Expression of 15-Lipoxygenase by Human Colorectal Carcinoma Caco-2 Cells during Apoptosis and Cell Differentiation*

We studied arachidonic acid metabolism and the expression of cyclooxygenase (Cox) and 15-lipoxygenase (15-LO) in the human colorectal carcinoma cell line, Caco-2, which undergo apoptosis and cell differentiation in the presence of sodium butyrate (NaBT). Caco-2 cells expressed very low levels of Cox-1 but highly expressed Cox-2. NaBT treatment shifted the arachidonic acid metabolites by cell lysates from prostaglandins to 15-hydroxyeicosatetraenoic acid, indicating the presence of a 15-LO. Linoleic acid, an excellent substrate for 15-LO, was metabolized poorly by the Caco-2 cells, but NaBT treatment shifted metabolism to 15-LO metabolite, 13(S)-hydroxyoctadecadienoic acid. Caco-2 cells expressed a 15-LO but only after treatment with NaBT, as determined by Northern blotting. Immunoblotting with anti-human 15-LO antibody detected a 72-kDa band in NaBT-treated Caco-2 cells. Expression of 15-LO mRNA was dependent on the duration of NaBT treatment, with the highest expression observed between 10 and 24 h. Results from expression and metabolism studies with arachidonic and linoleic acid cells indicated Cox-2 was responsible for the lipid metabolism in control cells, whereas 15-LO was the major enzyme responsible after NaBT induction of apoptosis and cell differentiation. The 15-LO in Caco-2 cells was characterized as human reticulocyte 15-LO by reverse transcription-polymerase chain reaction and restriction enzyme analysis. The expression of 15-LO and 15-hydroxyeicosatetraenoic acid or 13(S)-hydroxyoctadecadienoic acid formation correlates with cell differentiation or apoptosis in Caco-2 cells induced by NaBT. The addition of nordihydroguaiaretic acid, a lipoxygenase inhibitor, significantly increased NaBT-induced apoptosis, whereas the addition of indomethacin did not alter NaBT-induced apoptosis in the Caco-2 cells. However, indomethacin treatment decreased the expression of Cox-2 in NaBT-treated cells and significantly increased the expression of 15-LO during NaBT treatment. These studies suggest a role for 15-LO, in addition to Cox-2, in modulating NaBT-induced apoptosis and cell differentiation in human colorectal carcinoma cells.

The first genetic alteration in the multistep process that leads to the development of colon carcinogenesis seems to be the loss of APC gene function. Investigations of APC function and its mutations have provided important clues in understanding colon cancer (1). One of the important changes that results from the loss of the APC gene function is the overexpression of prostaglandin H synthase-2 (cyclooxygenase-2, Cox-2), 1 an inducible enzyme that converts arachidonic acid to prostaglandins. The link between APC, Cox-2, and polyp formation was firmly established in studies using APC (Apc ⌬716 ) and Cox-2 knockout mice (2). Mice carrying the APC mutation overexpress Cox-2 and develop intestinal polyps. When bred to mice with the disrupted Cox-2 gene, the offspring, homozygously deficient for Cox-2, had significantly less polyps than the mice with wild-type Cox-2. These results confirmed Cox-2 as a modulator after APC mutations in colon carcinogenesis and represent a key enzyme in tumor development. The elevated expression of Cox-2 was shown not only in polyps of Apc ⌬716 and Min mice (3) but also in human colorectal tumors (4,5) and immortalized cell lines of colorectal carcinoma (6). These findings provide an explanation for epidemiological, clinical, animal, and experimental studies which indicate that nonsteroidal anti-inflammatory drugs, which inhibit Cox-2, prevent colorectal cancer.
