Repeated isolation of an antibiotic-dependent and temperature-sensitive mutant of Pseudomonas aeruginosa from a cystic fibrosis patient

Abstract

Background

Bacteria adapt to survive and grow in different environments. Genetic mutations that promote bacterial survival under harsh conditions can also restrict growth. The causes and consequences of these adaptations have important implications for diagnosis, pathogenesis, and therapy.

Objectives

We describe the isolation and characterization of an antibiotic-dependent, temperature-sensitive Pseudomonas aeruginosa mutant chronically infecting the respiratory tract of a cystic fibrosis (CF) patient, underscoring the clinical challenges bacterial adaptations can present.

Methods

Respiratory samples collected from a CF patient during routine care were cultured for standard pathogens. P. aeruginosa isolates recovered from samples were analysed for in vitro growth characteristics, antibiotic susceptibility, clonality, and membrane phospholipid and lipid A composition. Genetic mutations were identified by whole genome sequencing.

Results

P. aeruginosa isolates collected over 5 years from respiratory samples of a CF patient frequently harboured a mutation in phosphatidylserine decarboxylase (psd), encoding an enzyme responsible for phospholipid synthesis. This mutant could only grow at 37°C when in the presence of supplemented magnesium, glycerol, or, surprisingly, the antibiotic sulfamethoxazole, which the source patient had repeatedly received. Of concern, this mutant was not detectable on standard selective medium at 37°C. This growth defect correlated with alterations in membrane phospholipid and lipid A content.

Conclusions

A P. aeruginosa mutant chronically infecting a CF patient exhibited dependence on sulphonamides and would likely evade detection using standard clinical laboratory methods. The diagnostic and therapeutic challenges presented by this mutant highlight the complex interplay between bacterial adaptation, antibiotics, and laboratory practices, during chronic bacterial infections.

Introduction

Pseudomonas aeruginosa is a ubiquitous, Gram-negative organism that inhabits a diverse range of environmental niches, including soil and aquatic environments, plants, animals, humans and abiotic surfaces. Its versatility is attributable to its large genome (5–7 Mbp), encoding for a remarkable degree of metabolic flexibility, a diverse array of virulence factors, and resistance to various external threats, including antibiotics, which promote survival of the organism when challenged.¹P. aeruginosa is also highly adaptable as a result of both genotypic and phenotypic alterations, often persisting in diverse and inhospitable environments.

P. aeruginosa frequently infects the respiratory tracts of people with cystic fibrosis (CF), adapting genetically and phenotypically during intermittent and persistent infections. These adaptations can impact many different bacterial characteristics, such as growth rate, exopolysaccharide production, motility, metabolism, antimicrobial susceptibility, cell-to-cell communication and virulence.² Evidence suggests that these adaptations occur in response to specific selective pressures, including the physicochemical state of airway secretions, immune challenge, and antibiotic exposure.^2,3 In recent studies, several in vitro P. aeruginosa adaptive phenotypes were linked to infection stage, increased pulmonary exacerbation rates, and failure to eradicate with antipseudomonal therapy,^4,5 indicating the important relationships between these adaptations and clinical outcomes. Chronic P. aeruginosa infection elicits a severe inflammatory response that contributes to progressive lung function deterioration and is associated with worse outcomes for these patients.^6–9 Therefore, optimizing detection and treatment of this organism are key goals towards improving health outcomes of CF patients.

P. aeruginosa mutants have been isolated from in vivo and in vitro sources exhibiting impaired in vitro growth rates, likely reflecting an in vitro fitness cost associated with a specific benefit. For example, metabolic mutants that require amino acid supplementation for in vitro growth have been detected in CF samples.^10,11 Mutants with alterations in LPS synthesis and severe growth impairments emerge during biofilm development in vitro.¹² A single case of a P. aeruginosa isolate that required polymyxin for continued or optimal growth, referred to as antibiotic-dependence, has even been reported.¹³ In this article, we describe the repeated isolation of P. aeruginosa from a CF patient with a mutation resulting in both an antibiotic-dependent and temperature-sensitive growth phenotype. These isolates illuminate important diagnostic and therapeutic challenges posed by P. aeruginosa adaptations.

