¹Dipartimento di Biologia e Patologia Cellulare e Molecolare 'L Califano', Centro di Endocrinologia ed Oncologia Sperimentale 'G. Salvatore' del Consiglio Nazionale delle Ricerche, Università 'Federico II', Napoli, 80131 Italy

²Dipartimento di Internistica Clinica e Sperimentale-Divisione di Gastroenterologia, II Ateneo di Napoli, Napoli, 80131 Italy

Correspondence to: A M Acquaviva, Dipartimento di Biologia e Patologia Cellulare e Molecolare 'L Califano', Università 'Federico II', Via S. Pansini 5, Napoli, 80131 Italy

Nonsteroidal anti-inflammatory drugs reduce the risk of colon cancer and this effect is mediated in part through inhibition of type 2 prostaglandin endoperoxide synthase/cyclo-oxygenase (COX-2). In the present study, we demonstrate that COX-2 expression and PGE₂ synthesis are up-regulated by an IGF-II/IGF-I receptor autocrine pathway in Caco-2 colon carcinoma cells. COX-2 mRNA and PGE₂ levels are higher in proliferating cells compared with post-confluent differentiated cells and in cells that constitutively overexpress IGF-II. Up-regulation of COX-2 expression by IGF-II is mediated through activation of IGF-I receptor because: (i) treatment of Caco-2 cells with a blocking antibody to the IGF-I receptor inhibits COX-2 mRNA expression; (ii) transfection of Caco-2 cells with a dominant negative IGF-I receptor reduces COX-2 expression and activity. Also, the blockade of the PI3-kinase, that mediates the proliferative effect of IGF-I receptor in Caco-2 cells, inhibits IGF-II-dependent COX-2 up-regulation and PGE₂ synthesis. Moreover, COX-2 expression and activity inversely correlate with the increase of apoptosis in parental, IGF-II and dominant-negative IGF-I receptor transfected cells. This study suggests that induction of proliferation and tumor progression of colon cancer cells by the IGF-II/IGF-I receptor pathway may depend on the activation of COX-2-related events. Oncogene (2000) 19, 5517-5524.

IGF-II; IGF-Ir; COX-2; PGE₂; colon cancer

IGF-II, insulin-like growth factor-II; IGF-Ir, IGF-I receptor; dnIGF-Ir, dominant negative IGF-I receptor; COX, cyclo-oxygenase; PG, prostaglandin; NSAIDs, nonsteroidal anti-inflammatory drugs; RT-PCR, reverse transcription polymerase chain reaction; MAPK-kinase, mitogen-activated protein kinase-kinase; PI3-kinase, phosphatidylinositol 3-kinase

Introduction

The turnover of the gastrointestinal epithelium is very rapid and is completed over a 3- to 5-day period. The mammalian intestinal mucosa undergoes a process of continual renewal characterized by active proliferation of stem cells in the crypts, progression of these cells up the crypt-villus axis with cessation of proliferation and subsequent differentiation. The differentiated enterocytes then undergo a process of programmed cell death (i.e., apoptosis) and extrusion into the gut lumen (Simon and Gordon, 1995). Aberrant regulation of cell proliferation, differentiation and apoptosis is thought to play an important role in the establishment of gut neoplasia (Kinzler and Vogelstein, 1996).

Prostaglandins (PG) are arachidonic acid derivates that mediate important functions involved in the regulation of intestinal epithelial cell homeostasis (Eberhart and DuBois, 1995). PGs synthesis depends on the activity of a constitutively expressed and an inducible PG endoperoxide synthase/cyclo-oxygenase (COX-1 and COX-2, respectively) (Williams and DuBois, 1996). Mounting evidence indicates that COX-2 may play an important role as an early event in colorectal carcinogenesis (Williams et al., 1997; Prescott and Fitzpatrick, 2000). In fact, COX-2 is overexpressed in 80-90% of colorectal adenocarcinomas and in 40-50% of premalignant adenomas (Williams et al., 1997), and inactivation of the COX-2 gene in mice is associated with decreased intestinal tumorigenesis (Oshima et al., 1996). Reduced prostaglandin biosynthesis through the inhibition of COX-2 activity is thought to be the molecular basis for the chemopreventive effects of nonsteroidal anti-inflammatory drugs (NSAIDs) on colorectal tumorigenesis in both human and rodents (Williams et al., 1997; Prescott and Fitzpatrick, 2000). COX-2 may facilitate colon cancer progression by stimulating cell proliferation and survival (Tsujii and DuBois, 1995; Sheng et al., 1998), tumor cell invasiveness (Tsujii et al., 1997) and the production of angiogenic agents in colon cancer cells (Tsujii et al., 1998).

Polypeptide growth factors are also known to regulate gastro-intestinal homeostasis by affecting cell proliferation, differentiation and apoptosis (Podolsky, 1993). In particular, insulin-like growth factors (IGF-I and IGF-II), are important modulators of growth in normal and transformed intestinal epithelial cells (Humbel, 1990; Singh and Rubin, 1993). IGF-I and IGF-II are small peptides structurally related to insulin and both exert their mitogenic activity through type I IGF receptor (IGF-Ir) (Humbel, 1990). A large percentage of tumors express IGF-II (Singh and Rubin, 1993; Zhang et al., 1997), and almost all primary human colon cancers and colon cancer cell lines are positive for IGF-I receptor (Guo et al, 1992; Singh and Rubin, 1993). Also, experimental evidences from our and other laboratories indicate that an autocrine IGF-II/IGF-Ir signal transduction pathway stimulates proliferation and inhibits differentiation of cultured colon cancer cells (Pommier et al., 1992; Garrouste et al., 1997; Zarrilli et al., 1994, 1996, 1999a).

