High-level nitrofurantoin resistance in a clinical isolate of Klebsiella pneumoniae: a comparative genomics and metabolomics analysis

ABSTRACT Nitrofurantoin is a commonly used chemotherapeutic agent in the treatment of uncomplicated urinary tract infections caused by the problematic multidrug resistant Gram-negative pathogen Klebsiella pneumoniae. The present study aims to elucidate the mechanism of nitrofurantoin action and high-level resistance in K. pneumoniae using whole-genome sequencing (WGS), qPCR analysis, mutation structural modeling and untargeted metabolomic analysis. WGS profiling of evolved highly resistant mutants (nitrofurantoin minimum inhibitory concentrations > 256 mg/L) revealed modified expression of several genes related to membrane transport (porin ompK36 and efflux pump regulator oqxR) and nitroreductase activity (ribC and nfsB, involved in nitrofurantoin reduction). Untargeted metabolomics analysis of total metabolites extracted at 1 and 4 h post-nitrofurantoin treatment revealed that exposure to the drug caused a delayed effect on the metabolome which was most pronounced after 4 h. Pathway enrichment analysis illustrated that several complex interrelated metabolic pathways related to nitrofurantoin bacterial killing (aminoacyl-tRNA biosynthesis, purine metabolism, central carbohydrate metabolism, and pantothenate and CoA biosynthesis) and the development of nitrofurantoin resistance (riboflavin metabolism) were significantly perturbed. This study highlights for the first time the key role of efflux pump regulator oqxR in nitrofurantoin resistance and reveals global metabolome perturbations in response to nitrofurantoin, in K. pneumoniae. IMPORTANCE A quest for novel antibiotics and revitalizing older ones (such as nitrofurantoin) for treatment of difficult-to-treat Gram-negative bacterial infections has become increasingly popular. The precise antibacterial activity of nitrofurantoin is still not fully understood. Furthermore, although the prevalence of nitrofurantoin resistance remains low currently, the drug’s fast-growing consumption worldwide highlights the need to comprehend the emerging resistance mechanisms. Here, we used multidisciplinary techniques to discern the exact mechanism of nitrofurantoin action and high-level resistance in Klebsiella pneumoniae, a common cause of urinary tract infections for which nitrofurantoin is the recommended treatment. We found that the expression of multiple genes related to membrane transport (including active efflux and passive diffusion of drug molecules) and nitroreductase activity was modified in nitrofurantoin-resistant strains, including oqxR, the transcriptional regulator of the oqxAB efflux pump. Furthermore, complex interconnected metabolic pathways that potentially govern the nitrofurantoin-killing mechanisms (e.g., aminoacyl-tRNA biosynthesis) and nitrofurantoin resistance (riboflavin metabolism) were significantly inhibited following nitrofurantoin treatment. Our study could help inform the improvement of nitrofuran derivatives, the development of new pharmacophores, or drug combinations to support the resurgence of nitrofurantoin in the management of multidrug resistant K. pneumouniae infection.

In contrast to this emerging understanding of resistance mechanisms in E. coli, to date, there remains a dearth of understanding of the nitrofurantoin killing and resistance mechanism(s) in K. pneumoniae.In the present study, we employed a combination of genetic analysis (whole-genome sequencing, qPCR analysis, and mutational mapping on molecular models) in combination with untargeted metabolomics to elucidate the mechanism(s) of nitrofurantoin action and resistance in a clinical K. pneumoniae isolate cultured from a hospital patient with UTI and experimentally evolved resistant mutants.The presented findings highlight that high-level nitrofurantoin resistance in K. pneumoniae is governed primarily by perturbations across several complex interrela ted metabolic pathways, and resistance is associated with mechanisms that decrease drug concentration inside the cell by increasing efflux and, potentially also, by reducing porin-mediated entry and limiting the reductive activation of nitrofurantoin.

