Characterization of Selected Attenuated Mutants 
Genomic DNA was recovered from the 12 group 1A isolates.
The transposon insertion site was determined through direct genomic DNA sequencing, using primers located upstream from each transposon inverted repeat (see Materials and Methods).
The sequences obtained were compared to the PAO1 genome sequence database (http://www.pseudomonas.com) [22] followed by comparison with the general sequence database.
The genes identified are shown in Table 1.
Two potential regulatory genes were found.
The first, PA2588, encodes an AraC-type regulatory protein, with 48% similarity to the PAO1 PA0831 gene product, OruR [36].
The second matched PA4380, which encodes a protein with 76% similarity to Pseudomonas fluorescens ColS, a histidine kinase sensor from a two-component regulatory system.
In P. fluorescens ColS has been implicated in root colonization while in Pseudomonas putida it has been reported to be involved in regulating TN4652 transposition and heavy metal resistance [37],[38],[39],[40].
Another gene was identified as PA4554, which encodes the type 4 fimbrial adhesin PilY1 [41].
We recovered two additional mutants with transposon insertion into genes which have functions related to motility.
First, PA4953 encodes a protein with similarity to the E. coli MotB, one of a pair of proteins, which contributes to flagellar rotation [23],[42],[43].
Secondly, we found an insertion in PA0173 (cheB2), one of four cheB gene homologues in the P. aeruginosa genome sequence.
PA0173 lies within the chemotactic gene cluster II found in the PAO1 genome (Figure 1) [21].
CheB proteins are responsible for removing methyl groups from MCPs [44].
The adaptation of the chemotaxis system in response to changes in attractant binding is dependent on the methylation state of the MCPs.
CheB2 has been proposed to be an essential component for an optimal chemotactic response [21].
In the remaining mutants, one contained a transposon insertion in PA5479, which encodes a protein similar to GltP from E. coli, a glutamate-aspartate carrier protein.
Another mutant had an insertion in PA2585 encoding the UvrABC endonuclease subunit UvrC, required for the excision of damaged DNA.
The uvrC gene is located downstream from the gacA gene encoding a response regulator known to be involved in P. aeruginosa virulence.
Since, in addition to the insertion in PA2585, we isolated a mutant with an insertion in a neighbor gene PA2588, it suggests that this cluster of genes might be required for virulence.
Another interrupted gene, PA2478, encodes a probable thiol:disulfide interchange protein of the DsbD family.
Finally, 4 interrupted genes, PA0946, PA3080, PA0260 and PA2769, encode proteins of unknown function.
In order to compare our liquid assay with the established C. elegans slow killing assay, we compared the lethality of the TB strain with the previously tested PA14 strain.
We observed that both strains kill worms with similar efficiency (data not shown).
Each of the 12 group 1A mutants derived from TB was further tested in the slow killing assay.
The parental TB isolate is able to kill 50% of exposed nematodes in the slow killing assay within 3 days.
Seven of the twelve group 1A isolates showed significant virulence attenuation (Table 1).
For example, three group 1A mutants TB0173s (cheB2, p<0.001), TB4554s (pilY1, p<0.001) and TB4380s (colS, p<0.001) each required at least one additional day (4 days) in order to reach 50% killing (data not shown).
Survival curves are considered significantly different from the control when p-values are <0.05 (see Materials and Methods).
