Several studies have shown that COX-2, VEGF and IL-8 are overexpressed in lung cancer compared to normal bronchial epithelium. Most studies have focused on NSCLC since clinical specimens of SCLC are difficult to obtain for research because SCLC patients are rarely operated on. Tas et al [
23,
24] recently reported that serum VEGF levels were significantly higher in lung cancer patients than healthy controls. Mean serum VEGF levels appeared to be slightly higher in SCLC (1350 pg/ml, range 170 – 3810 pg/ml, n = 34) than NSCLC patients (402 pg/ml, range 121 – 1800 pg/ml, n = 52), but this was not tested statistically.
Here, we have compared and contrasted SCLC and NSCLC using a panel of cell lines. Basal expression of VEGF, IL-8, COX-1, COX-2 and PGE
2 was determined in three NSCLC and five SCLC cell lines. VEGF was produced at similar levels in all cell lines. NSCLC cell lines produced much more IL-8 than SCLC cell lines. COX-1 was expressed at similar levels in all cell lines, whereas COX-2 was only expressed in NSCLC cell lines, albeit at low levels in H460. Interestingly, high levels of PGE
2 were not only produced in MOR/P and A549 cell lines as a result of strong COX-2 expression, but also in H460 cell line in which COX-2 expression was weak. This could result from reduced catabolism of PGE
2 due to low levels of 15-hydroxyprostaglandin dehydrogenase (15-PGDH), a primary enzyme responsible for PGE
2 metabolism [
25]. Previous studies showed that NSCLC cells expressed COX-2 protein [
17,
26]. However, one study showed that a subset of 20% of SCLC patients (11 out of 54) expressed COX-2 [
27]. These data, together with our results and those of Pold et al [
20], appear to suggest an association between COX-2 expression and IL-8 production in lung cancer cells. In contrast, VEGF production did not seem to be solely related to COX-2 expression as both COX-2 positive and negative cell lines produced similar levels of VEGF in our study.
COX-2 expression has previously been associated with VEGF in patients with NSCLC [
19,
20]. Two reports from Dubinett's lab showed that expressions of IL-8 and VEGF were enhanced in NSCLC cell lines transfected with retroviral COX-2 vector [
20,
28]. The aim of this study was to exam the causal relationship between COX-2 and IL-8 or VEGF in lung cancer cell through inhibition of endogenous COX-2 using either a potent and specific COX-2 inhibitor, NS398 or COX-2 siRNA. NSCLC cell lines were incubated with NS-398 for 24 h and 48 h. NS-398 exerts its effects by directly inhibits COX-2 activity. We demonstrated that COX-2 activity was inhibited by NS-398 as evidence of reduced production of PGE
2 in these cell lines. We decided to use a maximum concentration of 20 μM in this study because COX-2 inhibition had been demonstrated and cell growth was not significantly inhibited in NSCLC cell lines as shown in Figure
5 at this level. NS-398 has been shown to induce 16% apoptosis in H460 cells at 300 μM in another study [
29]. Unexpectedly, IL-8 production was not altered after treatment with NS-398 in these cell lines. However, VEGF was significantly decreased by NS-398 in A549 and MOR/P cell lines, but not in H460. Both of A549 and MOR/P cell lines expressed high levels of COX-2 whereas H460 expressed very low level of COX-2 in our experiments. In some reports, H460 was considered as COX-2-negative cell line [
29]. These results suggested that VEGF production was independent of COX-2 in H460 and other COX-2-negative cells such as SCLC cell lines. We further confirm the direct involvement of COX-2 on VEGF production using COX-2 siRNA, and similar results were found. VEGF was significantly reduced in both of A549 and MOR/P, but not in H460. IL-8 was not affected following treatment with COX-2 siRNA in all tested NSCLC cell lines. These results suggest that VEGF is, at least partly, COX-2 dependent in COX-2-expressing NSCLC cells such as A549 and MOR/P. This finding is in agreement with other findings [
19,
20,
28,
30]. However, VEGF production is clearly not depend on COX-2 status in other lung cancer cells such as SCLC cells, which expressing no COX-2, and some NSCLC cells including H460, which expressing very low level of COX-2. Our results also suggest that IL-8 is COX-2 independent in lung cancer cells. This finding seems to contradict others [
28]. The different findings may be due to the different experiments systems (e.g. to introduce ectopic overexpression of COX-2 or to knock down endogenous COX-2 by siRNA) employed in these studies. Interestingly, Raut et al [
31] reported that blocking COX-2 production by NS-398 in pancreatic cancer cell lines did not affect VEGF, bFGF and IL-8 production, but Singh et al [
32] found that NS-398 downregulated IL-8 by 30% in a COX-2 transfected MDA-231 breast cancer cell line. These contradictory findings suggest that the relationship between COX-2, VEGF and IL-8 is complex. More studies are required to establish the causal relationship between COX-2 and IL-8 and VEGF in lung cancer. COX-2 inhibitors have been shown to reduce angiogenesis and metastasis in lung cancer [
21], our results suggest that anti-angiogenic effects of COX-2 inhibitors might be, at least partly, mediated by inhibition of VEGF, but not IL-8. As IL-8 seems to be independent of COX-2, therefore, there is a strong rationale for combining treatments with inhibitors of COX-2 and IL-8 in NSCLC.