We have recently studied the effects of NSAIDs on proliferation and differentiation of the human colon carcinoma cell line Caco-2, an intestinal epithelial cell line that differentiates spontaneously in culture, showing features of small bowel enterocytes (Pinto et al., 1983). Aspirin treatment inhibits proliferation of Caco-2 cells and induces the enterocyte-like differentiation of the cells before they reach confluency. These effects are associated with a dose-dependent reduction in IGF-II mRNA expression (Ricchi et al., 1997).

In the present study, we investigated whether the IGF-II/IGF-Ir pathway may activate COX-2 expression and PGE₂ synthesis in Caco-2 cells. We show that COX-2 expression and activity are regulated during differentiation of the cells, being elevated when they actively proliferate and low when they cease to grow and differentiate. We demonstrate a direct correlation between COX-2 and IGF-II expression in Caco-2 cell transfectants that constitutively overexpress IGF-II. We also show that the blockade of an IGF-Ir-dependent pathway inhibits COX-2 expression and PGE₂ synthesis, and induces apoptosis in Caco-2 cells. We postulate that IGF-II/IGF-Ir-mediated growth and tumor progression of colon cancer cells may depend at least in part on the up-regulation of COX-2 expression and activity.

Results

Cyclo-oxygenase-2 (COX-2) mRNA expression and prostaglandin E₂ (PGE₂) synthesis in parental, control NEO and IGF-II-transfected Caco-2 cells

Our previous data demonstrated that aspirin dose-dependently inhibited proliferation and induced apoptosis of parental Caco-2 cells. The inhibition of growth was associated with a reduction in IGF-II mRNA expression and with the induction of the enterocyte-like differentiation of the cells (Ricchi et al., 1997). This finding suggested that aspirin by reducing IGF-II expression in turn downregulated cyclo-oxygenase activity and increased apoptosis in Caco-2 cells. To investigate whether an IGF-II/IGF-Ir-dependent pathway regulated cyclo-oxygenase expression in Caco-2 cells, we analysed COX-1 and COX-2 mRNA expression during differentiation of parental and IGF-II-transfected Caco-2 cells that constitutively overexpress IGF-II (Zarrilli et al., 1996). Northern blot analysis did not show any detectable mRNA levels of either COX-1 or COX-2 during all days of culture of parental or control NEO-transfected Caco-2 cells. On the contrary, COX-2, but not COX-1, mRNA expression was detectable in IGF-II-transfected Caco-2 cells (Figure 1a and data not shown). COX-2 mRNA expression in IGF-II-transfected cells was highest at day 4 of culture and decreased by twofold at day 8 and 14 of culture.

To study COX-1 and COX-2 mRNA expression by a more sensitive technique, we used a semiquantitative RT-PCR approach. COX-1 or COX-2-specific oligonucleotides were co-amplified with those complementary to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) for a number of cycles (30) at which the reaction was still linear. GAPDH primers were added after 10 cycles. Specific COX-1 transcripts were not amplified in cDNA samples from either parental or control NEO- or IGF-II-transfected Caco-2 cells (data not shown). On the contrary, COX-2 mRNA transcripts were detected in parental and NEO- or IGF-II-transfected Caco-2 cells by RT-PCR analysis (Figure 1b). COX-2 mRNA expression was regulated during differentiation of Caco-2 cells, being 10-fold higher in proliferating cells compared with post-confluent differentiated cells. COX-2 mRNA levels were 10-fold higher in proliferating IGF-II-transfected cells compared with parental or control NEO-transfected cells. Also, COX-2 mRNA expression was still elevated in post-confluent IGF-II-transfected Caco-2 cells, being only twofold reduced compared with preconfluent cells (Figure 1b).

To evaluate whether IGF-II dose-dependently stimulated COX-2 expression, we analysed COX-2 mRNA expression in four independent clones of IGF-II-transfected cells that expressed different amounts of IGF-II (Zarrilli et al., 1996). A direct correlation was found between COX-2 and IGF-II expression, the clones expressing highest (IGF-cl. 1), intermediate (IGF-cl. 4 and 9), or lowest (IGF cl. 2) levels of transfected IGF-II (Zarrilli et al., 1996) showing the highest, intermediate, or lowest COX-2 mRNA levels, respectively (r=0.94, p=0.016) (Figure 1c).

We also studied whether differences in COX-2 mRNA expression were associated with those in PGE₂ synthesis in parental, control NEO- and IGF-II-transfected Caco-2 cells. As shown in Figure 2, PGE₂ levels progressively decreased from day 4 (400 pg/mg protein) to day 14 of culture (30 pg/mg protein) in parental and control NEO-transfected Caco-2 cells. On the contrary, PGE₂ levels were fourfold higher in IGF-II-transfected cells compared with control NEO Caco-2 cells at day 4 of culture (1530 pg/mg protein) and still elevated at day 8 and 14 of culture (754 and 200 pg/mg protein, respectively).

