Polymyxin Susceptibility in Pseudomonas aeruginosa Linked to the MexXY-OprM Multidrug Efflux System

Abstract

The ribosome-targeting antimicrobial, spectinomycin (SPC), strongly induced the mexXY genes of the MexXY-OprM multidrug efflux system in Pseudomonas aeruginosa and increased susceptibility to the polycationic antimicrobials polymyxin B and polymyxin E, concomitant with a decrease in expression of the polymyxin resistance-promoting lipopolysaccharide (LPS) modification loci, arnBCADTEF and PA4773-74. Consistent with the SPC-promoted reduction in arn and PA4773-74 expression being linked to mexXY, expression of these LPS modification loci was moderated in a mutant constitutively expressing mexXY and enhanced in a mutant lacking the efflux genes. Still, the SPC-mediated increase in polymyxin susceptibility was retained in mutants lacking arnB and/or PA4773-74, an indication that their reduced expression in SPC-treated cells does not explain the enhanced polymyxin susceptibility. That the polymyxin susceptibility of a mutant strain lacking mexXY was unaffected by SPC exposure, however, was an indication that the unknown polymyxin resistance ‘mechanism’ is also influenced by the MexXY status of the cell. In agreement with SPC and MexXY influencing polymyxin susceptibility as a result of changes in the LPS target of these agents, SPC treatment yielded a decline in common polysaccharide antigen (CPA) synthesis in wild-type P. aeruginosa but not in the ΔmexXY mutant. A mutant lacking CPA still showed the SPC-mediated decline in polymyxin MICs, however, indicating that the loss of CPA did not explain the SPC-mediated MexXY-dependent increase in polymyxin susceptibility. It is possible, therefore, that some additional change in LPS promoted by SPC-induced mexXY expression impacted CPA synthesis or its incorporation into LPS and that this was responsible for the observed changes in polymyxin susceptibility.

INTRODUCTION

Pseudomonas aeruginosa is an opportunistic human pathogen that is historically difficult to treat owing to innate and acquired mechanisms of antimicrobial resistance (1). Polycationic antimicrobials such as aminoglycosides and polymyxins B and E (colistin) are used in the treatment of P. aeruginosa infections, with aminoglycosides being a mainstay, for example, in the treatment of lung infections in patients with cystic fibrosis (2, 3) and polymyxins used increasingly in the treatment of multidrug-resistant infections caused by this organism (4, 5). Aminoglycosides and polymyxins interact with and disrupt the lipopolysaccharide (LPS) layer as a first step in outer membrane permeabilization and entry into P. aeruginosa cells (6, 7). These antibiotics bind to common sites on the LPS typically occupied by Mg²⁺ (8), likely explaining the well-known Mg²⁺ antagonism of polymyxin (9, 10) and aminoglycoside (11) activity. Not surprisingly, LPS alterations and mutations are associated with resistance to these agents (12–17). Still, a major determinant of resistance to the polymyxins, the substitution of LPS lipid A with the cationic 4-aminoarabinose (4AA) (15–19), does not promote resistance to the aminoglycosides (19). Indeed, while mutations that promote 4AA substitution of lipid A are among the most common causes of colistin and polymyxin B resistance in clinical isolates (15–17, 20), the major endogenous mechanism of resistance to aminoglycosides is the MexXY-OprM multidrug efflux system (21–23).

The MexXY-OprM system is derived from the mexXY operon that is under the control of the MexZ repressor (24, 25) and the oprM gene of the mexAB-oprM multidrug efflux operon (26), with OprM serving as the outer membrane component of several multidrug efflux systems in P. aeruginosa (27). The mexXY operon is inducible by many of the antimicrobials that it exports, specifically those that target the ribosome (including aminoglycosides) (28). Antimicrobial induction of this efflux system is compromised by so-called ribosome protection mechanisms (29), suggesting that the MexXY efflux system is recruited in response to ribosome disruption and not antibiotics per se. Consistent with this, mutations in fmt (encoding a methionyl-tRNA-formyltransferase) (30), folD (involved in folate biosynthesis and production of the formyl group added to initiator methionine) (30), the ribosomal protein genes rplA (25) and rplY (31), and the rplU-rpmA ribosomal protein operon (32), all of which are expected to negatively impact protein synthesis, increase expression of mexXY. Upregulation of mexXY by antimicrobials (33) or mutations (fmt [30], folD [30], rplY [31], and rplU-rpmA [32]) is dependent upon a gene, armZ (PA5471), encoding a MexZ antirepressor (34, 35), whose expression is also promoted by ribosome-disrupting antimicrobials (33) and fmt (30), folD (30), or rplU-rpmA (32) mutations. Thus, mutational upregulation of ArmZ enhances mexXY expression and aminoglycoside resistance (36), while armZ mutations compromise aminoglycoside-inducible mexXY expression and aminoglycoside resistance (35).

The synthesis and attachment of 4AA to lipid A is carried out by the products of the arnBCADTEF-ugd (also known as pmrHFIJKLM-ugd and PA3552-59 and referred to here as arn) locus (18), a homologue of the well-studied pmr locus that is associated with resistance to cationic antimicrobial peptides (CAPs), including polymyxins, in Salmonella (37, 38). As in Salmonella, the arn locus and 4AA modification of LPS is inducible by low Mg²⁺ in P. aeruginosa (17, 19), which is consistent with earlier reports of Mg²⁺ deficiency promoting resistance to polymyxin B (39) and CAPs of innate immunity (40) in this organism. Mg²⁺ regulation of polymyxin and CAP resistance, in part via influences on arn expression (19), is mediated by the products of two two-component regulatory systems, PhoPQ (39, 40) and PmrAB (19), both of which are themselves inducible under conditions of Mg²⁺ deficiency (19, 39). Whereas low-Mg²⁺-dependent expression of phoPQ and pmrAB is mediated by the PhoP (40) and PmrA (19) response regulators, respectively, the expression of the arn locus under conditions of Mg²⁺ deficiency is predominantly dependent upon PhoP (19). Polymyxins and CAPs also induce pmrAB and arn expression (19), promoting so-called adaptive resistance to these agents, although PmrAB contributes minimally to this (19). Indeed, a recent study described a third two-component system, ParRS, that is responsible for polymyxin-/CAP-inducible arn expression and therefore adaptive polymyxin/CAP resistance (41). Moreover, mutations in parR can promote arn expression and polymyxin B resistance independently of PmrAB (42). Intriguingly, such mutations also promote mexXY expression and pan-aminoglycoside resistance (42), an indication that ParRS also regulates expression of this multidrug efflux operon. A CAP/polymyxin-responsive two-component system, CprRS, has also been described recently and it, too, contributes to Arn-mediated adaptive CAP/polymyxin resistance (43). A second CAP/polymyxin resistance-promoting locus that is both low Mg²⁺ inducible (19, 44) and CAP inducible (19, 45), PA4773-74, has been described in P. aeruginosa. Annotated as homologues of the spermidine biosynthesis genes, speDE, PA4773-74 are linked to the production of surface-localized spermidine, a polyamine that apparently functions similarly to Mg²⁺ and 4AA in protecting LPS from the action of CAPs, including polymyxins (45). CAPs also induce mexXY (via ParRS) and thus antagonize aminoglycoside activity (42), providing a link between these agents and the arn and mexXY loci. Limiting Mg²⁺ concentrations also enhance P. aeruginosa resistance to aminoglycosides, and mutations in phoPQ increase aminoglycoside resistance under high Mg²⁺ concentrations (40), although the nature of the resistance mechanism(s) involved and the link to PhoPQ remain unknown.