One possible explanation for the role of Cox-2 in colon carcinogenesis was obtained from studies with rat intestinal epithelial (RIE) cells engineered to overexpress Cox-2 (7). The overexpression of Cox-2 in RIE cells results in a resistance to sodium butyrate (NaBT) or adhesion-induced apoptosis. However, mechanisms that clearly explain how Cox-2 alters or inhibits apoptosis have not been determined. The addition of prostaglandins to cells in culture does not seem to alter cell proliferation and/or apoptosis (8), and other investigations indicate that nonsteroidal anti-inflammatory drug agents may act by a prostaglandin-independent pathway (9). The metabolites of arachidonic acid formed by Cox-2 are considered an important contributor, but we cannot ignore the possible involvement of linoleic acid metabolites, as it is also a substrate for Cox-2 (10). In addition, arachidonic acid and linoleic acid metabolites formed by the enzymatic activity of lipoxygenases may also need to be considered. Our laboratory has shown that both Cox-2 and 15-lipoxygenase (LO) metabolites of arachidonic acid and linoleic acid alter growth factor signaling pathways (11)(12)(13). For example, 13(S)-hydroxyoctadecadienoic acid (HODE)/13(S)-hydroperoxyoctadecadienoic acid (HpODE) is formed from linoleic acid in response to epidermal growth factor and contributes to epidermal growth factor-dependent mitogenesis in Syrian hamster fibroblast cells (13). The 13(S)-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  HpODE augmented the epidermal growth factor receptor signaling pathway by attenuation of receptor dephosphorylation (12). Furthermore, 15-LO metabolites of linoleic acid, 13(S)-HpODE inhibits starvation-induced apoptosis. 2 Furthermore, Tang et al. (14) showed that lipoxygenase metabolites, 12-hydroxyeicosatetraenoic acid (HETE) and 15-HETE, inhibit apoptosis in rat W256 cells. They suggest a role for 12-LO as an inhibitor of apoptosis (15).
These findings suggest a potential involvement of lipoxygenases in the apoptotic process. We have examined arachidonic acid and linoleic acid metabolism in the human colorectal cell line, Caco-2, which undergoes cell differentiation and apoptosis in the presence of NaBT (16). The metabolite profile was dramatically shifted from Cox-2 metabolites to 15-LO-derived metabolites by the NaBT treatment. The NaBT induced a 15-LO that was characterized as the reticulocyte 15-LO previously characterized from human bronchial epithelium (17). In this report, we show the first evidence that a 15-LO is clearly induced in colorectal carcinoma cells. Moreover, the 15-LO may act to modulate NaBT-induced apoptosis and cell differentiation in these cells.

EXPERIMENTAL PROCEDURES
Cell Culture-The human colorectal carcinoma cell line Caco-2 was obtained from the American Type Culture Collection (ATCC). Caco-2 was grown in Eagle's minimal essential medium, 15% fetal bovine serum (FBS) with 1 mM sodium pyruvate (Life Technologies, Inc.). The media for cells contained gentamicin (1 mg/100 ml, Life Technologies, Inc.). FBS was from Summit, and NaBT was obtained from Sigma. For treating cells with NaBT, cells were cultured in the appropriate media containing serum on 150 cm 2 round dishes until nearly confluent. The medium was removed and replaced with FBS medium containing either the solvent or 5 mM NaBT. The NaBT was solibilized by PBS and used. Cells were then harvested at the times stated in the figure legends.
DNA Fragmentation Assay-The floating cells and the cells attached to the dish were collected separately and sedimented. Washed cell pellets were resuspended in cell lysis buffer (10 mM Tris-HCl (pH 7.4), 10 mM EDTA (pH 8.0), 0.5% Triton X-100) and incubated. RNase A (0.5 mg/ml) and proteinase K (0.5 mg/ml) were added, respectively, and incubated for 2 h. DNA was precipitated by ethanol, and 3 g or 1 ϫ 10 6 cell numbers of water-diluted sample was run on a 2% agarose gel. Gels were stained with ethidium bromide, and DNA was visualized by UV transilluminater.
