lptG contributes to changes in membrane permeability and the emergence of multidrug hypersusceptibility in a cystic fibrosis isolate of Pseudomonas aeruginosa

Abstract Purpose In the lungs of cystic fibrosis patients, Pseudomonas aeruginosa is exposed to a myriad of antibiotics leading to alterations in antibiotic susceptibility. This study identifies mutations resulting in hypersusceptibility in isogenic mutants of a P. aeruginosa clinical isolate, PA34. Methods PA34 was exposed to subinhibitory concentrations of doripenem or meropenem during growth to mid‐log phase. Antibiotic susceptibility of surviving colonies was determined by agar dilution. Two carbapenem‐resistant colonies hypersusceptible to non‐carbapenem antibiotics were selected for further analysis. Antibiotic resistance gene expression was evaluated by RT‐rtPCR and OprD production by SDS‐PAGE. PA34 and isogenic mutants were evaluated with whole genome sequencing. Sequence variants were confirmed by Sanger sequencing, and cognate genes in eight carbapenem‐resistant clinical isolates hypersusceptible to non‐carbapenem antibiotics were sequenced. Lipopolysaccharide preparations of PA34 and hypersusceptible mutants were evaluated with ProQ‐Emerald stain. Results Isogenic mutants showed 4‐ to 8‐fold MIC increase for imipenem, meropenem, and doripenem. However, they were hypersusceptible (≥4‐fold MIC decrease) to aminoglycosides, fluoroquinolones, and non‐carbapenem β‐lactams. Expression of ampC or mex‐opr efflux pumps was unchanged, but OprD production was decreased. Mutations causing Q86H AlgU and G77C LptG amino acid substitutions and nonsense mutations within OprD were observed in both mutants. Lipopolysaccharide modifications were observed between isogenic mutants and PA34. Non‐synonymous mutations in LptF or LptG were observed in 6/8 hypersusceptible clinical isolates resistant to carbapenem antibiotics. Conclusion Evaluation of hypersusceptible mutants identified the association between lptG and a hypersusceptible phenotype. Modifications in lipopolysaccharide profiles suggests LptG modification interferes with lipopolysaccharide transport and contributes to hypersusceptibility.

Pseudomonas aeruginosa can survive antibiotic exposure through multiple mechanisms, including biofilm production, the presence of a relatively impermeable outer membrane, regulation of outer membrane porins, antibiotic-hydrolyzing enzymes, and efflux pumps capable of expelling antibiotics (Yoshimura and Nikaido, 1982). Biofilm production, mediated by multiple factors including AlgU, provides a barrier that inhibits accessibility of antibiotic to the organism. In the outer membrane, antibiotic uptake is inhibited by both membranebound porins and lipopolysaccharide (LPS) composition, as mediated by OprD and the Lpt LPS transport pathway (Fowler & Hanson, 2014;Nikaido & Hancock, 1986;Wolter, Hanson, & Lister, 2004). In the periplasm, an overproduced AmpC cephalosporinase can inactivate a broad range of β-lactam antibiotics while the Mex-Opr family of RND efflux pumps removes antibiotics from the periplasm, inner membrane, and cytosol (Wolter, Black, Lister, & Hanson, 2009). The carbapenems (doripenem, meropenem, and imipenem) in addition to colistin are considered last resort antibiotics for the treatment of P. aeruginosa infections (Manohar, Babu, Bozdogan, & Ramesh, 2018;Wi et al., 2017). However, mutations that result in truncations in OprD have been associated with imipenem resistance, and when coupled with increased expression of genes encoding the MexAB-OprM efflux pump lead to decreased susceptibility to meropenem and doripenem ).

