Non-interchangeable functions of efflux transporters of Pseudomonas aeruginosa in survival under infection-associated stress

ABSTRACT

Pseudomonas aeruginosa is a challenging opportunistic pathogen due to its intrinsic and acquired mechanisms of antibiotic resistance. A large repertoire of efflux transporters actively expels antibiotics, toxins, and metabolites from cells and enables growth of P. aeruginosa in diverse environments. In this study, we analyzed the roles of representative efflux pumps from the Resistance-Nodulation-Division (RND), Major Facilitator Superfamily (MFS), and Small Multidrug Resistance (SMR) families of proteins in the susceptibility of P. aeruginosa to antibiotics and bacterial growth under stresses imposed by human hosts during bacterial infections: an elevated temperature, osmotic stress, low iron, bile salts, and acidic pH. We selected five RND pumps MexAB-OprM, MexEF-OprN, MexCD-OprJ, MuxABC-OpmB, and TriABC-OpmH that differ in their substrate specificities and expression profiles, two MFS efflux pumps PA3136-3137 and PA5158-5160 renamed here into MfsAB and MfsCD-OpmG, respectively, and an SMR efflux transporter PA1540-1541 (MdtJI). We found that the most promiscuous RND pumps such as MexEF-OprN and MexAB-OprM are integrated into diverse survival mechanisms and enable P. aeruginosa growth under various stresses. MuxABC-OpmB and TriABC-OpmH pumps with narrower substrate spectra are beneficial only in the presence of the iron chelator 2,2′-dipyridyl and bile salts, respectively. MFS pumps do not contribute to antibiotic efflux but play orthogonal roles in acidic pH, low iron, and in the presence of bile salts. In contrast, MdtJI protects against polycationic antibiotics but does not contribute to survival under stress. Thus, efflux pumps play specific, non-interchangeable functions in P. aeruginosa cell physiology and bacterial survival under stresses.

IMPORTANCE

The role of multidrug efflux pumps in the intrinsic and clinical levels of antibiotic resistance in Pseudomonas aeruginosa and other gram-negative bacteria is well-established. Their functions in bacterial physiology, however, remain unclear. The P. aeruginosa genome comprises an arsenal of efflux pumps from different protein families, the substrate specificities of which are typically assessed by measuring their impact on susceptibility to antibiotics. In this study, we analyzed how deletions and overproductions of efflux pumps affect P. aeruginosa growth under human-infection-induced stresses. Our results show that the physiological functions of multidrug efflux pumps are non-redundant and essential for the survival of this important human pathogen under stress.

INTRODUCTION

Pseudomonas aeruginosa is an important opportunistic pathogen responsible for a myriad of infectious diseases in patients with a damaged cutaneous barrier and those with compromised immunity (1, 2). P. aeruginosa owes its high pathogenicity to its various virulence factors, a cellular envelope studded with active efflux pumps that is impermeable to most antibiotic drugs, a complex signaling network that allows it to adapt to various environments, and a variety of metabolic enzymes that make it suitable for occupying different niches (3–6). Persistent P. aeruginosa infections combined with high mortality and morbidity lead to extended hospital stays that incur further financial costs (7, 8). These features have led P. aeruginosa to be included in two categories of highly drug-resistant bacteria, the ESKAPE category and the TOTEM category (8, 9). In addition, the Center for Disease Control lists multidrug-resistant P. aeruginosa as a serious threat to public health and reported a 32% increase in the hospital-onset P. aeruginosa infections during the COVID-19 pandemic (10).

Eradicating infections caused by P. aeruginosa is challenging due to the intrinsic resistance mechanisms that the bacterium employs to protect itself against antimicrobial drugs. The major resistance mechanisms exploited by P. aeruginosa are the low permeability barrier of the outer membrane and an array of efflux transporters that actively expel antibiotics from cells (11). The synergistic working of these two mechanisms effectively prevents the antimicrobials from accessing the cellular targets (12, 13).

The efflux pumps are categorized into six different superfamilies of proteins: the ATP-Binding Cassette (ABC) family, the Small Multidrug Resistance (SMR) family, the Major Facilitator Superfamily (MFS), the Multidrug And Toxic compound Extrusion (MATE) family, Proteobacterial Antimicrobial Compound Efflux (PACE), and the Resistance Nodulation Division (RND) family (14). Among all these efflux pumps’ families, the RND transporters are the greatest contributors to antibiotic resistance phenotypes due to their broad substrate polyspecificity. In P. aeruginosa and other gram-negative bacteria, RND transporters form trans-envelope tripartite efflux complexes, consisting of an outer membrane factor/channel (OMP), a periplasmic membrane fusion protein (MFP), and an inner membrane RND transporter (15, 16). The major advantage of such complexes is that not only do they expel the broadest range of antibacterial agents but also are able to do this across the outer membrane, thus, taking advantage of synergy between the slow permeation and active efflux (13, 15).

In P. aeruginosa, twelve RND efflux pumps are characterized and found to provide a varying level of hyposusceptibility to various antibiotics (15, 17). In accordance with their expression profiles in vitro and in vivo, these efflux pumps fall into three categories: (i) expressed constitutively in vitro and in vivo (e.g., MexAB-OprM and TriABC-OpmH); (ii) regulated by stresses and during infections (e.g., MuxABC-OpmB); (iii) silent and mutationally selected during antibiotic exposure (e.g., MexEF-OprN and MexCD-OprJ) (17). All RND pumps are poly-specific but to a varying degree and based on antibiotic susceptibility studies, their substrate specificities partially overlap. In addition to protecting against antibiotics, the tripartite RND pumps of P. aeruginosa such as MexAB-OprM play important roles in metallophore pseudopaline transport across the outer membrane (18) as well as other physiological programs, but these roles still remain undefined (15, 19–22).

We previously reported that the inactivation of six major RND pumps MexAB-OprM, MexCD-OprJ, MexXY, MexEF-OprN, TriABC, and MexJK (PΔ6 strain) leads to significant changes in gene expression (15). Genes involved in iron starvation responses, modification of the bacterial outer membrane and several efflux and uptake transporters were all upregulated in RND-deficient cells (15). Among the efflux pumps with increased gene expression were the RND pump MuxABC-OpmB, the MFS pumps PA3136-3137 and PA5158-5160 renamed here as MfsAB (Major Facilitator Superfamily) and MfsCD-OpmG, respectively, and the SMR pump PA1540-41 (15). The Mfs pumps in P. aeruginosa are homologous to Escherichia coli EmrAB-TolC involved in efflux of ionophores and some antibiotics (23) and contain genes coding for the MFP (PA3136 and PA5159) and the drug efflux transporter (PA3137 and PA5160). In addition, the mfsCD-opmG operon contains a gene coding for an OMP OpmG (PA5158). Previously, both operons in P. aeruginosa were found to be induced in the presence of pentachlorophenol (24). PA1540-41 is homologous to MdtJI of E. coli. The overproduction of MdtJI in E. coli cells lacking the major RND pump AcrAB-TolC confers hyposusceptibility to the detergents SDS and deoxycholate and the antibiotics nalidixic acid and Fosfomycin (25). Also in enterobacteria, this pump was implicated in the secretion of polyamine spermidine (26, 27).