Alkaline Phosphatase Activity -Alkaline phosphatase activity was measured using the p-nitrophenyl phosphate (Sigma) as substrate. The standard assay mixture consisted of 0.5 ml of 2-amino-2-methyl-1propanol buffer (1.5 mol/liter) (Sigma), 2 mg of substrate, and 100 l of sonicated protein in a final volume of 1.1 ml. After 15 min at 37°C, the reaction was terminated by adding 10 ml of 0.05 N NaOH, and the absorbance of the color due to the formation of p-nitrophenol was measured spectrophotometrically at 410 nm. Enzyme activity was expressed in Sigma Unit/0.1 mg protein. The calibration was performed by using p-nitrophenol standard solution (Sigma).
Northern Blot Analysis-Total RNA from the attached cell was extracted by using TRI Reagent (Sigma) according to the procedure by Chomczynski (18). RNA samples (30 g per lane) were separated by electrophoresis in a formaldehyde-1% agarose gel. The RNA was transferred in 10 ϫ SCC by capillary action onto nylon membrane (Schleicher & Schuell) and UV-cross-linked with a Stratalinker UV light source (Stratagene). Human 15-LO, Cox-1 and Cox-2 (Oxford Biomedical Research), and human glyceraldehyde-3 phosphate dehydrogenase (CLONTECH) cDNA probes were labeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech) using the Prime-It-II random prime kit (Stratagene). After hybridization and washes, the blots were then exposed to x-ray film (Amersham) for autoradiography.
Analysis of Arachidonic and Linoleic Acid Metabolites-Caco-2 cells cultured in 150-cm 2 dishes at each condition were washed with PBS twice. Cells were scraped and collected in 1 ml of the lysis buffer (100 mM Tris-HCl, (pH 8.0), 1 g/ml leupeptin and pepstatin, 0.5 mM phenylmethylsulfonyl fluoride). Cells were sonicated four times for 20 s at 50% power by the sonic dismembrator for a total protein preparation. Then, protein content was quantified, and 0.8 mg of total cell lysate was used. The sonicates were then diluted with 1 ml of the reaction buffer (100 mM Tris-HCl, 10 mM CaCl 2 ) (2 ml total) and incubated [ 3 H]arachidonic acid (3 Ci, 25 M) (DuPont NEN) for 30 min or [ 14 C]linoleic acid (3 Ci, 25 M) (DuPont NEN) for 15 min at 37°C. Fatty acid compounds were extracted from the incubation buffer by acidification to pH 3.5 with acetic acid and applied to a C 18 -PrepSep solid phase extraction column (Waters) pretreated with methanol. The samples were then washed with acidified water, eluted with methanol, evaporated to dryness, and reconstituted with high pressure liquid chromatography (HPLC) solvent.
High Pressure Liquid Chromatography-Reverse-phase HPLC analysis was performed using an Ultrasphere ODS column (5 m; 4.6 ϫ 250 mm; Beckman). The solvent system consisted of a methanol/water gradient at flow rate of 1.1 ml/min as described previously (19). Radioactivity was monitored using a Flow Scintillation Analyzer (Packard) with EcoLume (ICN Biochemicals) as the liquid scintillation mixture. To separate 9-HODE and 13-HODE, straight-phase HPLC was used. The straight-phase HPLC used a Waters Porasil column (10 m; 3.9 ϫ 300 mm) with a flow rate of 2 ml/min and eluted with hexane:diethylether: acetic acid (1000:200:1) for 60 min. UV analysis was performed by monitoring absorbance at 235 nm with a Waters 990 photodioarray detector. Authentic standards of 9(R)-HODE, 13(S)-HODE, 15(S)-HETE, prostaglandin E 2 (PGE 2 ), and prostaglandin F 2␣ were obtained from Cayman Chemical and used.
