4-Hydroxyproline is found in collagens and collagen-like proteins in animals and in many glycoproteins in plants. Animal prolyl 4-hydroxylases (P4Hs) have been cloned and characterized from many sources, but no plant P4H has been cloned so far. We report here that the genome of Arabidopsis thaliana encodes six P4H-like polypeptides, one of which, a 283-residue soluble monomer, was cloned and characterized here as a recombinant protein. Catalytically critical residues identified in animal P4Hs are conserved in this P4H, and their mutagenesis led to complete or almost complete inactivation. The recombinant P4H effectively hydroxylated poly(l-proline) and many synthetic peptides corresponding to proline-rich repeats present in plant glycoproteins and other proteins. Surprisingly, collagen-like peptides were also good substrates, the V(max) with (Pro-Pro-Gly)(10) being similar to that with poly(l-proline). The enzyme acted in this peptide preferentially on prolines in Y positions in the X-Y-Gly triplets. Correspondingly, (Gly-Pro-4Hyp)(5) and (Pro-Ala-Gly)(5) were poor substrates, with V(max) values less than 5 and 20% of that obtained with (Pro-Pro-Gly)(10), respectively, the K(m) for the latter also being high. Peptides representing the N- and C-terminal hydroxylation sites present in hypoxia-inducible transcription factor alpha also served as substrates. As these peptides contain only one proline residue, a poly(l-proline) type II conformation was clearly not required for hydroxylation.