The elevated expression of Mfs pumps, as well as MdtJI and MuxABC-OpmB pumps in P. aeruginosa PΔ6 cells suggested that the functions of these transporters might be complementary to those of the inactivated RND pumps. However, deletions of mfsAB and mdtJI had no effect on the antibiotic susceptibility of P. aeruginosa, whereas the inactivation of muxABC-opmB increased the susceptibility but only to novobiocin (15). It remains unclear whether and how changes in antibiotic susceptibility of P. aeruginosa due to the inactivation of various efflux pumps correlate with their physiological functions.

In this study, we analyzed the growth phenotypes, antibiotic susceptibilities, and the cell envelope permeability of P. aeruginosa PAO1, the RND-deficient PΔ6 and their derivatives lacking and overproducing the RND pump MuxABC-OpmB, MFS pumps MfsAB, and MfsCD-OpmG and MdtJI and compared them to PΔ6 cells overproducing MexAB-OprM, MexCD-OprJ, MexEF-OprN and TriABC-OpmH, which represent the three categories of RND pumps based on their expression profiles. We focused on stress conditions typically associated with human infections and previously analyzed in transcriptomics studies (Fig. 1) (28). We found that PΔ6 cells have major growth deficiencies under conditions of acidic pH, low iron, and in the presence of bile salts that are further exaggerated by the inactivation of MuxABC-OpmB. The overproduction of RND transporters complements these growth phenotypes, albeit only partially and in a pump-specific manner. The impact of Mfs and MdtJI transporters both in the presence and absence of RND pumps is moderate and specific to stress conditions. Overall, the growth phenotypes of PAO1 strains are significantly different from the phenotypes of strains with PΔ6 background, suggesting a considerable impact of RND efflux pumps on the growth physiology of P. aeruginosa. Taken together, our results suggest that efflux pumps play specific non-overlapping physiological roles and that these roles do not correlate with their substrate specificities perceived by measuring antibacterial activities of various toxic agents.

FIG 1.

Stress conditions upon infection of mammalian hosts. Stress conditions are from reference (28).

RESULTS

Antibiotic susceptibility of efflux deletion and complemented mutants

In addition to the previously reported P. aeruginosa PAO1 and PΔ6 strains and their derivatives lacking MfsAB, MdtJI, and PΔ6 ΔmexVW ΔmuxABC-opmB (PΔ8 thereafter) (15), we constructed and characterized here PAO1 and PΔ6 strains carrying a deletion of mfsCD-opmG (Table S1). We also cloned the operons encoding various efflux pumps into a shuttle plasmid and expressed the pumps in the respective strains in trans (Table S1). Table 1 shows compounds, the antibacterial activities of which were previously reported to be affected by the expression of these transporters or their homologs in E. coli. All strains were next analyzed for their susceptibilities to several antibiotics representing different structural classes (Table 2).

TABLE 1.

Previously reported putative substrates of efflux pumps used in this study or their homologs from E. coli

Pump	Substrate	References
MexAB-OprM	β-lactams, fluoroquinolones, chloramphenicol, tetracycline, novobiocin, trimethoprim, macrolides, triclosan, ethidium bromide, SDS, aromatic hydrocarbons, thiolactomycin, cerulenin, acylated homoserine lactones	(15, 22, 29, 30)
MexCD-OprJ	β-lactams, fluoroquinolones, chloramphenicol, tetracycline, novobiocin, macrolides, crystal violet, ethidium bromide, acriflavine, SDS, aromatic hydrocarbons, triclosan, expanded-spectrum cephems	(29, 30)
MexEF-OprN	Fluoroquinolones, chloramphenicol, trimethoprim, aromatic hydrocarbons, triclosan	(16, 29)
TriABC-OpmH	Triclosan, SDS, trimethoprima	(31, 32)
MuxABC-OpmB	Aminocoumarins (novobiocin), bile saltsa	(15)
MdtJI	Spermidine (polyamines)b, detergents, bile salts	(27)
MfsAB	Heavy metalsb, carbonyl cyanide m-chlorophenylhydrazone (CCCP)b, nalidixic acidb	(23, 33)
MfsCD-OpmG	Heavy metalsb, carbonyl cyanide m-chlorophenylhydrazone (CCCP)b, nalidixic acidb	(23, 33)

^a This study.

^b Reported for E. coli homolog only.

TABLE 2.

MICs (μg/mL) of antibiotics against PAO1 and indicated efflux-deficient strains

Druga	NOV	CIP	CHF	TMP	AZI	TET	POL	TOB	AMI	CEF
PAO1	>1,024	0.0625	25	128	256	6.25	1	1–2	1–2	1–2
PAO1 (pUCP22)	>1,024	0.0625	25	128	128	6.25	0.5	4	2	R
PAO1 ΔMdtJI	1,024	0.0625	12.5	128	128–256	12.5	0.5–1	0.5	0.5–1	1
PAO1 ΔMfsAB	>1,024	0.0625	25	128	128	12.5	1	2	2	1
PAO1 ΔMfsCD	>1,024	0.125	25	128	128	12.5	1	2	2	1
PAO1 ΔMdtJI (pMdtJ1)	512	0.0625	6.25	128	64	6.25	2	4	2	R
PAO1 ΔMfsAB (pMfsAB)	1,024	0.0625	25	128	128	6.25	1	4	2	R
PAO1 ΔMfsCDG (pMfsCDG)	1,024	0.125	25	128	128	6.25	1	4	2	R
PΔ6	32	0.004	0.8	2	2	0.4–0.8	0.5–1	1–2	1–2	0.125
PΔ6 (pUCP22)	32	0.004	25	2	128	0.4	0.5	4	2	R
PΔ6 ΔMdtJI	32	0.004	0.4–0.8	2	2	0.4	0.5–1	1	1–2	0.0625
PΔ6 ΔMfsAB	32	0.004	0.8	2	2	0.8	1	2	2	0.125
PΔ6 ΔMfsCDG	32	0.004	0.8	2	2	0.4	1	2	2	0.125
PΔ6 ΔMdtJI (pMdtJ)	16	0.004	0.2	2	2	0.2	2	4	4	R
PΔ6 ΔMfsAB (pMfsAB)	32	0.004	0.8	2	2	0.4	1	4	2	R
PΔ6 ΔMfsCDG (pMfsCDG)	64	0.004	0.8	2	2	0.2	1	4	2	R
PΔ6 (pMexAB)	256	0.016	12.5	32	8	6.25	0.5	2	1	R
PΔ6 (pTriABC)	32	0.008	0.8	16	2	0.2	1	2	2	0.25
PΔ6 (pMexCD)	64	0.016	0.8	2	4	0.4	1	2	2	0.5–1
PΔ6 (pMexEF)	128	1–2	256	1,024	2	3.125	1	2	2	0.25
PΔ8	0.25	0.002	0.4	2	2	0.1–0.2	1	2	2	0.0625–0.125
PΔ8 (pMuxABC)	256	0.004	0.4	2	2	0.1–0.2	1	2	2	R