Materials and methods

CF sample culture

Respiratory samples were collected from CF patient 0122 during routine care at Seattle Children’s Hospital (SCH). Sputum samples were processed using sputolysin (Calbiochem) as described,¹⁴ serially diluted in PBS, and cultured on MacConkey agar (Remel). Plates were incubated at 35°C for up to 72 h for the isolation and enumeration of P. aeruginosa. During one of patient 0122’s clinic visits with SCH IRB approval, sputum was collected and processed as described above. The sample was cultured on two separate sets of MacConkey agar and a single set of trypticase soy agar with 5% (vol/vol) sheep blood (Remel, BAP). The first MacConkey set was incubated at 35°C for 72 h according to standard CF microbiological practices. The second set of MacConkey and BAP were incubated at 37°C for 72 h. Incubators were periodically monitored with temperature probes to ensure consistent temperatures. Representative morphotypes (100 colonies each) were selected from both media, and species were identified using biochemical methods and 16S rRNA sequencing using primers 8F and 1492R (Table S1, available as Supplementary data at JAC Online) with conditions according to published methods.¹⁵

PFGE

Clonality of the P. aeruginosa isolates was determined by PFGE as described.¹⁶ Genomic DNA was digested with 30 U of SpeI (Roche) and restriction fragments were separated in a 1% SeaKem Gold agarose gel using a CHEF DR-III (Bio-Rad) using conditions described previously.¹⁷ Restriction patterns were analysed using BioNumerics software (v6.5; Applied Maths).

Antibiotic susceptibility testing

Susceptibility testing was performed by disc diffusion using CLSI methods¹⁸ or by Etest (BioMérieux) according to the manufacturer’s instructions. In select experiments, sulfamethoxazole, trimethoprim or para-aminobenzoic acid (Sigma) were added to blank discs (BD) at various concentrations. Additional testing was performed on Trypticase soy, Mueller-Hinton (MH), Luria-Bertani, Brain-heart infusion, MacConkey, Cetrimide, and Pseudomonas Isolation (PI) agars. Plates were incubated at 30°C, 35°C, 37°C, or 42°C for 24 h. Isolates were inoculated into cation-adjusted Muller-Hinton broth in the presence or absence of sulfamethoxazole at a final bacterial concentration of ∼2 × 10⁵ cfu/mL. Viable counts were performed after incubation at different temperatures (30°C, 35°C and 37°C) for 24 h.

Phenotype microarray (PM)

The stimulator effects of different antibiotics on growth of 0122-320 was assessed in Biolog’s (Hayward, CA) 96-well phenotype microarray platform according to the manufacturer’s instruction and using chemical sensitivity plates PM11-PM20. Bacterial respiration was monitored for 24 h at 37°C using the redox sensitive dye, tetrazolium violet.

Genomic sequencing

Genomic DNA was isolated using the DNeasy Ultraclean Microbial Kit (Qiagen) according to the manufacturer’s protocol. Whole genomic shotgun sequencing libraries were prepared for isolates following published protocols,¹⁹ mutations were identified using a strategy of combined de novo genome assembly and short read mapping as described.²⁰ Briefly, draft genomes were assembled for 0122-320 and 0122-320R using ABySS²¹ and used to identify the closest matched reference genome available from GenBank as ascertained by BLAST,^19,22 followed by variant calling (nucleotide variants, insertions, deletions) as described.¹⁹

Complementation of isolate 0122-320 with psd from wild-type PACS2 and isolate 0122-15

psd and its ribosomal binding site from P. aeruginosa wild-type strain PACS2 and clinical isolate 0122-15 were amplified by PCR using primers EcoRI-PACS2psdRBS-F and XbaI-PACS2psdRBS6xHisStop-R (Table S1), and the products were cloned into vector pJN105 to create pJN105-psd and pJN105-psd (0122-15). Psd production was verified by Western blot to detect a His-tagged protein of the expected molecular weight (32 kDa). The empty and complementation vectors were electroporated into isolate 0122-320 and selected on LB with 100 μg/mL gentamicin to create strains 0122-320-psd, 0122-320-psd (0122-15), and 0122-320-empty (Table S1).