In the present study, mexXY induction by ribosome-perturbing agents such as spectinomycin (SPC) is shown to negatively impact arn and PA4773-74 expression and polymyxin resistance, and elimination of the efflux system substantially abrogates these effects. Still, reduced arn and PA4773-74 expression does not explain the increased polymyxin susceptibility of SPC-treated cells, which is still observable in mutants lacking these polymyxin resistance loci. Presumably, additional and as-yet-unknown polymyxin resistance-promoting LPS modifications are impacted by SPC and MexXY. This is supported by an observed SPC-driven, MexXY-OprM-dependent decline in LPS synthesis. MexXY-OprM clearly influences LPS and the expression of and need for LPS modification loci that promote polymyxin resistance, an indication that this efflux system has some link to LPS, and/or its operation impacts LPS structure.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in the present study are listed in Table 1. Plasmid pEX18Tc and its derivatives were maintained in Escherichia coli with 5 (in l broth) or 10 (on l agar) μg of tetracycline/ml and selected in Pseudomonas aeruginosa with 50 to 75 μg of tetracycline/ml.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmida	Relevant characteristicsb	Source or reference
Strains
E. coli
DH5α	ϕ80dlacZΔM15 endA1 hsdR17(rK− mK+) supE44 thi-1 recA gyrA96 relA1 F− Δ(lacZYA-argF)U169	87
S17-1	thi pro hsdR recA Tra+	88
P. aeruginosa
K767	PAO1 wild type
K2415	K767 ΔmexZ	33
K1525	K767 ΔmexXY	89
K2413	K767 ΔarmZ	33
K3232	K767 ΔarnB	This study
K3612	K767 ΔPA4774	This study
K3613	K767 ΔarnB ΔPA4774	This study
K3617	K767 ΔmexXY ΔarnB	This study
K3615	K767 ΔmexXY ΔPA4774	This study
PA14	Prototroph	90
PA14-galU	PA14 galU::MAR2×T7c (GalU−)	90
PA14-lptC	PA14 lptC::MAR2×T7 (LptC−)	90
PA14-wapR	PA14 wapR::MAR2×T7 (WapR−)	90
PA14-ssg	PA14 ssg::MAR2×T7 (Ssg−)	90
PAO1UBC*	Prototroph	91
K3613*	PAO1UBC Δwzz2	65
K3614*	PAO1UBC Δwzz2 wzz1::Gm	65
PAO1UW	Prototroph	66
PAO1-galU	PAO1UW galU::TnISphoA/hahd (GalU−)	66
PAO1-ssg	PAO1UW ssg::TnIS phoA/hah (Ssg−)	66
K3207	K767 ΔphoP	This study
K2860	K767 ΔpmrA	This study
K3206	K767 ΔparR	This study
K3233	K767 ΔcprR	This study
K3208	K767 ΔphoQ	This study
K3234	K767 ΔpmrA ΔphoP	This study
K3618	K767 ΔphoPQ	This study
K3616	K767 ΔphoQ ΔarnB	This study
K3621	K767 ΔcbrA	This study
K3619	K767 ΔcolR	This study
K3620	K767 ΔPA2572	This study
K3622	K767 ΔbrlR	This study
K3635*	PAO1UBC ΔwbpM (OSA−)	92
K3694*	PAO1UBC ΔPA5456 (CPA−)	93
K3695*	PAO1UBC ΔwbpM ΔPA5456 (OSA− CPA−)	Y. Hao, unpublished data
Plasmids
pEX18Tc	Broad-host-range gene replacement vector; sacB; Tcr	48
pCG017	pEX18Tc::ΔphoQ	This study
pCG018	pEX18Tc::ΔphoP	This study
pCG019	pEX18Tc::ΔparR	This study
pZC001	pEX18Tc::ΔpmrA	This study
pCG020	pEX18Tc::ΔarnB	This study
pCG021	pEX18Tc::ΔcprR	This study
pCG022	pEX18Tc::ΔPA4774	This study
pCG023	pEX18Tc::ΔphoPQ	This study
pCG024	pEX18Tc::ΔcolR	This study
pCG025	pEX18Tc::ΔPA2572	This study
pCG026	pEX18Tc::ΔcbrA	This study
pCG027	pEX18Tc::ΔbrlR	This study

^a*, Poole lab strain designation.

^bTc^r, tetracycline resistance; Ap^r, ampicillin resistance; Cb^r, carbenicillin resistance; OSA, O-specific antigen; CPA, common polysaccharide antigen.

^cTransposon (Tn) insertion mutant from the Harvard P. aeruginosa PA14 transposon MAR2 × T7 insertion library. The Tn-disrupted gene in each mutant is identified.

^dTn insertion mutant from the University of Washington P. aeruginosa PAO1 TnISphoA/hah insertion library. The Tn-disrupted gene in each mutant is identified.

DNA methods.

Standard protocols were used for restriction endonuclease digestions, ligations, transformations, and agarose gel electrophoresis, as previously described (46). Plasmid DNA was extracted from E. coli using a Fermentas GeneJET Plasmid Miniprep kit or a Qiagen Plasmid Midi kit according to protocols provided by the manufacturers. Chromosomal DNA was extracted from P. aeruginosa using a Qiagen DNeasy blood and tissue kit according to a protocol provided by the manufacturer. PCR products and restriction endonuclease digestion products requiring purification were purified using the Promega Wizard SV Gel and PCR Clean-Up system according to a protocol provided by the manufacturer. CaCl₂-competent E. coli (46) and electrocompetent P. aeruginosa (47) cells were prepared as previously described. Oligonucleotide synthesis was performed by Integrated DNA Technologies (Coralville, IA), and nucleotide sequencing was performed by ACGT Corp. (Toronto, Ontario, Canada).

Deletion strain construction.

Derivatives of P. aeruginosa deleted for various genes were generated by constructing deletions in plasmid pEX18Tc and mobilizing them into P. aeruginosa from E. coli S17-1 as described previously (33). Gene deletions were constructed by amplifying, via PCR, 1-kb fragments upstream and downstream of the sequences being deleted and cloning these individually into plasmid pEX18Tc for sequencing (to ensure that no mutations had been introduced during PCR) and then together into a single pEX18Tc vector. P. aeruginosa transconjugants harboring chromosomal inserts of the deletion vectors were selected on l agar plates containing tetracycline (50 μg/ml for deletions engineered into strain K767 or the ΔmexXY strain K1525; 75 μg/ml for deletions engineered into the ΔmexZ strain K2415) and chloramphenicol (5 μg/ml; to counterselect E. coli S17-1). These were subsequently streaked onto l agar containing sucrose (10% [wt/vol]) as described previously (48), with sucrose-resistant colonies screened for the appropriate deletion using colony PCR with 2.5 U of Taq polymerase in 10% (vol/vol) dimethyl sulfoxide (DMSO) (49) and the respective Up-F and Down-R primers for each deletion unless otherwise indicated.

