Use of functional interactions with MarA to discover chromosomal genes affecting antibiotic susceptibility in Escherichia coli

Sir,

Antibiotic action and cell responses to antibiotics are very complex and include not only the direct target of an antibiotic but also secondary targets, signalling effects and indirect effects such as antibiotic-induced generation of hydroxyl radicals or changes in gene expression and cell metabolism. Identifying genes important for antibiotic action in the cell or antibiotic resistance would uncover new targets for potentiating existing antibiotics or for discovering new ones. Several approaches have been used to identify these genes, including studying antibiotic-resistant strains of clinical origin or selected in the laboratory and, more recently, screening the entire Keio collection of 4000 single-gene knockouts of Escherichia coli for increased antibiotic susceptibility [1]. Such approaches have been successful but have limitations.

Here we propose a new approach to identify genes affecting antibiotic resistance or susceptibility based on screening genes that have functional interactions with other genes already known to be important for drug resistance. Such an approach was suggested by recent studies [2] with the multiple antibiotic resistance regulator MarA, a transcription factor in E. coli that upregulates multidrug efflux and downregulates membrane permeability [2]. The effect of inactivation of 24 unrelated chromosomal genes using the EZ-Tn5™ <Kan-2> transposome (Epicentre, Madison, WI) on antibiotic resistance mediated by MarA was studied [2]. Fifteen of these genes affected MarA-mediated antibiotic resistance, whereas nine of them did not. Here we examine whether these 24 inactivated genes affect antibiotic susceptibility per se in the absence of MarA.

Antibiotic susceptibility to cefoxitin, norfloxacin, chloramphenicol and minocycline of the parental strain CR2000 (E. coli BW25113 ΔmarRA [2]) and of its 24 gene-inactivated mutant derivatives was determined by Etest as described previously [2]. Briefly, the different strains were grown overnight in Luria–Bertani (LB) medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) at 37 °C with agitation and their minimum inhibitory concentrations were determined using Mueller–Hinton plates and Etest strips (AB BIODISK, Solna, Sweden) according to the manufacturer’s specifications following incubation for 20 h at 37 °C. The results for each mutant were compared with those of the parental strain by t-test (two independent samples with equal variance, two-tailed distribution, performed using Microsoft^® Excel 2003 software) to determine statistically significant differences.

Nine (60%) of the fifteen genes functionally related to MarA-mediated antibiotic resistance also affected antibiotic susceptibility/resistance independently from MarA. Such a double role might help cells to have a more co-ordinated response to antibiotics. In contrast, only two (22%) of the nine genes that do not affect MarA-mediated antibiotic resistance affected antibiotic susceptibility/resistance (Table 1).

Table 1.

Effect of inactivation of chromosomal genes functionally related or unrelated to MarA-mediated antibiotic resistance on antibiotic susceptibility in Escherichia coli

Gene	Locusa	Functional categoryb	Protein/function	n-fold change in MIC compared with parental strainc
				FOX	NOR	CHL	MIN
Genes whose inactivation affects MarA-mediated antibiotic resistance
crp	b3357	R	Cyclic AMP receptor protein	1.3	1.4	1.2	0.7
hns	b1237	R	DNA-binding protein H-NS	1.4	1.2	0.3	0.2
cyaA	b3806	E	Adenylate cyclase	1.1	1.4	0.9	0.9
degP	b0161	E	Serine protease Do	0.7	0.8	1.1	0.8
maoC	b1387	E	Putative ring-cleavage enzyme of phenylacetate degradation	1.1	0.8	0.9	1.4
pcnB	b0143	E	Poly(A) polymerase I; RNA modification	1.7	1.0	1.2	1.3
recD	b2819	E	RecBCD DNA helicase/exonuclease	1.2	0.8	0.9	1.4
treC	b4239	E	Trehalose-6-phosphate hydrolase	1.1	1.2	0.9	1.2
acrA	b0463	T	AcrAB–TolC multidrug efflux system	0.4	0.4	0.1	0.2
acrB	b0462	T	AcrAB–TolC multidrug efflux system	0.5	0.4	0.1	0.2
tolC	b3035	T	TolC outer membrane channel	0.5	0.3	0.1	0.2
nikD	b3479	T	Subunit of nickel ABC transporter	1.1	1.0	0.9	1.8
damX	b3388	O	Predicted membrane-anchored protein	1.0	1.0	1.0	1.0
yibLpd	b3602	O	Conserved protein	1.0	0.8	0.9	1.2
yniDpd	b4535	O	Predicted unconserved protein	1.0	0.9	0.9	1.2
Genes whose inactivation does not affect MarA-mediated antibiotic resistance
cspGpd	b0990	R	Cold-shock protein CspG	0.9	0.8	0.8	1.2
ompR	b3405	R	OmpR transcriptional dual regulator	5.8	1.6	1.2	1.3
appA	b0980	E	Acid phosphatase	1.0	1.0	0.9	1.1
metL	b3940	E	Aspartate kinase/homoserine dehydrogenase	0.7	1.0	1.0	1.3
ynfE	b1587	E	Oxidoreductase subunit paralogue of DmsA	1.0	1.0	0.9	1.1
mhpT	b0353	T	Putative propionic acid transporter	1.0	0.8	1.0	0.9
alx	b3088	O	Putative membrane protein	1.0	1.0	1.0	1.0
metV	b2816	O	tRNA–methionine	1.1	0.8	1.0	1.1
yfdG	b2350	O	CPS-53 (KpLE1) prophage; bactoprenol-linked glucose translocase (flippase	1.0	1.0	0.9	1.3

