markers of apoptotic events, including the unique morphological features of apoptotic cells(Fig. 2 C), were observed in matrix-detached cells only. Not one of these markers was detected in the remaining adherent cells, even after a 24-h apoptotic process.

Fig. 3 A shows the dose-response curves examining the survival of IFN-sensitized cells after an incubation with TNF or anti-Fas Ab for 24 h in the presence of graded concentrations of des-(1, 2, 3)-IGF-I and insulin. Des-(1, 2, 3)-IGF-I inhibited TNF-mediated cell apoptosis with an IC₅₀ at 2.5 ng/ml. The survival effect elicited by insulin was slightly less potent with an IC₅₀ at 10 ng/ml. Quite similar rates of protection were obtained when TNFR1 Ab was used as an agonistic Ab instead of TNF to induce apoptosis (not shown). In contrast, apoptosis induced by Fas cross-linking was minimally inhibited by des-(1, 2, 3)-IGF-I (∼25% at 100 ng/ml), whereas insulin did not promote any significant survival effect. Des-(1, 2, 3, 4, 5, 6)-IGF-II was as potent as des-(1, 2, 3)-IGF-I to prevent IFN/TNF-induced apoptosis,whereas IGF-I and IGF-II were slightly less protective with an IC₅₀ at 5.0 and 10 ng/ml, respectively (not shown). IGF-binding proteins secreted by HT29-D4 cells may explain this slight lower potency of IGFs compared with their truncated counterparts(27, 28).

These dose-response curves are consistent with mediation of cell survival by interaction of IGF-I and insulin with their cognate receptors. To further confirm which receptor was responsible for mediating the antiapoptotic effect of IGF-I and insulin, we used theα-IR3 and B6 mAbs raised against IGF-IR and IR, respectively. We have shown previously that these Abs inhibited the binding of ¹²⁵I-labeled IGF-I and ¹²⁵I-labeled insulin to their cognate receptors(28). Fig. 3 B shows that neither α-IR3 nor B6 Ab reversed the cell survival effect elicited by IGF-I and insulin. In fact, α-IR3 and B6 Abs acted as IGF-I- and insulin-mimetic Abs in term of induction of survival; at a concentration of 10.0 μg/ml,α-IR3- and B6-mediated cell survival was as large as that seen with des-(1, 2, 3)-IGF-I and insulin at 10 and 50 ng/ml, respectively.

Fig. 4,A indicates that the protective effect of des-(1, 2, 3)-IGF-I was maximum when it was added to the cells prior to or within 2 h of IFN/TNF addition. Addition of des-(1, 2, 3)-IGF-I at later time points offered less cell protection. However, the prevention of cell death was yet highly effective (∼80% of the maximal protective effect) when des-(1, 2, 3)-IGF-I was added with a delay of 6 h after IFN/TNF stimulation, a time that correlated with the beginning of cell-matrix detachment. As shown on Fig. 4 B, it is interesting to note that a 1-min pulse of des-(1, 2, 3)-IGF-I treatment was quite sufficient to induce a potent survival effect against IFN/TNF-mediated apoptosis. Pulse treatment prior to or after IFN cell sensitization did not modify the rate of IGF-I protection (not shown).

To determine whether TNFR1- and Fas-activated pathways altered IGF-IR signal transduction, we assessed tyrosine phosphorylation of IRS-1. IGF-I alone induced, within 1 min, an optimal tyrosine phosphorylation of a band with a M_r∼180,000 identified as IRS-1 (Ref. 28; Fig. 5,A). Preincubation of the cells for 1 h with TNF or anti-Fas Ab alone did not induce any tyrosine phosphorylation of IRS-1(Fig. 5,B, Lanes b and c). In addition, neither TNF nor anti-Fas Ab significantly altered the extent of IGF-I-induced IRS-1 tyrosine phosphorylation (Fig. 5,B, Lanes e and f). However, preincubation with TNF caused a delay in the electrophoretic mobility of IRS-1 (Fig. 5 B, Lane e),presumably because of serine phosphorylation of IRS-1(34). IGF-I also induced tyrosine phosphorylation of proteins at M_r 97,000 and M_r 102,000 identified as two isoforms of IGF-IR β-subunits in HT29-D4 cells (35). However,neither the extent of phosphorylation nor the mobility of these bands was altered by pretreatment of cells with TNF and anti-Fas Ab (not shown).

