Antimicrobial Studies with the Pseudomonas aeruginosa Two-Allele Library Require Caution

The increasing prevalence of antimicrobial resistance requires a continuous effort to find new antimicrobials. The availability of genome sequences for a variety of microbes has facilitated the development of whole-genome, single-gene knockout libraries for systematic gene function analysis. These libraries, such as the Keio library of Escherichia coli (1) and the University of Washington two-allele library of Pseudomonas aeruginosa (3), can be used in high-throughput phenotypic array studies (7); thus, they seem well suited for identifying new antimicrobial targets. However, we recently noticed with the P. aeruginosa two-allele mutant library that intrinsic, unidentified factors may affect susceptibility for some antimicrobials with many randomly picked mutants, thereby obscuring effects attributed to specific candidate mutants.

In E. coli, a phoU knockout mutation confers hypersensitivity to a variety of antimicrobials and other stresses (5). To determine whether inactivation of a phoU homologue in P. aeruginosa confers a similar phenotype, we obtained a phoU mutant and the parental wild-type strain (PAO1, lacking a transposon insertion) from the University of Washington two-allele library. When quinolone susceptibility was tested, a 4- to 16-fold decrease in MIC was observed for both ciprofloxacin and oxolinic acid (Table 1), as expected (5). However, when 10 unrelated mutants were tested, each of those mutants also exhibited hypersusceptibility to quinolones (Table 1). Moreover, PCR analysis failed to confirm transposition being inside the phoU allele. Thus, increased susceptibility of the phoU mutant could not be specifically attributed to the phoU mutation.

TABLE 1.

Antimicrobial susceptibility of mutants obtained from the P. aeruginosa two-allele library

Straina	Allele/PA no.b	Insertion sitec	Colony colord	MIC (μg/ml)e
				CIP	OXO	TOB	CHL	MEM	CAZ
PAO1	Wild type/NA	None	+	0.8	64	0.2	800	0.38	2
33410	phoU/PA5365	259 (729)	+++	0.05	4	0.4	12.5	0.38	1.5
30848	phoU/PA5365	−50/135 (729)	+++	0.2	16	0.4	200	0.5	3
32455	phzS/PA4217	313/523 (1,209)	Red	0.1	4	0.4	12.5	0.25	2
3208	phzM/PA4209	551 (1,005)	Yellow	0.1	4	0.2	12.5	0.38	2
3434	NA/PA0125	218 (228)	++	0.05	4	0.4	25	0.38	2.5
3894	NA/PA0729	109 (348)	++	0.05	8	0.4	12.5	0.19	2.5
1494	phnA/PA1001	790 (1,593)	+++	0.1	8	0.4	12.5	0.38	1.5
40198	nirJ/PA0511	899 (1,164)	+	0.1	8	0.4	12.5	0.25	1.5
8564	pyrX, C′/PA0401	+750/+224	+	0.1	8	0.2	12.5	0.38	1.5
9256	ybhO/PA2155	47 (1,206)	+	0.1	8	0.4	12.5	0.25	2
8666	NA/NC	NA	+	0.1	8	0.2	12.5	0.25	2
3465	tsaA/PA3529	347 (603)	++	0.1	4	0.2	12.5	0.25	2
312	PBPP/PA0617	34 (327)	+++	0.1	8	0.2	12.5	0.25	2
46396	katA/PA4236	454 (1,449)	++	0.1	8	0.2	12.5	0.25	2

^aThese P. aeruginosa strains were chosen for having transposon insertions at a variety of chromosome locations, either inside an open reading frame or outside (both up- and downstream) an open reading frame. The numbers listed are University of Washington two-allele library identification numbers.

^bP. aeruginosa (PA) numbers indicate open reading frames of the P. aeruginosa genome; no PA number was assigned to the wild-type strain (PAO1) or to the transposition site-uncharacterized strain, 8666. The same results as those for the wild-type strain were obtained with a second PAO1 isolate from an independent source. PA5365 (strain 33410) and PA0125 were originally proposed to have transposon insertions in/near them, but this could not be confirmed in the present study. Abbreviations: NA, not available; NC, transposition site not characterized (served as a random control in the present study); PBPP, probable bacteriophage protein.

^cInsertion site at the nucleotide number within the open reading frame is listed; the gene length (number of nucleotides) is in parentheses. The originally proposed insertion locations for strains 33410 and 3434 could not be confirmed. The transposon insertion sites for strains 30848, 32455, and 8564 were confirmed in/near the proposed allele but at a different position from the original report (3). In such cases, the number before the slash is the approximate location revealed in the present study, and the number after the slash is the proposed location reported originally. The transposon insertion sites for strains 3208, 3894, 1494, 3465, and 312 were confirmed by PCR in the present study. The transposon insertion sites for strains 3208, 3894, 40198, 8564, 9256, 3465, 312, and 46396 were listed as confirmed on the original study group website. + and − indicate that the insertion site is downstream and upstream, respectively, of the open reading frame listed. NA, not available.

^d+, light greenish blue; ++, greenish blue; +++, dark greenish blue.

^eMICs for ciprofloxacin (CIP), oxolinic acid (OXO), tobramycin (TOB), and chloramphenicol (CHL) were determined by broth dilution in triplicate. MICs for meropenem (MEM) and ceftazidime (CAZ) were obtained by Etest.

Colony color with most of these mutants was darker (bluish green) than with the wild-type strain, indicating production of more phenazine pigments in the mutants. Since phenazines, such as pyocyanin, may be involved in oxidative stress and thereby affect antimicrobial susceptibility (2), we tested two additional mutants (the phzS and phzM mutants) that do not produce pyocyanin (6). Reduced MICs, similar to those seen with the other mutants, were observed. Thus, variation in pyocyanin production was not responsible for reduced quinolone susceptibility. Since the mutants tested had transposon insertions at a variety of chromosomal locations, some of which were far outside open reading frames, transposon-mediated polarity is also an unlikely explanation for reduced MIC. When other antimicrobial classes were tested, hypersusceptibility was observed with chloramphenicol (4- to 64-fold) but not with tobramycin, meropenem, or ceftazidime (Table 1). Thus, the mutants tested were universally hypersusceptible to some, but not all, antimicrobials for reasons that are unknown. This lack of specificity may not be unique to the University of Washington library, because several lux-insertional mutants obtained from a different library (4) also conferred quinolone hypersusceptibility (data not shown).

The results described above suggest that the transposition/insertion of a DNA fragment into the P. aeruginosa chromosome may itself have a general effect on susceptibility for some, but not all, antimicrobials. Had we simply compared the phoU mutant with the parental wild-type strain, which is standard practice with many antimicrobial studies, we would have been misled. Thus, interpretation of antimicrobial experiments involving mutants from the P. aeruginosa insertional libraries requires caution.