The above data therefore demonstrated that COX-2 expression and activity were regulated during differentiation of Caco-2 cells and suggested that they were stimulated by an IGF-II-dependent pathway.

Effect of IGF-I receptor blocking antibody -IR3 on COX-2 mRNA expression in Caco-2 cells

We have shown previously that the effects of IGF-II on cell growth and differentiation of Caco-2 cells are mediated by the IGF-I receptor (Zarrilli et al., 1994, 1996). To investigate whether the effect of IGF-II on COX-2 up-regulation was mediated through activation of IGF-I receptor, we analysed COX-2 mRNA expression in preconfluent Caco-2 cells following treatment with a monoclonal antibody (IR3) that blocks the binding sites of IGF-I receptor (Flier et al., 1986). IR3 monoclonal antibody inhibited COX-2 mRNA expression to 50% compared with control untreated cells, while the same concentration of the isotypic control antibody MOPC-21 had no effect (Figure 3).

Effect of dominant negative IGF-I receptor transfection on COX-2 mRNA expression in Caco-2 cells

To directly demonstrate that IGF-II/IGF-Ir pathway might stimulate COX-2 expression, we inhibited the endogenous IGF-I receptor by stably transfecting Caco-2 cells with a dominant negative type I IGF receptor (dnIGF-Ir) harboring a mutation in the ATP binding site (Kato et al., 1993).

A pool of clones and several independent clones of dnIGF-Ir-transfected cells were isolated and characterized for the expression of COX-2 mRNA levels and PGE₂ synthesis during their proliferative phase (day 4 of culture). COX-2 mRNA expression was reduced by 2-3-fold in a pool of dnIGF-Ir-transfectants compared with a pool of control NEO-transfectants. Also, COX-2 mRNA levels were reduced by 4-10-fold in seven out of 10 dnIGF-Ir-transfected independent clones compared with parental or control NEO-transfected cells. An example of COX-2 mRNA levels in a pool and three independent clones of dnIGF-Ir-transfected cells is shown in panel (a) of Figure 4. We also studied whether the decrease of COX-2 mRNA levels in dnIGF-Ir-transfected cells was associated with a reduction of PGE₂ synthesis. As shown in panel (b) of Figure 4, PGE₂ levels were threefold and 3-5-fold reduced in a pool and in three independent clones of dnIGF-Ir-transfectants compared with a pool and one independent clone of control NEO-transfected cells, respectively (108 and 126 to 78 pg/mg protein versus 397 and 375 pg/mg protein, respectively). COX-2 mRNA and PGE₂ levels were undetectable in dnIGF-Ir-transfectants at day 8 and 14 of culture (data not shown).

Effect of MAPK-kinase or PI3-kinase inhibition on COX-2 mRNA expression and PGE₂ synthesis in Caco-2 cells

We have shown previously that Caco-2 cell proliferation is stimulated by autocrine IGF-II synthesis through activation of type I IGF-I receptor (Zarrilli et al., 1994). To identify the signal transduction pathway that mediates IGF-II/IGF-Ir autocrine stimulus of cell growth, we have studied the effects of either PI3-kinase or MAPK-kinase inhibition on IGF-Ir-dependent proliferation. Both basal and exogenous IGF-I (10^-8 M)-dependent ³H-thymidine incorporation of Caco-2 cells were reduced by more than 90% following incubation with the PI3-kinase inhibitor LY294002 at the concentration of 50 M for 24 h. On the contrary, basal and exogenous IGF-I-dependent ³H-thymidine incorporation were reduced by 60% in the presence of the MAPK-kinase inhibitor PD098059 at 40 M for 24 h (data not shown). PI3-kinase and MAPK-kinase inhibitors at the doses used reduced cell growth without affecting cell viability because: (i) no increase in the percentage of apoptosis was observed at the end of both treatments; (ii) cells were able to restart to proliferate after drug withdrawal (data not shown).

To evaluate whether the signal transduction pathways that mediate the proliferative stimulus by IGF-Ir also regulate COX-2 expression and PGE₂ synthesis, we analysed in the same set of experiments the effects of MAPK-kinase and PI3-kinase inhibition on COX-2 mRNA expression and PGE₂ synthesis. COX-2 mRNA expression in control NEO- and IGF-II-transfected Caco-2 cells treated with the PI3-kinase inhibitor decreased by 40 and 70%, respectively. Also, COX-2 mRNA expression in control NEO- and IGF-II-transfected Caco-2 cells treated with the MAPK-kinase inhibitor decreased by 30 and 40%, respectively (Figure 5a). PGE₂ synthesis was reduced by approximately 80% in both NEO- and IGF-II-transfected Caco-2 cells treated with the PI3-kinase inhibitor, while decreased by 30 and 40% in NEO- and IGF-II-transfected Caco-2 cells following treatment with the MAPK-kinase inhibitor (Figure 5b). Altogether, these data suggested that the PI3-kinase pathway mediated both the mitogenic effect of IGF-Ir and the IGF-II/IGF-Ir-dependent up-regulation of COX-2 expression and activity.

