Non-steroidal anti-inflammatory drugs (
NSAIDs) inhibit colorectal
carcinogenesis and prevent or revert the growth of premalignant
colonic polyps. They inhibit
cyclooxygenase (COX) but recent data indicate that this is not the only or even the most important mechanism of inhibition in
colorectal tumor cells. We have used colonic
carcinoma and
adenoma cell lines to study the effects of the
NSAID sulindac sulfide, its COX-inactive metabolite,
sulindac sulfone, and the
isoenzyme-specific inhibitors
SC58125,
SC236 and
SC58560 on
tumor cell growth in relation to COX-2 expression and
prostaglandin production. To establish the role of COX-2 in
NSAID action, we constructed clones expressing different levels of COX-2 from SW480 cells. All five compounds inhibited
DNA synthesis and/or induced apoptosis, each with a characteristic pattern. ID(50)s were very similar in all the cell lines and were independent of COX expression, except for the COX-1 inhibitor
SC58560, which was least effective in HT29/HI1, the cell line expressing the highest level of COX-1 (ID(50) 70 microM; in other cells lines the ID(50) was 15 microM). For all other compounds ID(50) concentrations varied less than two-fold: 25-40, 40-90 and 150 microM for
SC236,
sulindac sulfide and
sulindac sulfone, respectively.
SC58125 was the weakest inhibitor, never causing >50% cell loss. All compounds modulated expression of Bcl-2 and Bak and activated
caspase 3. Overexpression of COX-2 in SW480 cells protected them against induction of apoptosis by
sulindac sulfide. The effect was restricted to clones producing high levels of
prostaglandin E(2). In summary, our data indicate that both COX-dependent and COX-independent mechanisms are involved in
NSAID-induced growth in
colorectal tumor cells. The concentrations necessary to inhibit growth were higher than serum concentrations that can be obtained in vivo, indicating that the
therapeutic effect of
NSAIDs cannot be explained by a direct effect of
NSAIDs on the epithelial cells alone. For therapeutic purposes, compounds using different targets could be used to minimize side effects while optimizing
therapeutic effect.