Amino acid sequence and composition data of Escherichia coli dnaG primase protein and its tryptic peptides have confirmed that the dnaG gene contains an unusually high number of codons that are not frequently used in most E. coli genes. In 25 E. coli proteins analyzed the codons AUA, UCG, CCU, CCC, ACG, CAA, AAT, and AGG are infrequently used, occurring as 4% of the total codons in the reading frame and 11% and 10% in the nonreading frames. In dnaG they occur as 11% in the reading frame and 12% in the nonreading frames. The rpsU and rpoD genes, which flank the dnaG gene [Smiley, B. L., Lupski, J. R., Svec, P. S., McMacken, R. & Godson, G. N. (1982) Proc.Nati Acad Sci USA 79, 4550-4554], however, have normal codon usage. Translational modulation using isoaccepting tRNA availability may therefore be part of the mechanism of keeping the dnaG gene expression low, while expression of the adjacent rpsU and rpoD genes on the same mRNA transcript is high.Nucleotide sequences, which are becoming available in increasing numbers, have clearly indicated that codon usage is not random and that, within an organism, certain synonymous codons are preferred to others. The spectrum ofsynonymic codon preference appears to differ from organism to organism and this has now been shown to reflect the pattern of tRNA isoaccepting species expressed in that organism (1).Because there are frequently and less-frequently expressed members of tRNA isoaccepting species within a cell, the lessfrequently expressed tRNAs can theoretically be used to modulate translation of protein expression. This was proposed by Fiers and Grosjean from an examination of the codon usage of the bacteriophage MS2 RNA (2) but has never been experimentally verified.It has also been observed that codon usage within the known coding reading frame ofa nucleotide sequence is different from that observed in the two noncoding reading frames. Analysis of codon usage had therefore been proposed by several groups to be a method of identifying the correct coding reading frame within a nucleotide sequence and also as a method ofidentifying nucleotide sequencing errors (3, 4). This paper, however, demonstrates that the Escherichia coli dnaG primase gene does not conform to the codon usage of either the organism or its flanking genes in the macromolecular synthesis operon [dnaG and rpoD (5) and rpsU4K]. The dnaG gene contains an unusually large number of infrequently used synonymic codons (rare codons) and within the gene there are long stretches of nucleotide sequence that contain more rare codons in the coding reading frame than in the noncoding reading frames. The use ofrare codons in the dnaG gene, therefore, may be part of the mechanism to maintain its expression at low copy number compared with the flanking genes on the same mRNA transcripts as shown by Smiley and colleagues (5,6). This is supported by the observation that the E. coli repressor genes lacI, trpR, and araC, which are also expressed in low amounts, contain unusually high numbers of rare codons.MATER...