Exposure to subinhibitory concentrations of antibiotics has been
shown to select for decreased susceptibility to multiple classes of antibiotics (Kumari, Balasubramanian, Zincke, & Mathee, 2014).
In this study, we identified a panel of P. aeruginosa clinical isolates showing resistance to doripenem, meropenem, and imipenem, but susceptible to non-carbapenem antibiotics. We also evaluated how laboratory exposure to subinhibitory concentrations of doripenem or meropenem resulted in the emergence of hypersusceptibility to non-carbapenem antibiotics concurrent with reduced susceptibility to doripenem and meropenem in a cystic fibrosis isolate of P. aeruginosa. To our knowledge, this is the first report of carbapenem exposure resulting in a hypersusceptible phenotype to multiple classes of non-carbapenem antibiotics in P. aeruginosa. The purpose of this study was to use phenotypic, molecular and genomic data to characterize mutations that lead to a hypersusceptible phenotype in isogenic mutants of a P. aeruginosa clinical isolate collected from a patient suffering from cystic fibrosis.

| Generation of isogenic mutants
Isogenic mutants were generated from a P. aeruginosa clinical isolate (PA34) collected from a patient with cystic fibrosis through exposure to subinhibitory concentrations of meropenem and doripenem (Harrison & Hanson, 2017). Briefly, PA34 was challenged during log-phase growth with either meropenem or doripenem at concentrations equal to 1/2 the MIC as previously described (Fowler, 2014;Thomson, Sanders, & Hayden, 1991). The bacteria were pelleted and resuspended in saline before being inoculated into molten Mueller-Hinton agar containing doripenem or meropenem at the MIC to select for resistant mutants and incubated for 24-48 hr at 37°C. Surviving colonies were subcultured onto blood agar plates and incubated for 24 hr at 37°C before MICs of the isogenic mutants were determined. A 4-fold increase in MIC to doripenem or meropenem defined each mutant colony and two mutants were selected for further study: PA34-812M and PA34-822D. The laboratory strain, PAO1 and a susceptible strain collected from a cystic fibrosis patient, PA443, were used as controls.

| Pulsed-field electrophoresis
Genetic relatedness of the hypersusceptible isogenic mutants was confirmed by pulsed-field gel electrophoresis (PFGE) analyses using SpeI digestion according to the CDC PulseNet protocol with modifications (Goering, Ribot, & Gerner-Smidt, 2011;Swaminathan, Barrett, Hunter, & Tauxe, 2001). These modifications included culturing P. aeruginosa on blood agar plates and in brain-heart infusion broth, increasing wash duration of post in situ lysis agarose plugs to four cycles of 30 min each and the addition of thiourea (50 µg/ml) in the running buffer during electrophoretic separation of chromosome fragments (Fowler, 2014;Goering et al., 2011;Swaminathan et al., 2001). Eight clinical isolates of P. aeruginosa were collected from the King Fahad Specialist Hospital-Dammam, Saudi Arabia, and susceptibility profiles determined by VITEK2.

| Whole genome sequencing
DNA was harvested from overnight cultures of PA34, PA34-822D, and PA34-812M grown in MHB at 37° using the Qiagen Blood and Tissue DNA extraction kit and purified in an Amicon 0.5 ml filter prior to sequencing. Whole genome libraries were prepared using the NexteraXT kit (Illumina) according to the manufacturer's instructions and paired-end reads (2 × 300) were sequenced on an Illumina MiSeq (Creighton University). Initial de novo assembly of the P. aeruginosa genomes was performed using the SPAdes 3.10.1 small genome assembler (Bankevich et al., 2012). Post-assembly scaffolding and gap closure were performed using RAGOUT, ICORN2, and AlignGraph while PROKKA was used to annotate the finished assemblies (Bao, Jiang, & Girke, 2014;Kolmogorov, Raney, Paten, & Pham, 2014;Otto, Sanders, Berriman, & Newbold, 2010;Seemann, 2014). Chromosomal rearrangements were evaluated using MAUVE (Darling, Mau, Blattner, & Perna, 2004). Single-nucleotide polymorphisms (SNPs) were identified by mapping reads from the hypersusceptible isogenic mutants of PA34 back to the parent isolate using the bowtie2 (v2.2.6) assembler and the Genome Assembly Toolkit (Langmead & Salzberg, 2012). SNPs were evaluated using both the variant call file metrics and through visual inspection of mapped reads using IGV (Thorvaldsdottir, Robinson, & Mesirov, 2013). Mutations identified were confirmed using Sanger sequencing. The PA34 draft genome is available at DDBJ/ENA/GenBank under the accession PDLR00000000 (Harrison & Hanson, 2017).