Phospholipid and lipid A analysis

Isolate 0122-320, revertant 0122-320R, and laboratory strain PAO1 were grown on MHA with or without supplementation with 15 mM MgCl₂, 2% glycerol (v/v), or sulfamethoxazole at concentrations of 48 mg/L (0122-320), 0.5 mg/L (0122-320R), or 2 mg/L (PAO1). Membrane phospholipids were extracted using the Bligh and Dyer method²³ while LPS and lipid A extractions were performed as described.²⁴ Lipid A content was analysed by MALDI-TOF) in negative ion mode using the matrix norharmane (Bruker microflex) and compared between the clinical isolate, phenotypic revertant, and laboratory strain PAO1. Modified structures were validated by fragmentation as described.²⁵ Phospholipids were analysed by MALDI Fourier transform ion cyclotron resonance (FTICR) in negative ion mode and identified within 5 ppm supplemented with MSn using collision-induced dissociation, lipid identities were based on the sum composition from predominant fragments.²⁶ Phospholipids were compared as relative abundances with a lower limit threshold cutoff of 1.5%. Averages of relative abundances from triplicate spectra were reported, representative of two experiments. Relative abundances of phosphatidylserine in 0122-320 were confirmed by thin-layer chromatography of the total lipid extract compared with phospholipid headgroup class standards.

Results

Isolation from CF sputum of P. aeruginosa unable to grow on MacConkey at 37°C

P. aeruginosa was isolated from a CF patient’s respiratory sample on MacConkey incubated at 35°C, a temperature used in the SCH CF clinical laboratory, whereas other laboratories use 37°C.²⁷ Upon subculturing a representative isolate, 0122-7, onto MacConkey and BAP and incubating at a recorded temperature of 37°C, the isolate failed to grow on MacConkey and grew as pinpoint colonies on BAP. To determine if P. aeruginosa would have been missed if cultured on MacConkey at 37°C, sputum was collected and inoculated onto duplicate sets of MacConkey and a set of BAP. The first set of MacConkey plates were incubated at 35°C, and the second MacConkey set was incubated at 37°C along with the BAP. Two colony morphotypes were observed after incubation of MacConkey at 35°C and were identified as Stenotrophomonas maltophilia and P. aeruginosa at absolute abundances of 2.6 × 10⁷ and 4.6 × 10⁷ cfu/mL, respectively. By comparison, after 72 h of incubation at 37°C on MacConkey, a single non-fermenting colony morphotype was observed, reflecting a sputum abundance of 4.5 × 10⁶ cfu/mL. All 100 randomly selected colonies growing on MacConkey at 37°C from different sample dilutions were oxidase-negative, and representative colonies were identified as S. maltophilia. Two predominant colony morphotypes were observed on BAP at 37°C. The first morphotype was an oxidase-negative, Gram-negative bacillus present at an abundance of 7.8 × 10⁶ cfu/mL that was identified as S. maltophilia. The second morphotype consisted of pinpoint colonies at an absolute abundance of 5.0 × 10⁷ cfu/mL. A representative isolate of the latter morphotype, 0122-320, was an oxidase-positive, Gram-negative bacillus identified as P. aeruginosa by 16S rRNA gene sequencing that was genetically related to 0122-7 by PFGE analysis (Figure S1A).

Isolate 0122-320 is temperature-sensitive and antibiotic-dependent at 37°C

0122-320 produced larger colonies on BAP and MacConkey at 30°C than when incubated at 35°C (Figure S2A and S2B). 0122-320 was unable to grow on MacConkey at 37°C or BAP at 42°C. Temperature not only affected growth but also this isolate’s antibiotic susceptibilities. 0122-320 was more susceptible to several anti-pseudomonal antibiotics at 35°C compared with 30°C (Figure 1a and c), with the exception of trimethoprim/sulfamethoxazole (SXT). A zone of inhibition was evident around a trimethoprim/sulfamethoxazole disc at 30°C but not 35°C (Figure 1a). Surprisingly, the organism only grew around a disc (Figure 1a) or Etest strip (Figure 1b) containing trimethoprim/sulfamethoxazole at 37°C, indicating 0122-320 exhibited a drug-dependent phenotype at physiological temperature. 0122-7 demonstrated this same antibiotic-dependence for growth at 37°C (Figure S3).