We first considered the possible role of PKC because of reports that these kinases could be modulators of apoptosis in colon cancer cells(36). Fig. 6,A shows that BIM, a PKC inhibitor affecting all PKC isoforms,inhibited by ∼50% IFN/TNF-induced apoptosis. In contrast, CPHC, a specific inhibitor of diacylglycerol-dependent (conventional and novel)PKC isoforms did not, suggesting the involvement of an atypical PKC. Because we detected the atypical PKC λ/τ in HT29-D4 cells(29), we suggest that it is a likely candidate for participating in the apoptotic pathway induced by IFN/TNF in these cells. Both BIM and CPHC were, however, unable to reverse the IGF-I protective effect on IFN/TNF-induced apoptosis (Fig. 6 A),thus indicating that activation of PKC is not necessary to convey the IGF-I antiapoptotic signaling.

The PI3′K pathway has been widely reported to play a central role in signal transduction by IR and IGF-IR (19, 20, 21). Moreover,this kinase has been shown to sustain survival in many cell models via activation of the downstream serine/threonine kinase PKB/AKT (37, 38). Fig. 6,A shows that two specific PI3′K inhibitors, LY294002, a synthetic inhibitor (39), and WMN,a fungal protein, enhanced the degree of apoptosis induced by IFN/TNF. However, when the rate of protection conferred by the addition of des-(1, 2, 3)-IGF-I in combination with LY294002 was calculated with respect to the level of apoptosis induced by IFN/TNF plus LY294002, it was not significantly different from the one measured in the absence of the drug (68% versus 80%; Fig. 6,A). The use of WMN as a PI3′K inhibitor gave essentially the same results. Finally, Fig. 6 A shows that rapamycin, a specific inhibitor of pp70^S6K (40),a kinase lying downstream from PI3′K, did not interfere with the IGF-I antiapoptotic effect. Thus, the PI3′K pathway does not appear to play a key role in the protection conferred by IGF-I against IFN/TNF-induced apoptosis in HT29-D4 cells.

The MAPK/ERK pathway mediates many of the known effects of IGF-I(19, 20, 21). To block this pathway, we used PD098059, a specific inhibitor of the activation of MAPK/ERK kinase by Raf-1, thus suppressing ERK activation (41). Two other MAPK subfamilies, the JNK and p38 kinases, have been also reported to be involved in the regulation of TNF-induced apoptosis(7, 8, 9, 10, 11). The specificity of the SB203580 inhibitor(42) allowed us to define the specific requirement for MAPK/p38 as a mediator of IGF-I survival signaling. Fig. 6,Bshows that PD098059 increased by ∼30%, whereas SB203580 decreased by∼40%, the extent of apoptosis induced by IFN/TNF. Thus, MAPK/ERK and MAPK/p38 pathways convey opposite signals on survival during IFN/TNF exposure. PD098059 and SB203580 partially blocked the protective effect of des-(1, 2, 3)-IGF-I on IFN/TNF-induced apoptosis (55 and 50%,respectively, versus. 80%; Fig. 6,B). These results suggest that the MAPK/ERK and MAPK/p38 pathways convey in part the IGF-I-induced antiapoptotic signaling. Reversal of the IGF-I antiapoptotic effect was not further improved by incubation of the cells with PD098059 and SB203580 together (Fig. 6 B). This latter observation suggests that common downstream survival targets are used by these two MAPK subfamilies.