MIC, minimum inhibitory concentration; FOX, cefoxitin; NOR, norfloxacin; CHL, chloramphenicol; MIN, minocycline.

^aThe locus tags refer to the accession number (http://biocyc.org/ecocyc/index.shtml).

^bR, transcriptional regulator; E, enzyme; T, transport; O, other/unknown function.

^cThe MICs of the different strains are shown as the average (parental, n = 14; mutants, n = 3 6) and normalised to (divided by) those of the parental strain CR2000 (cefoxitin MIC = 3.1 mg/L, norfloxacin MIC = 0.08 mg/L, chloramphenicol MIC = 7.7 mg/L and minocycline MIC = 2.5 mg/L); that is, they are shown as the n-fold change in MIC (n-fold = 1 corresponds to no difference in the MIC between a mutant and the parental strain). Values in bold indicate when inactivation of a gene significantly (P < 0.01) affected the MIC compared with the parental strain. The MICs obtained using bactericidal antibiotics (FOX and NOR) were less variable than those obtained using bacteriostatic antibiotics (CHL and MIN), which might be related to the different general mechanism of action of the two groups of antibiotics. This different variability explains why MIC differences of ≥30% were significant for cefoxitin and norfloxacin, whereas only differences of ≥50% were significant for chloramphenicol and minocycline.

^dInsertion of the transposome is in the promoter of the gene. All the other insertions are within the open-reading frame (ORF) of the genes [2].

Those genes that affect both antibiotic resistance mediated by MarA and antibiotic susceptibility/resistance independently from MarA were as follows. First, acrA, acrB and tolC, whose inactivation strongly increased susceptibility to all the antibiotics tested (Table 1), as expected [1]. Second, crp, cyaA, hns and degP, whose inactivation increased resistance to cefoxitin (crp and hns) or norfloxacin (crp and cyaA), increased susceptibility to cefoxitin (degP) and strongly increased susceptibility to chloramphenicol and minocycline (hns) (Table 1). Inactivation of crp and cyaA was previously shown to produce resistance to some β-lactams [3] and fosfomycin [4]. Inactivation of hns or degP was found to increase susceptibility to rifampicin, sulphonamide and vancomycin (hns), or rifampicin and sulfamethoxazole (degP) in the Keio collection screen [1]. And finally, pcnB and nikD, two genes with no previous known role in antibiotic resistance whose inactivation increased resistance to cefoxitin (pcnB) or minocycline (nikD) (Table 1).

Only two genes that did not affect MarA-mediated antibiotic resistance (ompR and metL) affected antibiotic susceptibility/resistance (Table 1). Inactivation of ompR affected resistance to cefoxitin (in agreement with previous studies [3,5]) and norfloxacin. Inactivation of metL, a gene with no previous known role in antibiotic susceptibility, produced a small but statistically significant increase in susceptibility to cefoxitin (Table 1).

We have found that genes known to affect MarA-mediated antibiotic resistance were three times more likely to affect antibiotic susceptibility per se than genes that do not affect MarA-mediated antibiotic resistance. Moreover, these experiments have allowed us to confirm and expand the role of crp, cyaA, hns, degP and ompR in antibiotic susceptibility in E. coli and to identify three new genes (pcnB, nikD and metL) previously not known to affect antibiotic susceptibility in any bacterium. These findings support our proposed approach of studying genes that have functional interactions with other genes associated with drug resistance as a method to identify genes that independently affect antibiotic susceptibility.