Figure 1.

Antibiotic susceptibility testing of P. aeruginosa isolate 0122-320 by disc diffusion (a) and Etest (b) on MHA incubated at the temperatures indicated above each plate. Bacterial growth was only observed around the trimethoprim/sulfamethoxazole (SXT) disc and Etest when plates were incubated at 37°C. Antibiotics are labelled above their respective discs on the 30°C MHA plate and are in the same position on MHA incubated at the other indicated temperatures. Zone diameters (c) were compared for 0122-320 when grown at different temperatures, and with those for control strain ATCC 27853. Shown are disc diffusion diameters for piperacillin/tazobactam (TZP), ceftazidime (CAZ), cefepime (FEP), aztreonam (ATM), imipenem (IPM), meropenem (MEM), tobramycin (NN), ciprofloxacin (CIP), and trimethoprim/sulfamethoxazole (SXT).

Sulfamethoxazole supports growth at 37°C

Trimethoprim/sulfamethoxazole consists of two antifolates that target different enzymes in folate catabolism. Of the two antibiotics, sulfamethoxazole alone supported growth of 0122-320 in a concentration-dependent manner (Figure 2a). At high sulfamethoxazole concentrations (192 and 384 μg discs), a zone of inhibition occurred inside the zone of growth (Figure 2b), similar to the Etest results (Figure 1b). These results were confirmed in microbroth assays, in which bacterial growth increased with sulfamethoxazole in a concentration-dependent manner (Figure S4). A statistically significant increase in bacterial growth was observed with the addition of sulfamethoxazole at a concentration of 4.8 mg/L (P = 0.0008) compared with incubation without sulfamethoxazole, and maximum growth was observed at 19 mg/L of sulfamethoxazole (P < 0.0001) (Figure S4). In addition, Biolog phenotype microarray profiling testing the effects of a diverse collection of compounds showed that respiratory activity, a surrogate for bacterial growth, of 0122-320 was only detected in wells containing sulphonamide antibiotics (Figure S5).

Figure 2.

Antibiotic-dependent growth of isolate 0122-320 on MHA after 24 h of incubation at 37°C. Growth with trimethoprim (TMP) or sulfamethoxazole (SMX) (a) was tested separately at amounts 2-fold below, equal to, and 2-fold above the amounts present on commercially available BD trimethoprim/sulfamethoxazole (SXT) discs (TMP 1.25 μg/SMX 23.75 μg). The response of 0122-320 to higher amounts of SMX was tested on a separate plate (b). The antibiotic added to each disc is indicated on the left of the plate and the amount is labelled above each disc (a) or above and below each disc (b).

Para-aminobenzoic acid antagonizes the effect of sulfamethoxazole

Sulfamethoxazole is a structural analogue of para-aminobenzoic acid (PABA), a tetrahydrofolate intermediate that competitively inhibits the bacterial enzyme dihydropteroate synthetase, thereby blocking folate synthesis. Addition of PABA (5 μg) onto a sulfamethoxazole disc (Figure S6A) or a blank disc in close proximity to the sulfamethoxazole disc (Figure S6B) prevented sulfamethoxazole restoring growth. This PABA inhibition was overcome with increased sulfamethoxazole concentrations (Figure S6C), indicating direct competition between PABA and sulfamethoxazole to influence growth. While PABA is toxic to P. aeruginosa at high concentrations,²⁸ an 80 μg PABA disc did not inhibit PAO1 growth, indicating its effect on sulfamethoxazole complementation was not due to growth inhibition. 0122-320 grew on nearly all media tested at 37°C with sulfamethoxazole, except for MacConkey (Figure S7). However, 0122-320 grew at 37°C on other Pseudomonas selective media, PIA and cetrimide, without sulfamethoxazole (Figures S7 and S8). Both media contain magnesium (Mg²⁺), potassium sulphate, and glycerol, which are absent from MacConkey. Supplementation of MHA with Mg²⁺ or glycerol supported growth of 0122-320 at 37°C (Figure S8).

