Nucleic Acids Research

To investigate the possibility that the unusual dominant rifampicinresistance characteristic of the riP ^ allele of E.coli rpoB is due to a secondary, regulatory mutation, we have determined the nudeotide sequence of a 1.1 Kbp wild-type DNA fragment, including the transcriptional attenuator and translational start-site of rpoB. We have also re-investigated the previously published sequences of this region in X£if18 and Xrif47 DNA. Our results indicate that all three sequences are identical, and reveal some, errors in the published data. We discuss the basis of dominance of rif 18. INTRODUCTION The entire DNA nudeotide sequence of the rplKAJLrpoBC operon, encoding ribosomeand RNA polymerase subunits in Escherichia coli, has been reported by Post et al (1), with extension and correction by Delcuve et al (2); together with Gurevitch et al (3), Ovchinnikov and Colleagues (4,5), and Squires et al (6). The Soviet groups (3-5) studied DNA derived from xrif47 (7); the rest, xrlf/18 (8). These rpoBC-transducing phages carry (different) mutations of the rpoBC operon which have the unusual property of conferring not only resistance to rifampicin, but dominance over the wildtype (rpoB , RifS) allele in terms of drug response. The sequence of wildtype DNA has been reported (9) only for the EcoRI-generated internal fragment of rpoB, basepairs 3528 through 6401(Fig.1). ( We use the Post et al scale (1) as corrected in (2), and adjusted for two other discrepancies discussed below). The only reported differences between wild-type, rif 18 and ri^47 DNA within this fragment are shown in Table 1 (base pairs 4516 and 4561), and are believed to generate the amino acid changes in the B-subunit of RNA polymerase which lead to rifampicin-resistance. The published work also suggests that, upstream of the above EcoRI fragment, there is a surprising deletion of AGC in rif*H8 as compared with rif°^47, and that the "tail" sequences of the transcriptional attenuator which lies between rplL and rpoB (10) may differ as between these two alleles (Table 1). i IRL Press Limited, Oxford, England. 5465 Nucleic Acids Research We have previously reported evidence (11) which suggested that the dominance property of \rifi8 rpoBC DNA might be genetically separable from the rifampicin-resistance mutation at base pair 4561. Specifically, a xderivative carrying the HindIIIrplJL rpoBC fragment of Arif 18 (constructed in vitro) was recombined r̂i vivo into rpoB (RifS) DNA originally derived from E.coli AJ1 (12). The lysogen was then induced, and aberrant excision products were selected; among these was AAJN261, which had inherited a functional promoter for rplJL rpoBC from the chromosome. This phage expresses the rifampicin-resistance, but not the dominance property of ArifH8 (11). This finding, together with other unpublished, preliminary data led us to consider the possibility that dominance arises from a closely linked, cis-acting regulatory "up" mutation, located upstream of the rifampicin resistance mutation. This could cause dominance by increasing the proportion of drug-resistant molecules among the RNA polymerase population. Qie possible mutation of this type would lead to an improved ribosomebinding sequence for initiation of rpoB translation (or reduced feedback regulation of this translation); another possible class would reduce the efficiency of transcriptional termination at the attenuator between rplL and rpoB. The reported difference in attenuator sequence between Arii: 18 and A£if/47 DNA (1, 3) lent some credibility to the latter model (see Table 1). We have tested these ideas by sequencing of the appropriate DNA. EXPERIMENTS AND DISCUSSION We have compared the DNA sequences of the rpoBC-attenuator region from four existing clones in X: xrif^47 (7); ArjLf 18 (8); AAJN63, carrying a Hindlll-'rplJL rpoBC fragment from the wild-type (RifS) chromosome of E.coli K12 strain CR63 (11); and AAJN261 (11), obtained as described above via recombination between DNA derived from X£i£ 18 and from the chromosome of E.coli AJ1 (rpoB , Rifs). Appropriate EcoRI Sail fragments were subcloned into M13 mpiO and -mp11 (13), and sequenced by a recent modification of Sanger's dideoxynucleotide approach (14). In this way sequences were determined for at least one strand throughout the EcoRI (' rpllz-attenuatorrpoB') fragment (basepairs 2444 through 3533) of Ari^TB, AAJN63, and AAJN261; and for both strands between the following limits: wild-type (CR63) DNA, basepairs 2645 to 2731 and 3175 to 3318; xrjLf̂ iS DNA, 2534 to 2836 and 3200 to 3348; and AAJN261 DNA, 2659 to 2785 and 3188 to 3291. Note that the attenuator region (circa 2660 to 2725) and the AQC at 3286 to 3288 have been determined on both strands. Note also that two independent pre-