Introduction

Top Abstract Introduction Methods Results Discussion References

NSAIDs inhibit colorectal tumor growth in vitro (Narisawa et al., 1981; Pollard and Luckert, 1981) and in vivo (Giardiello et al., 1993). For example, several epidemiological studies have revealed that NSAIDs, and in particular aspirin, reduce approximately 50% the risk of colorectal carcinoma (Kune et al., 1988) and other cancers of the gastrointestinal tract (Thun et al., 1993). The antiinflammatory effects of NSAIDs result from inhibition of cyclooxygenases (Mitchell et al., 1994), and studies with selective antagonists have indicated that the inducible isozyme cyclooxygenase-2 is also one of the targets for the inhibitory effects of NSAIDs on colorectal carcinoma growth in vivo (Oshima et al., 1996). However several lines of evidence suggest that the in vitro antiproliferative effects of NSAIDs are mediated by different targets. For example, the sulfone derivative of the NSAID sulindac does not inhibit cyclooxygenases in vitro (Shen and Winter, 1977), but is an effective inhibitor of the growth of colorectal carcinoma cell lines (Hixson et al., 1994; Piazza et al., 1995). Furthermore cyclooxygenase-2-selective antagonists have no effect on the growth of human colorectal carcinoma cell lines (Murphy et al., 1998). Evidence is also accumulating that sulindac and its sulfone derivative can induce apoptosis both in colorectal carcinoma cell lines (Piazza et al., 1995; Shiff et al., 1995) and in the colorectal epithelium (Pasricha et al., 1995). The pathway of long chain fatty acid oxidation has recently been suggested as a novel alternative target, based on the observation that several NSAIDs, including sulindac sulfone, inhibit palmitate oxidation in colorectal carcinoma cell lines, with potencies similar to their potencies for inhibition of cell growth (Yang et al., 1998).

Two consecutive steps in the oxidation of long chain fatty acids, namely hydration of enoyl-CoA and dehydrogenation of 3-hydroxyacyl-CoA, are catalyzed by the TP (Uchida et al., 1992). The close relationship between the amino acid sequences of rat (Kamijo et al., 1993) and human (Zhang and Baldwin, 1994; Kamijo et al., 1994a) TP, and the amino acid sequence of a 78-kDa porcine GBP (Mantamadiotis et al., 1993), suggested that TP and the GBP were the product of the same gene in different species. Further evidence for a functional similarity between TP and the GBP was provided by the observations that cross-linking of iodinated gastrin to the GBP was blocked by both enoyl-CoAs and 3-ketoacyl-CoAs (Baldwin, 1994), and both enzyme activities of TP were inhibited by gastrin (Hashimoto et al., 1996). The observation that both halves of TP bound gastrin when expressed separately in Escherichia coli confirmed that gastrin bound to both hydratase and dehydrogenase active sites (Murphy et al., 1996). Hence competition for iodinated gastrin binding provides a convenient assay for inhibitors of TP activities.

The identification of hereditary mutations in the  subunit has clearly established the metabolic importance of the TP (Wanders et al., 1989, 1992; Jackson et al., 1992; Kamijo et al., 1994b). Patients present in early childhood with a spectrum of clinical features, including hypoketotic hypoglycemia progressing to coma, muscle weakness, cardiomyopathy and microvesicular fat deposition in the liver. Symptoms are often triggered by fasting or by upper respiratory tract or gastrointestinal infections. In most cases the primary defect is in the dehydrogenase domain of the -subunit (Wanders et al., 1992; Jackson et al., 1992; Kamijo et al., 1994b), and by far the commonest mutation results in substitution of glutamic acid at position 474 with glutamine (Ijlst et al., 1994). The observation that TP mutation is also the cause of acute fatty liver of pregnancy (Treem et al., 1994; Sims et al., 1995) confirms that TP is a critical metabolic enzyme.

TP has also been identified as the target for the inhibitory effects of the gastrin/cholecystokinin receptor antagonists benzotript and proglumide on colorectal carcinoma cell growth (Baldwin, 1994). Benzotript has also been shown to inhibit oxidation of the long chain fatty acid palmitate in fibroblast homogenates, and all three activities of TP in vitro (Hashimoto et al., 1996). Since NSAIDs inhibit long chain fatty acid oxidation in colorectal carcinoma cell lines (Yang et al., 1998), we wished to determine whether or not TP was the target. We therefore determined the affinities of NSAIDs for TP by measuring their ability to compete with ¹²⁵I-[Nle¹⁵]-gastrin_2-17 in a covalent cross-linking assay, and investigated the effect of NSAIDs on the 3-hydroxyacyl CoA dehydrogenase activity intrinsic to TP. We also compared the IC₅₀ values obtained with the potencies of NSAIDs as inhibitors of colorectal carcinoma cell proliferation.