Effect of NS398 COX-2 inhibitor on apoptosis of parental, dnIGF-Ir- and IGF-II-transfected Caco-2 cells

It has been shown previously that one of the mechanisms responsible for COX-2 induction of colon carcinogenesis relies on the impairment of apoptosis (Tsujii and DuBois, 1995; Sheng et al., 1998). To evaluate the effect of IGF-II/IGF-I receptor pathway-dependent COX-2 activity on apoptosis in Caco-2 cells, we analysed the percentage of sub-G1 population by DNA flow cytometry in parental and dnIGF-Ir-transfected cells. Spontaneous apoptosis occurred in 6 and 8% of parental and dnIGF-Ir-transfected Caco-2 cells at day 4 of culture, respectively. The percentage of cells in the sub-G1 fraction increased to 11% both in parental and dnIGF-Ir-transfected cells following treatment with NS398 specific COX-2 inhibitor at 50 M for 72 h (Figure 6a). At day 14 of culture, spontaneous apoptosis occurred in 15% of parental cells, and this value increased to 28% following treatment with NS398. On the contrary, dnIGF-Ir-transfected cells at day 14 of culture showed higher levels of spontaneous apoptosis than parental cells and low or no increase of apoptosis following treatment with NS398 (Figure 6b). Thus, either the decrease of COX-2 activity during the days of culture or that caused by treatment with NS398 stimulated apoptosis of parental and dnIGF-Ir-transfected Caco-2 cells. We also analysed the effect of NS398 COX-2 inhibitor on apoptosis in IGF-II-transfected Caco-2 cells because they expressed high levels of COX-2 and PGE₂. As shown in Table 1, NS398 COX-2 inhibitor at 50 M increased apoptosis in parental but not in IGF-II-transfected cells. A higher concentration of NS398 (100 M) was necessary to increase apoptosis in IGF-II transfectants, thus suggesting that COX-2 overexpression rendered IGF-II-transfected cells more resistant to apoptosis than parental Caco-2 cells.

Discussion

This study demonstrates that IGF-II up-regulates COX-2 expression and PGE₂ synthesis in Caco-2 colon carcinoma cells and that this effect is mediated by the activation of IGF-I receptor.

Cyclo-oxygenase activity has been indicated as the rate-limiting step in colon carcinogenesis (Prescott and Fitzpatrick, 2000). Experimental evidences suggest that COX-2 may facilitate colon cancer progression by increasing cell growth and survival (Tsujii and DuBois, 1995; Sheng et al., 1998), tumor cell invasiveness (Tsujii et al., 1997) and tumor angiogenesis (Tsujii et al., 1998). We used Caco-2 cells as a model system to study the mechanisms that regulates COX-2 expression in colon cancer cells. Caco-2 cells express COX-2, but not COX-1, mRNA levels, expression being variable in different studies probably due to different clones of Caco-2 cells used and/or different experimental conditions (Kutchera et al., 1996; Tsujii et al., 1997, 1998; Kamitani et al., 1998).

Our data show that COX-2 mRNA expression and PGE₂ synthesis are regulated during differentiation of Caco-2 cells, being 10-fold higher in proliferating cells compared with differentiated post-confluent cells. We have previously shown that IGF-II and IGF-Ir expression is co-ordinately regulated during differentiation of Caco-2 cells being elevated in proliferating pre-confluent cells and low in quiescent post-confluent cells (Zarrilli et al., 1994). Thus, COX-2 expression positively correlates with IGF-II and IGF-Ir expression in Caco-2 cells. We therefore investigated whether the IGF-II/IGF-Ir pathway may regulate COX-2 expression and PGE₂ synthesis in Caco-2 cells. Our data show that COX-2 expression in Caco-2 cells that constitutively overexpress IGF-II is 10-fold higher compared with parental cells. Moreover, a direct correlation exists between COX-2 and IGF-II expression by the analysis of four independent IGF-II-transfected clones expressing different amounts of IGF-II. Also, COX-2 mRNA expression and PGE₂ synthesis are still elevated in post-confluent IGF-II-transfected cells that continue to overexpress IGF-II after confluence. The above data indicate that autocrine IGF-II up-regulates COX-2 expression and activity in Caco-2 cells.

The biological effects of IGFs in colon cancer cells are mediated through the activation of the IGF-I receptor (Pommier et al., 1992; Singh and Rubin, 1993; Garrouste et al., 1997; Zarrilli et al., 1994, 1996, 1999a; Remacle-Bonnet et al., 2000). We find that treatment of Caco-2 cells with a blocking antibody to the IGF-I receptor inhibits COX-2 mRNA expression by 50% and that stable transfection of Caco-2 cells with a dominant negative IGF-I receptor reduces COX-2 mRNA expression and PGE₂ synthesis. These data indicate that IGF-II-dependent up-regulation of COX-2 expression in Caco-2 cells is mediated through activation of IGF-I receptor.

Constitutive IGF-II overexpressing Caco-2 cells show increased proliferative rate and delayed inhibition of growth at confluence (Zarrilli et al., 1996). Conversely, dnIGF-Ir-transfected Caco-2 cells have reduced growth rate in the absence of serum and are not able to respond to exogenously added IGF-I or IGF-II (data not shown). The data reported herein show that COX-2 expression is elevated in IGF-II overexpressing cells and low in dnIGF-Ir transfected cells. This suggests that COX-2 expression in Caco-2 cells is up-regulated by the proliferative stimulus driven by autocrine IGF-II through IGF-Ir. This finding is in agreement with previous data showing that COX-2 expression in intestinal epithelial cells can be stimulated by transforming growth factor- (DuBois et al., 1994). In addition, a recent report has shown that epidermal growth receptor activation by TGF- induces COX-2 protein translocation to the nucleus and the release of prostaglandins into the basolateral medium of polarized colon cancer cells (Coffey et al., 1997).