For the ΔparR deletion, the upstream and downstream fragments were amplified using the primers parRUp-F (5′-CGATGAATTCCCGAACAGAGCGAAGAACTC-3′; the EcoRI site is underlined) and parRUp-R (5′-CGATGGTACCCTTGCTGAGGGTAGGGCAG-3′; the KpnI site is underlined) and the primers ParRDown-F (5′-CGATGGTACCCTCTGAATGCTGCGCCTG-3′; the KpnI site is underlined) and ParRDown-R (5′-CGATTCTAGACGTCGTTGACCCGTATCTG-3′; the XbaI site is underlined), respectively. The 50-μl PCR mixtures contained 1 μg of chromosomal P. aeruginosa K767 DNA as the template, 0.5 μM concentrations of each primer, 0.2 mM concentrations of each deoxynucleoside triphosphate (dNTP), and 1 U of Phusion DNA polymerase (New England BioLabs, Ltd., Pickering, Ontario, Canada) in 1× Phusion GC buffer. After an initial denaturation step at 98°C for 30 s, the mixture was subjected to 30 cycles of heating at 98°C for 30 s, 65°C (annealing temperature) for 30 s, and 72°C for 30 s, before finishing with a 5-min incubation at 72°C. For colony PCR, samples were heated at 95°C for 3 min, followed by 35 cycles of 95°C for 45 s, 65°C for 45 s, and 72°C for 3 min, before finishing with 5 min at 75°C.

For the ΔphoQ deletion, the upstream and downstream fragments were amplified using the primers phoQUp-F (5′-CTAGGGATCCCAGGACGGCCAAGGGCTTC-3′; the BamHI site is underlined) and phoQUp-R (5′-GTACTCTAGACAGGGAACGGATCACCGG-3′; the XbaI site is underlined) and the primers phoQDown-F (5′-GTACTCTAGAGTCTGAGACTTGGCGGCCG-3′; the XbaI site is underlined) and phoQDown-R (5′-CTAGAAGCTTCGATGGCCACCTGATGCTG-3′; the HindIII site is underlined), respectively. The reaction mixture for the upstream fragment was formulated as described above except 2.5 U of EXACT DNA polymerase (5 PRIME, Inc., Gaithersburg, MD) was used in 1× EXACT PCR buffer containing 1× 5P-Solution, and primers and dNTPs were included at 1 μM and 0.3 mM, respectively. The mixture was heated at 95°C for 30 s, followed by 30 cycles of 94°C for 30 s, 65°C for 30 s, and 72°C for 1 min, before finishing with a 7-min incubation at 72°C. The downstream fragment was amplified using the same reaction mixture (except for primers) and cycling parameters as for the ΔparR fragments. For colony PCR, samples were heated at 95°C for 3 min, followed by 35 cycles of 95°C for 45 s, 61.3°C for 45 s, and 72°C for 2.5 min, before finishing with 5 min at 75°C.

For the ΔpmrA deletion, the upstream and downstream fragments were amplified using the primers pmrAUp-F (5′-AATCGGGGTACCGCCTGCTTCGTGATGGAC-3′; the KpnI site is underlined) and pmrAUp-R (5′-AACTCGTCTAGACAGCAGTATTCTCATGGCAG-3′; the XbaI site is underlined) and the primers pmrADown-F (5′-AACTCGTCTAGAGGCTACGGCATCGACCAG −-3′; the XbaI site is underlined) and pmrADown-R (5′-AATTCCCAAGCTTCTCCTTGTCGATGGCCTGG-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments except 1 U of Vent DNA polymerase (New England BioLabs) was used in 1× Thermopol buffer and DMSO (10% [vol/vol]) was included. For colony PCR, samples were heated at 94°C for 2 min, followed by 35 cycles of 94°C for 1 min, 67°C for 1 min, and 72°C for 2.5 min, before finishing with 5 min at 75°C.

For the ΔphoP deletion, the upstream and downstream fragments were amplified using the primers phoPUp-F (5′-CTAGGGATCCGAACAGGCAGATCACGAGAA-3′; the BamHI site is underlined) and phoPUp-R (5′-CTAGTCTAGACAGTTTCATGAGGTTCCTCCG-3′; the XbaI site is underlined) and the primers phoPDown-F (5′-CTAGTCTAGAGAGCGCTGCCGGTGATC-3′; the KpnI site is underlined) and phoPDown-R (5′-CTAGAAGCTTGAGAAGTCCCGCTGCAGG-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments except that DMSO (2.5% [vol/vol]) was included. For colony PCR, primers pmrAColony-F (5′-CGACCTACGAGAACATCTCC-3′) and pmrAColony-R (5′-GACGTCGAACACGAAGAACT-3′) were used and samples were heated at 95°C for 3 min followed by 35 cycles of 95°C for 45 s, 63.1°C for 45 s, and 72°C for 1.5 min, before finishing with 5 min at 75°C.

For the ΔcprR deletion, the upstream and downstream fragments were amplified using the primers cprRUp-F (5′-CAGTGAATTCTGGTGATCTTCCTGCACG-3′; the EcoRI site is underlined) and cprRUp-R (5′-CAGTGGATCCGATATGCATGGTATCTGTTC-3′; the BamHI site is underlined) and the primers cprRDown-F (5′-CAGTGGATCCGAGCGCAAATGAAACGCG-3′; the BamHI site is underlined) and cprRDown-R (5′-CAGTAAGCTTCGTCGAGGTGGAAGGTCA-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated as described above for the ΔphoQ fragments and were heated at 95°C for 5 min, followed by 30 cycles of 94°C for 30 s, 60.9°C for 30 s, and 72°C for 1 min, before finishing with 7 min at 75°C. For colony PCR, primers cprRcolony-F (5′-CCCGCAGTATCCGAAGGAAG-3′) and cprRcolony-R (5′-ATGAACCAGACCAGCAGGC-3′) were used, and samples were heated as described above for the ΔpmrA colony PCR except that an annealing temperature of 56.2°C (instead of 67°C) was used.

For the ΔphoPQ deletion, the upstream fragment from the phoP deletion and the downstream fragment for the phoQ deletion were cloned into pEX18Tc. For colony PCR, primers PQcolony-F (5′-TCGGCGACAACAACACCAAG-3′) and PQcolony-R (5′-CGGCGAGAAAGAACAGCACA-3′) were used, and samples were heated as described for the ΔpmrA colony PCR except that the extension time at 75°C was 2.5 min instead of 3 min.

For the ΔarnB deletion, the upstream and downstream fragments were amplified using the primers arnBUp-F (5′-CTAGGAATTCCACGCCAAGGACGCCAAC-3′; the EcoRI site is underlined) and arnBUp-R (5′-CTAGGGATCCCCGGGAGAATGGCAGAAAGTC-3′; the BamHI site is underlined) and the primers arnBDown-F (5′-CTAGGGATCCCGTCGATGAAGCCCTATCCG-3′; the BamHI site is underlined), and arnBDown-R (5′-CTAGAAGCTTCGACGGCTTTCGAGGTCATG-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments. For colony PCR, samples were also heated as described above for the ΔparR deletion.