| Sanger sequencing
The sequences of lptG and lptF amplicons for PA34, the isogenic mutants and panel of hypersusceptible clinical isolates were determined offsite at Functional Biosciences, Inc. (Madison, WI), by Sanger sequencing (BigDye version 3.1, ABI 3730xl). Samples were prepared according to company submission standards, and the nucleotide quality was called at a Phred score greater than 20.

| Protein modeling
Models for proteins of unknown structure were generated using the I-TASSER suite 4.4 (Zhang, 2008). Ten candidate models were generated for proteins of unknown structure and a representative model for each was selected based on similarity to structure of proteins with known function (e.g., RpoE as a model for AlgU). Normal mode analysis was used to evaluate the flexibility of protein structures based on the interaction of amino acid residues and a differential plot of mobility was used to evaluate changes in flexibility between LptG found in PA34 and G77C-LptG found in the hypersusceptible isogenic mutants (Ma, 2005).

| LPS profiles
Lipopolysaccharide was extracted and diluted to 30 ng/µl using a hot phenol method followed by membrane dialysis (Perry, MacLean, Schollaardt, Bryan, & Ho, 1995). Twenty microliters of the LPS extract was separated using a 12% Tris-Glycine polyacrylamide gel and visualized using Pro-Q Emerald 300 stain (Invitrogen).

| Gene expression
RNA was extracted as described previously ).
Gene expression was evaluated using Qiagen OneStep RT-PCR Kit and the Qiagen Rotor-Gene Q 5-Plex thermocycler. Fold change in gene expression was calculated using the 2 −ΔΔCt method with rpoD as the endogenous control.
We then characterized known mechanisms of antibiotic resistance in P. aeruginosa using RT-rtPCR, SDS-PAGE analysis, and Sanger sequencing. Expression of ampC, oprD, and the mex-oprD efflux pump genes was compared between the susceptible P. aeruginosa control strains, the parent isolate PA34, and the hypersusceptible isogenic mutants (Table 2) Having identified a known mechanism of carbapenem resistance in the isogenic mutants, our next task was to identify the mechanism of non-carbapenem antibiotic hypersusceptibility. Whole genome sequencing (WGS) was used to identify genetic differences between isogenic mutants and appraise potential mechanisms for the hypersusceptible phenotype. All mutations resulting in amino acid substitutions or truncations are provided in Table 3. In addition to confirming the mutations in oprD, WGS revealed SNPs in genes encoding LptG and AlgU at conserved positions in both mutants. These SNPs resulted in a Q86H substitution in the AlgU transcriptional regulator and a G77C substitution in the LptG subunit of the LPS transport complex LptB 2 FG.
The transcription factor AlgU is responsible for activating a myriad of genes involved in the mucoid phenotype (Firoved & Deretic, 2003). When evaluating mucoid phenotype for PA34 and hypersusceptible isogenic mutants, it was noted that PA34 had a mucoid phenotype but the hypersusceptible isogenic mutants producing AlgU with a Q86H substitution did not (Figure 1). Among the panel of clinical isolates susceptible to non-carbapenem antibiotics but resistant to doripenem and meropenem, none of the isolates presented a mucoid phenotype.
LptG is a subunit of the LptB 2 FG complex that is involved in the transport of LPS from the outer leaflet of the inner membrane to the outer membrane (DeVries & Ohman, 1994). As an initial evaluation of the impact of a G77C substitution in LptG, we compared the structural mobility projections of sequence-derived models of LptG to the G77C LptG using normal mode analysis. Briefly, normal mode analysis of proteins informs the flexibility of the protein model by evaluating the mobility and environment of each amino acid residue in the structure. Normal mode analysis of a sequence-derived LptG model protein revealed that the G77C substitution could alter the structural mobility of the protein (Figure 2). This analysis indicated that LptG with the G77C substitution affected the model structure through decreased amino acid mobilities near residues 199-203 and