	Methods

Top Abstract Introduction Methods Results Discussion References

NSAIDs were from Sigma Chemical Co. (St. Louis, MO), with the exception of sulindac sulfide and sulfone (Merck, Sharp and Dohme, West Point, PA) and tiaprofenate (Roussel UCLAF, Melbourne, Australia), which were generous gifts from the indicated companies. The human colon carcinoma cell lines LIM 1215 (Whitehead et al., 1985) and LIM1899 (Whitehead et al., 1992) were provided by Dr. R. H. Whitehead, Ludwig Institute for Cancer Research, Melbourne, Australia.

Cross-linking of iodinated gastrin to TP. TP was partially purified from detergent extracts of porcine gastric mucosal membranes by sequential chromatography on concanavalin-A- and DEAE-Sepharose as described previously (Baldwin et al., 1994). Cross-linking of ¹²⁵I-[Nle¹⁵]-gastrin_2,17 to TP in the presence of increasing concentrations of NSAIDs was measured at least three times for each NSAID, as described previously (Baldwin, 1994). Briefly, ¹²⁵I-[Nle¹⁵]-gastrin_2,17 (0.4 nM) was reacted with 0.2 mM disuccinimidyl suberate in 50 mM Na⁺ HEPES, pH 7.6, for 15 min at 0°C. Aliquots (25-µl, 10 fmol, approx. 30,000 cpm) were added to 25-µl aliquots of TP that had been preincubated in the same buffer containing 0.1% Triton X-100 and twice the desired concentration of NSAID for 15 min at 0°C. After 20 min at 0°C, the reaction was stopped by addition of 50 µl Laemmli loading buffer and the samples were treated for 5 min at 95°C and electrophoresed on sodium dodecyl sulfate-10% polyacrylamide gels. After staining with Coomassie blue and drying, the radioactivity associated with TP was detected and quantitated with a Phosporimager (Molecular Dynamics, Sunnyvale, CA). Estimates of IC₅₀ values and of the levels of iodinated gastrin bound in the absence of NSAIDs were obtained with the programs EBDA and LIGAND (Baldwin, 1994).

Assay of 3-hydroxyacyl-CoA dehydrogenase activity. Aliquots (4 µg) of TP were assayed in duplicate for the reverse reaction of 3-hydroxyacyl-CoA dehydrogenase at 25°C by measuring the increase in NAD⁺ absorption at 340 nm in the presence of 20 µM 3-ketopalmitoyl-CoA, 100 µM NADH, 250 mM K⁺ phosphate, pH 7.4 and increasing concentrations of NSAIDs.

Cell growth assay. A colorimetric assay (Mosmann, 1983) was used to measure cell growth. Briefly 10⁴ cells were seeded in a 96 well plate in RPMI 1640 medium containing 10 µM thioglycerol, 25 U/ml insulin, 1 mg/ml hydrocortisone and 10% fetal calf serum. Fresh medium containing the above additives, 10% fetal calf serum and the substance under investigation was added 24 hr later. After incubation at 37°C for 20 hr in a humidified atmosphere of 10% CO₂, 10 µl of 5 mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, Sigma Chemical Co., St. Louis, MO) was added per well, and the plate was incubated for a further 4 hr before the medium was discarded. 200 µl 0.04 M HCl in isopropanol was added to lyse the cells, and the absorbance at 560 nm was read on a Titertek Multiscan MCC 1340 (Labsystems, Helsinki, Finland).

	Results

Top Abstract Introduction Methods Results Discussion References

Binding of NSAIDs to the trifunctional protein. The trifunctional protein, which catalyses three of the four reactions of mitochondrial long chain fatty acid oxidation, appeared to be a likely target for the inhibitory effects of NSAIDs on palmitate oxidation (Yang et al., 1998). We therefore determined whether or not NSAIDs would bind to the TP. Because most NSAIDs are not available in a radioactively labeled form, we measured the potency of NSAIDs as inhibitors of the previously described binding of ¹²⁵I-[Nle¹⁵]-gastrin_2,17 to TP (Baldwin, 1994). All NSAIDs tested inhibited cross-linking of iodinated gastrin to porcine TP (fig. 1), with IC₅₀ values which ranged from 40 µM for sulindac sulfide to 9.3 mM for aspirin (table 1).