Mutation of phosphatidylserine decarboxylase causes antibiotic-dependence and temperature-sensitivity

Although both phenotypes (sulfamethoxazole-dependence and temperature-sensitivity) of 0122-320 were retained after repeated passaging, colonies were occasionally observed on media away from sulfamethoxazole discs (Figure S9), suggesting phenotypic reversion. A selected revertant, 0122-320R, exhibited an identical PFGE pattern to its parental isolate (Figure S1B), grew on MacConkey at 37°C (Figure S2A and S2B), and was susceptible to sulfamethoxazole at this temperature (Figure S10A). Genomic sequence comparison between 0122-320 and 0122-320R identified a missense mutation in psd (PA4957) resulting in a leucine for proline amino acid substitution at position 155 of Psd in the revertant (Table 1). Threonine is conserved at this position among P. aeruginosa genomes deposited in Genbank (Table 1), indicating that 0122-320 carried a mutation in this gene of functional importance. Complementation of 0122-320 with wild-type psd (0122-320-psd) enabled growth of the isolate on MacConkey at 37°C and BAP at 42°C and eliminated dependence on sulfamethoxazole for growth at 37°C (Figure S2A and S2B), while transformation with vector alone (0122-320-empty) did not. In fact, 0122-320-psd was hypersusceptible to sulfamethoxazole (Figure S10B) at concentrations to which P. aeruginosa is intrinsically resistant.²⁹ The MexAB-OprM efflux pump is responsible for this intrinsic resistance, and genomic analysis of this operon in both 0122-320 and 0122-320R revealed a single nucleotide deletion at position 19 of mexA, resulting in a frameshift, which likely explains sulfamethoxazole hypersusceptibility. The interplay between temperature and sulfamethoxazole on growth of the antibiotic-dependent mutant and revertant was confirmed using microbroth assays. sulfamethoxazole enhanced growth of 0122-320 at 37°C but was inhibitory at lower growth temperatures (Figure S11). Revertant 0122-320R had the highest bacterial densities at 37°C, but growth was inhibited with the addition of sulfamethoxazole at all temperatures (Figure S11).

Table 1.

Comparison of Psd amino acid sequences between Patient 0122 P. aeruginosa isolates 0122-320 and 0122-15, revertant 0122-320R and sequences published in GenBank

	Psd amino acid positiona
Isolate	62	155	195	289	GenBank reference
PAO1	Phe	Thr	Arg	Ser	NC_002516
UCBPP-PA14	Tyr	Thr	Arg	Ala	NC_008463
2192	Tyr	Thr	Arg	Ala	NZ_CH482384
39016	Tyr	Thr	Arg	Ser	NZ_CM001020
B136-33	Tyr	Thr	Arg	Ser	NC_020912
C3719	Tyr	Thr	Arg	Ser	NZ_CH482383
DK2	Tyr	Thr	Arg	Ser	NC_018080
LESB58	Tyr	Thr	Arg	Ser	NC_011770
M18	Tyr	Thr	Arg	Ser	NC_017548
NCGM2.S1	Tyr	Thr	Arg	Ser	NC_017549
RP73	Tyr	Thr	Arg	Ser	NC_021577
PACS2	Tyr	Thr	Arg	Ser	NZ_AAQW01000001
ATCC 15442	Tyr	Thr	Arg	Ala	NZ_AYUC01000060
ATCC 27853	Tyr	Thr	Arg	Ser	NZ_MTGC01000006
ATCC 14886	Tyr	Thr	Arg	Ser	NZ_AKZD01000186
0122-320	Tyr	Pro	Arg	Ser	This study
0122-320R	Tyr	Leu	Arg	Ser	This study
0122-15	Tyr	Pro	Leu	Ser	This study

The table only includes amino acid differences between isolates.

^aAmino acid at the specified position.