255-265 and increased mobility near residues 204-210.
To test whether the hypersusceptible isogenic mutants displayed altered LPS composition compared to the parent isolate, we evaluated LPS mobility profiles of the organisms with polyacrylamide gel electrophoresis. LPS profiles of the hypersusceptible isogenic mutants were different compared to those of the
TA B L E 3 Mutations selected for by exposure to subinhibitory concentrations of carbapenems F I G U R E 1 Comparison of colony growth in the mucoid clinical isolate PA34 (top) and the non-mucoid isogenic mutants PA34-822D (middle) and PA34-812M (bottom) aeruginosa typically has low permeability to rifampicin (Nikaido & Vaara, 1985;Vaara, 1992). Therefore, we used rifampicin to eval-  (Table 5). These amino acid substitutions in the LptFG concurrent with antibiotic hypersusceptibility in both the laboratory derived mutants and clinical isolates highlight the importance of these components in the LPS transport pathway with regard to antibiotic resistance.

| D ISCUSS I ON
Several publications have explored the hypersusceptible phenotype in P. aeruginosa (Linares et al., 2010;Mulet et al., 2011;Wolter et al., 2009). These papers have identified target pathways and transcription factors to be investigated but few gave insight into which effector genes contribute to this phenotype. Identifying candidate genes that contribute to antibiotic hypersusceptibility enables targeted research to develop novel therapeutics and restore the potency of current antibiotics to which strains of P. aeruginosa have become resistant.
Whole genome sequencing analysis revealed mutations in two genes that indirectly mediate entry of the antibiotic into P. aeruginosa, algU and lptG; algU through regulation of biofilm production and lptG through the transport of lipopolysaccharide to the outer membrane ( Figure 5). The observed AlgU Q86H substitution in the hypersusceptible isogenic mutants resides in region 2 of the sigma factor, a location responsible for recognizing the −10 promoter element and binding to the RNA polymerase core. Conversion from a mucoid to a non-mucoid phenotype has been shown to result from mutations in algU, consistent with the phenotype observed in the hypersusceptible mutants (Damron, Qiu, & Yu, 2009;DeVries & Ohman, 1994). selected the isogenic mutants in this study and mutations observed in the PAO1 mutants favored loss-of-function mutations in oprD.
However, meropenem selected for different mutations in addition to large deletions within the PAO1 genome (Cabot et al., 2016). These mutations or deletions observed in PAO1 however, did not affect the Lpt transport pathway, but did result in hypersusceptibility to colistin which was associated with a mutation in galU.
The use of rifampicin disk diffusion assays indicated that outer membrane permeability for the PA34 mutants was modified.
Permeabilization of the outer membrane affects the retention of molecules located in the periplasm. A previous study compared extracellular AmpC activity from cultures of P. aeruginosa with alterations in outer membrane composition (Mulet et al., 2011 substitution in LptG found in multiple hypersusceptible isogenic mutants indicates one potential mechanism by which clinical isolates are able to alter outer membrane permeability and develop antibiotic hypersusceptibility in response to antibiotic exposure.
In addition, this pathway was also targeted in clinical isolates of P.
aeruginosa. It is possible that using a combination therapy consisting of a carbapenem and another anti-pseudomonal antibiotic to treat P. aeruginosa could be beneficial in preventing the emergence of carbapenem resistance.

ACK N OWLED G M ENTS
We thank Dr. Paul Brett for the assistance in providing the LPS ex-

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
LH assembled genomes and identified SNPs, generated protein models and mobility analysis, performed LPS profiles and rifampicin disk susceptibilities, and contributed to experimental design. RF generated and characterized isogenic mutants, performed outer membrane protein, gene expression and PFGE analyses, and contributed to experimental design. BA provided the carbapenem-resistant clinical isolates with a hypersusceptible phenotype. AS generated sequences for genome assembly and strain selection. NH acquired strains and funding, coordinated research and contributed to experimental design.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
All data are provided in the results of the manuscript with the exception of pulsed-field electrophoresis gels and the outer membrane protein assay showing the loss of OprD, which will be made available upon request. Draft genome sequence for PA34 is located at https://www.ncbi.nlm.nih.gov/nuccore/ under the accession number PDLR00000000.