Evidence for the participation of Rel/NF-kappaB transcription factors in long-term memory has recently been reported in the context-signal learning paradigm of the crab Chasmagnathus, in which a high correlation between long-term memory formation and NF-kappaB activation was observed. Two components of the NF-kappaB pathway in the crab brain have now been identified by cross-immunoreactivity using mammalian antibodies for IkappaB-alpha and IkappaB kinase alpha. Furthermore, IkappaB kinase-like phosphotransferase activity, which was inhibited by the IkappaB kinase inhibitor sulfasalazine, was detected in brain extracts. We have evaluated the effect of sulfasalazine administration on long-term memory tested at 48 h. Amnesia was found when sulfasalazine was administered pre-training and 5 h after training but not at 0 or 24 h after training. Thus, two periods for sulfasalazine-induced amnesia were found in coincidence with the two phases of NF-kappaB activation previously described (immediately and 6 h after training). The cyclooxygenase inhibitor indomethacin did not induce amnesia when administered pre-training. Thus, the possibility that sulfasalazine induces amnesia by means of cyclooxygenase inhibition is unlikely to be tenable. In vivo sulfasalazine inhibition of basal NF-kappaB activity was found between 30 and 45 min after injection, as assessed by electrophoretic mobility shift assay. On the other hand, in vivo sulfasalazine administration 6 h after training inhibited the second phase of training-induced NF-kappaB activation, providing evidence that the sulfasalazine effect on memory is due to a direct effect of the drug on the NF-kappaB pathway. These results provide the first evidence that IkappaB kinase and NF-kappaB activation are necessary for memory formation.