The blockade of the PI3-kinase pathway, that mediates most of the proliferative effects of IGF-Ir in Caco-2 cells, inhibits IGF-II-dependent COX-2 up-regulation and PGE₂ synthesis. Concordingly, it has been demonstrated that arachidonic acid of platelet microparticles induces COX-2 expression through a PI3-kinase-dependent pathway in a human monocytoid cell line (Barry et al., 1999). Our data are also in agreement with those obtained in different experimental systems showing that a PI3-kinase-dependent pathway mediates the proliferative effects of IGF-I receptor (Dufourny et al., 1997; Rubin and Baserga, 1995). It has been recently demonstrated that COX-2 expression in colon cancer cells is mediated by the activation of nuclear factor kappa B (NF-B) (Plummer et al., 1999; Kojima et al., 2000). Because the PI3-kinase signaling pathway through AKT kinase is able to activate NF-B (Romashkova and Makarov, 1999), we can speculate that COX-2 up-regulation by IGF-II/IGF-I receptor in Caco-2 cells might be mediated through a PI3-kinase/AKT/NF-B-dependent pathway. Further experiments will be necessary to validate this hypothesis.

We also found that either the decrease in COX-2 activity occurring during the days of culture or that induced by NS398 COX-2 inhibitor is accompanied by an increase in apoptosis in parental and dnIGF-Ir-transfected Caco-2 cells. Interestingly, the highest values of spontaneous apoptosis are obtained in dnIGF-Ir transfectants at day 14 of culture in which COX-2 mRNA and PGE₂ levels are undetectable. Also, IGF-II-transfected cells that overexpress COX-2 are more resistant to NS398-induced apoptosis compared with parental Caco-2 cells. Mounting evidence indicates that COX-2 over-expression through PGE₂ synthesis protects colon cancer cells from apoptosis (Tsujii and DuBois, 1995; Sheng et al., 1998). Concordingly, the activation of IGF-I receptor protects colon cancer cells from death factor-induced apoptosis (Remacle-Bonnet et al., 2000). Therefore, based on our data we can speculate that COX-2 activity mediates the antiapoptotic effect of IGF-I receptor in colon cancer cells.

In conclusion, our study suggests that increased COX-2 expression and PGE₂ synthesis mediate the effects of the IGF-II/IGF-Ir pathway activation on growth and tumor progression of colon cancer cells.

Materials and methods

Cell culture

Caco-2 cells were cultured as described previously (Pinto et al., 1983; Pignata et al., 1994; Zarrilli et al., 1994). Cells were seeded at 5´10⁴ cells/ml and were routinely subcultured when about 80% confluent. The culture medium was changed every other day. Confluence was reached 6-8 days after inoculum and the stationary phase on day 10. Cells were always >90% viable, as shown by trypan blue exclusion. Generation and mantainance of NEO- and IGF-II-transfected Caco-2 cells have been described previously (Zarrilli et al., 1996). Briefly, cells were transfected with a plasmid carrying the neomycin resistance gene fused to the Rous Sarcoma Virus promoter or cotransfected with a 1 : 10 ratio of the same plasmid and a vector expressing exons 4-6 cDNA sequences of rat IGF-II under the control of the Rous Sarcoma Virus promoter (Zarrilli et al., 1996). Dominant negative IGF-I receptor (dnIGF-Ir) expressing cells were generated by cotransfecting Caco-2 cells with a 1 : 20 ratio of a plasmid carrying the neomycin resistance gene fused to the Rous Sarcoma Virus promoter and a vector expressing the cDNA of human IGF-I receptor harboring a mutation in the ATP binding site under the control of Bovine Papilloma Virus promoter (Kato et al., 1993). Cells were seeded at 2.5´10⁵ cells/ml and after 24 h transfected by the Lipofectamine plus procedure as indicated by the manufacturer (Gibco-BRL, Life Technologies, Milano, Italy). Stable transfectants were isolated as single clones in the presence of the neomycin analog G418 (0.6 mg/ml) and further maintained in the presence of 0.4 mg/ml G418 during the course of all the experiments.

RNA isolation and Northern analysis

Total RNA was isolated from Caco-2 cells by the guanidinium thiocyanate acid-phenol procedure and subjected to Northern analysis as described previously (Romano et al., 1998). In brief, 10 g of total RNA per lane was separated by electrophoresis in 1% agarose-formaldehyde gels. RNA was transferred to Hybond-N⁺ (Amersham Italia Srl, Milano, Italy), cross-linked (UV Stratalinker-1800, Stratagene, La Jolla, CA, USA), and hybridized to ³²P-labeled cDNA probes. ³²P-labeled isotopes were from Amersham Corp., Milano, Italy. The COX-1 probe was a HindIII/PstI 514-bp fragment corresponding to the 3' end of the human COX-1 cDNA. The COX-2 probe was a 276-bp EcoRI/EcoRI fragment of the human COX-2 cDNA. An oligonucleotide (5'-AACGATCAGAGAGTAGTGGTATTTCACC-3') complementary to human 28 S rRNA was used as a loading control.