Mutations arising in K. pneumoniae INF348 strains during nitrofurantoin exposure
The nitrofurantoin-susceptible clinical isolate K. pneumoniae INF348 (MIC 32 mg/L) was passaged in increasing nitrofurantoin concentrations, and as a result, we obtained six related mutants that were stable in the presence of high levels of nitrofurantoin (MICs ≥ 256 mg/L).These high-level resistant mutants were designated as INF348-01, INF348-02, INF348-03, INF348-04, INF348-05, and INF348-06.Comparative mapping of the genome sequences of each resistant mutant vs the wild-type parent strain K. pneumoniae INF348 revealed three insertion sequence (IS)-mediated disruptions and seven non-synonymous single-base substitutions, affecting four genes (oqxR, ompK36, ribC, and nfsB; see Table 1).Changes in gene expression, in the resistant mutants vs wildtype progenitor, were measured via qPCR for oqxB, acrAB, ompK36, ribC, and nfsB (Fig. 1).All of the nitrofur antoin-resistant mutants overexpressed oqxB and acrAB [log 2 fold change (FC) ≥0.59, corresponding to a fold change of approximately 1.5 in linear scale, relative to the wildtype progenitor], and downregulation of ompK36 and ribC was also observed in selected mutants (Fig. 1).Two resistant mutants harbored substitution mutations in nfsB, resulting in amino acid substitutions at positions 46 or 198 of the NfsB protein sequence (Table 1).
The oqxR gene was altered in two resistant mutants: one via ISKpn26 insertion and one via non-synonymous mutation (N38K).OqxR is a transcriptional repressor of the oqxAB efflux pump, such that disruptions in oqxR are expected to result in de-repression of oqxAB and increased efflux (24).The oqx operon is a core locus present in most K. pneumoniae genomes (25).Overexpression of oqxAB has been previously reported to be involved in nitrofurantoin resistance in K. pneumoniae and has also been noted as a determinant of resistance to nitrofurantoin and other antimicrobials in E. coli and K. pneumoniae (25)(26)(27).Here, while the elevation of oqxB expression was observed in all high-level nitrofurantoin-resistant mutants, it is noteworthy to underscore that those with oqxR mutations showed the greatest overexpression (log 2 FC > 4.2), which aligns with the removal of oqxR-mediated repression (Fig. 1) (24).How ever, it remains imperative to acknowledge that other strains, such as INF-348-04, also displayed a notable 4.0-log 2 FC increase in oqxR expression.This suggests a wider spectrum of heightened expression occurrences, which might necessitate a comprehen sive assessment of the intricate interplay of determinants influencing oqxR expression in strains tested.
The ompK36 gene encodes an outer membrane porin (homologous to E. coli ompC) (28,29) and was disrupted in five of the six resistant mutants (Table 1).Disruption of ompC homologs has been associated with reduced susceptibility to carbapenems and cephalosporins in K. pneumoniae (30) and other Enterobacteriaceae (31,32) and to nitrofurantoin in E. coli (33); however, we could not identify any prior reports of ompK36/ ompC mutations related to nitrofurantoin resistance in K. pneumoniae.INF348-06 harbored an IS1 × 2 insertion 47 bp upstream of ompK36, which is within the promoter region defined for E. coli ompC (34).Consistent with this, INF348-06 showed significantly reduced ompK36 expression (log 2 FC FC = −2.49).It has previously been reported that an IS5 insertion 37 bp upstream of ompK36 reduced its expression in K. pneumoniae ST258, resulting in resistance to carbapenems but not colistin (nitrofurantoin was not tested) (35).Notably, ompK36 expression was also reduced in four other isolates, including three (INF348-03, INF348-04, and INF348-05) that shared a common substitution resulting in premature termination of translation of the OmpK36 protein after just four amino acids (Table 1; Fig. 1).
The nfsB gene encodes a nitroreductase that directly interacts with nitrofurantoin, as outlined above.Here, we detected substitution mutations in nfsB in two resistant mutants (INF-348 02 and INF-348 04), resulting in amino acid substitutions at positions 46 or 198 of the NfsB protein sequence (Table 1).Both of these isolates, plus two others with wildtype nfsB, showed evidence of downregulation of nfsB expression (log 2 FC ≥ −0.68); the other two isolates showed minor upregulation relative to the wildtype progenitor isolate (log 2 FC ≥ 0.68).Modification or loss of NfsB is a well-known mech anism of nitrofurantoin resistance in E. coli, leading to inactivated or poor catalytic  function, which in turn cripples the cells' ability to reduce nitrofurantoin into its toxic intermediate compounds (22).
The resistant mutant INF348-01 acquired only one mutation (ribC disruption) to which its increase in MIC (to 256 mg/L) can be attributed.The other five resistant mutants harbored two mutations each, making it difficult to draw conclusions about the effects of individual mutations; however, these isolates all had an even higher MIC (512 mg/L), and all mutations were in genes with plausible relevance to nitrofurantoin resistance as outlined above.Taken together, these findings provide supportive evidence that multiple genetic mutations are associated with nitrofurantoin resistance in K. pneumo niae, spanning across two key resistance tiers: (i) reducing the amount of drug accu mulating inside the bacterial cell by repression of drug influx (disruption of ompK36) and/or enhancement of drug efflux (de-repression of the oqxAB efflux pump via oqxR disruption); and (ii) crippling the enzymatic activity of the activating nitroreductases and thus reducing the accumulation of reactive metabolites, either by direct modification of nfsB or via inhibition of FMN cofactor synthesis.