RT-PCR and Restriction Enzyme Analysis-First-strand complementary DNA (cDNA) was generated using 1 g of total RNA as template. Oligo(dT) 18 primer (20 pmol) was used to prime a standard reversetranscription (RT) reaction. Advantage TM RT-for-polymerase chain reaction (PCR) kit was purchased from CLONTECH and used for this cDNA synthesis. As a control for possible contamination of genomic DNA, the same RNA was also prepared to react without reverse transcriptase. The reaction solution was diluted to a final volume of 100 l by DEPC-treated water, and the 5 l was used as PCR. PCR mixture consisted of 10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl 2 , 0.2 mM each dNTP, 2.0 units of Taq DNA polymerase (Pharmacia), and 0.4 M each of the following primers: P1, 5Ј-GAGTTGACTTTGAGGTTTCGC-3Ј; P2, 5Ј-GCCACGTCTGTCTTATAGTGG-3Ј; P3, 5Ј-CCCTGTGGATGAGC-GATTTC-3Ј; and P4, 5Ј-AAGTGTCCCCTCAGAAGATG-3Ј in a total volume of 50 l. Thermocycling was performed according to the following profile: 94°C for 45 s, 60°C for 45 s, and 72°C for 2 min, repeated 35 times followed by a final extension at 72°C for 7 min. cDNA fragments generated by RT-PCR using P1 and P2 primers were digested with PstI or HindIII (Boehinger Mannheim). Analysis of the fragments was performed on a 1.8% agarose gel electrophoresis and then photographed.
Inhibition of 15-LO by Nordihydroguaiaretic Acid (NDGA) and Apoptosis-To inhibit the activity of 15-LO in Caco-2 cells treated with NaBT, NDGA (Sigma), a inhibitor of lipoxygenase, was added in addition to NaBT. Before assessing the effect of attenuating the 15-LO on NaBT-induced apoptosis, the optimal concentration of NDGA for inhibiting 15-LO without affecting Cox activity was determined by studying arachidonic acid metabolism. The procedure for metabolism study and HPLC analysis were described above. As described previously, the floating Caco-2 cells were apoptotic, and thus the population of cells floating can be used as a measure of apoptosis. Cells were cultured on 6-well dishes (10 cm 2 ) in FBS containing medium until subconfluent. Then, the media were replaced by the following four conditions: serumcontaining medium (FBS), 5 mM of NaBT in FBS, NDGA (10 M) in FBS, and NDGA (10 M) containing 5 mM NaBT in FBS. NDGA was added to the medium every 24 h. NDGA was dissolved in ethanol, and the final concentration of ethanol in medium was less than 0.01%. The numbers of attached and floating cells were respectively counted with a hemocytometer in each condition at every 24 h until 96 h. Five dishes were performed in each condition. The ratio of the numbers of floating cells per total cells was calculated as the indicator of apoptosis. Also, the attached cells and the floating cells were harvested, and genomic DNA was prepared and assayed for the fragmentation of DNA.
Inhibition of Cox-2 by Indomethacin and Apoptosis -To inhibit the Cox activity of Caco-2 cells, indomethacin (Sigma) was added to the cells during NaBT treatment. Cells were suspended and cultured on 6-well dishes (10 cm 2 ) until subconfluent. Then, the media were replaced with media containing FBS under the following conditions: control, 5 mM NaBT, indomethacin (1 M), and 5 mM NaBT and indomethacin (1 M). Indomethacin was administered at every 12 h. The numbers of attached and floating cells were counted with a hemocytometer in each condition at every 24 h until 96 h. Cells were also lysed at 24, 48, and 72 h after treatment for immunoblot analysis.