RT-PCR analysis

RT-PCR analysis was performed as described previously (Nardone et al., 1996; Zarrilli et al., 1999b). First strand cDNA was prepared using 200 units of reverse transcriptase (Supertranscript RT, Gibco-BRL), 1 g of total RNA as template and 10 pM of random hexamers in the presence of 0.1 mM dithiothreitol, 0.5 mM dNTP (Pharmacia) and 20 units of RNase inhibitor (Promega). The reaction profile was 37°C´10 min followed by 42°C´60 min. To control for contamination by genomic DNA, all RNA samples were run in duplicate with or without addition of reverse transcriptase. Semiquantitative RT-PCR coamplification of COX-2/GAPDH and COX-1/GAPDH transcripts was performed using COX-1 (sense, CCACCTACAACTCAGCACATG; antisense, CATTTCTCCATCCAGCACCTG), COX-2 (sense, TTCAAATGAGATTGTGGGAAAAT; antisense, AGATCATCTCTGCCTGAGTATCTT), and GAPDH (sense, CACCATCTTCCAGGAGCGAG; antisense, TCACGCCACAGT-TTCCCGGA) -specific primers. PCR amplifications were performed for 30 cycles using 50 ng of cDNA in the presence of 0.2 mM dNTP (Boehringer Mannheim), 0.3 l Ampli Taq DNA polymerase (Perkin Elmer) and 1.5 Ci -³²P dGTP, and GAPDH primers were added after 10 cycles. PCR products were phenol-chloroform extracted, ethanol precipitated and subjected to 5% polyacrylamide gel electrophoresis and autoradiography. Sizes of the amplified fragments were estimated from migration of the 1 Kb ladder molecular weight marker (Gibco-BRL) and identity was assessed by restriction enzyme digestion. To test for contamination by genomic DNA, samples were run in duplicate with (+RT) or without (-RT) the addition of reverse transcriptase. COX-1 and COX-2 transcripts levels were normalized to those of GAPDH. PCR products were quantitated by densitometric scanning of the autoradiograph using a Howtek Scanmaster-3 densitometer with RFL Print-TM software (Pharmacia, Biotech Inc, Cologno Monzese, Italy).

PGE₂ assay

Caco-2 cells grown in complete medium (DMEM supplemented with 10% FBS) at different days of culture were collected in ice cold 0.1 M phosphate buffer pH 7.4 and passed through a 23 gauge hypodermic needle. PGE₂ concentration in the cell extracts was measured directly in triplicate with a highly sensitive ¹²⁵I radioimmunoassay kit according to the procedure indicated by the manufacturer (Amersham Italia Srl, Milano, Italy). Briefly, cell extracts were subjected to solid phase extraction and extracted PGE₂ was converted into its methyl oximate derivative using the methyl oximation reagent provided by the manufacturer. Samples were then stored at -70°C in N₂ atmosphere before analysis. Methyl oximate standards in the range of 1.25 to 160 pg PGE₂ per tube were used and the curve calculated by regression analysis. The limit of detection was 1.0 pg PGE₂ in 0.1 ml volume per assay tube. For PGE₂, cross reactivity with PGE₁ was 5% and with all other PGs less than 0.01%. Results were expressed as pg/mg protein. Protein concentration of cell extracts was determined with the Bradford's dye binding test.

FACS analysis

Cells were trypsinized, pelleted, washed twice with phosphate-buffered saline (PBS), fixed, and permeabilized in 70% ethanol. After washing with PBS, cells were stained in 50 g/ml propidium iodide plus 0.5 mg/ml ribonuclease A in PBS at 4°C for 30 min. The DNA content of cells was analysed with a flow cytometer (Becton Dickinson, Mountain View, CA, USA) coupled to a Hewlett-Packard computer. For each sample, 20 000 events were stored in list mode.

Statistical analysis

Significance of differences was assessed by one-way analysis of variance and, when the F value was significant, by Duncan's multiple range test. Differences were considered significant if P<0.05.

Acknowledgements

We thank Dr D LeRoith (NIH, Bethesda, MD, USA) for kindly providing the dnIGF-Ir vector, Drs D Zanzi and G Ruggiero for help in FACS analysis, and Dr M Romano for critical reading of the manuscript. We also thank M Berardone for the artwork and R Cerillo for technical support. This work was supported by grants from Associazione Italiana per la Ricerca sul Cancro (AIRC) and Ministero dell'Università per la Ricerca Scientifica e Tecnologica (MURST). Dr A Apicella was recipient of a fellowship from Fondazione Italiana per la Ricerca sul Cancro (FIRC).

Barry OP, Kazanietz MG, Praticò D and FitzGerald GA. (1999). J. Biol. Chem. 274, 7545-7556. MEDLINE

Coffey RJ, Hawkey CJ, Damstrup L, Graves-Deal R, Daniel VC, Dempsey PJ, Chinery R, Kirkland SC, DuBois RN, Jetton TL and Morrow JD. (1997). Proc. Natl. Acad. Sci. USA 94, 657-662. MEDLINE

DuBois RN, Awad J, Morrow J, Roberts RJ II and Bishop PR. (1994). J. Clin. Invest. 93, 493-498. MEDLINE

Dufourny B, Alblas J, van Teeffelen HAAM, van Schaik FMA, van der Burg B, Steenbergh PH and Sussenbach JS. (1997). J. Biol. Chem. 272, 31163-31171. MEDLINE

