Main conclusionAnthoceros agrestis hydroxycinnamoyltransferase accepts shikimic and 3-hydroxyanthranilic acids while hydroxycinnamoylester/amide 3-hydroxylase (CYP98A147) preferred p-coumaroyl-(3-hydroxy)anthranilic acid compared to the shikimic acid derivative. Alternative pathways towards rosmarinic acid have to be considered.Rosmarinic acid (RA) is a well-known ester of caffeic acid and 3,4-dihydroxyphenyllactic acid. In the search for enzymes involved in RA biosynthesis in the hornwort Anthoceros agrestis, the hydroxycinnamoyltransferase sequence with the highest similarity to rosmarinic acid synthase from Lamiaceae has been amplified and heterologously expressed in Escherichia coli. In parallel, the single cytochrome P450 sequence belonging to the CYP98 group in Anthoceros agrestis was isolated and expressed in Saccharomyces cerevisiae which did not result in protein formation. Codon optimization and co-expression with NADPH:cytochrome P450 reductase (CPR) from Coleus blumei resulted in the formation of active enzymes. Both, the hydroxycinnamoyltransferase and CYP98 were characterized with respect to their temperature and pH optimum as well as their substrate acceptance. The hydroxycinnamoyltransferase (AaHCT6) readily accepted p-coumaroyl- and caffeoyl-CoA with a slightly higher affinity towards p-coumaroyl-CoA. The best acceptor substrate was shikimic acid (Km 25 µM with p-coumaroyl-CoA) followed by 3-hydroxyanthranilic acid (Km 153 µM with p-coumaroyl-CoA). Another accepted substrate was 2,3-dihydroxybenzoic acid. Anthranilic acid and 4-hydroxyphenyllactic acid (as precursor for RA) were not used as substrates. p-Coumaroylesters and -amides are substrates hydroxylated by CYP98 hydroxylases. The only CYP98 sequence from Anthoceros agrestis is CYP98A147. The best substrates for the NADPH-dependent hydroxylation were p-coumaroylanthranilic and p-coumaroyl-3-hydroxyanthranilic acids while p-coumaroylshikimic and p-coumaroyl-4-hydroxyphenyllactic acids were poor substrates. The biosynthetic pathway towards rosmarinic acid thus still remains open and other enzyme classes as well as an earlier introduction of the 3-hydroxyl group to afford the caffeic acid substitution pattern must be taken into consideration.