Apoptosis and Cell Differentiation in Caco-2 Cells during
NaBT Treatment-During culture of colorectal cell lines, some cells detach from the dish and float in the medium. Evidence indicates the floating cells have undergone apoptosis (20). NaBT is reported to induce apoptosis and cell differentiation in various human colorectal carcinoma cells in culture (16,(21)(22)(23) including Caco-2 cells. In our experiments, we observed, microscopically, Caco-2 cells floating in the media with increasing time of treatment with NaBT. Two procedures were performed to evaluate the cell status during treatment with NaBT of Caco-2 cells. Fragmentation of DNA, an index of apoptosis, was determined by an examination of DNA ladder formation in the floating and attached cells. Cell differentiation was measured by alkaline phosphatase activity. The characteristic apoptotic DNA ladders were detected in the floating cells after treatment with NaBT at 72 and 96 h as shown in Fig. 1A, and we could not clearly observe DNA ladders in the attached cells. Thus, NaBT induced apoptosis in the Caco-2 cell characterized by the detachment of cells from the dishes and the presence of cells floating in the media. These observations and conclusions are consistent with the previous reports (16,(21)(22)(23). The alkaline phosphatase activity, a marker of cell differentiation (24), was clearly increased in the attached cells by the NaBT treatment ( Immunoblot analysis was used to confirm the expression of Cox-2 and 15-LO in control and NaBT-treated cells (Fig. 3B). Protein expression could only be determined in attached cells; the protein was degraded in the apoptotic floating cells. Cox-2 protein was detected in the FBS-cultured cells at all time points examined, and the surviving cells also expressed similar amounts of Cox-2. Furthermore, Cox-2 expression was decreased by NaBT induction, a result that is consistent with the Northern data. We did not observe any protein that reacts to the human 15-LO antibody in the cells grown in FBS. In contrast, treatment with NaBT increased the expression of 15-LO protein. We began to detect expression of 15-LO between 6 and 14 h after treatment with NaBT. After 24 -48 h, significant expression of 15-LO was observed. Taken together, these results support the conclusion for the expression of 15-LO during NaBT-induced cell differentiation and apoptosis in Caco-2 cells.
Metabolites of Arachidonic Acid in NaBT-treated Caco-2 Cells-To study arachidonic acid metabolism, NaBT-treated cells and FBS-cultured cells were sonicated, and proteins were estimated. The lysates, 0.8 mg of protein, were then incubated with exogenous [ 3 H]arachidonic acid (3 Ci; 25 M) for 30 min. Reverse-phase HPLC analysis of the extracts is shown in Fig.  4. For FBS-treated cells, PGE 2 (retention time 20ϳ23 min) and prostaglandin F 2␣ (retention time 23ϳ29 min) were the major metabolites, and HETE (retention time 72ϳ78 min) was the minor metabolite (Fig. 4A). The formation of the metabolites was inhibited by indomethacin (5 M), indicating metabolism by Cox-2 activity (Fig. 4B). In contrast, NaBT-treated cells showed a significant increase in the formation of 15-HETE (retention time 73ϳ76 min) and moderate formation of PGE 2 (Fig. 4C). Moreover, the formation of the PGE 2 was inhibited by 5 M of indomethacin, but the formation of 15-HETE peak was not inhibited (Fig. 4D), indicating metabolism by both Cox-2 Also, our studies with fibroblasts indicate 13(S)-HpODE and 13(S)-HODE as potent modulators of signaling pathways, which control cell growth, apoptosis, and other biological process (11,12). Thus, [ 14 C]linoleic acid metabolism studies were performed on cell sonicates (0.8 mg of total protein) prepared from FBS-and NaBT-treated cells. The metabolites of linoleic acid are shown in Fig. 5. In NaBT-treated cells, the 10ϳ12% of linoleic acid was converted to a HODE metabolite(s) at 73 min retention time by reverse-phase HPLC (Fig. 5B). These metabolites coeluted with 9/13-HODE standard and had a UV absorption at 234 nm because 13-HODE and 9-HODE have the same retention time by our reverse-phase HPLC system. Moreover, this peak was not suppressed by addition of indomethacin (5 M) (data not shown). FBS-treated cells converted only 2ϳ4% of linoleic acid to metabolite peaks at 73 min retention time, and several metabolites around the predicted 9/13-HODE peak could be seen (Fig. 5A). To identify 9-HODE and 13-HODE from FBS-and NaBT-treated cells, straight-phase HPLC was performed. Metabolite peaks were collected from reverse-phase HPLC and then analyzed by straight-phase HPLC with 9(R)-and 13(S)-HODE standards. 9-HODE (retention time 11 min) is the dominant metabolite formed in FBStreated cells (Fig. 5C), whereas 13-HODE (retention time 7 min) is the major metabolite in NaBT-treated cells (Fig. 5D). These results are consistent with the conclusion that Cox-2 is responsible for the metabolism in FBS-treated cells, whereas 15-LO is responsible for the metabolism in Caco-2 cells during NaBT-induced apoptosis. The data also support the conclusion that a 15-LO is induced by the NaBT treatment. Moreover, these results indicated that linoleic acid is a better substrate than arachidonic acid for this inducible 15-LO. PstI digestion, and two fragments (648 and 304 bp) are predicted by HindIII digestion. After synthesis of first strand cDNA from total RNA of NaBT-treated Caco-2 cells, PCR was performed with the above primers. At the same time, the RNA samples without reverse transcriptase were amplified to neglect the contamination of the genomic DNA. Two PCR fragments and restriction analysis gave a restriction map that agreed with the predicted size as shown in Fig. 6B. This experiment was repeated twice with the same findings. These results confirm the presence of the reticulocyte type 15-LO expression in human colorectal carcinoma cell line Caco-2 by treatment with NaBT.