An essential role for NF-κB in preventing cell death induced by TNF has been reported in several cell models (10, 12, 13, 14). To determine the role of NF-κB in the mediation of the IGF-I antiapoptotic effect, we used BAY 11-7082, a drug inhibiting IκBαphosphorylation (43), which prevents proteasome-mediated IκBα degradation and in turn release of free NF-κB. Fig. 6,B shows that treatment of the cells with BAY 11-7082 increased the extent of IFN/TNF-induced apoptosis by ∼40% but also partially blocked the des-(1, 2, 3)-IGF-I antiapoptotic effect (47%versus 80%; Fig. 6 B). Thus, activation of the NF-κB pathway appears to be required for both TNF- and IGF-I-mediated survival signaling.

Lastly, we determined whether MAPK- and NF-κB-dependent antiapoptotic signals may act synergistically to mediate the IGF-I antiapoptotic function. Fig. 6 B shows that the simultaneous blocking of these pathways by adding BAY 11-7082 plus PD098059 or SB203580 indeed induced a total reversal of the IGF-I antiapoptotic effect. In contrast, no additive effect was observed when the PI3′K inhibitor,LY294002, was added with inhibitors of either NF-κB or MAPK pathways(not shown). These results indicate that the coordinate activation of MAPK and NF-κB pathways is required for IGF-I to confer a full resistance against IFN/TNF-induced apoptosis in HT29-D4 cells.

Fig. 7 shows protein synthesis requirement for IGF-I to induce survival signaling. As reported previously (36), addition of CHX(10 μg/ml) allowed TNF to induce apoptosis to a death rate of 42%. Surprisingly, the sensitizing effect of IFN on TNF-induced apoptosis was not altered by CHX. In contrast, the antiapoptotic capacity of des-(1, 2, 3)-IGF-I was totally abrogated in the presence of CHX (Fig. 7). Thus, new protein synthesis is a prerequisite for IGF-I to function as an antiapoptotic factor in HT29-D4 cells.

TNF has been reported to stimulate in colon cancer cells the synthesis of the proinflammatory chemokine IL-8 (44, 45, 46). IL-8 expression requires the recruitment, among others, of the transcription factors NF-κB and AP-1 (47). Because the above reported results suggest that NF-κB and MAPK/AP-1 pathways are essential to convey the IGF-I-antiapoptotic effect, we next examined the ability of des-(1, 2, 3)-IGF-I to modulate IL-8 production. As previously reported for the parental HT29 cells (44, 45, 46), HT29-D4 cells produced low levels of IL-8 in the absence of added stimuli, and this production was slightly up-regulated by the addition of des-(1, 2, 3)-IGF-I (Fig. 8). The basal production of IL-8 was markedly increased by the addition of TNF, and TNF-induced IL-8 production was further strongly increased(∼2.5-fold) upon addition of des-(1, 2, 3)-IGF-I. The sensitization of the cells with IFN also enhanced the TNF-induced IL-8 production, and addition of des-(1, 2, 3)-IGF-I further increased this synthesis by a same 2-fold factor (not shown). As shown on Fig. 8, treatment with PD098059,SB203580, and BAY 11-7082 completely abrogated basal (not shown),TNF-induced, and TNF-induced/IGF-I-enhanced IL-8 production (Fig. 8). In contrast, LY294002 had no appreciable effect on IL-8 production. Thus, IGF-I is a potent stimulator of TNF-induced MAPK/ERK-, MAPK/p38-,and NF-κB-dependent IL-8 production in HT29-D4 cells.