For the ΔPA4774 deletion, the upstream and downstream fragments were amplified using the primers PA4774Up-F (5′-CAGTGAATTCGGGTGGTCTACCTGGTGGAC-3′; the EcoRI site is underlined) and PA4774Up-R (5′-CAGTGGATCCGAAGTGGTTCGGCGGCAG-3′; the BamHI site is underlined) and the primers PA4774Down-F (5′-CGATGGATCCGCACAATTGCCGGCCCTG-3′; the BamHI site is underlined) and PA4774Down-R (5′-CAGTAAGCTTCAGCAGGTCGAACTCGTCG-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments with the exception that DMSO (5% [vol/vol]) was included and an annealing temperature of 68.1°C was used. For colony PCR, samples were heated as described above for the ΔparR deletion.

For the ΔcolR deletion, the upstream and downstream fragments were amplified using the primers colRUp-F (5′-CGATGAATTCGCGGCTGTTCTACATGAAGG-3′; the EcoRI site is underlined) and colRUp-R (5′-CGATGGATCCCAGTATTCGCATGTCCCACTC-3′; the BamHI site is underlined) and the primers colRDown-F (5′-CGATGGATTCCGAATGGAGTATAAGCAGAGCC-3′; the BamHI site is underlined) and colRDown-R (5′-CGATAAGCTTCAACAGGTTGCCCATCAC-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments. For colony PCR, samples were heated as described above for the ΔparR deletion.

For the ΔcbrA deletion, the upstream and downstream fragments were amplified using the primers cbrAUp-F (5′-CGATGAATTCGCGCTACGGCTTCGAATGG-3′; the EcoRI site is underlined) and cbrAUp-R (5′-CGATGGATCCCTGGGTCAGGCTAAAGCTCG-3′; the BamHI site is underlined) and the primers cbrADown-F (5′-CTAGGGATCCCAGATAACCATCGAGAGCCCG-3′; the BamHI site is underlined) and cbrADown-F (5′-CGATAAGCTTGCTTCAGCGAGATCACATGC-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments. For colony PCR, the primers cbrAcolony-F (5′-TCCTGGTCGTGGGCATCTAT-3′) and cbrAcolony-R (5′-TTCTCCTGGCGGTCCTTGA-3′) and LongAmp Taq polymerase (New England BioLabs; 1 U) were used in a buffer containing 300 μM concentrations of each dNTP and 0.4 μM concentrations of each primer, with samples heated as described above for the ΔparR deletion except for an extension time of 3.5 min at 75°C.

For the ΔPA2572 deletion, the upstream and downstream fragments were amplified using the primers PA2572Up-F (5′-CGATGAATTCCGACGCGGGAGAAGTTCTTC-3′; the EcoRI site is underlined) and PA2572Up-R (5′-GATCGGATCCGTTCATGAGTGTCTGTCCACC-3′; the BamHI site is underlined) and the primers PA2572Down-F (5′-GATCGGATCCGAAGCGATCGAAGGCGCAC-3′; the BamHI site is underlined) and PA2572Down-R (5′-CGATTCTAGACGCTGCGCAAACTGGTCTC-3′; the XbaI site is underlined). The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments. For colony PCR, samples were heated as described above for the ΔparR deletion except for an extension time of 3.5 min at 75°C.

For the ΔbrlR deletion, the upstream and downstream fragments were amplified using the primers brlRUp-F (5′-CAGTGGTACCCAATGTTCGCAAGCACCGC-3′; the KpnI site is underlined) and brlRUp-R (5′-CAGTGGATCCCAGTTGGCCGATGGTGAGC-3′; the BamHI site is underlined) and the primers brlRDown-F (5′-CAGTGGATCCCCCATCTACTGAATCAGCGC-3′; the BamHI site is underlined) and brlRDown-R (5′-CAGTAAGCTTGCTGCCGGTGGAGGAATTC-3′; the HindIII site is underlined), respectively. The reaction mixtures for both fragments were formulated and heated as described above for the ΔparR fragments. For colony PCR, samples were heated as described above for the ΔparR deletion.

Antimicrobial susceptibility testing.

The susceptibility of bacterial strains to various antimicrobials was assessed using the 2-fold serial dilution technique in 96-well microtiter plates as previously described (50). In experiments where the impact of various antibiotics on resistance to polymyxin B and colistin was assessed, the antibiotics were included at 1/4 MIC.

Quantitative RT-PCR.

RNA was prepared from log-phase cells grown in l broth without or with 1/4 MICs of various antimicrobials (SPC, 128 μg/ml for wild-type strain K767 and 16 μg/ml for MexXY⁻ strain K1525; chloramphenicol, 8 μg/ml; tetracycline, 4 μg/ml; amikacin, 0.5 μg/ml; norfloxacin, 0.125 μg/ml), added 90 min prior to harvesting, as described previously (32). RNA conversion to cDNA and assessment of mexX, rpoD, arnB, PA4773, and PA4774 expression using quantitative reverse transcription-PCR (RT-PCR) was assessed as described previously (32), except that the cDNA template was diluted 100-fold and the primer pairs qPCR-arnB-F (5′-TTTCATCGCCAGTCACCTGCAC-3′) and qPCR-arnB-R (5′-GAACAGCGGGATGGAGCACA-3′) (amplification efficiency = 93.3%; r² = 0.995), qPCR-PA4773-F (5′-GTGATCGCCGAATCGCAC-3′) and qPCR-PA4773-R (5′-TACGTCGCCGCAGGTGAACA-3′) (amplification efficiency = 97.4%; r² = 0.997), and qPCR-PA4774-F (5′-ATCGCCGACACCTACAAC-3′) and qPCR-PA4774-R (5′-AGCTCGTGGTACAGCGAC-3′) (amplification efficiency = 98.5%; r² = 0.994) were used to assess arnB, PA4773, and PA4774 expression, respectively.

Preparation and visualization of LPS.

LPS was prepared from cells of P. aeruginosa grown to log-phase and then exposed or not to SPC (1/4 MIC; 128 μg/ml for stain K767 and 16 μg/ml for K1525; overnight exposure) using the sodium dodecyl sulfate (SDS) and proteinase K method described by Hitchcock and Brown (51). LPS samples were resolved by electrophoresis in Tricine-SDS-PAGE gels as described by Maskell (52) and visualized using either the Ultrafast silver staining method (53) or immunoblotted as described previously (54) using monoclonal antibodies specific for the common polysaccharide antigen (N1F10 [55]), the O-specific antigen (MF15-2 [56]), or LPS outer core (5c-101 [57]). Densitometry analysis of immunoblots was carried out using NIH ImageJ software (http://imagej.nih.gov/ij/).

RESULTS AND DISCUSSION

mexXY induction enhances susceptibility to polymyxin antimicrobials.