Inhibition of 15-LO in NaBT-treated Caco-2 Cells-The next question is whether the inducible 15-LO plays a role in the NaBT-induced apoptosis. To address this question, we treated cells with an inhibitor of 15-LO, NDGA. NDGA is well known as the inhibitor of lipoxygenases, but it can inhibit cyclooxygenase at high concentrations (29). The optimum concentration of NDGA for inhibition of the lipoxygenase without inhibiting Cox-2 activity was determined by HPLC analysis of arachidonic acid metabolism. NDGA, 10 M, which inhibited lipoxygenase activity without significant effects on Cox activity (data not shown), was used for the following apoptosis experiment. To examine if the 15-LO modulated NaBT-induced apoptosis, we measured the number of floating (apoptotic) and attached cells during the course of NaBT treatment in the presence of the 15-LO inhibition NDGA. This method has been extensively used to estimate apoptosis in colorectal cells (30). Apoptosis of the floating cells was also confirmed by DNA fragmentation analysis. The ratio of apoptosis to attached cells in NaBTtreated Caco-2 cells gradually increased with duration of NaBT treatment. NDGA treatment significantly enhanced the NaBTinduced apoptosis compared with that of NaBT-treated Caco-2 cells without NDGA at each time point from 48 to 96 h (apoptotic ratio: at 96 h, 0.51 Ϯ 0.025 S.E. in NaBT, 0.85 Ϯ 0.009 S.E. in NaBT plus NDGA) (Fig. 7A). Treatment with NDGA alone did not enhance apoptosis in the Caco-2 cells. Moreover, the nucleosomal DNA ladder formation could only be seen in floating cells from the Caco-2 cells treated with NaBT or NaBT with NDGA (Fig. 7B). The data support the conclusion that inhibition of 15-LO by NDGA enhances apoptosis in NaBTtreated Caco-2 cells.