In view of the above reported results, we hypothesized that the poor capacity of IGF-I to protect HT29-D4 cells against IFN/Fas- and Fas-mediated apoptosis (Fig. 1) could be attributable to an insufficient triggering of the MAPK- and NF-κB-dependent inflammatory/survival pathways by Fas cross-linking. We therefore investigated the ability of anti-Fas Ab to modulate, with or without IGF-I, the production of IL-8, used as a marker of activation of MAPK/AP-1 and NF-κB pathways. Fig. 9,A shows that anti-Fas Ab did induce only a weak stimulation of IL-8 production, and addition of des-(1, 2, 3)-IGF-I raised this level to ∼5% of that induced by the combination of TNF and IGF-I (compare Figs. 9,A and 8). This ratio remained identical when the assays were performed with IFN-sensitized cells (not shown). In addition, Fig. 9,A indicates that the signaling requirement for Fas- and TNF-mediated IL-8 production was identical. Thus, an insufficient level of activation of the MAPK and NF-κB signaling pathways may be the reason for the incapacity of IGF-I to protect HT29-D4 cells from Fas-induced apoptosis. To further address this issue, cells were exposed to anti-Fas Ab in combination with TNF and/or des-(1, 2, 3)-IGF-I. These experiments were done in the absence of IFN to activate the TNF-dependent survival but not the TNF-dependent proapoptotic signaling. Fig. 9 B shows that individually addition of either TNF or des-(1, 2, 3)-IGF-I to anti-Fas Ab-stimulated cells did not alter the extent of apoptosis. However, the simultaneous addition of TNF and des-(1, 2, 3)-IGF-I together induced a potent resistance against Fas-induced cell apoptosis. Taken together, these data suggest that resistance to Fas-induced apoptosis requires the synergistic activation of TNF- and IGF-I-dependent antiapoptotic signals in HT29-D4 cells.

Using the HT29-D4 human colonic cancer cell line as a model, we report in this study that IGF-I and insulin induce a total resistance of cells to apoptosis mediated by a combination of IFN and TNF. This antiapoptotic function is relatively selective in that Fas- and IFN/Fas-mediated apoptosis is poorly protected by these peptides. In addition, we show that the antiapoptotic signaling from the IGF-IR requires de novo protein synthesis and is via enhancement of the proinflammatory/survival signaling generated by the TNF activation itself. Specifically, the protective signals involve activation of ERK and p38 MAPK in combination with NF-κB. This antiapoptotic mechanism is further strengthened by the capacity to induce cell resistance against Fas-mediated apoptosis provided that TNFR1- and IGF-IR-dependent antiapoptotic signals are simultaneously triggered.

Our experiments show that HT29-D4 cells, which contain a nonfunctional p53 protein (48), express in a constitutive manner Fas,TNFR1 and TNFR2. However, preincubation of the cells with IFN was necessary to induce apoptosis in response to TNF, which correlates well with several reports done with the parental HT29 cells (36, 44, 49, 50). Although HT29-D4 cells express about 2-fold more TNFR2 than TNFR1, only the TNFR1 could mediate TNF-induced apoptosis, which agrees with most of the reports using cells expressing physiological TNFR2 levels (51). In contrast, HT29-D4 cells were susceptible to anti-Fas Ab-induced apoptosis, which indicates that their intracellular death pathway is functional. However, IFN further increased the cell sensitivity to Fas-mediated apoptosis. Strikingly, a very short pulse of IFN treatment (<5 min) was sufficient to induce an optimal sensitization to TNFR1- and Fas-mediated apoptosis, and this effect was not abrogated by treatment with CHX. Thus, the IFN-induced proapoptotic sensitization, in contrast to other systems (36, 49), does not appear to regulate the expression of select genes in HT29-D4 cells. Such a yet unknown signaling pathway might use select signal transducer and activator of transcription factors exerting a function unrelated to gene expression, e.g., adaptor function, as reported recently (52).