Although only ribosome-disrupting antimicrobials induce mexXY expression (28, 29, 33), the MexXY-OprM multidrug efflux system accommodates a broad spectrum of antimicrobials (58), so its induction by ribosome perturbation can and will impact the susceptibility of P. aeruginosa to antimicrobials that do not target the ribosome. Aminoglycosides and polymyxins share a mechanism of uptake that involves LPS binding and outer membrane permeabilization (8, 59). To assess the impact of mexXY induction on resistance to non-ribosome-targeting agents such as the polymyxins, steps were first taken to define the “best” inducer of mexXY. Thus, wild-type P. aeruginosa strain K767 was exposed to chloramphenicol, tetracycline, a representative aminoglycoside (amikacin), and a related aminocyclitol, spectinomycin (SPC), and the expression of mexXY was assessed using quantitative RT-PCR. Interestingly, the aminoglycoside was the poorest inducer (2-fold), whereas SPC was the best inducer (18-fold), with chloramphenicol and tetracycline showing intermediate levels of induction (5- to 7-fold) that were reminiscent of expression levels seen for a mexZ deletion mutant (Fig. 1A). Using SPC, the impact of mexXY induction on resistance to polymyxin B and colistin was assessed. Intriguingly, exposure of P. aeruginosa strain K767 to SPC enhanced its susceptibility to polymyxin B and colistin (2- to 4-fold) (Table 2). Consistent with the SPC driving mexXY expression, resistance to norfloxacin, a known MexXY-OprM substrate (58), increased 4-fold in the presence of the aminocyclitol (Table 2). To assess whether the impact of SPC on polymyxin susceptibility was a unique property of the aminocyclitol or was more generally related to mexXY induction by ribosome-perturbing agents, the impact of chloramphenicol, tetracycline, and amikacin on the susceptibility of P. aeruginosa to the polymyxins was also assessed. Chloramphenicol and tetracycline had a more modest impact on polymyxin or colistin susceptibility, consistent with their more modest induction of mexXY, which was also reflected in their more modest enhancement of norfloxacin MICs (2-fold; Table 2), whereas amikacin, a poor inducer of mexXY, minimally impacted polymyxin and colistin MICs (Table 2). Thus, mexXY expression seems to correlate with an enhanced susceptibility of P. aeruginosa to polycationic antimicrobials.

FIG 1.

Impact of ribosome-targeting antimicrobials on expression of mexXY and the arn and PA4773-74 loci of polymyxin resistance in P. aeruginosa. mexX (A), arnB (B), and PA4773-74 (C) expression was assessed in wild-type P. aeruginosa PAO1 strain K767 in the absence (−) or presence of the indicated antimicrobial (SPC, spectinomycin; AMI, amikacin; TET, tetracycline; CAM, chloramphenicol) at 1/4 MIC using quantitative RT-PCR. The results for the mexZ deletion strain K2415 unexposed to antimicrobials are also shown. Expression was normalized to rpoD and is reported relative (fold change) to P. aeruginosa strain K767. Values represent the mean ± the standard errors of the mean (SEM) from at least three independent determinations, each performed in triplicate.

TABLE 2.

Influence of mexXY-inducing antimicrobials on polycation resistance in P. aeruginosa^a

Antimicrobial added	MIC (μg/ml)
	NOR	COL	PXB
None	0.5	2	2
SPC	2	0.5	0.5
AMI	0.5	2	1
TET	1	1	1
CAM	1	0.5	0.5

^aThe susceptibility of wild-type P. aeruginosa strain K767 to norfloxacin (NOR), colistin (COL), or polymyxin B (PXB) was determined in the absence (none) or presence of 1/4 MIC of the mexXY-inducing antimicrobials spectinomycin (SPC; 128 μg/ml), amikacin (AMI; 0.5 μg/ml), tetracycline (TET; 4 μg/ml), and chloramphenicol (CAM; 8 μg/ml).

mexXY expression compromises expression of the arn locus encoding an LPS modification system linked to polycation resistance.

A major contributor to resistance to polycationic antimicrobials such as colistin and polymyxin B in P. aeruginosa is an LPS modification system that attaches 4AA onto lipid A (18). Encoded by the variously named pmrHFIJKLM-ugd (60, 61) or arnBCADTEF-PA3559 (62) locus, the positively charged 4AA modification serves to block polycation binding to LPS (61), an otherwise necessary first step for these agents to cross the outer membrane and enter the bacterial cell (8, 59). One possibility is that a basal level of arn gene expression and, thus, 4AA decoration of LPS occurs in wild-type cells grown in l broth, with the induction of mexXY by ribosome-perturbing agents somehow compromising this and reducing the 4AA levels on LPS. To assess this, the impact of SPC treatment on arnB expression (as a measure of expression of the arn locus) was assessed. As seen in Fig. 1B, SPC promoted a roughly 10-fold reduction in arnB expression. In examining the impact of other ribosome-targeting agents, amikacin, which had no impact on polycationic antimicrobial susceptibility, did not influence arn gene expression while chloramphenicol and tetracycline, which had a modest impact on polycationic antimicrobial susceptibility, had a modest impact on arnB expression (2- to 3-fold reduction). A mexZ mutant strain derepressed (5-fold) for mexXY expression also showed a modest (<2-fold) but reproducible reduction in arnB expression (Fig. 1B), although this did not translate into any change in polymyxin susceptibility (Table 3). Notably, exposure of the mexZ mutant to SPC further enhanced mexXY expression (Fig. 2A) as well as polymyxin and colistin susceptibility (Table 3), the latter concomitant with a decrease in arnB expression (Fig. 2B). Although these results support the MexXY-OprM status of the cell influencing expression of the arn locus, clearly expression of this efflux operon alone does not explain the reduction in polymyxin MICs seen for antibiotic-treated wild-type P. aeruginosa K767, since mexXY expression levels for the mexZ mutant were comparable to that seen for chloramphenicol- and tetracycline-treated K767 (Fig. 1A), although only the drug-treated K767 showed a reduction in polymyxin MICs. It may therefore be that MexXY-OprM and additional effects of drug treatment together function to compromise polymyxin resistance.

TABLE 3.