Inhibition of Cox-2 in NaBT-treated Caco-2 Cells-Several questions arise from these data. Is NaBT-induced apoptosis regulated by Cox-2 expression in Caco-2 cells as is reported for overexpressing Cox-2 in RIE cells (7), and is the expression of 15-LO linked to Cox-2 in these cells? Cox-2 in Caco-2 cells was decreased by NaBT treatment as indicated by Northern and immunoblot analysis at different times after NaBT treatment. In contrast, 15-LO expression increased during the NaBT treatment (Fig. 3). To address these questions, indomethacin was added to the cells to inhibit Cox-2 activity, and based upon reports in the literature (7), we expected Cox-2 inhibition to result in an increase in apoptosis in the NaBT-treated cells. However, the addition of indomethacin (1 M) at every 12 h did not alter the number of apoptotic floating cells in the NaBTtreated cells or FBS-treated cells at any time point examined, as shown in Fig. 8A. This result seems to suggest the decreasing Cox-2 is not modulating NaBT-induced apoptosis in Caco-2 cells. To further examine this issue, we measured the expression of Cox-2 and 15-LO in the attached cells under these conditions. Protein expression could not be estimated in apo-

DISCUSSION
Colorectal tumor formation results from the accumulation of specific genetic changes. The homeostasis of the colorectal epithelial cells, which give rise to the tumor, depends not only on the rate of cell proliferation but also on apoptosis. Mutation in the APC gene is an early event in the development of colorectal cancer and is associated with a decrease or inhibition in apoptosis (30). Studies with mice clearly link the loss of APC gene function with the induction of Cox-2 and polyp growth (3). In addition, other studies using mice with the disrupted Cox-2 gene provide evidence to support the hypothesis that Cox-2 expression is an early rate-limiting step in the development of adenoma formation (2). The most compelling evidence linking the overexpression of Cox-2 to altered apoptosis and cell differentiation was provided by studies with RIE cells (7). The overexpression of Cox-2 in RIE cells prevented NaBT and adhesioninduced apoptosis and cell differentiation. Apoptosis was restored by treatment with a Cox inhibitor (7). Human colorectal cell lines have been useful in investigations of the cellular mechanisms associated with Cox-2 expression and altered apoptosis. Apoptosis in human colorectal carcinoma cell lines can be induced by inhibitors of Cox-2 by both Cox-2 dependent (31,32) and independent (8,9) pathways. However, the addition of exogenous prostaglandins failed to reverse inhibitor-induced apoptosis (8).
We have examined unsaturated fatty acid metabolism and the expression of the enzymes responsible for lipid metabolism during NaBT-induced apoptosis and cell differentiation in human colorectal carcinoma Caco-2 cells. These cells have been frequently used, undergo apoptosis and cell differentiation in response to NaBT, and have a mutated APC gene expressing the truncated APC protein product (30,33). The arachidonic acid metabolism by Caco-2 cells was dramatically shifted during the NaBT treatment from prostaglandins to metabolites of 15-LO, mainly 15-HETE. Likewise, the metabolism of linoleic acid, a predominate unsaturated fatty acid present in cells and the preferred substrate for the reticulocyte 15-LO, was dramatically shifted from 9-HODE to 13-HODE. Furthermore, the NaBT treatment resulted in a significant enhancement of overall cis-unsaturated fatty acid metabolism with the lipoxygenase-derived metabolites predominating. Caco-2 cells poorly expressed Cox-1, but high expression of Cox-2 was observed. Treatment with NaBT modestly attenuated the expression of Cox-2 and significantly increased the expression of a 15-LO. The NaBT-induced 15-LO in Caco-2 cells was confirmed by RT-PCR and restriction enzyme analysis as the human reticulocyte 15-LO isolated by Sigal et al. (27). Ample evidence is presented in this report to support the conclusion that during NaBT-induced apoptosis and cell differentiation in Caco-2 cells, significant expression of the human reticulocyte 15-LO occurs, which shifts the metabolism from prostaglandin to lipoxygenase-catalyzed metabolites. This is the first report demonstrating the expression and enzymatic activity of human reticulocyte 15-LO in a human colorectal carcinoma cell line. Whether 15-LO expression also occurs in human colon tumors is currently under investigation.
The relationship between Cox-2 expression, which is modulated by the APC gene alteration, and 15-LO expression during apoptosis is apparent. During the course of NaBT treatment, there seems to be an inverse relationship between Cox-2 expression and 15-LO expression. At times during the apoptosis and cell differentiation process where the lowest expression of Cox-2 was detected, the highest 15-LO expression was observed. In addition, treatment of the cells with an inhibitor of Cox-1 and -2, indomethacin, attenuates the expression of Cox-2 in NaBT-treated cells and enhances the expression of 15-LO by NaBT at times during the course of apoptosis. These findings suggest Cox-2 and 15-LO are working in parallel to modulate the process of apoptosis and cell differentiation in the Caco-2 cells. The regulation of Cox-2 has been extensively studied, and several studies have indicated either the up-regulation (34,35) or down-regulation (36,37) by the stable metabolites formed from arachidonic acid by this enzyme. Studies suggest a regulation via the SP-1, NFB, NF-IL6, AP-1, AP-2, and CRE sites present in the Cox-2 gene (38,39), but regulation by NaBT has not been reported. The regulation of the human 15-LO is poorly understood. The inflammatory cytokines IL-4 and IL-13 enhance the expression of 15-LO in the human lung carcinoma A-459 (40), human monocytes (41)(42)(43), and human bronchial epithelial cells (25). The expression of the 15-LO is most notable in human airway epithelial cells (44,45) but has also been detected in monocytes and skin (46). There is no report that NaBT, the inducer of cell differentiation or apoptosis by the mechanism of histone hyperacetylation (47), regulates the expression of 15-LO.
