Previous studies suggest that the enhanced expression of the osteoblastic phenotype exhibited by MG63 osteoblast-like cells on rough Ti surfaces (R(a) 4-5 microm) involves increased production of prostaglandin. Inhibition of prostaglandin synthesis by indomethacin blocks surface-roughness-dependent decreases in cell proliferation and increases in alkaline phosphatase activity and the production of osteocalcin and TGF-beta1. This study examined the hypothesis that the increase in expression of the osteoblastic phenotype noted in MG63 cells cultured on rough Ti surfaces is mediated by inducible cyclooxygenase-2 (Cox-2) whereas Cox-1 modulates prostaglandin production and phenotypic expression of the cells under standard conditions and on smooth Ti surfaces. MG63 cells were cultured on tissue culture plastic, smooth Ti (PT, R(a) = 0.60 microm), and two rough Ti surfaces with differing morphologies (SLA, R(a) = 3.97 microm and TPS, R(a) = 5.21 microm). At 24 h after plating, media were replaced with media containing the general Cox inhibitor indomethacin (10(-7)M), the Cox-1 inhibitor resveratrol (1 or 10 microM), or the Cox-2 inhibitor NS-398 (1 or 10 microM). Media were changed again after 48 h. Five days after plating, osteocalcin, PGE(2), and TGF-beta1 content of the conditioned media were determined. Cell numbers were assessed in the same cultures used for determination of osteocalcin production. Cell layer protein and alkaline phosphatase specific activity were assessed in cultures used to measure PGE(2) and TGF-beta1. Indomethacin, resveratrol, and NS-398 had no effect on cell number. Indomethacin blocked the surface-roughness-dependent increase in PGE(2) production by up to 80%. Similarly, resveratrol inhibited up to 50% of the PGE(2) production on smooth surfaces and up to 80% on rough surfaces. In contrast, NS-398 had no effect on PGE(2) production by cells on smooth surfaces but caused a 60% reduction in cultures on rough surfaces. Indomethacin reduced alkaline phosphatase on all surfaces below basal levels. However, neither resveratrol nor NS-398 had an effect. Indomethacin blocked the stimulatory effect of surface roughness on osteocalcin production while resveratrol only partially reduced osteocalcin production, and NS398 completely blocked the surface-dependent increase. TGF-beta1 production on rough surfaces was blocked by indomethacin. The effects of resveratrol and NS-398 were dose dependent, but neither agent caused total inhibition of the increase noted on SLA, and only resveratrol blocked the increase on TPS. These results indicate that both Cox-1 and Cox-2 are involved in the response of osteoblasts to surface roughness with respect to production of PGE(2), TGF-beta1, and osteocalcin. While prostaglandin mediates the effects of surface roughness on alkaline phosphatase, neither Cox-1 nor Cox-2 appears to be involved, at least with respect to the two inhibitors used.