Mapping the mutations onto the modeled structure of K. pneumouniae nfsB nitroreductase
A structural model of NfsB was constructed in Molsoft using the E. coli crystal structure of NfsB nitroreductase 1YKI bound with the nitrofuran antibiotic nitrofurazone and FMN as the modeling template (38).The modeled K. pneumouniae structure had 84% sequence identity with the E. coli protein and near-identical antibiotic and cofactor binding domains, with the exception of His11 and Asn67 (Tyr11 and Thr67 for K. pneumoniae; Fig. 2A).In order to help understand the functional significance of the nfsB mutations, INF 348-02 and INF 348-04 mutational analysis was also carried in Molsoft to quantify the potential change in protein stability (Fig. 2B ; Table 2) (39,40).The substitution mutations were localized proximal to the NfsB active site.Importantly, the mutations are settled on an interface region of the pocket where FMN cofactor and nitrofuran binding occurs.Thus, it is likely that these mutations attenuate the ability of nitroreductase to bind FMN and nitrofuran drugs.The modeled structure indicates there are several indispensable amino acids responsible for complexation with FMN that were proximal to and thus likely impacted by the aforementioned active site mutations (Fig. 2A).These included Arg207 and Lys205 that form hydrogen bonds with phosphate of FMN and ensure that the homodimer is in the active conformation.Leu145 and Tyr144 form hydrophobic interactions with the benzenoid portion of FMN.Glu165 and Gly166 contain amide groups that interact with N5 and O4 in the isoalloxazine ring system of FMN.Furthermore, residues Arg10, Ser12, and Ser39 are involved in interactions with ribityl chain of FMN.Residues Lys14, Asn72, Glu73, Arg74, and Lys75 interact with O2, N3, and O4 in pyrimidine of FMN (Fig. 2A).
The comparative sequence and structural alignments show that the aforementioned key amino acids in NfsB are identical between E. coli and the K. pneumoniae isolates in our study, indicating that these conserved amino acids are important for cofactor binding (Fig S1) (40).All NfsB mutations identified in this study were situated in positions in or next to an α-helix or β-sheet; therefore, they could potentially perturb these secondary structures.In INF 348-04, the mutation D198Y is next to an α-helix composed of amino acids fraction of conserved native contacts (Fig S1; Fig. 2B).When Asp198 is replaced by the bulkier Tyr sidechain, we suspect that the mutation perturbs the formation of the α-helix immediately preceding the mutation, in turn solvent exposing the hydrophobic residue of Phe199.This mutation in part could also affect the loop region (residues 10-14) immediately adjacent to the residue, in turn affecting a number of hydrogen bonds with FMN.Both scenarios could also affect the distance of Arg207 and Lys205 to FMN and render them unable to form hydrogen bonds.
In addition to Arg207 and Lys205, other amino acids from positions 203−209, e.g., Leu or Pro could not offer an amine or hydroxyl group to form hydrogen bonds with FMN.The Trp46Cys mutation in INF 348-02 immediately precedes a β-sheet comprising a series of hydrophobic residues and partially π-stacks with Gln44 which forms hydrogen bond contacts with the amides of the adjacent monomer and residue Leu208 (Fig S1; Fig. 2B).In the β-sheet, amino acids are stabilized by mutual hydrogen bonds and interaction of their side chains, and in this case, it is a water excluding hydrophobic patch.When position 46 is harbored by Trp, the benzenoid portion in its side chain helps to stabilize this patch and stacking arrangement with Gln44 enabling hydrogen bonds to be maintained.Replacement with a Cys could affect the overall stability of the homodimer and the positioning of the FMN-binding loop residues of Leu208 and Arg207.Overall, the stability of the dimeric enzyme is predicted to be readily affected with an increase in delta delta G (ddG) for each modeled mutant (Table 2), in particular, W46C (with an average ddG of 5.69 kcal/mol), indicating that dimerization is no longer spontaneous.In conclusion, the Asp198Tyr and W46C mutations in NfsB are likely to negatively influence the binding to FMN and in particular the overall fold and stability of the enzyme itself.