Influence of SPC on polymyxin resistance in P. aeruginosa^a

Strain	Phenotype	Presence or absence of SPC	MIC for (μg/ml)
			COL	PXB
K767	WT	–	1	2
		+	0.5	0.5
K2415	MexZ−	–	1–2	2
		+	0.5	0.5
K1525	MexXY−	–	2	2
		+	2	2
K2413	ArmZ−	–	2	2
		+	2	2
K3232	ArnB−	–	1	2
		+	0.5	0.5
K3612	PA4774−	–	1	1
		+	0.25	0.25
K3613	ArnB− PA4774−	–	1	1
		+	0.25	0.25
K3617	MexXY− ArnB−	–	1	1–2
		+	1	1
K3615	MexXY− PA4774−	–	1	1–2
		+	1	1
PA14	WT	–	1	1
		+	0.25	0.25
PA14-galUb	GalU−	–	1	1
		+	0.25	0.25
PA14-lptC	LptC−	–	0.25	0.5
		+	0.25	0.5
PA14-wapR	WapR−	–	1	1
		+	0.5	0.5
PA14-ssg	Ssg−	–	1	1
		+	0.25	0.25
PAO1UW	WT	–	2	2
		+	0.5	0.5
PAO1UW-galUc	GalU−	–	1-2	2
		+	0.5	0.5
PAO1UW-ssg	Ssg−	–	2	2
		+	0.5	0.5
PAO1UBC	WT	–	2	2
		+	1	1
PAO1UBC Δwzz2	Wzz2−	–	2	2
		+	0.5	1
PAO1UBC Δwzz2 wzz1::Gm	Wzz1− Wzz2−	–	2	2
		+	1	1
K3207	PhoP−	–	1	2
		+	0.5	0.5
K2860	PmrA−	–	1	2
		+	0.5	0.5
K3206	ParR−	–	1	2
		+	0.5	0.5
K3233	CprR−	–	1	2
		+	0.5	0.5
K3208	PhoQ−	–	2–4	2
		+	2	2
K3234	PhoP− PmrA−	–	1	2
		+	0.5	0.5
K3618	PhoQ− PhoP−	–	2	2
		+	0.5	0.5
K3616	PhoQ− ArnB−	–	2	2
		+	0.5	0.5
K3621	CbrA−	–	2	2
		+	0.5	0.5
K3619	ColR−	–	2	2
		+	0.5	0.5–1
K3620	PA2572−	–	2	2
		+	0.5	0.5
K3622	BrlR−	–	2	2
		+	0.5	0.5
K3635	WbpM− (OSA−)	–	2	2
		+	1	1
K3694	PA5456− (CPA−)	–	2	2
		+	1	1
K3695	WbpM− PA5456− (OSA− CPA−)	–	2	2
		+	1	1

^aThe susceptibility of the indicated strains to colistin (COL) and polymyxin B (PXB) was assessed in the absence (–) and presence (+) of 1/4 MIC (128 μg/ml) SPC. In strains where SPC has no effect on COL or PXB, the MIC values are indicated in boldface. WT, wild type.

^bTransposon insertion mutant from a PA14 Tn mutant library (90). The mutated gene is indicated.

^cTransposon insertion mutant from a PAO1 Tn mutant library (66). The mutated gene is indicated.

FIG 2.

Influence of mexXY on expression of the arn and PA4773-74 loci of polymyxin resistance. mexX (A), arnB (B), and PA4774 (C) expression was assessed in the indicated strains whose relevant genotypes are highlighted. Expression was normalized to rpoD and is reported relative (fold change) to the wild-type P. aeruginosa PAO1 strain K767. Values represent the means ± the SEM from at least three independent determinations, each performed in triplicate.

Consistent with the need for mexXY induction for the observed increased in colistin and polymyxin B susceptibility upon exposure to ribosome-targeting agents such as SPC, a P. aeruginosa mutant strain lacking mexXY, K1525, failed to show any increase in susceptibility to these polycationic antimicrobials upon exposure to SPC (Table 3). Indeed, the loss of mexXY actually enhanced arnB expression (2-fold; Fig. 2B), and while SPC treatment did reduce this, the reduction was only 3-fold (versus 10-fold for wild-type) and arnB expression levels for SPC-treated K1525 were comparable to that seen for untreated wild type (Fig. 2B). Thus, MexXY-OprM is generally needed for the SPC-dependent reduction in polymyxin/colistin resistance and, to a lesser but still significant extent, arn gene expression. Consistent with this, strain K2413 lacking the MexZ antirepressor AmzR, whose expression is also induced by ribosome-perturbing agents (33) and is required for SPC-inducible mexXY expression (Fig. 2A), also showed no decrease in polymyxin/colistin resistance upon SPC exposure (Table 3), and a modest reduction only in arn gene expression, which is the same as for untreated wild-type (Fig. 2B). This may explain why polymyxin/colistin MICs remain unchanged in the SPC-treated MexXY⁻ and ArmZ⁻ mutants relative to SPC-treated wild type. Interestingly, the increase in arnB expression in the MexXY⁻ and ArmZ⁻ strains paralleled a modest (2-fold) increase in colistin resistance although no change in polymyxin B MICs was observed (Table 3). There appears, therefore, to be an inverse relationship between mexXY and arn gene expression/polymyxin resistance, suggesting a link between the efflux system and LPS modification, although, clearly, the impact of SPC on arn expression and polymyxin resistance is not fully explained by, or dependent on, MexXY.

SPC- and MexXY-dependent increase in polymyxin susceptibility is not explained by reduced expression of known LPS modification genes.

To confirm that the increased susceptibility to polymyxins promoted by SPC exposure or mexXY expression resulted from reduced arn expression and, presumably, reduced 4AA decoration of LPS, the arnB gene was deleted in both the wild-type P. aeruginosa strain K767 and the ΔmexXY strain K1525. The expectation was that loss of arnB (and thus 4AA modification) would render both strains polymyxin sensitive, with SPC treatment then having no effect on polymyxin susceptibility. Surprisingly, loss of arnB in the K767 derivative, K3232, had no impact on polymyxin susceptibility, and SPC treatment still provided for a 2- to 4-fold reduction in polymyxin MICs (Table 3). Loss of arnB in the K1525 derivative, K3617, also had no impact on polymyxin susceptibility, and the lack of an effect of SPC on polymyxin MICs seen in the MexXY⁻ strain K1525 was retained in K3617 (Table 3). Thus, although SPC and mexXY expression clearly reduced expression of the arn locus, this does not explain their impact on polymyxin susceptibility.

A second locus that is linked to polymyxin resistance in P. aeruginosa and that might explain the SPC and mexXY effects on polymyxin susceptibility is the PA4773-74 locus responsible for the synthesis of the polyamine spermidine, whose binding to LPS appears to stabilize and protect the outer membrane from the action of polycationic antimicrobials such as the polymyxins (45). As with arnB, expression of this locus was markedly reduced (ca. 20-fold) in wild-type strain K767 upon exposure to SPC (Fig. 1C). Again, however, deletion of PA4474 in wild-type strain K767 or MexXY⁻ strain K1525 did not alter susceptibility to polymyxin (Table 3, see K3612 and K3615), with SPC still reducing polymyxin MICs in the K767 ΔPA4774 derivative, K3612, and having no impact on polymyxin MICs in the K1525 ΔPA4774 derivative, K3615 (Table 3). SPC treatment of the MexXY⁻ K1525 did reduce PA4774 expression, although as with arnB expression it remained above that seen for SPC-treated wild-type strain K767 (Fig. 2C). As with arnB, too, PA4774 expression decreased in the mexXY-overexpressing ΔmexZ strain (Fig. 1C and 2C) and increased in the ΔmexXY mutant (Fig. 2C). Again, this confirms that mexXY expression is inversely related to expression of LPS-decorating genes that impact polymyxin resistance. The elimination of both arnB and PA4774 in strain K767 had no impact on polymyxin susceptibility, and SPC treatment still yielded a decrease in polymyxin MICs (see Table 3, strain K3613), a finding consistent with the SPC-/MexXY-mediated reduction in polymyxin resistance not simply explained by a reduction in these known LPS-decorating polymyxin resistance loci. Nonetheless, these results do indicate that SPC and MexXY-OprM impact expression of these LPS-decorating genes, an indication, perhaps, that they are affecting some aspect of LPS synthesis and/or structure that is subsequently impacting the need for the arn and PA4773-74 gene products.

