AUTHOR'S CORRECTION
Volume 193, no. 17, pages 4346–4360, 2011. Page 4355: We erroneously stated, “The only genes encoding polyamine biosynthetic enzymes in L. pneumophila were metK (methionine adenosyltransferase) and speA (arginine decarboxylase).” As recently pointed out by Dr. A. J. Michael (University of Texas Southwestern Medical Center), L. pneumophila possesses the enzyme homospermidine synthase (F. L. Shaw et al., J. Biol. Chem. 285:14711–14723, 2010) and accumulates the polyamine sym-homospermidine (K. Hamana and M. Takeuchi, Microbiol. Cult. Collect. 14:1–14, 1998). The other two enzymes encoded in the arginine decarboxylase (speA) locus in L. pneumophila genomes are agmatine deiminase and N-carbamoyl putrescine amidohydrolase. We did not account for these three enzymes, mainly because in our bioinformatics analysis we looked only for the conventional polyamine biosynthetic enzymes present in Escherichia coli and Vibrio cholerae. We thus acknowledge here the presence of a complete putrescine-homospermidine alternative biosynthetic pathway in Legionella. Our suggestion that “L. pneumophila cannot synthesize all polyamines” (end of p. 4355) still stands, and our finding that L. pneumophila grows better in the presence of putrescine, spermidine, and/or spermine is not affected by this correction.