Metabolite profiling of wild-type K. pneumoniae INF348 following nitrofuran toin treatment
To explore the impact of nitrofurantoin exposure on the metabolism of K. pneumoniae, we performed metabolite profiling on INF348 before and after exposure to 48 mg/L    59, corresponding approximately to a 1.5-fold change in linear scale] was used to compare post-exposure levels to pre-exposure levels to identify which metabolites were significantly altered in response to nitrofurantoin treatment.The reproducibility of all sample groups was within acceptable limits at both experimental time points, wherein the median relative SD across was 19%-26% for the untreated control groups and 18%-26% for the treated groups, which is consistent with some baseline varia bility in the dynamics of ordinary bacterial metabolism irrespective of nitrofurantoin treatment (Table S1).Principal component analysis illustrated that, for each post-expo sure timepoint analyzed, the first principal component (responsible for >70% of the variation in each comparison) separated the treated and untreated samples.Importantly, a gap between the control and treated samples was seen at 1 and 4 h, which reflected the significant variance in the abundance between the two groups (Fig S2; Table S2).
Heatmaps of metabolite profiles indicate that nitrofurantoin induced marked changes in the metabolite (mainly reduced) intensities compared to the control group; particularly at 4 h (Fig S3).Nitrofurantoin treatment significantly perturbed the bacterial metabolome with 433 (1 h) and 518 (4 h) significantly altered metabolites relative to the untreated control group (Fig S4A ).The classification of the significantly perturbed metabolites indicated that peptides, amino acids, lipids, and carbohydrates were largely perturbed compared to other metabolites classes; and the abundance of almost all of the impacted metabolite classes declined at both 1 and 4 h, except for an elevation of lipid abundance (Fig S4B).The Venn diagram data plot showed that there were 89 and 174 unique significantly impacted metabolites of in response to nitrofurantoin treatment at 1 and 4 h, respec tively (Fig S4C).Notably, 344 significant metabolites were in common across the 1 and 4 h time points (Fig S4C).

Pathways analysis of K. pneumoniae the wild-type INF348 induced by nitrofurantoin treatment at 1 and 4 h
The most significantly perturbed pathways in the K. pneumoniae INF348 wild-type metabolome following nitrofurantoin treatment at both time points (1 and 4 h post-exposure) were aminoacyl-tRNA biosynthesis, purine, riboflavin, nicotinate and nicotinamide metabolism, pantothenate and CoA biosynthesis, and tricarboxylic acid cycle (TCA cycle) (Fig S5).In the interest of brevity, central carbohydrate metabolism (TCA and glycolysis), pantothenate, and CoA biosynthesis dysregulations are discussed in detail in the supplementary document.
The observed perturbations of aminoacyl-tRNA biosynthesis and purine metabolism are in line with the primary mode of bacterial killing instigated by the unstable nitrofur antoin metabolites, which involves disruption of bacterial DNA, RNA, and protein synthesis (14,(41)(42)(43).

Riboflavin metabolism
Riboflavin metabolism is the main hub for the production of FMN and FAD which are employed as cofactors in a plethora of cellular flavoprotein enzymatic reactions (45,46).The deletion of the ribE gene, which encodes riboflavin synthetase, is associated with the development of nitrofurantoin resistance in Gram-negative bacteria; this comes about due to the abolition of synthesis of riboflavin/FMN, a crucial cofactor for nitroreductases NfsA and NfsB (21).

Conclusions
The study is the first to integrate the genomics and metabolomics platforms to investi gate the genes that govern nitrofurantoin resistance as well as the potential metabolic pathways underlying the bacterial killing effect in K. pneumoniae.The comparative genomics analysis showed for the first time that nitrofurantoin resistance in K. pneumo niae spans across two major resistance levels, first, by limiting the amount of drug accumulating inside the bacterial cell (enhancing drug efflux and potentially also by repressing influx); and second, inhibition of FMN cofactor synthesis and mutations targeting NfsB nitroreductase that abolish its ability to enzymatically activate nitrofuran toin.The metabolomics findings suggested that nitrofurantoin exerts its antibacterial killing effects by inhibiting aminoacyl-tRNA biosynthesis and purine metabolism, which ultimately terminates bacterial ribosomal RNA, DNA, and protein synthesis.