^a AMI, amikacin; AZI, azithromycin; CEF, cefepime; CHF, chloramphenicol; CIP, ciprofloxacin; POL, polymyxin B; R, plasmid resistance; TET, tetracycline; TOB, tobramycin; TMP, trimethoprim; NOV, novobiocin, MfsCDG, MfsCD-OpmG; MexAB, MexAB-OprM; MexCD, MexCD-OprJ; MexEF, MexEF-OprN; MuxABC, MuxABC-OpmB; PAO1, WT; PΔ6, PAO1 ΔmexAB-oprM ΔmexCD-oprJ ΔmexEF-oprN ΔmexJKL ΔmexXY ΔtriABC; PΔ8, PΔ6 ΔmexVW ΔmuxABC-opmB. Values that differ significantly from parental strains are shown underlined and in bold.

The inactivation of neither MfsAB nor MfsCD-OpmG in PAO1 genetic background affected MICs of antibiotics. Likewise, no changes in MICs were found for the MSF pump deletion mutants complemented with the plasmid-borne respective pumps. Thus, neither of these two pumps contribute to the efflux of tested antibiotics.

Surprisingly, PAO1 ΔMdtJI cells became two–four fold more susceptible to cationic antibiotics polymyxin B, tobramycin, and amikacin (Table 2). This susceptibility has been complemented by the plasmid-borne MdtJI, as seen from the MIC values of these cationic antibiotics, which are at or two–fourfold above the values in the parental PAO1 strain. At the same time, cells overproducing MdtJI became more susceptible to novobiocin, chloramphenicol, and azithromycin. Thus, this SMR pump protects the cells against the action of cationic antibiotics, but the mechanism compromises the susceptibility to other antibiotics.

As expected, PΔ6 is notably more susceptible than PAO1 to novobiocin, ciprofloxacin, trimethoprim, tetracycline, azithromycin, and chloramphenicol, whereas no changes in MICs were found for polymyxin B, tobramycin, and amikacin (Table 2). The deletions of MdtJI and Mfs pumps in PΔ6 had no significant (>twofold) effect on MICs of antibiotics. Among the complemented strains, the overproduction of MdtJI again increased the MICs of the cationic antibiotics, suggesting that this function of the MdtJI pump is independent of the presence of major RND pumps. As seen for PΑΟ1 ΔMdtJI (pMdtJI), the MIC of chloramphenicol but not novobiocin or azithromycin decreased in PΔ6 ΔMdtJI (pMdtJI) cells by fourfold as well.

We next complemented the PΔ6 phenotype by the plasmid-borne RND pumps from the three categories: (i) the constitutively expressed MexAB-OprM and TriABC-OpmH, (ii) the regulated by stresses MuxABC-OpmB, and (iii) the mutationally selected by antibiotics MexEF-OprN and MexCD-OprJ (17). The overproduction of MexAB-OprM in PΔ6 has increased the MIC values of all tested antibiotics except the cationic polymyxin B, tobramycin, and amikacin (Table 2). However, this complementation was only partial, with MICs of antibiotics in the wild-type PAO1 being significantly higher than in PΔ6 (pMexAB). This result suggested that additional transporters contribute to the hyposusceptibility of PAO1 to antibiotics, albeit to a lesser extent. The overproduction of another constitutively expressed RND pump, TriABC-OpmH, increased MICs of trimethoprim only and decreased MIC of tetracycline, and the result is consistent with the previously reported narrow substrate specificity of this pump (Table 1) (31).

MexCD-OprJ and MexEF-OprN, the two pumps that are not typically expressed in vitro, also differed dramatically in their substrate specificities (Tables 1 and 2). The overproduction of MexCD-OprJ increased the MIC of the only tested fluoroquinolone ciprofloxacin, whereas the overproduction of MexEF-OprN elevated MICs of novobiocin, ciprofloxacin, chloramphenicol, trimethoprim, and tetracycline. The effect of MexEF-OprN was particularly strong on susceptibilities to chloramphenicol and trimethoprim with MICs of these antibiotics against PΔ6 (pMexEF) significantly higher than the MICs against PAO1 strain.

As previously reported, in comparison to PΔ6, the PΔ8 derivative became hypersusceptible specifically to novobiocin, whereas differences in other antibiotics were within the twofold (Table 2). As, MexVW requires OprM for its function (34), which is deleted in the parental PΔ6 strain, this pump does not contribute to the observed phenotypes. Hence, MuxABC-OpmB defines the PΔ8 phenotype. In agreement, the overproduction of MuxABC-OpmB has complemented the hypersusceptibility of PΔ8 to novobiocin to the level of PΔ6 (pMexAB) but did not affect MICs of other antibiotics.

Thus, based on MIC measurements MuxABC-OpmB, TriABC-OpmH, and MexCD-OprJ are relatively narrow-spectrum efflux pumps, whereas MexAB-OprM and MexEF-OprN can expel a broader spectrum of antibiotics. MdtJI orthogonally affects susceptibility to cationic polymyxin B and aminoglycosides and other antibiotics. No effect on antibiotic susceptibility is found for MfsAB and MfsCD-OpmG pumps.

