Proteins secreted by Mycobacterium tuberculosis may play a key role in virulence and may also constitute
antigens that elicit the host immune response. However, the M.
tuberculosis protein export machinery has not been characterized. A library of M.
tuberculosis H37Rv genomic
DNA fragments ligated into a
signal sequence selection vector that contained a leaderless
beta-lactamase gene and an upstream Tac promoter was constructed. Transformation of Escherichia coli with the M.
tuberculosis DNA library and selection on plates containing 50-100 micrograms
ampicillin ml-1 resulted in the identification of 15 Ampr clones out of a total of 14,000 transformants. Twelve of the
beta-lactamase gene fusions conferred high levels of Ampr (up to 1 mg
ampicillin ml-1); insert sizes ranged from 350 to 3000 bp. Of ten inserts that were completely sequenced, two were identified as fragments of the genes for M.
tuberculosis antigens 85A and 85C, which are the major secreted
proteins of this pathogen. Seven of the remaining inserts were > or = 97% identical to hypothetical ORFs in the M.
tuberculosis genome, one of which encoded a
protein with 35% identity to a low-affinity
penicillin-binding protein (PBP) from Streptomyces clavuligerus. Four of the seven hypothetical ORFs encoded putative exported
proteins with one or more membrane interaction elements, including
lipoprotein attachment sites and type I and II transmembrane (TM) segments. All of the inserts encoded typical
signal sequences, with the exception of a possible type II
membrane protein. It is concluded that expression of
beta-lactamase gene fusions in E. coli provides a useful system for the identification and analysis of M.
tuberculosis signal-sequence-encoding genes.