Bacterial strains
K. pneumoniae INF348 was cultured from the urine of a female patient in her 40s at the Alfred Hospital in Melbourne, in 2014, as part of routine diagnostic procedures.Its complete genome sequence (GenBank accession, GCA_904863425) was previously reported (47) and belongs to multi-locus sequence type (ST) 34 and carries capsule (K) locus KL22 and no acquired antimicrobial resistance or hypervirulence genes.The isolate was stored in tryptone soy broth (Oxoid) with 20% glycerol (Ajax Finechem, Seven Hills, NSW, Australia) in cryovials at −80°C.It was sub-cultured onto nutrient-rich agar plates (Media Preparation Unit, University of Melbourne, Melbourne, VIC, Australia) and incubated at 37°C for 24 h prior to use.

Determination of MICs
MICs were assessed in triplicates using the broth microdilution method as described earlier (48).A final concentration of DMF in the growth media of 0.5% (vol/vol) showed no inhibitory effect on the tested strains.

Induction and selection of nitrofurantoin-resistant strains
Two sequential selection steps were employed to generate nitrofurantoin-resistant K. pneumoniae derived from isolate INF348.After preparing the overnight culture, a fresh bacterial solution was prepared by adjusting the overnight OD 600 value to achieve an inoculum size of OD 600 ~ 0.1.Then 100 µL was taken and distributed evenly onto CAMHB plates containing 64 mg/L nitrofurantoin (2× MIC) and incubated for 24-48 h at 37°C.The surviving bacterial colonies were collected and then verified as resistant on the CAMHB plates containing the same concentration of nitrofurantoin.The surviving mutants were then progressed to the second selection round, wherein an overnight culture was prepared from each mutant, and 100 µL was spread evenly into CAMHB plates containing a higher nitrofurantoin concentration of 256 mg/L and incubated for 24 and 48 h.The surviving colonies were then verified as highly resistant on the 256 mg/L nitrofurantoin CAMHB plates.

Whole-genome sequencing
The genomes of the six nitrofurantoin-resistant mutants (INF348-01, INF348-02, INF348-03, INF348-04, INF348-05, and INF348-06) and nitrofurantoin-susceptible parent strain K. pneumoniae INF348 were sequenced on an Illumina HiSeq instrument (Illumina Australia, Melbourne, VIC, Australia) to over 30× coverage, following Nextera library preparation.INF348 was also subjected to long-read sequencing via Oxford Nanopore Technologies as previously described (49).The Illumina short and Nanopore long reads for INF348 were used to generate a hybrid assembly using Unicycler v0.4.7 using default parameters, and the resulting completed genome was annotated with Prokka v1.13.3 (50).Illumina short reads for each mutant were mapped to the completed INF348 genome using RedDog to detect single nucleotide variants (SNVs; available at https:// github.com/katholt/RedDog). Briefly, RedDog employs Bowtie v2.2.9 (51) with a sensitive local algorithm and a maximum insert length of 2,000 bp, and variant sites are called with SAMtools v1.3.1 (52).RedDog also produces depth and coverage statistics for each gene in the reference genome; these statistics were interrogated to identify missing genes relative to INF348 in each mutant genome.Genes were considered missing if the coverage across the gene was <95%, or the read depth was <5×.The completed INF348 genome was screened for IS using ISSaga (53).Seven IS types were identified in INF348 (IS1203, IS1 × 2, IS903B, ISEc52, ISKpn21, ISKpn26, and ISSty2).Illumina readsets for all mutant genomes were screened for these seven IS using ISMapper v2.0 (54) with default parameters and the INF348 reference genome.IS insertion sites that varied from INF348, as well as detected gene deletions, were annotated in each respective mutant genome and visually inspected using Artemis (55).All genes carrying mutations (SNVs, deletions, or IS insertions) not present in the founder strain were identified using BLAST searches of the NCBI database and by consulting the literature.

RNA extraction and reverse transcription
The six nitrofurantoin-resistant isolates (INF348-01, INF348-02, INF348-03, INF348-04, INF348-05, and INF348-06) and nitrofurantoin-susceptible parent strain K. pneumoniae INF348 were cultured on CAMHB plates overnight.The total RNA was extracted using RNAdvance Tissue Kit (Beckman Coulter, U.S.A), and contaminating DNA was removed using DNaseI (ThermoFisher Scientific) according to the manufacturer's protocol.The extracted RNA was stored in −80°C for further use.The cDNA was synthesized using 4 µL of the reverse transcription buffer with random primers (Promega), 2 µL of GoScript reverse transcriptase, and 100 ng of RNA, according to the manufacturer's protocol.