Contribution to the permeability barrier of P. aeruginosa

We next analyzed whether the constructed deletions and their complemented derivatives affect the permeability barrier of P. aeruginosa. For this purpose, we used a hydrophobic membrane probe N-phenyl-1-naphthylamide (NPN), which is highly fluorescent when incorporated into the inner membrane. The probe reports on both its active efflux and the permeability of the outer membrane in a bacterial growth-independent manner (35, 36). The permeability analyses were done with exponentially growing cells that were induced with IPTG for the same periods of time to insure reproducibility and comparison of different strains. The PAO1 cells are essentially impermeable to NPN, as seen from the lack of any time-dependent changes in its fluorescence (Fig. 2). In contrast, the fluorescence of NPN increases in a time-dependent manner in both PΔ6 and PΔ8 cells, demonstrating that the permeability barrier is compromised in these strains (Fig. 2).

FIG 2.

Intracellular uptake of the NPN in P. aeruginosa and its derivatives. (A) MdtJI deletion and complemented strains. (B) MfsAB deletion and complementation strains. (C) MfsCD-OpmG deletion and complemented strains. (D) PAO1 and PΔ6 strains and PΔ6 producing MexAB-OprM, MexCD-OprJ, MexEF-OprN, TriABC-OpmH, and MuxABC-OpmB pumps. Data represent real-time kinetics of changes in NPN fluorescence (8 µM final external concentration). Error bars refer to data from at least two independent experiments. MfsCDG, MfsCD-OpmG; MexAB, MexAB-OprM; MexCD, MexCD-OprJ; MexEF, MexEF-OprN; MuxABC, MuxABC-OpmB; PAO1, WT; PΔ6, PAO1 ΔmexAB-oprM ΔmexCD-oprJ ΔmexEF-oprN ΔmexJKL ΔmexXY ΔtriABC; PΔ8, PΔ6 ΔmexVW ΔmuxABC-opmB.

Surprisingly, the accumulation of NPN in PAO1 ΔMdtJI cells was significantly lower than in PAO1, suggesting that either the efflux of NPN increased upon deletion of MdtJI or the outer membrane of these cells is less permeable to the probe (Fig. 2A; Table 3). The complemented PAO1 ΔMdtJI (pMdtJI) accumulated even less amounts of NPN suggesting that both the deletion and overproduction of MdtJI reduce the permeability of P. aeruginosa cells for NPN probe. As neither the effect of MdtJI deletion nor its overproduction is seen in PΔ6 ΔMdtJI, the effect is dependent on the presence of RND pumps (Fig. 2A).

TABLE 3.

Steady state accumulation levels of NPN in P. aeruginosa and its derivatives

Strain	Steady state (µM)
PAO1a	0.139 ± 0.0163
PAO1 ΔMdtJI	0.078 ± 0.0073
PAO1 ΔMfsAB	0.119 ± 0.0009
PAO1 ΔMfsCDG	0.142 ± 0.0072
PAO1 ΔMdtJI (pMdtJI)	0.064 ± 0.0009
PAO1 ΔMfsAB (pMfsAB)	0.146 ± 0.0003
PAO1 ΔMfsCDG (pMfsCDG)	0.109 ± 0.0006
PΔ6	0.746 ± 0.0666
PΔ6 ΔMdtJI1	0.833 ± 0.0213
PΔ6 ΔMfsAB	0.749 ± 0.0694
PΔ6 ΔMfsCDG	0.898 ± 0.0159
PΔ6 ΔMdtJI (pMdtJI)	0.712 ± 0.0011
PΔ6 ΔMfsAB (pMfsAB)	0.811 ± 0.0223
PΔ6 ΔMfsCDG (pMfsCDG)	0.713 ± 0.0064
PΔ6 (pMexAB)	0.199 ± 0.0152
PΔ6 (pTriABC)	0.296 ± 0.0338
PΔ6 (pMexCD)	0.367 ± 0.0216
PΔ6 (pMexEF)	0.195 ± 0.0137
PΔ8	0.747 ± 0.1336
PΔ8 (pMuxABC)	0.904 ± 0.0428

^a For strain descriptions, see Table 2.

The PΔ6 ΔMfsCD but not PAO1 ΔMfsCD cells have accumulated higher levels of NPN, as seen from the increased NPN fluorescence in PΔ6 ΔMfsCD cells (Fig. 2C). This result suggested that NPN could be a substrate of MfsCD-OpmG pump, but the presence of RND pumps hides its activity. Indeed, the overproduction of MfsCD-OpmG reduced NPN accumulation in both PAO1 ΔMfsCD and PΔ6 ΔMfsCD cells, albeit modestly. In contrast, neither deletion nor overproduction of its homolog MfsAB affected the accumulation of this probe (Fig. 2B).

The overproduction of the RND pumps had the strongest effect on the intracellular accumulation of NPN (Fig. 2D). The overproduction of either MexAB-OprM or MexEF-OprN restored the barrier back to the PAO1 levels because NPN is a known substrate of these two pumps. The reduction in the NPN accumulation in the cells overproducing either MexCD-OprJ or TriABC-OpmH was intermediate between the PAO1 and PΔ6 levels. In contrast, the overproduction of MuxABC-OpmB had no effect on the NPN accumulation when compared to the parent strain PΔ8. Thus, the efficiency in efflux of NPN correlates with the MIC measurements. Both assays show that RND pumps with the broadest substrate specificities are the most efficient in protecting the permeability barrier of P. aeruginosa. In addition, our results show that MfsCD-OpmG pump contributes to efflux of NPN.

Growth at high temperature and osmolarity

To compare growth phenotypes of various strains, we first analyzed their growth curves under typical laboratory conditions (LB broth, pH = 7.2, 37°C, with aeration). As with susceptibility measurements, we found no effect of an empty vector on growth of PAO1 and PΔ6 strains under normal conditions and under stresses (Fig. S1). To quantitatively compare the growth curves, we calculated the area under the curve (AUC) values for each curve (Table 4). We found that the AUC values of MfsAB, MfsCD-OpmG, and MdtJI deletion strains in the PAO1 background were all within 10% of the wild type (Table 4; Fig. 3A through C). The growth of PΔ6 was somewhat reduced with the AUC value 77% of that for PAO1. The growth of PΔ6 ΔMfsAB cells further decreased to 60% of the PAO1 primarily due to early entry into the stationary phase, suggesting that MfsAB transporter contributes to the PΔ6 growth under typical laboratory conditions (Fig. 3B).

TABLE 4.