Sulfamethoxazole alters the membrane composition to support 0122-320 growth at 37°C

Psd encodes for phosphatidylserine decarboxylase, which converts phosphatidylserine to phosphatidylethanolamine, the predominant phospholipid in membranes of Pseudomonas species.³⁰ Therefore, mutants with dysfunctional Psd are predicted to accumulate phosphatidylserine at 37°C, a condition noted to be toxic.³¹ Using mass spectrometry, two ions (m/z 734.5 and 760.5), identified as deprotonated phosphatidylserines (PS): [PS(32 : 0)-H]⁻ and [PS(34 :0 )-H]⁻, respectively, were observed in lipid extracts from 0122-320 grown at 37°C with Mg²⁺ or glycerol that were absent in 0122-320R and PAO1. Based on sum composition of fragmentation, these parent phosphatidylserines were defined as PS 16 : 0/16 : 0 and PS 16 : 0_18 : 1; fragments from odd-numbered acyl chain-containing phosphatidylglycerol isobars were detected in low abundance and minor structural isomers are present. The relative abundance of these phosphatidylserines in 0122-320 when grown at 37°C with Mg²⁺ or glycerol was 13–15-fold higher than when grown at 30°C without supplementation (Figure 3), indicating increased phosphatidylserine levels at 37°C. When 0122-320 was grown at 37°C on media containing sulfamethoxazole, the amount of phosphatidylserine was similar to levels observed after growth at 30°C (Figure 3). Therefore, the growth phenotypes of 0122-320 paralleled the abundance of phosphatidylserine. Phosphatidylserine levels were below the detection threshold in revertant 0122-320R and PAO1 at all conditions tested (Figure 3).

Figure 3.

Relative abundance of phosphatidylserine (PS) in the membranes of isolate 0122-320, revertant 0122-320R, and laboratory strain PAO1 grown under the indicated conditions and temperatures. The graph shows the abundance of two prominent peaks identified as PS by fragmentation analysis, m/z 734.5 and 760.5, relative to the total phospholipid composition in the membrane. Isolates were grown in MHA with and without 15 mM MgCl₂ (Mg++), 2% v/v glycerol (GLY), or sulfamethoxazole (SMX), as indicated. Subinhibitory concentrations of sulfamethoxazole were used for the revertant 0122-320R (2 μg) and PAO1 (40 μg).

Studies have suggested a mechanistic link between membrane phospholipid and lipopolysaccharide (LPS) biosynthesis.^32–34 For this reason, and because LPS forms a significant portion of the P. aeruginosa outer membrane,³⁵ the essential component of LPS, lipid A, was examined in 0122-320 by mass spectrometry. Lipid A from reference strains of P. aeruginosa, including PAO1, contain penta-acylated structures corresponding to ions m/z 1446 and 1462, lacking the 3-O-acylation.³⁶ However, 0122-320 had higher abundances of modified lipid A structures, characterized by peaks m/z 1616 and 1632, when grown at 37°C on media containing Mg²⁺compared with growth on sulfamethoxazole media (Figure 4). These two peaks were characterized as hexa-acylated lipid A structures with all primary acylations present. We detected few or none of these modified lipid A structures in membranes from the revertant 0122-320R when grown at 37°C with either Mg²⁺ or sulfamethoxazole.

Figure 4.

Lipid A profile of isolate 0122-320 (a and b) and revertant 0122-320R (c and d) after growth on MHA at 37°C supplemented with sulfamethoxazole (SMX) (a and c) or MgCl₂ (b and d). Peak intensities are expressed in arbitrary units (a.u.). Subinhibitory concentrations of sulfamethoxazole were used to supplement the media for revertant 0122-320R.

