Pools of mutants of minimal complexity but maximal coverage of genes of interest facilitate screening for genes under selection in a particular environment. We constructed individual deletion mutants in 1,023 Salmonella enterica serovar Typhimurium genes, including almost all genes found in Salmonella but not in related genera. All mutations were confirmed simultaneously using a novel amplification strategy to produce labeled RNA from a T7 RNA polymerase promoter, introduced during the construction of each mutant, followed by hybridization of this labeled RNA to a Typhimurium genome tiling array. To demonstrate the ability to identify fitness phenotypes using our pool of mutants, the pool was subjected to selection by intraperitoneal injection into BALB/c mice and subsequent recovery from spleens. Changes in the representation of each mutant were monitored using T7 transcripts hybridized to a novel inexpensive minimal microarray. Among the top 120 statistically significant spleen colonization phenotypes, more than 40 were mutations in genes with no previously known role in this model. Fifteen phenotypes were tested using individual mutants in competitive assays of intraperitoneal infection in mice and eleven were confirmed, including the first two examples of attenuation for sRNA mutants in Salmonella. We refer to the method as Array- based analysis of cistrons under selection (ABACUS).
One strategy to define bacterial genes with a role during infection involves the screening of a pool of random transposon insertion mutants, where each mutant is identifiable by a specific motif or unique transcript. Changes in the survival of each mutant indicate a role of the mutated region during infection. To ensure coverage of most genes of interest, a large number of random transposon mutants would be needed. However, when a pool of bacterial mutants spreads into different sites in an animal host, some mutants are lost at random if the founder population (i.e. the bacteria initially reaching this site, before expansion) is smaller than the number of mutants introduced into the animal. This random loss severely obscures mutants that are truly at a disadvantage. In order to minimize the number of mutants to be screened and thus minimize random loss of mutants from an infecting pool, we generated targeted specific deletions in each gene of interest. Furthermore, we inserted a promoter in each mutant that allows simultaneous monitoring of a pool of these mutants on a novel inexpensive microarray. To demonstrate the utility of our technique, a pool of over 1,000 gene-targeted Salmonella mutants was injected into mice and recovered from spleens. We identified mutants that were less fit than wild-type Salmonella in this model. Eleven mutants in genes that were not previously known to affect Salmonella fitness in intraperitoneal infection were confirmed using the individual mutants in competition with wild-type bacteria. These new phenotypes include the first two examples of attenuation and one of hypervirulence in Salmonella due to mutations in small stable RNAs, a class of regulators that bind to other RNAs and proteins.