Tuberculosis (TB) still remains a worldwide health concern. Efforts to understand the complex biology of Mycobacterium tuberculosis, the causative agent of TB, are important for new antitubercular drug development. Despite the completion of the genome sequence and the development of new genetic tools to manipulate this organism, the availability of sufficient amounts of mycobacterial proteins still remains an essential and laborious step to study the biochemical features of this pathogen. The T7-RNA polymerase-based pET system has been largely employed to express mycobacterial proteins in Escherichia coli, but it presents some limitations. To overcome problems with unstable expression of an M. tuberculosis inhA-encoded enoyl reductase mutant protein and lack of expression of two mabA-encoded ketoacyl reductase mutants, a sub-population of E. coli BL21(DE3) host cells was selected from a small-opaque colony. This empirically selected host, named BL21(DE3)NH, allowed stable expression of these mutant proteins. Although the mechanism that led the BL21(DE3)NH host to express the recombinant mutant proteins remains unknown, the persistent phenotype points to a stable genetic switch. This genetic alteration resulted in a tight control of the highly processive T7 RNA polymerase. Moreover, the absolute requirement for IPTG to obtain protein expression in the BL21(DE3)NH host cells suggests that no inherent defect in the transcriptional activity of the T7 promoter is present. Empirical host selection requires no further genetic manipulation of recombinant plasmids and may represent a means of obtaining tailor-made E. coli strains that overcome toxic effects associated with heterologous protein expression. (C) 2003 Elsevier Inc. All rights reserved.