AUC for indicated strains under host-relevant stress conditions^a

	LB	Bile salts 0.125%	Bile salts 0.5%	Iron 125 µM	Iron 250 µM	pH 4.6	NaCl 0.5M	41°C
PAO1	10.00	7.79	3.73	9.94	9.08	5.69	6.69	10.43
PAO1 ΔMfsAB	10.08	8.26	4.51	9.28	8.61	6.35	7.31	11.18
PAO1 ΔMfsAB (pMfsAB)	9.85	8.15	4.09	9.32	8.83	5.04	7.27	10.91
PAO1 ΔMfsCDG	9.97	7.96	4.27	9.10	8.55	3.97	6.09	9.76
PAO1 ΔMfsCDG (pMfsCDG)	9.19	6.88	3.89	8.41	7.74	2.60	5.20	9.50
PAO1 ΔMdtJI	9.01	6.46	2.21	9.00	8.46	5.04	5.01	9.51
PAO1 ΔMdtJI (pMdtJI)	8.42	5.99	2.84	7.94	7.29	5.05	4.33	8.90
PΔ6	7.67	2.81	0.67	3.24	0.08	2.56	4.78	8.18
PΔ6 ΔMfsAB	4.97	2.44	2.22	1.41	0.07	2.97	5.54	7.65
PΔ6 ΔMfsAB (pMfsAB)	6.18	2.83	1.89	2.70	0.09	2.38	3.42	8.50
PΔ6 (pMexAB)	6.57	5.64	3.85	5.63	4.06	2.66	6.59	8.37
PΔ6 (pMexCD)	8.53	4.49	1.20	6.77	2.37	3.23	4.03	8.81
PΔ6 (pMexEF)	7.40	6.49	4.15	7.12	6.88	3.00	5.24	7.46
PΔ6 (pTriABC)	7.11	5.02	1.25	4.18	0.01	2.01	5.19	6.93
PΔ6 ΔMfsCDG	7.98	2.43	0.69	2.25	ng	0.44	4.25	7.40
PΔ6 ΔMfsCDG (pMfsCDG)	7.67	3.68	1.66	2.10	ng	0.20	3.82	6.04
PΔ6 ΔMdtJI	7.94	2.89	0.40	2.86	0.01	3.30	4.74	7.67
PΔ6 ΔMdtJI (pMdtJI)	8.03	2.03	0.15	3.33	ng	3.38	4.18	7.21
PΔ8	7.97	ng	0.05	2.59	0.04	2.33	5.53	7.49
PΔ8 (pMuxABC)	8.58	4.08	1.03	3.47	1.95	1.89	5.26	8.05

^a For strain descriptions, see Table 2. ng, no growth.

FIG 3.

Growth curves of P. aeruginosa and its derivatives in LB. (A) MdtJI deletion and complemented strains. (B) MfsAB deletion and complementation strains. (C) MfsCD-OpmG deletion and complemented strains. (D) PΔ6 and PΔ8 strains carrying the indicated plasmid-borne RND pumps. Overnight cell cultures were diluted into fresh LB medium to OD₆₀₀ of 0.001 and inoculated at 5 × 10³ cells into a 96-well plate. Optical densities were monitored in the Tecan Spark 10M plate reader at 37°C every 30 min. The complemented strains were induced with 0.1 mM isopropyl β-D-1 thiogalactopyranoside (IPTG) for the expression of the plasmid-borne pumps. Error bars represent the standard error of at least two biological replicates, each with at least two technical replicates. For strains, see Fig. 2 legend and Table S1.

We next analyzed the growth at the elevated 41°C temperature and found that most of the strains grew slightly better at 41°C than at 37°C (Table 4; Fig. S2). Only PΔ6 ΔMfsCD (pMfsCD) strain had the AUC value ~25% lower than that of the parent in PΔ6 ΔMfsCD, suggesting that the overproduction of MfsCD-OpmG has a negative effect on P. aeruginosa growth at high temperatures.

The high osmolarity (0.5 M NaCl) reduced the growth of most strains by ~30% (Table 4; Fig. S3). In some cases, for example, in PAO1 ΔMdtJI, PAO1 ΔMdtJI (pMdtJI), or PΔ6 ΔMfsAB (pMfsAB), the effect was stronger, and cells lost 50% of their AUC values. In contrast, the overproduction of MexAB-OprM has complemented the growth deficiency of PΔ6, suggesting that the activity of this efflux pump is beneficial to cells under osmotic stress.

Growth in the presence of bile salts

Bile salts, as other detergents, are known to be the substrates of the RND efflux pumps. In agreement, at both tested concentrations 0.125% (Fig. S4) and 0.5% (Fig. 4), PΔ6 growth is notably slower than PAO1 (Table 4). Inactivation of MdtJI but not Mfs pumps exaggerated the PΔ6 growth deficiency in the presence of bile salts, as seen from the loss in PΔ6 ΔMdtJI growth in the presence of 0.5% bile salts (Fig. 4A through C). In contrast, PΔ6 ΔMfsAB became less susceptible than the parental PΔ6 to 0.5% bile salts. Neither the plasmid-borne MdtJI nor MfsAB was able to complement the phenotypes of the corresponding deletion strains, suggesting that these two pumps cannot expel bile salts, and phenotypes are caused by other functions. The changes in the expression of MfsCD-OpmG led to a different outcome. Although the growth of PΔ6 ΔMfsCD cells matched that of the parental PΔ6, the growth of the pump overproducer PΔ6 ΔMfsCD (pMfsCD) in 0.5% bile salts was notably improved (Fig. 4C). Thus, MfsCD-OpmG could contribute to efflux of bile salts.

FIG 4.

Growth curves of P. aeruginosa and its derivatives exposed to 0.5% bile salts (bile stress). (A) MdtJI deletion and complemented strains. (B) MfsAB deletion and complementation strains. (C) MfsCD-OpmG deletion and complemented strains. (D) PΔ6 and PΔ8 strains carrying the indicated plasmid-borne RND pumps. Experiments were carried out as described in Fig. 3 legend and Methods but in the presence of 0.5% bile salts. Error bars represent the standard error of at least two biological replicates, each with at least two technical replicates. For strains see Fig. 2 legend.

The overproduction of either MexEF-OprN or MexAB-OprM fully complemented the PΔ6 hypersusceptibility to bile salts and even made the cells more resistant than PAO1 (Fig. 4D). In contrast, neither MexCD-OprJ nor TriABC-OpmH was able to improve the growth of PΔ6. Surprisingly, PΔ8 cells were even more susceptible to bile salts than PΔ6 and failed to grow not only at 0.5% (Fig. 4D) but also at 0.125% bile salts (Fig. S4). The overproduction of MuxABC-OpmB has rescued the hypersusceptibility of PΔ8 to the bile salts, suggesting that this pump is able to expel bile salts.