The antiapoptotic function of insulin and IGFs peptide family against IFN/TNF-stimulated apoptosis was powerful and obtained for quite physiological concentrations. It was observed whatever the early- and late-stage apoptotic marker we used. These observations also correlate well with the ability of αIR-3 and B6 Abs, which interfere with the binding to their cognate receptors of IGF-I and insulin, respectively,to exert a potent agonist antiapoptotic effect. Therefore, we conclude that endogenous IGF-IR (∼25,000/cell) and IR (∼5,000/cell; Ref.35) are effective for survival signaling in HT29-D4 colonic cancer cells. The IR-dependent antiapoptotic function is in contrast to several studies (40), which report a survival effect of insulin exclusively at high concentrations thought to trigger the IGF-IR. It is, however, in agreement with recent reports demonstrating that IR can protect cells from apoptosis in several models (19, 53). This finding may also be related to the emerging concept suggesting a role for hyper-insulinemia in colon cancer pathogenesis (54). The capacity of IGF-I to prevent IFN/TNF-stimulated apoptosis was rapidly and irreversibly delivered because a 1-min IGF-I pulse treatment was sufficient to induce an optimal antiapoptotic effect. This correlates well with the delay required for IGF-I to optimally induce tyrosine phosphorylation of IRS-1, the most proximal substrate in the IGF-I signaling pathway(20, 21). TNF, at the concentrations here used to induce apoptosis, had no significant effect on IGF-I-induced tyrosine phosphorylation of IRS-1, but induced its serine phosphorylation. Because serine phosphorylation of IRS-1 by TNF has been reported to impair insulin and IGF-I biological actions (34), the mechanism of protection by the IGF-IR, at this point, is obscure. However, D’Ambrosio et al. (55) have reported that IGF-I was quite able to protect cells from apoptosis induced by okadaic acid, a drug inducing serine phosphorylation of IRS-1, provided the drug was used at a concentration that does not decrease the IGF-I-induced IRS-1 tyrosine phosphorylation. Although the full significance of these observations awaits additional experimentation,we suggest that unaltered IGF-I-induced tyrosine phosphorylation of IRS-1 by TNF allows IGF-I to exert a potent protective action in our system, even in the face of a substantial TNF-induced serine phosphorylation of IRS-1. Alternatively, protection by IGF-I may be mediated via signaling pathways that are distinct from those requiring IRS-1 because it was suggested by using IGF-IR and IR mutated in their cytoplasmic domain (55, 56).

Although the aim of this study was not to examine the intracellular mechanisms by which IFN/TNF signals apoptosis in the absence of IGF-I,we report that inhibition of atypical PKC, presumably the λ/τisoform (29), and MAPK/p38 markedly decreased IFN/TNF-induced apoptosis. This suggests that these enzymes are required for a successful apoptotic response. In contrast, inhibitors of NF-κB, MAPK/ERK, and PI3′K enhanced apoptosis induced by IFN/TNF,thus suggesting that these signaling molecules are mediators of endogenous protective signals. However, it was not possible to determine whether these sets of molecules belong to IFN- or TNF-dependent signaling pathways or both, because neither TNF nor IFN stimulated apoptosis in the HT29-D4 system.

Taking into account the reported role for the PI3′K/PKB pathway for signaling several IGF-I/insulin-induced biological responses (20, 21) and survival against a wide variety of stimuli (37, 38, 40, 57, 58, 59), it was surprising to find that the PI3′K inhibition failed to block the ability of IGF-I to protect cells from IFN/TNF-induced apoptosis. These findings are, however, in agreement with reports showing that PI3′K/PKB-independent survival signaling pathways may also be used by IGF-I (56, 60). It is therefore likely that context-dependent parameters should influence the nature of the pathway that conveys the IGF-I antiapoptotic message. Furthermore, no PKC activity was involved in the IGF-I antiapoptotic signaling. Because we recently reported that PI3′K, PKC δ and PKC γwere involved in IGF-I-induced HT29-D4 cell migration(29), the current data illustrate the complexity of the signaling pathways used by IGF-I to exert a balanced control on a variety of biological functions in the same cell.

