Mycobacterium marinum is closely related to Mycobacterium tuberculosis, the cause of tuberculosis in humans. M. marinum has become an important model system for the study of the molecular mechanisms involved in causing tuberculosis in humans. Through molecular genetic analysis of the differences between pathogenic and nonpathogenic mycobacteria, we identified two loci that affect the ability of M. marinum to infect macrophages, designated mel 1 and mel 2 . In silico analyses of the 11 putative genes in these loci suggest that mel 1 encodes secreted proteins that include a putative membrane protein and two putative transglutaminases, whereas mel 2 is involved in secondary metabolism or biosynthesis of fatty acids. Interestingly, mel 2 is unique to M. marinum and the M. tuberculosis complex and not present in any other sequenced mycobacterial species. M. marinum mutants with mutations in mel 1 and mel 2 , constructed by allelic exchange, are defective in the ability to infect both murine and fish macrophage cell lines. These data suggest that the genes in mel 1 and mel 2 are important for the ability of M. marinum to infect host cells.Although mycobacteria were among the first organisms associated with disease in humans (19, 23), they remain possibly the most important cause of death due to a single infectious agent throughout the world. Possible reasons for this are the relative difficulty of manipulating mycobacteria in the laboratory and the relatively low growth rate (ϳ20-h generation time) of the most important mycobacterial species, Mycobacterium tuberculosis. Investigators have sought appropriate models that are both relevant and easy to manipulate to speed progress in tuberculosis research. Recently, there has been a great deal of interest in Mycobacterium marinum as a model for study of M. tuberculosis pathogenesis because of its ease of genetic manipulation (2, 17, 33, 37), close genetic relationship to M. tuberculosis (36,42,46), relatively high growth rate (ϳ4-h generation time) (10), and the presence of a number of useful laboratory models for in vitro (16) and in vivo (14,34,38,45) virulence studies.Pathogenic mycobacterial species, such as M. tuberculosis, differ from nonpathogenic species, such as Mycobacterium smegmatis, in that they invade mammalian cells more efficiently (6, 13, 39), block lysosomal fusion (4, 20), and replicate well in eukaryotic cells (29,40,47,48). Investigators have taken advantage of these differences to identify the genes involved in host cell interactions by cloning genomic DNA from pathogenic species into nonpathogenic mycobacterial species (6,29,47,48). Although these studies have resulted in identification of genes that are potentially important, further analyses have been slowed by the fact that they were isolated from slowgrowing pathogenic mycobacterial species. As a result, no strains have been constructed with mutations in these genes, and it remains unclear whether mutants would be defective for host cell interactions. Our group recently found that the relatively...