Mycobacterium tuberculosis continues to pose a serious health threat to people throughout the world. With an estimated 2.9 million people having died from tuberculosis in 1997, M. tuberculosis remains the infectious agent responsible for the most deaths worldwide (64). In order to develop new drugs, treatments, and vaccines for tuberculosis, a better understanding of M. tuberculosis pathogenesis and M. tuberculosis antigens that elicit a protective immune response is required. Secreted and cell envelope-associated proteins are likely to play a critical role in M. tuberculosis disease. Research on several bacterial pathogens has revealed that the majority of virulence factors are secreted (22), and it was recently shown that the secreted ERP protein contributes to the virulence of M. tuberculosis (6). A key component of M. tuberculosis infection is the ability of the bacillus to survive within phagocytic cells. Since secreted and surface proteins are ideally positioned to interact with the host, they could facilitate this survival by influencing phagosome maturation, by enabling access to the cytoplasm, or by countering the antimicrobial attacks of the phagocyte. Secreted proteins of M. tuberculosis also play an important role in the generation of a protective immune response. The most striking demonstration of this property comes from experiments in which mice or guinea pigs were immunized with extracellular proteins and significant protective immunity ensued (2, 32, 33, 45, 51). It has been reported that M. tuberculosis secretes an extensive number of proteins, and there are even more proteins that must be cell surface associated (3, 50, 56). Many of these proteins have not yet been identified and even fewer have been tested for a role in virulence and protective immunity. PhoA (Escherichia coli alkaline phosphatase) protein fusions have been used in many different organisms to identify exported proteins, including ones that are important to bacterial virulence (7, 17, 26, 36, 49). Importantly, PhoA has been shown to function as a reporter for secreted proteins in M. smegmatis (60). Moreover, when a multicopy plasmid library of phoA fusions to M. tuberculosis genomic DNA was screened in M. smegmatis, four active PhoA fusions were identified, one of which involved the ERP protein (see above) (40). Typically, phoA fusions are made either by using in vivo transposition with transposons such as TnphoA (a derivative of Tn5) or by cloning genomic DNA upstream of a truncated ′phoA gene (43). The ′phoA gene in both of these cases lacks signals for expression and export. Since PhoA is active only when it is located outside of the cytosol, enzymatically active PhoA fusions identify proteins that have export signals. A simple and efficient in vitro transposition system for Tn552, a transposon first identified in Staphylococcus aureus (52), has recently been developed (25, 39). An extensive analysis of transposon insertion sites produced by this in vitro system demonstrated that the transposon inserts randomly within different target DNAs, including mycobacterial DNA (25). Thus, it appears to be ideal for constructing large complex libraries of gene fusions. We set out to use this system to create a random library of phoA fusions in cosmids containing M. tuberculosis genomic DNA to identify genes encoding exported proteins. We constructed Tn552′phoA transposons and inserted them into cosmids containing M. tuberculosis DNA by using in vitro transposition reactions. The resulting population of transposon-containing cosmids was integrated into the M. smegmatis genome in single copy, and colonies were screened for active phoA fusions. To our knowledge, this is the first report of an in vitro transposition system for producing phoA fusions. It provides a simple alternative to constructing libraries by the standard methods. Initial screening of our M. tuberculosis DNA-′phoA fusion libraries identified 31 secreted and membrane proteins. An additional feature of this system is that cosmids containing a transposon insertion can later be used as a substrate for allele exchange in M. tuberculosis. This should enable a relatively rapid transition from the identification of an exported protein to the construction of the corresponding M. tuberculosis mutant and evaluation of its role in pathogenesis and protective immunity.