The importance of the 15-LO expression in the process of NaBT-induced apoptosis and cell differentiation is not clear. Inhibition of the lipoxygenase activity by NDGA at concentrations determined to attenuate only the 15-LO activity in these cells enhanced the NaBT-induced apoptosis, which supports the hypothesis for 15-LO metabolites acting as inhibitors of apoptosis. Several reports in the literature (14,15) and in this laboratory 2 also suggest that lipoxygenase metabolites including 15-LO metabolites can act as inhibitors of apoptosis. However, NDGA is a general inhibitor of lipoxygenases and can influence the redox state of cells (48,49), so this conclusion must be viewed with caution. We considered directly testing the 15-LO metabolites of arachidonic acid and linoleic acid, but the high binding of the metabolites to proteins and other materials in the tissue culture system severely compromises this approach. An alternate approach is to prepare Caco-2 cells, which highly express 15-LO, as a tool to study the importance of the lipoxygenase in apoptosis. These experiments are currently in progress. Indomethacin at concentrations that inhibit Cox-2 did not attenuate NaBT-induced apoptosis, which was an unexpected result in view of other reports in the literature that clearly indicate inhibition of Cox-2 enhances apoptosis in colorectal carcinoma cells (50). An explanation for this observation was provided by analysis of the expression of Cox-2 and 15-LO under these conditions. Indomethacin enhanced the expression of 15-LO, which should increase the formation of 15-LO-derived metabolites. These metabolites would act to replace the prostaglandins and attenuate the apoptosis. Thus, the net response to indomethacin treatment is lack of effect on NaBTinduced apoptosis. Thus, Cox-2 and 15-LO may act in concert to attenuate apoptosis in the Caco-2 cells. The inverse relationship between Cox-2 and 15-LO expression (note Fig. 3) is further support for this notion.
Linoleic acid and not arachidonic acid is the preferred substrate of the reticulocyte 15-LO, and thus linoleic acid metabolites must be considered as potentially biologically active. Considerable 13-HODE was produced from the cell lysates from NaBT-treated cells during cell differentiation and apoptosis. Because NaBT induced cell differentiation as well as apoptosis, we must also consider if these lipids are modulators of cell differentiation. In the human airway epithelium, the expression of 15-LO is observed during retinoid-induced cell differentiation (25). Furthermore, increased expression of rabbit reticulocyte 15-LO is observed during reticulocyte maturation (51). The overexpression of peroxisome proliferator-activated receptor (PPAR)␥ in rat intestinal tumors was recently reported, and Caco-2 cells were found to express the highest levels of PPAR␥ of all the human colorectal cell lines tested (52). Furthermore, 13(S)-HODE and 9(S)-HODE bind to and activate the PPAR␥ (53). These lipid metabolites induce monocyte maturation and regulate gene expression. The induction of 15-LO during NaBT treatment of Caco-2 cells may be a response to rescue the cells from apoptosis. One can speculate that the 15-LO acts as an anti-apoptotic signal that directs the cell from apoptosis to a newly differentiated cell. The overexpression of Cox-2 observed after APC mutations acts as a more prolonged or constitutive signal to attenuate apoptosis. Thus, both Cox-2 and 15-LO may act to modulate apoptosis and cell differentiation in colorectal carcinoma cells.