Eberhart CE and DuBois RN. (1995). Gastroenterology 109, 285-301. MEDLINE

Flier JS, Usher P and Moses AC. (1986). Proc. Natl. Acad. Sci. USA 83, 664-668. MEDLINE

Garrouste FL, Remacle-Bonnet MM, Lehman MM-A, Marvaldi JM and Pommier GJ. (1997). Endocrinology 138, 2021-2032. MEDLINE

Guo Y-S, Narayan S, Chandrasekhar Y and Singh P. (1992). Gastroenterology 102, 1101-1108. MEDLINE

Humbel RE. (1990). Eur. J. Biochem. 190, 445-462. MEDLINE

Kamitani H, Geller M and Eling T. (1998). J. Biol. Chem. 273, 21569-21577. MEDLINE

Kato H, Faria TN, Stannard B, Roberts CT and LeRoith D. (1993). J. Biol. Chem. 268, 2655-2661. MEDLINE

Kinzler KW and Vogelstein B. (1996). Cell 87, 159-170. MEDLINE

Kojima M, Morisaki T, Izuhara K, Uchiyama A, Matsunari Y, Katano M and Tanaka M. (2000). Oncogene 19, 1225-1231. MEDLINE

Kutchera W, Jones DA, Matsunami N, Groden J, McIntyre TM, Zimmerman GA, White RL and Prescott SM. (1996). Proc. Natl. Acad. Sci. USA 93, 4816-4820. Article MEDLINE

Nardone G, Romano M, Calabrò A, Pedone PV, De Sio I, Persico M, Budillon G, Bruni CB, Riccio A and Zarrilli R. (1996). Hepatology 23, 1304-1312. MEDLINE

Oshima M, Dinchuk JE, Kargman SL, Oshima H, Hancock B, Kwong E, Trzaskos JM, Evans JF and Taketo MM. (1996). Cell 87, 803-809. MEDLINE

Pignata S, Maggini L, Zarrilli R, Rea A and Acquaviva AM. (1994). Cell Growth Differ. 5, 967-973. MEDLINE

Pinto M, Robine-Leon S, Appay MD, Kedinger M, Triadou N, Dussaulx E, Lacroix B, Simon-Assman P, Haffen K, Fogh J and Zwelbaum A. (1983). Biol. Cell. 47, 323-330.

Plummer SM, Holloway KA, Manson MM, Munks RJL, Kaptein A, Farrow S and Howells L. (1999). Oncogene 18, 6013-6020. MEDLINE

Pommier GJ, Garrouste FL, El Atiq F, Roccabianca M, Marvaldi JL and Remacle-Bonnet MM. (1992). Cancer Res. 52, 3182-3188. MEDLINE

Podolsky DK. (1993). Am. J. Physiol. 264, G179-G186. MEDLINE

Prescott SM and Fitzpatrick FA. (2000). Biochim. Biophys. Acta 1470, M69-M78. MEDLINE

Remacle-Bonnet MM, Garrouste FL, Heller S, Andrè F, Marvaldi JL and Pommier GJ. (2000). Cancer Res. 60, 2007-2017. MEDLINE

Ricchi P, Pignata S, Di Popolo A, Memoli AM, Apicella A, Zarrilli R and Acquaviva AM. (1997). Int. J. Cancer 73, 880-884. Article MEDLINE

Romano M, Ricci V, Memoli A, Tuccillo C, Di Popolo A, Sommi P, Acquaviva AM, Del Vecchio Blanco C, Bruni CB and Zarrilli R. (1998). J. Biol. Chem. 273, 28560-28563. MEDLINE

Romashkova JA and Makarov SS. (1999). Nature 401, 86-90. Article MEDLINE

Rubin R and Baserga R. (1995). Lab. Invest. 73, 311-331. MEDLINE

Sheng H, Shao J, Morrow JD, Beauchamp RD and DuBois RN. (1998). Cancer Res. 58, 362-366. MEDLINE

Simon TC and Gordon JI. (1995). Curr. Opin. Gen. Dev. 5, 577-586.

Singh P and Rubin N. (1993). Gastroenterology 105, 1218-1237. MEDLINE

Tsujii M and DuBois RN. (1995). Cell 83, 493-501. MEDLINE

Tsujii M, Sunao K and DuBois RN. (1997). Proc. Natl. Acad. Sci. USA 94, 3336-3340. Article MEDLINE

Tsujii M, Kawano S, Tsujii S, Sawaoka H, Hori M and DuBois RN. (1998). Cell 93, 705-716. MEDLINE

Williams CS and DuBois RN. (1996). Am. J. Physiol. 270, G393-G400. MEDLINE

Williams CS, Smalley W and DuBois RN. (1997). J. Clin. Invest. 100, 1325-1329. MEDLINE

Zarrilli R, Pignata S, Romano M, Gravina A, Casola S, Bruni CB and Acquaviva AM. (1994). Cell Growth Differ. 5, 1085-1091. MEDLINE

Zarrilli R, Romano M, Pignata S, Casola S, Bruni CB and Acquaviva AM. (1996). J. Biol. Chem. 271, 8108-8114. MEDLINE

Zarrilli R, Pignata S, Apicella A, Di Popolo A, Memoli A, Ricchi P, Salzano S and Acquaviva AM. (1999a). Gastroenterology 116, 1358-1366.