Antibiotic-dependent and temperature-sensitive P. aeruginosa chronically infected patient 0122

For almost 1.5 years prior to the detection of the first temperature-sensitive and sulfamethoxazole-dependent isolate (0122-7), all cultures from patient 0122 were negative for P. aeruginosa, and records documented no known sulphonamide usage during that time (Table S2). After the initial detection of 0122-7, this patient was often culture-positive for P. aeruginosa over a 5 year period and periodically received sulphonamides (Table S2), although P. aeruginosa detection fluctuated during this time (Table S2). All tested P. aeruginosa isolates recovered from respiratory samples of this patient were clonally related by PFGE analysis (Figure S1A). With the exception of a single isolate, 0122-15, all P. aeruginosa were temperature-sensitive and sulfamethoxazole-dependent at 37°C (Figure 1a and Figure S3). The isolate without these phenotypic characteristics, 0122-15, had an additional missense mutation in psd causing an amino acid substitution from arginine to leucine at position 195 (Table 1). This additional mutation compensated for the proline substitution at site 155, as transformation of 0122-320 with psd from 0122-15 restored growth at 37°C on BAP and MacConkey (Figure S12).

Discussion

In this report, we describe the detection of a bacterium exhibiting dependence on sulfamethoxazole for growth at physiological temperature. P. aeruginosa isolates with this phenotype were repeatedly cultured from respiratory samples of this patient for approximately 5 years and were often the most abundant organism cultured, reaching sputum abundances of 5.0 × 10⁷ cfu/mL. The procedure for P. aeruginosa isolation from CF samples at SCH includes incubation of MacConkey at 35 ± 2°C, likely explaining detection of this mutant when incubation temperatures were below 37°C. However, this mutant was unculturable from this patient’s sputum on MacConkey at 37°C, and therefore, its detection would have been missed at this temperature. MacConkey is used by clinical and research laboratories for P. aeruginosa isolation from polymicrobial sources, such as CF infections. Studies have described using 37°C to cultivate P. aeruginosa from sputa on selective media^14,37–39 and 37°C is within an accepted range for incubation at most clinical labs. Adaptations creating extreme phenotypes such as antibiotic-dependence and temperature-sensitivity, may partially explain reports of P. aeruginosa detection by molecular methods without positive cultures.^39,40 Failure to identify P. aeruginosa delays treatment of a significant pathogen associated with poor health outcomes for CF patients.^6,7

The outer membrane of Gram-negative bacteria is an asymmetric bilayer composed of phospholipids (inner leaflet) and LPS (outer leaflet) and protects against external threats,⁴¹ including bile salts in MacConkey and antibiotics. For the isolate lineage described here, the antibiotic-dependent, temperature-sensitive phenotype was caused by a mutation in Psd, which converts phosphatidylserine to phosphatidylethanolamine, the latter representing the most abundant phospholipid in the bacterial membrane.³⁰ Hawrot et al.³¹ reported the isolation of Escherichia coli psd mutants unable to grow at elevated temperatures in the absence of divalent cations (Mg²⁺ or Ca²⁺). In wild-type E. coli, phosphatidylserine is present in the membrane at trace amounts, but a dysfunctional Psd results in phosphatidylserine accumulation at concentrations considered toxic for cells due to the negatively charged lipid headgroup.³¹ Cations presumably alleviate this toxicity by shielding the negative charge. Isolate 0122-320 contained high phosphatidylserine levels and modified lipid A structures, consisting of hexa-acylated species, m/z 1616 and 1632,²⁴ in its membrane at 37°C on Mg²⁺-supplemented media. A decrease in growth temperature reduced membrane phosphatidylserine levels in both E. coli³¹ and P. aeruginosa psd mutants, allowing growth of the organisms without divalent cations. In previous work, bacteria changed their phospholipid and LPS membrane compositions in response to growth temperature.^42,43

Surprisingly, sulfamethoxazole supported growth of 0122-320 on laboratory media at 37°C and reduced membrane phosphatidylserine and modified lipid A species levels, suggesting a mechanistic link between these activities. Thus, alterations in membrane composition at different growth temperatures likely explains the shifts in antibiotic susceptibility for isolate 0122-320. While membrane-targeting compounds such as polymyxin B induce transcription of genes involved in membrane modifications,⁴⁴ it is unclear how sulfamethoxazole, which targets folate metabolism, also modified the membrane composition. The competitive inhibition of sulfamethoxazole-dependent growth by PABA, a tetrahydrofolate intermediate, suggests the involvement of the folate pathway in the effect of sulfamethoxazole on the membrane. While a link between folate, phospholipid, and LPS metabolic pathways has not been reported in bacteria, mammalian liver cells with folate deficiencies have diminished membrane phosphatidylcholine levels, potentially through disruption in methylation activity.⁴⁵ However, trimethoprim, which also targets the folate metabolic pathway, was not able to rescue 0122-320 from growth inhibition at 37°C, raising doubts about a direct connection between folate metabolism and phospholipid synthesis in this mutant. Rather, sulfamethoxazole may directly or indirectly influence bacterial membrane composition through an as-yet-unidentified pathway or mechanism.