Taken together, the three RND pumps MexEF-OprN, MexAB-OprM, and MuxABC-OpmB and the MfsCD-OpmG pump can contribute to survival in the presence of bile salts by expelling these detergents from cells. MdtJI and MsfAB are unlikely to expel bile salts, and the identified phenotypes of strains carrying deletions of these pumps are caused by the loss of other functions of these pumps.

Adaptation and survival in acidic pH

We next analyzed the growth curves of the constructed strains at acidic pH (pH 4.6) (Fig. 5) (28). We found that at pH 4.6 the AUC of the wild-type PAO1 cells was reduced by 43% (Table 4; Fig. 5), because of the extended lag phase. Once resumed, the growth rate was the same as at pH 7.2 (Fig. 3). The growth of PΔ6 has declined by 66%, suggesting that RND pumps contribute to the acidic pH adaptation but are not required for it. No further effect on the growth at pH 4.6 was seen for PΔ8. In agreement, we found that the plasmid-borne overproduction of MuxABC-OpmB had no effect under this stress. Similarly, the growth phenotypes of PΔ6 (pMexEF) and PΔ6 (pMexCD) were indistinguishable from PΔ6 cells (Fig. 5D). Although PΔ6 (pTriABC) cells had the slowest overall growth among the complemented strains, the differences were not statistically significant (Fig. 5D). The phenotype of PΔ6 (pMexAB) was somewhat different; that is, these cells had the longest lag phase ~12 h, but once recovered the growth rate was comparable to the wild-type PAO1 cells. Thus, the overproduction of MexAB-OprM is stressful during adaptation to acidic pH but beneficial for the growing cells.

FIG 5.

Growth curves of P. aeruginosa and its derivatives exposed to pH = 4.6 (acidic stress). (A) MdtJI deletion and complemented strains. (B) MfsAB deletion and complementation strains. (C) MfsCD-OpmG deletion and complemented strains. (D) PΔ6 and PΔ8 strains carrying the indicated plasmid-borne RND pumps. Experiments were carried out as described in Fig. 3 legend and Methods. Error bars represent the standard error of at least two biological replicates, each with at least two technical replicates. For strains, see Fig. 2 legend and Table S1.

Neither inactivation nor overproduction of MdtJI in the wild-type and PΔ6 backgrounds had a significant effect on adaptation to pH 4.6 (Fig. 5A). The same no effect was found for MfsAB pump (Fig. 5B). In contrast, the MfsCD-OpmG deletion extended the lag phase of PAO1 to ~12 h, whereas PΔ6 ΔMsfCD cells inoculated into the pH 4.6 medium required more than 15 h to resume growing (Fig. 5C). Thus, MsfCD-OpmG contributes to the pH 4.6 adaptation independently from RND pumps. The overproduction of MsfCD-OpmG from the plasmid did not rescue the phenotype and even had a negative effect on the pH 4.6 adaptation.

Low-iron stress

One of the strongest transcriptional responses to the inactivation of RND pumps was the overproduction of siderophores, indicating that PΔ6 cells are under a low-iron stress (15). To remove iron from the medium, we added the iron chelator 2,2′-dipyridyl to the growth medium, and the approach is broadly used to induce iron starvation in vitro (37, 38). Despite the overproduction of siderophores and activation of the iron starvation response, PΔ6 cells failed to grow in the presence of 250 µM 2,2′-dipyridyl (Fig. 6) and were notably slower than PAO1 in the medium supplemented with 125 µM 2,2′-dipyridyl (Fig. S5). This growth deficiency was further exaggerated in PΔ8 lacking the MuxABC-OpmB pump. In addition, the inactivation of MdtJI or MfsCD-OpmG in either PAO1 or PΔ6 genetic backgrounds and MfsAB only in PΔ6 further diminished the ability of cells to grow in the presence of 2,2′ dipyridyl. Thus, various efflux pumps contribute to growth under low-iron conditions. However, only selected pumps were able to complement these growth deficiencies.

FIG 6.

Growth curves of P. aeruginosa and its derivatives 250 µM 2,2′-dipyridyl (limited iron stress). (A) MdtJI deletion and complemented strains. (B) MfsAB deletion and complementation strains. (C) MfsCD-OpmG deletion and complemented strains. (D) PΔ6 and PΔ8 strains carrying the indicated plasmid-borne RND pumps. Experiments were carried out as described in Fig. 3 legend and Methods. Error bars represent the standard error of at least two biological replicates, each with at least two technical replicates. For strains, see Fig. 2 legend and Table S1.

The expression of MfsAB has complemented the growth defects of PΔ6 ΔMfsAB cells in low-iron conditions to the level of the parental PΔ6 strain (Fig. 6B). Hence, MfsAB plays a specific role under this condition. In contrast, the plasmid-borne expression of MdtJI and MsfCD-OpmG had a negative effect as seen from the declined growth of PAO1 ΔMdtJI (pMdtJI) and PAO1 ΔMsfCD (pMsfCD) cells (Fig. 6A and C; Fig. S5).

Among the RND pumps, MexAB-OprM, MexEF-OprN, MexCD-OprJ, and MuxABC-OpmB all promoted the growth of PΔ6 and PΔ8 strains, respectively, in low-iron conditions, albeit to various extent (Fig. 6D). In contrast, the expression of TriABC-OpmH pump had no effect under these conditions. We conclude that substrate specificities of efflux pumps are important for their functions under stress.

Principal component analysis (PCA) of P. aeruginosa fitness under stress conditions

We next carried out the PCA of the collected AUC values to uncover relationships between P. aeruginosa phenotypes associated with the inactivation and the overproduction of efflux pumps. We found that the first two principal components (PC1 and PC2) cover ~82% of the total variance in the data set (Fig. 7). PC1 separates largely the growth phenotypes of the strains with PAO1 background from those constructed in PΔ6 and PΔ8, highlighting a significant impact of RND inactivation on the growth physiology of P. aeruginosa under different conditions. PAO1 ΔMdtJI, PAO1 ΔMdtJI (pMdtJI), PΔ6 ΔMdtJI (pMdtJI), PΔ6 ΔMfsAB, PΔ6 (pMexAB), PΔ6 (pMexEF), and PΔ6 (pMexCD) strains are clearly separated from each other and the rest of the strains. Thus, these genetic modifications and plasmid complementation generate phenotypes that are distinct from both the parental PAO1 and PΔ6 strains. The PΔ8 phenotype and that of its complemented derivative PΔ8 (pMuxABC) cluster together with PΔ6 and its derivatives PΔ6 ΔMfsCD and PΔ6 ΔMdtJI, all of which had very similar growth physiology.