View larger version (34K): [in this window] [in a new window]   Fig. 1.   NSAIDs inhibit cross-linking of gastrin to TP. Cross-linking of 125I-[Nle15]-gastrin2,17 to TP was measured by phosphorimager (insets) in the presence of increasing concentrations of NSAIDs, and expressed as a percentage of the value obtained in the absence of competitor. Points are the means (± S.D.) of duplicate determinations, and lines of best fit were obtained with the programs EBDA and LIGAND. Values for IC50 and for the predicted ordinate intercept were as follows: sulindac sulfide, 28 µM, 108%; indomethacin, 105 µM, 121%; ibuprofen, 1.7 mM, 116%; aspirin, 16 mM, 110%. These values were averaged with the results of at least two similar experiments to obtain the mean values presented in table 1.

View this table: [in this window] [in a new window]   TABLE 1 IC50 Values for Inhibition of Colon Carcinoma Cell Growth and TP Cross-linking by NSAIDs

Effect of NSAIDs on 3-hydroxyacyl-CoA dehydrogenase activity. To confirm that binding of NSAIDs affected the enzyme activities intrinsic to TP, we next tested the effect of NSAIDs on 3-hydroxyacyl-CoA dehydrogenase activity. Diclofenac and ibuprofen both inhibited dehydrogenase activity (fig. 2), with IC₅₀ values of 4.1 ± 1.8 and 24.4 ± 1.9 mM, respectively. Possible reasons for the discrepancy between these values and the IC₅₀ values for inhibition of gastrin cross-linking to TP are considered "Discussion."

View larger version (22K): [in this window] [in a new window]   Fig. 2.   NSAIDs inhibit 3-hydroxyacyl-CoA dehydrogenase. Hydrogenation of 3-ketoacyl-CoA by TP (4 µg) was assayed in the presence of increasing concentrations of NSAIDs. Activities were expressed as a percentage of the value obtained in the absence of inhibitor. Points are the means (± S.D.) of duplicate determinations. Because the data could not be fitted satisfactorily with the programs EBDA and LIGAND, IC50 values [diclofenac (open triangles), 4.1 mM; ibuprofen (closed triangles), 24 mM] were determined graphically. These IC50 values were averaged with the results of two similar experiments to obtain the mean values presented in the text.

Inhibitory effects of NSAIDs on colorectal carcinoma cell growth. Several NSAIDs have been reported to inhibit the proliferation of the human colon carcinoma cell lines HT 29, SW 480 and DLD-1 (Hixson et al., 1994) and LIM 1215 (Murphy et al., 1998). The IC₅₀ values for inhibition of growth of the human colon carcinoma cell line LIM 1215 by NSAIDs are similar (table 1). To determine whether or not there was any connection between inhibition of cell growth and inhibition of long chain fatty acid oxidation, we next compared the IC₅₀ values for inhibition of LIM 1215 cell growth with the IC₅₀ values for inhibition of cross-linking of iodinated gastrin to porcine TP (fig. 3). The good correlation observed is consistent with the hypothesis that inhibition of TP contributes to inhibition of cell growth.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Antiproliferative effects of NSAIDs are mediated by the TP. The mean IC50 values (table 1) for the inhibition of cross-linking of 125I-[Nle15]-gastrin2,17 to TP by NSAIDs (identified by the first three letters of the names given in table 1, with the exception of sulindac sulfide (SUS), sulindac sulfoxide (SUX) and sulindac sulfone (SUN)) were determined as described in the legend to figure 1, and compared with the IC50 values for inhibition of proliferation of LIM 1215 cells measured as described in the table 1 legend. The line of best fit (y = 0.49x + 1.24, r = 0.81, P < .0001) was obtained by linear regression.

	Discussion

Top Abstract Introduction Methods Results Discussion References

We have identified a novel mechanism that may contribute to the inhibitory effects of NSAIDs on colorectal carcinoma cell growth. We have previously reported that several NSAIDs, including sulindac sulfide, indomethacin, ibuprofen and aspirin, inhibit oxidation of the long chain fatty acid palmitate by the human colorectal carcinoma cell line LIM 1215 (Yang et al., 1998). We now report that inhibition of mitochondrial long chain fatty acid oxidation appears to be due, at least in part, to binding of NSAIDs to TP, which catalyses two of the four reactions involved. All NSAIDs tested inhibit cross-linking of iodinated gastrin to TP (fig. 2). Furthermore, the interaction between TP and either diclofenac or ibuprofen resulted in inhibition of 3-hydroxyacyl-CoA dehydrogenase activity in vitro (fig. 3).