Bacterial mutants requiring antibiotics for growth have been reported,^13,46–50 and their selection followed treatment of the infected patient with the respective antibiotic, representing an extreme type of resistance. In contrast, records indicate patient 0122 was culture-negative for P. aeruginosa and had not received sulphonamide treatment for at least 1.5 years prior to isolation of the first sulfamethoxazole-dependent mutant (0122-7). It is therefore unclear how and why this mutant was selected. Given the in vivo persistence of this mutant for many years, it may have been selected by sulfamethoxazole treatment well before its first detection, prior to the first records available to us, and was simply missed because of its in vitro growth restriction. The mexA mutation in isolate 0122-320 indirectly supports sulfamethoxazole’s role in selecting for this mutation. P. aeruginosa is intrinsically resistant to sulfamethoxazole by efflux through a constitutively produced efflux pump, MexAB-OprM,²⁹ and loss of any pump component, including MexA, confers hypersusceptibility to sulfamethoxazole and other antibiotics.^29,51,52 Therefore, this psd mutant would likely have outcompeted its parental lineage during sulphonamide exposure. Alternatively, this mutant may have been transmitted from another patient or selected in patient 0122 by pressures unrelated to sulfamethoxazole, such as antimicrobial peptides. Lipid A alterations have been described for P. aeruginosa CF isolates that confer resistance to antimicrobial peptides in the CF lung.⁵³

In other infection cases with antibiotic-dependent bacteria, treatment simply involved removal of the offending drug. However, this mutant was not solely dependent on sulfamethoxazole for survival, as this organism could grow at 37°C on BAP, albeit poorly, and on media containing Mg²⁺. Divalent cations have been detected in CF sputum at average concentrations of 1.7 and 0.6 mM for Ca²⁺ and Mg²⁺, respectively.⁵⁴ While lower than the 15 mM Mg²⁺ we supplemented with, these cation concentrations, or possibly the presence of other compounds mimicking glycerol, may have supported growth of this organism in the CF lung. Similar to other antibiotic-dependent bacteria, the prevalence of this adaptation remains unknown given the limited detection under most laboratory conditions. CF lung infections are polymicrobial, and sulphonamides are frequently used to treat several of diverse pathogens that cause these infections, including Staphylococcus aureus, Stenotrophomonas maltophilia, Burkholderia cepacia complex, and Achromobacter species,^55,56 but not P. aeruginosa. Because P. aeruginosa often co-infects CF airways with these pathogens, P. aeruginosa is frequently exposed to sulphonamides collaterally. For example, in a recent multicentre, longitudinal study of 230 children with CF (age 6–16 years), the rates of sulphonamide treatment during the 2 year study period were similar between all study subjects (92/230 subjects, 40%) and subjects culture-positive for P. aeruginosa (24/66 subjects, or 36.4%).⁵⁷ Therefore, the probability of encountering P. aeruginosa psd mutants may be higher among patients with polymicrobial infections, such as CF, than patients with single-taxon P. aeruginosa infections, for which sulphonamide treatment would not be indicated. However, a future study to screen a collection of CF P. aeruginosa isolates cultured at 35°C for the temperature-sensitive and sulphonamide-dependent phenotype would be required to determine the prevalence of this mutation among CF isolates. Persistence of this mutant for a long period of time suggests the psd mutation was not as disadvantageous in vivo as it appeared in vitro. Thus, in vitro culture assays may not sufficiently reflect in vivo environments to accurately predict the behaviour of adaptive mutants in chronic infections.