FIG 7.

PCA of the growth fitness of P. aeruginosa strains under different stress conditions. PCA plot of AUC values obtained for PAO1 and PΔ6 and their derivatives. For abbreviations, see in Fig. 2 legend and Table S1.

Taken together, our results suggest that efflux pumps play specific, non-interchangeable functions in P. aeruginosa cell physiology. The expression of a single RND efflux pump, even with such a broad substrate specificity as MexAB-OprM or MexEF-OprN, does not fully complement the PΔ6 phenotype but enables bacterial growth under certain stress conditions. The narrow specificity pumps are beneficial only under specific conditions.

DISCUSSION

One of the important questions in P. aeruginosa physiology is why this bacterium possesses such a large arsenal of drug efflux pumps. In this study, we analyzed growth phenotypes associated with deletions and overproduction of efflux transporters from three highly structurally diverse families of proteins. MdtJI from the SMR family acts across the inner membrane, whereas MFS and RND transporters are multi-component pumps expelling their substrates across both the inner and the outer membranes. The substrate specificities of these pumps as assessed from antibacterial activities of various antibiotics are divergent, yet some MFS and RND pumps recognize and expel the same substrates. We found that the most promiscuous RND pumps such as MexEF-OprN and MexAB-OprM are also integrated into more diverse survival strategies, whereas the narrow spectrum pumps contribute under specific conditions (Table 4).

MdtJI was reported to expel intracellular polyamines, such as spermidine as well as some detergents including bile salts and antibiotics. The extracellular spermidine is thought to bind to polyanionic lipopolysaccharides(LPS) and modify the permeability of the outer membrane. Indeed, we found that PAO1 ΔMdtJI cells are more susceptible and reciprocally PAO1 ΔMdtJI (pMdtJI) less to polycationic antibiotics such as polymyxin B and aminoglycosides (Table 1). As efflux across the inner membrane is unlikely to protect against polymyxin B, the MdtJI-dependent changes in antibiotic susceptibilities are likely due to efflux of spermidine across the outer membrane and the reduced negative charge of LPS. Interestingly, both PAO1Δ MdtJI and the complemented PAO1 ΔMdtJI (pMdtJI) strains are less permeable to the non-ionic membrane probe NPN (Fig. 2). This effect was diminished in PΔ6 background, suggesting that MdtJI activity could be synergistic with RND-dependent efflux.

The phenotypes of MFS tripartite pumps were modest but specific to MfsAB or MfsCD-OpmG pumps. Our results suggest that MfsAB engages an OMP component of an RND pump. The deletion of MfsAB was beneficial in both PAO1 and PΔ6 cells growing in the presence of bile salts or in pH 4.6, pointing onto increased activity of alternative pumps (Fig. 4B and 5B). However, PΔ6 ΔMfsAB cells were less fit than the parental PΔ6 under normal laboratory conditions and in the presence of 125 µM 2,2′-dipyridyl (Fig. 3B; Fig. S4B). Unlike with bile salts and pH 4.6, the low iron defect was complemented by the MfsAB expression in trans. Thus, in the absence of major RND pumps, the efflux activity of MsfAB contributes to growth under low-iron conditions.

The phenotypes associated with the deletion and overproduction of MfsCD-OpmG were different from those of MsfAB. The inactivation of MfsCD-OpmG notably diminished the growth of both PAO1 and PΔ6 cells at pH 4.6, and its overexpression improved the growth of PΔ6 (pMfsCD) in the presence of bile salts. This finding is similar to the recently reported non-overlapping roles of the tri-partite MFS-type efflux pumps in the physiology and stress survival of Acinetobacter baumannii (39). Interestingly, E. coli and Shigella spp. also contain two MFS-type efflux pumps EmrAB-TolC and EmrKY-TolC that do not contribute to intrinsic levels of antibiotics resistance, but their overproduction is needed for survival under various stresses (23, 25, 40–42). It remains unclear what are the substrates of these pumps under different physiological conditions, but their conservation across gram-negative bacteria points onto the non-overlapping but fundamental functions.

In contrast, some RND pumps of P. aeruginosa such as MuxABC-OpmB and TriABC-OpmH are not conserved in genomes of gram-negative bacteria and found only in its close relatives. Unlike Mex transporters, these pumps have peculiar substrate specificities and selectively change P. aeruginosa susceptibilities only to a few agents (Tables 1 and 2). These pumps also differ from Mex pumps in their sequences and structures, with MuxABC-OpmB comprising the two non-interchangeable RND transporters MuxB and MuxC and TriABC-OpmH comprising the two non-interchangeable periplasmic MFPs TriA and TriB. The RND components of these complexes are evolutionary distant from each other and Mex pumps and form the cores of two distinct clusters (43). The strongest phenotype of MuxABC-OpmB is that it enables the growth of PΔ8 under low iron conditions (Fig. 6D). This result is consistent with previous reports that the homolog of MuxABC-OpmB in Pseudomonas putida (named MdtABC-OpmB) might be involved in recycling of the siderophore pyoverdine (44). We also found that it recovers the PΔ8 growth in the presence of bile salts but only to the level of PΔ6, suggesting that it does not complement or replaces the functions of other pumps (Fig. 4D).

Unlike MuxABC-OpmB, TriABC-OpmH pump is constitutively expressed in the wild-type PAO1 (15), and its overproduction is selected by exposure to triclosan (32). Here, we found that the overproduced TriABC confers hyposusceptibility to trimethoprim (Table 1) but does not protect against various stresses and may interfere with the adaptation of PΔ6 cells to pH 4.6 (Fig. 5D). Both triclosan and trimethoprim are excellent substrates of the broad-spectrum Mex pumps, and MexAB-OprM pump alone confers high-level triclosan resistance (MIC, >1 mg/mL) (32). Thus, it is unlikely that TriABC-OpmH contributes to protection against these antibacterials in PAO1 cells.

The three Mex pumps analyzed here, MexAB-OprM, MexEF-OprN and MexCD-OprJ, are homologous to each other and belong to the same evolutionary cluster (43), but only MexAB-OprM is expressed constitutively in PAO1 strain (15). These pumps protect cells from various antibiotics (Tables 1 and 2), and their broad substrate specificity is also reflected in their contributions to various infection-related stresses (Table 4), albeit to a different extent. MexEF-OprN and MexAB-OprM were equally and highly effective in protection against bile salts, but on the one hand, their roles diverged at pH 4.6 or low iron, with MexEF-OprN being more efficient in the presence of iron chelator and MexAB-OprM supporting the growth at pH 4.6. MexCD-OprJ, on the other hand, showed only weak protection against bile salts and at low 125 µM concentration of 2.2′-dipyridyl (Table 4).