Zarrilli R, Tuccillo C, Santangelo M, Nardone G and Romano M. (1999b). Am. J. Gastroenterol. 94, 3376-3378.

Zhang L, Zhou W, Velculescu VE, Kern SE, Hruban RH, Hamilton SR, Vogelstein B and Kinzler KW. (1997). Science 276, 1268-1272. Article MEDLINE

Figure 1 Cyclo-oxygenase-2 (COX-2) mRNA expression in parental, control NEO- and IGF-II-transfected Caco-2 cells. Total RNA was isolated from cells grown in complete medium (DMEM supplemented with 10% FBS) at different days of culture. (a) Northern blot analysis of COX-2 transcripts in control NEO- and IGF-II-transfected Caco-2 cells at days of culture indicated. Twelve g of total RNA were loaded. The same filters were sequentially hybridized to human COX-2 and human 28 S rRNA probes. (b) RNA samples extracted at days of culture indicated from parental, control NEO- and IGF-II-transfected Caco-2 cells were coamplified by RT-PCR analysis with COX-2- and GAPDH-specific primers. PCR amplifications were performed for 30 cycles, and GAPDH primers were added after 10 cycles. Amplified PCR products are indicated on the left side of the autoradiograms. A representative autoradiograph of four separate experiments is shown. (c) RNA samples extracted at day 8 of culture from one control NEO- and four IGF-II-transfected independent clones of Caco-2 cells were coamplified by RT-PCR analysis with COX-2- and GAPDH-specific primers. RT-PCR profiles were identical to those shown in (b). Amplified PCR products are indicated on the left side of the autoradiograms. A representative autoradiograph of four separate experiments is shown

Figure 2 PGE₂ synthesis in parental, control NEO- and IGF-II-transfected Caco-2 cells. Cells were grown in complete medium (DMEM supplemented with 10% FBS) at days of culture shown and PGE₂ concentration in the cell extracts was assessed as described in the Materials and methods section. Values shown are means±s.d. of four experiments run in triplicate

Figure 3 Effect of IGF-I receptor blocking antibody -IR3 on COX-2 mRNA expression in Caco-2 cells. Caco-2 cells at day 3 of culture were washed and grown in serum-free medium. Two days later, cells were incubated in serum-free medium alone (DMEM), or with the addition of 10 g/ml -IR3 IGF-Ir-blocking monoclonal antibody (Flier et al., 1986), or with the same concentration of the isotypic control antibody MOPC-21 for 48 h. (a) RT-PCR analysis of COX-2 mRNA expression. RT-PCR profiles were identical to those shown in panel (b) of Figure 1. A representative autoradiograph of four separate experiments is shown. (b) Densitometric analysis of the autoradiographs. mRNA levels from cells incubated in serum-free medium alone (DMEM) were arbitrarily taken as 100. Results are the mean±s.d. of four experiments

Figure 4 Effect of dominant negative IGF-I-receptor transfection on COX-2 mRNA expression and PGE₂ synthesis in Caco-2 cells. (a) RT-PCR analysis of COX-2 mRNA expression. RNA samples extracted from cells grown in complete medium (DMEM supplemented with 10% FBS) at day 4 of culture from control NEO- and dnIGF-Ir-transfected pools and independent clones of Caco-2 cells were coamplified by RT-PCR analysis with COX-2- and GAPDH-specific primers. PCR amplifications were performed for 30 cycles, and GAPDH primers were added after 12 cycles. Amplified PCR products are indicated on the left side of the autoradiograms. A representative autoradiograph of four separate experiments is shown. (b) PGE₂ synthesis in control NEO- and dnIGF-Ir-transfected pools and independent clones of Caco-2 cells. PGE₂ concentration in the total extracts of cells at day 4 of culture was assessed as described in the Materials and methods section. Values shown are means±s.d. of four experiments run in triplicate

Figure 5 Effect of MAPK-kinase or PI3-kinase inhibitors on COX-2 mRNA expression and PGE₂ synthesis in control NEO- and IGF-II-transfected Caco-2 cells. Cells at day 4 of culture were incubated with either 40 M PD098059 (PD) or 50 M LY294002 (LY) for 24 h in DMEM supplemented with 10% FBS. (a) RT-PCR analysis of COX-2 mRNA expression. RT-PCR profiles were identical to those shown in panel (b) of Figure 1. Densitometric analysis of the autoradiographs is shown. mRNA levels from untreated NEO- and IGF-II-transfected cells (controls) were expressed as arbitrary units. Results are the mean±s.d. of four experiments. (b) PGE₂ concentration in the cell extract was assessed as described in the Materials and methods section. Values shown are means±s.d. of four experiments run in triplicate

Figure 6 Effect of NS398 COX-2 inhibitor on apoptosis of parental and dnIGF-Ir-transfected Caco-2 cells. Parental and dnIGF-Ir-transfected Caco-2 cells at day 4 (a) or 14 (b) of culture were incubated with standard culture medium (control) or with 50 M NS398 for 72 h. At the end of the incubation time, cells were fixed, stained with propidium iodide, and analysed for DNA content. The percentage of apoptotic cells was assessed by the sub-G₁ population in DNA flow cytometry. Results are the mean±s.d. of four experiments

Table 1 Effect of NS398 COX-2 inhibitor on apoptosis in parental and IGF-II-transfected Caco-2 cells