A recent study that attempted to define the polymyxin resistome in P. aeruginosa PA14 identified several genes that contribute to intrinsic polymyxin resistance, including a number that are linked to LPS biosynthesis: galU, wapR, ssg, and lptC (63). Encoding proteins implicated in core biosynthesis (galU, wapR, and ssg) and LPS transport to the outer membrane (lptC), transposon insertion mutants individually lacking these genes showed increased susceptibility to polymyxin B, concomitant with the expected gross changes in LPS banding patterns (as visualized on polyacrylamide gels) (63). To assess the possible involvement of these genes in the SPC-mediated reduction in polymyxin resistance, the impact of SPC on the polymyxin susceptibility of available PA14- and PAO1-derived transposon insertion mutants disrupted in these genes (Table 1) was assessed. Unexpectedly, all mutants, except for the PA14-derived LptC⁻ mutant, failed to show any increase in polymyxin susceptibility relative to their PA14 or PAO1 wild-type parents (Table 3), in contrast to previous results (63). This may reflect differences in the growth media used in the two studies. Nonetheless, the SPC-promoted reduction in polymyxin resistance was observed in the PA14 and PAO1 parent strains, as well as all mutants, with the exception of the LptC⁻ strain (Table 3), an indication that the SPC-promoted changes that enhanced polymyxin susceptibility in P. aeruginosa were lacking in the absence of this LPS transport gene.

LPS is comprised of a membrane-imbedded lipid A linked to a core oligosaccharide that is substituted distally with a repeating sugar polymer, the O-polysaccharide or O-antigen (O-Ag) (64). P. aeruginosa produces two O-Ags, a homopolymer of d-rhamnose (the so-called common polysaccharide antigen [CPA]) and a heteropolymer of different sugars organized into repetitive units (the O-specific antigen [OSA]) of various numbers, that in P. aeruginosa PAO1 shows a modal distribution around long and very long O-Ag lengths (64). Despite the annotation of lptC as an LPS transport gene, however, the P. aeruginosa PA14 LptC⁻ mutant expressed presumably outer membrane-localized LPS, displaying a loss, only, of the very long O-antigen chains and increased levels of capped core oligosaccharide (LPS with a single O-antigen unit) (63). This was suggestive of a possibly partial defect only in LPS synthesis/transport in this mutant, one that is specific to the longer O-Ag-containing LPS. Whether and how this might be related to the loss of an SPC impact on polymyxin susceptibility is unclear, although the observation that the galU, wapR, and ssg mutants showed a much greater LPS deficiency (no O-Ag and a shorter, truncated core) (63) indicates that the O-Ag deficiency itself in the lptC mutant cannot explain this. In agreement with this, mutant derivatives of wild-type strain PAO1 lacking the genes for very long (wzz2) or both long and very long (wzz1 wzz2) O-Ag synthesis (65), showed no increase in polymyxin susceptibility relative to PAO1 (Table 3). Moreover, the SPC-promoted increase in polymyxin B and colistin susceptibility was still observable in these mutants (Table 3). Presumably, the unique O-Ag changes in the lptC mutant reflect additional, as yet undefined, changes in the LPS of this mutant, possibly in the core region (where some gross differences between wild-type and the lptC mutant are discernible in LPS gels [63]), and these are related to SPC- and MexXY-OprM-promoted changes in polymyxin susceptibility. Interestingly, attempts to construct a ΔlptC mutant in a K767 background were unsuccessful, despite repeated attempts, suggesting that lptC may be essential in P. aeruginosa PAO1. Consistent with this, an lptC transposon insertion mutant was not available in the PAO1 transposon mutant library (66) (http://www.pseudomonas.com). The corresponding gene in E. coli is also essential, as expected for a gene whose product functions in LPS transport to the outer membrane (67). Presumably, then, P. aeruginosa PA14 possesses a gene(s) that can compensate for the loss of lptC, or else this gene is not essential for LPS transport in PA14.

SPC-promoted increase in polymyxin susceptibility is not mediated by known regulators of polymyxin resistance.

A variety of regulators are known to positively impact arn gene expression, including PhoP and PmrA, which mediate the Mg²⁺ limitation-dependent induction of arn (19, 60), and ParR and CprR, which mediate the CAP inducibility of this locus (41, 43). To determine whether any of these play a role in the SPC-/MexXY-dependent decline in polymyxin resistance, mutants lacking each of these individually were constructed and the impact on the SPC-driven reduction in resistance was assessed. In all cases, the mutants retained the SPC-mediated reduction in polymyxin resistance (Table 3). Thus, these regulators do not mediate the SPC and MexXY impact on susceptibility to these agents. The SPC-promoted decline in polymyxin/colistin resistance was, however, lost in a mutant, K3208, deleted for the phoQ gene encoding a sensor kinase known to regulate PhoP (directly) and PmrA (indirectly), and likely other regulators (40) (Table 3). Since PhoQ controls both PmrA and PhoP, such that loss of one or the other of these response regulators might not obviate a PhoQ influence on polymyxin resistance in response to SPC, the impact of deleting both regulators on the SPC-dependent reduction in polymyxin resistance was assessed. Again, however, SPC exposure still reduced resistance to the polymyxins (Table 3), suggesting that another PhoQ-controlled regulator might mediate the SPC/MexXY impact on polymyxin susceptibility. Alternatively, since a phoQ mutant has previously been shown to display elevated expression of phoP (40) and the arn locus (19) under noninducing conditions (i.e., high Mg²⁺), PhoP being a positive regulator of the arn locus (19), it is possible that a SPC-mediated reduction in polymyxin resistance in the ΔphoQ mutant strain K3208 is simply masked by resistance promoted by an increase in PhoP-mediated arn expression and thus 4AA modification of LPS. In agreement with the latter possibility, loss of either phoP (see K3618) or arnB (see K3616) in K3208 restored the SPC-mediated reduction in polymyxin MICs (Table 3). Thus, the SPC-promoted increase in polymyxin susceptibility in P. aeruginosa is not mediated by any PhoQ-controlled regulators. Although SPC-mediated decreases in arn and PA4773-74 expression do not explain the increased polymyxin susceptibility of SPC-treated P. aeruginosa, it is interesting that none of the aforementioned regulators mediated the SPC-promoted decline in arn/PA4773-74 expression: all regulator mutants retained the SPC-promoted increase in mexXY expression and decrease in arn/PA4773-74 expression (Fig. 3).

FIG 3.

Influence of two-component system deletions on expression of the arn and PA4774 loci of polymyxin resistance. mexX (A), arnB (B), and PA4774 (C) expression was assessed in the indicated strains whose relevant genotypes are highlighted. Expression was normalized to rpoD and is reported relative (fold change) to the wild-type P. aeruginosa PAO1 strain K767. Values represent the mean ± the SEM from at least three independent determinations, each performed in triplicate.