The current results highlight the significance of efflux pumps in bacterial physiology and their role in the survival of bacteria when exposed to host-relevant stress conditions. The Mex efflux pumps with the broadest substrate specificities appear to support P. aeruginosa growth under more diverse conditions, whereas the narrow specificity efflux pumps of RND, MFS, and SMR families are integrated into specific physiological responses. One of the limitations of this study was that we do not have experimental means to determine the number of active efflux complexes assembled from their components. Therefore, some of the effects could be hidden or exaggerated by differences in the assembly or numbers of active efflux pumps. In addition, the overproduction of some but not other efflux pumps due to their numbers or efficiencies could trigger non-specific cell envelope stress responses that contribute to the observed phenotypes. Future studies of the roles of efflux pumps in stress responses and their integration into regulatory networks of P. aeruginosa could lead to the identification of specific molecules transported by these important molecular machines.

MATERIALS AND METHODS

Strains and plasmids

Strains and plasmids used in these studies are shown in Table S1. Unless indicated otherwise, P. aeruginosa strains were grown in Luria Bertani Broth (LB) (10 g tryptone, 5 g yeast extract, 5 g NaCl per liter, pH 7.0) at 37°C with shaking. Complemented strains carrying plasmid, were grown in LB-containing gentamicin (30 µg/mL WT, 15 µg/mL Δ6). To overexpress selected genes, media were supplemented with 0.1 mM IPTG.

Construction of P. aeruginosa knockout and complemented strains

Deletion mutants PAO1 ΔMfsAB, PΔ6 ΔMfsAB, PAO1 ΔMdtJI, PΔ6 ΔMdtJI, and PΔ8 (Table S1) were constructed previously (15). Deletion mutants ΔMfs were generated from PAO1 WT and PΔ6 P. aeruginosa through two-step allelic exchange as described by Hmelo et al. (45). Suicide vectors pEXG2 carrying PCR fragments located up- and downstream of targeted genes, were constructed using the Gibson Assembly technique. Overlapping flanking primers were designed to amplify the upstream and the downstream 500 bp DNA sequences of MfsCD (Table S2). Two fragments for the construct were then assembled into the prepared vector pEXG2 linearized by PCR (pEXG2_vector_FOR/REV; Table S2) following the Gibson Assembly reaction procedure (GeneArt Gibson Assembly HiFi Cloning Kit, Thermo Fisher Scientific, USA). Constructed suicide vectors (Table S1) were then conjugationally transferred into recipient P. aeruginosa strains PAO1 WT and PΔ6 to create the desired deletions (45). Homologous recombination resulted in Gm-resistant single-crossover mutants in which the plasmids were integrated site-specifically into the chromosome. Subsequently, double-crossover mutants were isolated directly using sucrose-mediated counter-selection on the no salt LB agar plates containing 15% sucrose (8). Deletions were finally confirmed by PCR.

Complementation plasmids for MdtJI, MfsAB, MfsCD, MexAB-OprM, MuxABC-OpmB, MexCD-OprJ, MexEF-OprN, TriABC-OpmH were constructed in the pUCP22 or pBSPII (46). To amplify genes indicated above and involved in transport in P. aeruginosa, sets of primers were designed with the following restriction sites as indicated in Table S2. DNA fragments were amplified by PCR using the genomic DNA of P. aeruginosa PAO1 as a template. The corresponding products were cloned into the pUCP22 vector (Table S1). Electroporation was used to construct PAO1 and PΔ6 cells overproducing chosen genes.

Growth under infection relevant stress conditions

In order to investigate stress response of P. aeruginosa and related mutants (Table S1), we checked their growth profiles under host-related conditions as reported by Avican et al. with modifications (47). Overnight cultures were diluted to OD600 of 0.001 and inoculated at 5 × 10³ cells into a 96-well plate containing LB broth and exposed to five stresses separately (Table S3). Optical densities were monitored in Tecan Spark 10M plate reader every 30 min at 37°C or 41°C when indicated. Experiments were performed in at least two replicates, and data were collected from two independent experiments. AUC (growth potential) was used as an additional growth metric which correlated with both the growth rate and the cell density (48, 49) and was calculated using Formula 1. Multivariate statistical techniques (PCA) were applied to investigate the association between strains depending on their physiological properties (growth under stress conditions, NPN uptake, MICs). Data for conducting PCA were normalized by scaling and centering to the mean and dividing by standard deviation. All codes, unless indicated, were written in house using RStudio 2022.07.2 Build 576, and R version 4.2.2.

$${\displaystyle \mathrm{A}\mathrm{U}\mathrm{C}=\left(\mathrm{O}{\mathrm{D}}_{\mathrm{n}+1}+\mathrm{O}{\mathrm{D}}_{\mathrm{n}}\right)/2\times \left({\mathrm{T}}_{\mathrm{n}+1}-{\mathrm{T}}_{\mathrm{n}}\right)}$$ (1)

MIC

MIC determination was carried out using the twofold broth dilution method described previously (16). When needed, the expression of transport-related genes was induced by the addition of 0.1 mM IPTG. Overnight-grown cells induced with IPTG are reinoculated into a fresh medium supplemented with an antibiotic and the inducer.

NPN uptake

The fluorescence-based permeability assay was performed according to the protocol published by Krishnamoorthy et al. (12). Briefly, cells from frozen stocks were inoculated into LB medium and incubated for 16 h at 37°C. Cells were then subcultured into a fresh 20 mL volume of LB medium and grown at 37°C to an optical density at 600 nm (OD600) of 0.3. The cells were then induced with 0.1 mM IPTG and grown to an OD600 of 1.0, collected by centrifugation at 4,000 rpm for 20 min at room temperature, and washed in 15 mL HEPES-KOH buffer (50 mM; pH 7.0) containing 1 mM magnesium sulfate and 0.4 mM glucose (HMG buffer). The OD of cells in HMG buffer was adjusted to ~1.0 and kept at room temperature during the experiment. The experiment was performed in fluorescence mode in a temperature-controlled microplate reader (Tecan Spark 10M multimode microplate reader equipped with a sample injector). Fluorescence data collected using intensities from NPN uptake experiments were plotted against time and normalized to the emission before cells were added.

DATA AVAILABILITY

All data generated during this study are included in the paper. Strains and plasmids are available upon request.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jb.00054-24.

Supplemental material. jb.00054-24-s0001.pdf.

Tables S1 to S3; Figures S1 to S5.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.