The possibility that the firA gene of Escherichia coli (Dicker, I. B., and Seetharam, S. (1991) Mol. Microbiol. 6, 817-823) might function in lipid A biosynthesis was examined based on its homology to the lpxA gene, which encodes UDP-N-acetylglucosamine O-acyl-transferase, the first enzyme in lipid A formation. Extracts of a temperature-sensitive firA mutant, RL-25, were assayed for their ability to acylate UDP-GlcNAc, using a coupled assay. The results suggested that extracts of RL-25 might be defective in the third enzyme of this pathway, the UDP-3-O-(R-3-hydroxymyristoyl)-glucosamine N-acyltransferase. Living cells of RL-25 also displayed a 5-fold decreased rate of lipid A biosynthesis at the nonpermissive temperature as judged by a 32Pi incorporation assay. In order to examine N-acyltransferase activity directly, the substrate [alpha-32P]UDP-3-O-(R-3-hydroxymyristoyl)-GlcN was synthesized enzymatically. N-Acyltransferase specific activity in RL-25 extracts was reduced to less than 10% of wild-type. When the wild-type firA gene was cloned into a T7-based expression vector, N-acyltransferase specific activity increased almost 360-fold relative to wild-type extracts, demonstrating that firA is the structural gene for the enzyme. The N-acyltransferase displays absolute specificity for the R-3-OH moiety of R-3-hydroxymyristoyl-ACP, as does the O-acyltransferase, consistent with the placement of R-3-hydroxymyristate in E. coli lipid A.