The data reported in the present study show that ERK and p38 MAPK mediate in part the IGF-I-antiapoptotic effect. The absence of additive effect we obtained by simultaneously inhibiting MAPK/ERK together with MAPK/p38 suggests that these MAPK cascades converge at the level of common downstream survival effectors as reported previously(8, 9, 10, 11). It is also likely that this observation accounts for the MAPK/JNK that the pharmacological approach here used does not permit us to discriminate. In agreement with our observations, IGF-I has been reported recently to interfere with p38 and JNK MAPK(61, 62). The role of MAPK cascades in the regulation of apoptosis is, however, full of contradictions. Depending on the cell type, its state of activation, and the context, the activities of these kinases are seen as a cause of apoptosis, a consequence of stress, or a survival force (7, 8, 9, 10, 11). Such a versatility was also observed in HT29-D4 cells, where the MAPK/p38 appears to be able to convey both IFN/TNF-induced proapoptotic stimuli and the IGF-I-mediated antiapoptotic signal.

Several reports (12, 13, 14) have shown that activation of NF-κB protects numerous tumor and embryonic cells from death induced by TNF and various stimuli including Fas ligation, ionizing radiation,and chemotherapeutic drugs. In HT29-D4 cells, our results indicate that blocking NF-κB activation by inhibiting IκBα phosphorylation prevented a substantial part of the protection afforded by IGF-I. Thus,induction by IGF-I of NF-κB-driven protective programs appears to also be a way for the colonic cancer cells to acquire resistance against IFN/TNF-induced cell death. Activation of the NF-κB pathway was reported recently to be involved in the IGF-I-mediated neuronal survival against oxidative stress (63) and the insulin antiapoptotic activity in IR-overexpressing CHO cells(53).

A particularly interesting finding is that the combined inhibition of MAPK and NF-κB-dependent antiapoptotic signals synergistically induced the complete reversal of the IGF-I capability to rescue cells from IFN/TNF-induced apoptosis. To our knowledge, this is the first report showing that the combined activation of MAPK and NF-κB pathways is a key regulatory mechanism to protect cells after IGF-IR engagement. At this point, it should be reminded that activation of MAPK/AP-1 and NF-κB pathways are also key events by which TNFR1 initiates the proinflammatory signaling and the protection against apoptosis (1, 2, 3, 10, 14). Thus, it should be noticed that IGF-IR and TNFR1 share common signaling pathways. This overlap was further confirmed in this study by showing that IGF-I strongly enhanced TNF-induced IL-8 production whether cells were pretreated or not with IFN. Moreover, IL-8 production was totally abrogated by blocking either MAPK or NF-κB pathway, thus confirming its dependence on activation of AP-1 and NF-κB transcription factors (47). IGF-IR activation alone did induce only a weak increase in IL-8 production. Thus, we conclude that IGF-I functions as a potent enhancer of TNF-induced NF-κB- and MAPK/AP-1-dependent inflammatory/survival signaling in HT29-D4 cells. Although a PI3′K dependence of IGF-I-induced NF-κB (63) and MAPK/JNK (61)activities has been reported by others, we observed no alteration in IGF-I-enhanced IL-8 production by inhibiting PI3′K. This indicates that in HT29-D4 cells, this pathway is not involved in the IGF-I activation of MAPK and NF-κB and agrees with the above-discussed incapacity of PI3′K inhibitors to block the IGF-I antiapoptotic effect. The synergistic roles of MAPK and NF-κB we observed in mediating IGF-I-induced cytoprotection suggest that the IGF-I protective signals are separately conveyed by MAPK and NF-κB, and that convergence of these two pathways exists at a more distal point. This is in agreement with the finding that TRAF2/MAPK and NF-κB initiate distinct antiapoptotic pathways that act in concert to prevent TNF-induced apoptosis (14). We suggest that such a convergence may operate on the regulation of the expression of inflammatory/survival genes controlled by NF-κB and AP-1 transcription factors. The fact that the antiapoptotic effect of IGF-I required new protein synthesis is in favor of such a gene-driven protective mechanism.