More-recently described regulators that have been linked to polymyxin resistance and which may play a role in the SPC-mediated decrease in polymyxin resistance include the following: CbrA, a response regulator that positively influences phoPQ, pmrAB, and arn expression (68); ColRS, a two-component system linked to 4AA modification of LPS (69); PA2572, an orphan response regulator that positively influences pmrAB expression (70); and BrlR, a MerR type repressor of phoPQ (71). Deletions of each of these were individually engineered into P. aeruginosa strain K767, but again their loss did not obviate the SPC-promoted decline in polymyxin MICs (Table 3).

SPC- and MexXY-mediated changes in LPS.

The observed impact of SPC and the MexXY status on susceptibility to the LPS-targeting polymyxins and on expression of LPS modification loci suggests that SPC and the efflux system may be influencing LPS synthesis and/or structure. To identify any gross changes in LPS that may be manifest by SPC treatment or the presence or absence of MexXY, the LPS profiles of wild-type P. aeruginosa K767 and its ΔmexXY derivative, K1525, grown in the absence or presence of SPC were determined. As seen in Fig. 4A (compare lane 2 to lane 1), growth of K767 in the presence of mexXY-inducing SPC yielded a reduction in longer chain LPS synthesis. Immunoblotting with antibodies specific for OSA (Fig. 4B) and CPA (Fig. 4C) revealed that this reduction was generally due to a marked decline (ca. 50%) in CPA synthesis (Fig. 4C, compare lane 2 to lane 1). Interestingly, this reduction was minimal (ca. 20%) in SPC-treated K1525 (Fig. 4A and C, compare lane 4 to lane 3), an indication that it was in some way dependent on SPC-promoted MexXY production. The SPC-dependent reduction in CPA synthesis in K767 correlates nicely with the SPC-mediated decline in polymyxin MICs, as does the lack of an influence of SPC on CPA synthesis and polymyxin MICs in the MexXY⁻ strain K1525, an indication that CPA levels are somehow reflecting polymyxin susceptibility. Still, an available mutant strain devoid of CPA did not show any increase in polymyxin susceptibility relative to its wild-type parent strain (Table 3, compare K3694 with PAO1_UBC), and its polymyxin susceptibility was enhanced upon exposure to SPC just as it was for the parent (Table 3). This was also true for a mutant lacking OSA (K3635) or both O polysaccharides (K3695) (Table 3). Thus, the reduced polymyxin resistance of SPC-treated K767 cannot be explained by a decline in CPA levels. More likely, this decline simply reflects an additional change(s) in LPS structure that is manifest by SPC treatment, and this impacts CPA synthesis and perhaps polymyxin susceptibility. Apparently, these changes are dependent on the presence of the MexXY-OprM multidrug efflux system. The potential for MexXY-OprM operation to be impacting LPS structure in ways that affect polymyxin susceptibility likely explains the lack of involvement of regulatory pathways that typically influence CAP susceptibility. As such, too, the impact of SPC and MexXY-OprM on the expression of the arn and PA4773-74 LPS modification genes may simply be a downstream response to LPS changes manifest by SPC and MexXY-OprM.

FIG 4.

Influence of spectinomycin (SPC) on LPS production by P. aeruginosa. P. aeruginosa strains K767 (MexXY⁺; lanes 1 and 2) and K1525 (MexXY⁻; lanes 3 and 4) were cultured without (−) or with (+) SPC as indicated, and LPS was extracted, electrophoresed, and detected after silver staining (A) or immunoblotting with antibodies specific for the serotype O-5 OSA (B), the CPA (C), and the outer core (D). The latter served as a control to ensure equal loading of LPS in all lanes. The numbers shown in panel C are the results of densitometry analysis of the CPA bands, normalized to outer core LPS in each case, with the numbers in parentheses indicating the percentages of untreated K767 values.

Concluding remarks.

Although SPC-promoted decreases in arn and PA4773-74 expression do not explain the enhanced polymyxin susceptibility of SPC-treated P. aeruginosa, these results do suggest that the expression of these LPS-modifying gene products is influenced by SPC or that the availability of specific LPS targets for the Arn- and PA4773-74-promoted modifications is impacted by SPC, and this indirectly influences arn and PA4773-74 expression. The regulator responsible for this SPC-promoted decline in gene expression is as yet unknown. Although the MexXY-OprM status of the cell also impacts arn and PA4773-74 gene expression and SPC's influence on this, an indication that some of the SPC impact is MexXY dependent, SPC is also clearly able to influence arn and PA4773-74 expression independent of this efflux system. Possibly, SPC influences LPS structure in multiple ways. Antimicrobial-promoted LPS changes, including a reduction in LPS synthesis, have been noted previously in P. aeruginosa (72), as well as in other organisms (for example, E. coli [73, 74]), although the mechanistic details were not elucidated. In the case of P. aeruginosa, this was seen for the ribosome-targeting, mexXY-inducing macrolides, erythromycin, and azithromycin (72) but not for ciprofloxacin (75), which does not induce mexXY. Again, this is suggestive of a link between MexXY-OprM and LPS.

The observation that the decline in polymyxin MICs mirrors a decline in CPA levels but that CPA loss does not explain polymyxin susceptibly suggests that the resistance-altering change must occur elsewhere on the LPS and is specifically impacting either CPA synthesis or CPA attachment to the core polysaccharide. Why CPA synthesis specifically would be impacted is unclear, since it is unknown what purpose CPA serves in P. aeruginosa versus OSA. Certainly, CPA and OSA production in P. aeruginosa can be differentially regulated; growth at elevated temperatures, for example, specifically compromises OSA synthesis (76). Similarly, alterations in dissolved oxygen tension impact OSA but not CPA levels in P. aeruginosa (77). As such, changes to LPS, likely in the core region that is the site of CPA (and OSA) attachment, may differentially impact the need for or ability to accommodate CPA versus OSA. Despite the long-appreciated importance of lipid A as a binding site for polymyxins (78, 79) and a target for resistance-promoting modification in Gram-negative bacteria (80), there is also evidence of core polysaccharide involvement, with core modifications also linked to polymyxin resistance in both E. coli (81, 82) and Salmonella enterica serovar Typhimurium (81, 83). Similarly, core truncations have been shown to reverse the intrinsic polymyxin resistance of Burkholderia cenocepacia (84, 85), another indication that this region of LPS can influence polymyxin susceptibility. Still, whether and what polymyxin susceptibility-altering core changes are manifest by SPC and MexXY-OprM remains, at present, a mystery. A recent report highlighting the high-affinity binding of rough core LPS to the purified MacA component of the MacABC macrolide exporter of E. coli and a functional link between LPS binding and pump operation (86) does provide precedence for drug efflux systems interacting with and thereby influencing LPS. Possibly, MexXY-OprM exports some constituent of LPS and in so doing impacts LPS structure. Still, it cannot be ruled out that SPC and/or MexXY-OprM indirectly impact LPS structure and, ultimately, polymyxin resistance by altering some aspect of P. aeruginosa physiology that necessitates changes to the LPS.