Our data are consistent with previous reports in the literature that individual NSAIDs, including pirprofen (Geneve et al., 1987), ibuprofen (Freneaux et al., 1990; Zhao et al., 1992), aspirin (Deschamps et al., 1991), salicylic acid (Deschamps et al., 1991) and flurbiprofen (Zhao et al., 1992), inhibit oxidation of long chain fatty acids by isolated liver mitochondria and in whole animals. Thus the inhibitory effects reported herein are not restricted to colorectal carcinoma cells, but are also observed in normal tissue. Ibuprofen has also been reported to inhibit short chain fatty acid oxidation in isolated rat colonocytes (Roediger and Millard, 1995), and the observation that normal colonocytes rely heavily on fatty acid oxidation for their metabolic energy supply (Ardawi and Newsholme, 1985) suggests that inhibition of the pathway of long chain fatty acid oxidation might have a significant effect on the growth of normal colonocytes.

Technical difficulties prevented the demonstration of inhibition of 3-hydroxyacyl CoA dehydrogenase activity by the other NSAIDs listed in table 1. In particular, the absorption of the NSAIDs at 340 nm (the wavelength used for measurement of NADH oxidation) interfered with the dehydrogenase assay, so that the maximum concentrations tested were generally less than the IC₅₀ values for inhibition of gastrin cross-linking. Similarly the strong absorption of the NSAIDs at 280 nm (the wavelength used for measurement of enoyl-CoA hydration) precluded assay of the effects of NSAIDs on hydratase activity. However, because both gastrin and the gastrin/cholecystokinin receptor antagonist benzotript inhibited both hydratase and dehydrogenase activities (Hashimoto et al., 1996) it seems reasonable to assume that NSAIDs will also inhibit hydratase activity. The lower IC₅₀ values, and the absence of spectral interference, clearly establish the gastrin cross-linking assay as the method of choice for screening of potential TP antagonists.

The IC₅₀ values in the dehydrogenase assay were 10- to 25-fold higher than the IC₅₀ values for inhibition of gastrin cross-linking to TP. This discrepancy presumably reflects the fact that the IC₅₀ value determined by kinetic methods is always greater than the dissociation constant of the complex between an enzyme catalyzing a two substrate reaction and a competitive inhibitor by a factor of (1 + A/K_a)(1 + B/K_b), where K_a and K_b are the dissociation constants for substrates A and B, respectively (Cheng and Prusoff, 1973). Because substrate concentrations are saturating in the in vitro assay (i.e., A > K_a, B > K_b), the factor is considerably more than 1, and the IC₅₀ value determined by cross-linking in the absence of substrates is a more accurate estimate of the dissociation constant. Because almost all of the NSAIDs tested inhibit cross-linking of gastrin to the active sites of TP, it seems likely that inhibition of 3-hydroxyacyl-CoA dehydrogenase activity by other NSAIDs would also be observed, if it was technically possible to measure dehydrogenase activity with substrate concentrations lower than the dissociation constants K_a and K_b.

Comparison of the inhibitory potencies of NSAIDs in the assays reported herein is consistent with our previous hypothesis (Yang et al., 1998) that inhibition of long chain fatty acid oxidation may contribute to inhibition of cell growth. In particular the IC₅₀ values obtained for inhibition of cross-linking of iodinated gastrin to TP (table 1) correlate well with the IC₅₀ values for inhibition of proliferation of the human colon carcinoma cell line LIM 1215 by NSAIDs (fig. 3). The failure of acetaminophen to inhibit cross-linking is particularly interesting, because epidemiological studies have revealed that acetaminophen use is not associated with a decreased risk of any gastrointestinal tract carcinoma (Thun et al., 1993). The fact that the slope of the correlation (fig. 3) is less than one may be explained by several factors, including the difference in species, the differences in reaction conditions between the in vitro cross-linking assay and the interior of LIM 1215 cells, or by concentration of some NSAIDs (e.g., ibuprofen) by cells (Shen and Winter, 1977).

In conclusion inhibition of TP may contribute to the effects of NSAIDs on colorectal carcinoma cell growth. We have demonstrated previously that the inhibitory effects of gastrin receptor antagonists are also mediated by TP (Baldwin, 1994). We now report that NSAIDs not only bind to TP but also inhibit the enzyme activities intrinsic to TP. We postulate that inhibition of TP results in the previously reported reduction in long chain fatty acid oxidation in colorectal carcinoma cells (Yang et al., 1998), and that the consequent decrease in available energy may contribute to a reduced proliferation rate. Recognition of TP as a possible inhibitory target of NSAIDs should permit the rational design of more potent and selective TP antagonists which may ultimately provide novel cancer therapies.