The constitutive activation of MAPK/ERK in colon tumors(64) and increased NF-κB activation in inflamed colonic epithelial cells (65) have been reported. We also observed in HT29-D4 cells a constitutive MAPK/ERK activity.⁵In addition, as shown in this study and reported by others(44, 45, 46), HT29 cells produce small amounts of IL-8 in the absence of added stimuli. Thus, one would have thought that the constitutive activation of MAPK and NF-κB constitutes a significant factor in resistance of colon cancer cells to TNF-induced apoptosis. However, this hypothesis can be rejected because inhibition of MAPK and/or NF-κB basal activities resulted only in a modest apoptosis in response to TNF. Moreover, such a constitutive activation was also insufficient to prevent anti-Fas Ab-induced apoptosis.

In agreement with the above reported model, the incapacity of IGF-I to mediate a successful resistance against Fas- and IFN/Fas-induced apoptosis is consistent with the minimal capacity of anti-Fas Ab, with or without IGF-I, to enhance IL-8 production used as a marker of MAPK/AP-1 and NF-κB activation. This observation is in agreement with our general understanding of Fas signaling, which initiates the Fas-associated death domain protein-dependent apoptotic pathway, but unlike TNFR1, does not recruit TRAF molecules, thus being unable to efficiently lead to NF-κB and MAPK activation (1, 2, 3). We have further investigated whether the activation state of MAPK and NF-κB would be able to influence Fas-induced apoptosis. Indeed,sensitization to apoptosis by NF-κB blockade has been reported to be restricted to the context of TNF signaling in some systems (14, 66) but not in others (13). In HT29-D4 cells, we report that although treatment with TNF or IGF-I did not alter Fas-induced apoptosis, the combination of these stimuli induced a marked protective effect. Thus, both MAPK and NF-κB pathways, provided they achieve a sufficient state of activation, can protect colonic cancer cells from death factor-induced apoptosis in general. It remains to test other proapoptotic stimuli to generalize this conclusion.

In this study, the IGF-I antiapoptotic function has been assayed in most of the experiments in IFN-sensitized cells. Therefore, it could be argued that the IGF-I survival signaling interferes with IFN-mediated rather than TNF-mediated pathways. Although this question cannot be directly assessed because of the incapacity of TNF and IFN to individually induce apoptosis, we propose that IGF-I specifically interferes with the TNF antiapoptotic signaling for the following reasons: (a) IGF-I was able to potentiate the TNF-mediated proinflammatory pathway, even in the absence of IFN; (b)IGF-I was unable to protect cells against Fas-induced apoptosis, even in IFN-sensitized cells; and (c) the combination of IGF-I and TNF did not require IFN sensitization to protect cells from Fas-mediated apoptosis.

Finally, an interesting finding of this study is that IGF-I, in addition to its antiapoptotic function, is also a potent inflammatory mediator contributing to the amplification of the TNF-induced secretion of IL-8 in colonic cancer cells. The role of IGF-I should be then in vivo more complex because the enhanced release of inflammatory mediators by colon cancer cells may promote further inflammation, neutrophil migration, and tumor destruction. The result of such a dual effect on tumor growth may be an important determinant for the outcome of colon cancer cells in patients. Whether IGF-I enhances secretion of other inflammatory cytokines in intestine epithelial cells also remains an open question.

Taken together, the findings reported in this study underline the synergism between the signaling pathways shared by IGF-IR and TNFR1 to induce resistance in colon cancer cells against death factors. Here exists a therapeutic challenge because the IGF-I and IGF-II autocrine/paracrine loops, which have been reported to be key components in the intestine epithelial cell physiology (24, 27, 28, 29), may exert a dual proinflammatory and survival-promoting effect interfering with: (a) the undesirable cell loss observed during the inflammatory bowel diseases; and (b) the expected efficiency of immune- and drug-based anticancer therapies.

The authors thank R. Rance for excellent technical assistance. The help of M. Roccabianca for the electron microscopy and C. Prevot for the flow cytometry experiments is gratefully acknowledged.