COX-3Oa virtual pain target in humans? JAN M. SCHWAB, 1,2 THOMAS BEITER, 1 JUERGEN U. LINDER,* ,1 STEFAN LAUFER,* JOACHIM E. SCHULZ,* RICHARD MEYERMANN, AND HERMANN J. SCHLUESENER Institute of Brain Research, Medical School, D-72076 Tuebingen; and *Institute of Pharmacy, University of Tuebingen, D-72074 Tuebingen, Germany A long-sought molecule (1) was recently identi- fied as acting independent of COX-1/-2 (2). The potent analgesic and anti-pyretic actions of acetamino- phen lacking anti-inflammatory potency suggested the presence of an additional cyclo-oxygenase that could be directly responsible for acetaminophen-sensitive gener- ation of prostanoids in neuronal systems. This could now be understood at least in part as modulation of the recently identified COX-3 (2). In canine, alternative splicing generates four different mRNA variants de- rived from the COX-1 geneOCOX-3, COX-1, and two partial COX-1 (denoted as PCOX-1 proteins), PCOX-1a and PCOX-1b, encoding novel members of the COX-1 protein family. Among them only canine COX-1 and canine COX-3 possessed glycosylation-dependent cyclo- oxygenase activity. The discovery by Dan Simmons and colleagues of COX-3 will most surely herald more interest in these pathways and enzymes as recently illustrated by the number, scholarship, and import of several commentaries (3– 6). Predicted canine COX-3 is composed of COX-1 and the retained intron 1 (2). Identification of canine COX-3 might have a significant effect on understand- ing inflammatory messengers, characterizing new routes of prostanoid formation and pain conduction (3– 6). These findings may be crucial to the develop- ment of new therapeutic agents targeting both COX-2 and COX-3 in order to reduce CNS inflammatory reactions, fever, and pain. However, this may hold true only for the dog. Database analysis of human COX-1 (Fig. 1) showed a frameshift induced by intron 1, possibly revealing COX-3 to be a virtual protein in humans. However, Western blot analysis of human aorta tissue using polyclonal antibodies directed against the first 13 amino acids of the predicted human and dog/mouse COX-3 detected (surprisingly) a 65 kDa protein postulated to be human COX-3 (2). The high sequence similarity of the canine insert to intron 1 of either human or mouse COX-1 genes and the presence of 5' and 3' consensus splice sites indi- cated the complete retention of intron 1 in this tran- script. The predicted retention of a complete intron 1 sequence in human COX-1 (OMIM accession #176805) from a screening of several databases revealed a frame- shift due to insertion of intron 1 consisting of 94 nucleotides or 98 nucleotides in the murine COX-1 gene, respectively. The frameshift after the insert (see Fig. 2) results 1) in a TGA stop codon 48 amino acids later and 2) a completely changed protein sequence not related to human COX-1. The compensating mo- lecular events suggested by Chandrasekharan et al. are unlikely. COX-1 polymorphisms or sequencing errors concerning the nucleotides of intron 1 are implausible since recent screening of the human COX-1 gene, including the complete intron 1 region, revealed no polymorphisms or deletions in intron 1 (7). Moreover, compensation by “-1” ribosomal frame shifting has not been reported in eukaryotic mRNAs to date. COX-3 is considered a new and important lead in the generation of anti-inflammatory and analgesic agents. However, a frameshift caused by complete retention of intron 1 in the human sequence questions its relevance to human pathophysiology. REFERENCES 1. Botting, R. M. (2000) Mechanism of action of acetaminophen: is there a cyclooxygenase 3? Clin. Infect. Dis. Suppl. 5, 202–210 2. Chandrasekharan, N. V., Dai, H, Roos, K. L., Evanson, N. K., Tomsik, J., Elton, T. S., and Simmons, D. L. (2002) COX-3, a cyclooxygenase-1 variant inhibited by acetaminophen and other analgesic/antipyretic drugs: cloning, structure, and expression. Proc. Natl. Acad. Sci. USA 99, 13926 –13931 3. Warner, T. D., and Mitchell, J. A. (2002) Cyclooxygenase-3 (COX-3): filling in the gaps toward a COX continuum? Proc. Natl. Acad. Sci. USA 99, 13371–13373 1 These authors contributed equally to the work. 2 Correspondence: Institute of Brain Research, University of Tuebingen, Calwer Str. 3, D-72076 Tuebingen, Germany. E-mail: jmschwab@med.uni-tuebingen.de doi: 10.1096/fj.03-0595lte Figure 1. Structure of putative human COX-3. CDS = coding sequence. 2174 0892-6638/03/0017-2174 © FASEB