Quantitative polymerase chain reaction
To measure the relative mRNA levels, qPCR was used to detect the fluorescent signal of the BRYT Green Dye binding to double-stranded DNA [fold change (FC) ≥2, relative to the wildtype progenitor].The PCR mixture (20 µL) consists of 16 µL of Go Taq qPCR Master Mix, 2 µL of 0.8 µM each primer, and 2 µL of cDNA.The PCR cycle reaction was performed using a PCR Thermal cycler (ThermoFisher Scientific) under the following conditions: 95°C for 2 min for start, then 95°C for 15 seconds, 60°C for 1 min, and run 40 cycles, according to manufacturer's protocol.The amplification efficiency of each primer pairs was tested; all the values were within 90%-100%.All primers are listed in (Table S3), and rpoB was used as a reference gene.

Molecular modeling
Utilizing the ICM-Homology modeling algorithm and refinement tools (56)(57)(58) available in the ICM-Pro modeling suite (Molsoft LLC; molsoft.com), the target sequence of K. pneumouniae NfsB was constructed.ICM-Pro was also used for template searches for our candidate protein, allowing for automated alignment and inspection, prior to modeling the target protein.Mutational analysis focusing on protein stability was also carried using ICM-Pro.

Bacterial culture preparation for metabolomics
K. pneumoniae INF348 was cultured on Mueller-Hinton agar and incubated overnight at 37°C for 20 h.Prior to each experiment, an overnight culture was prepared by inoculation of one colony in 10 mL of CAMHB in 50 mL Falcon tubes (Thermo Fisher, Australia), which were immediately incubated at 37°C in a shaking water bath (180 rpm).The following day, a fresh bacterial solution was prepared in a 500 mL conical flask containing fresh CAMHB, which was incubated at 37°C at a shaking speed of 180 rpm for 2 h to achieve a log-phase of OD600 ~0.5 (~10 8 colony-forming units (CFU) per mL).Freshly prepared nitrofurantoin solution in milliQ water was then added to the correspondence flask to give a final concentration of 48 mg/L (1.5× MIC), and a drug-free control flask was also prepared.The flasks were returned to the rotary shaker at 37°C and a shaking speed of 180 rpm.At each time point (1 and 4 h), 15 mL of each sample was collected, quenched, and normalized to the required OD600 ~ 0.5 with fresh CAMHB.Thereafter, the samples were centrifuged at 3,220 × g at 4°C for 10 min.Lastly, the supernatants were discarded, and the pellets were stored at −80°C for further metabolite extraction.The experiment was conducted in four biological replicates to minimize the bias from inherent random variation.

Cellular metabolite extraction
The bacterial pellets were rinsed with cold 0.9% NaCl followed by centrifugation at 3,220 × g at 4°C for 5 min to eliminate residual extracellular metabolites.Extraction solvent of chloroform:methanol:water [1:3:1, (vol/vol)] containing 1 µM of generic internal standards (CHAPS, CAPS, PIPES, and TRIS) was then added.The samples were immedi ately frozen in liquid nitrogen and allowed to thaw on ice, and the freeze-thaw process was repeated three times to lyse the cells and release cellular metabolites.The extracted samples were centrifuged for 10 min at 3,220 × g at 4°C, and the supernatant was collected and further centrifuged at 14,000 × g for 10 min at 4°C.The final supernatant samples (200 µL) were collected into injector vials for Liquid Chromatography-Mass Spectrometry (LC-MS) analysis.Quality control samples were prepared by combining an equal volume of each sample (59).

LC-MS analysis of metabolites
Samples were analyzed on a Q-Exactive Orbitrap mass spectrometer (Thermo Fisher), coupled to a Dionex high-performance liquid chromatograph (U3000 RSLC HPLC, Thermo Fisher) with a ZIC-pHILIC column (5 µm, polymeric, 150 × 4.6 mm; SeQuant, Merck).The MS system operated at 35,000 resolutions in both positive and negative electro-spray ionization mode (rapid switching) and a detection range of 85-1,275 m/z.
The LC solvent consisted of 20 mM ammonium carbonate (A) and acetonitrile (B) with a multistep gradient system from 80% B to 50% B over 15 min, then to 5% B at 18 min, followed by a wash with 5% B for 3 min, and re-equilibration for 8 min with 80% B at a flow rate of 0.3 mL/min.The injection sample volume was 10 µL, and the run time was 32 min.All samples were analyzed in the same run and the chromatographic peaks, signal reproducibility, and analyte stability were monitored by assessment of pooled biological quality control samples (i.e., an aliquot of 10 µL from each sample, including both footprints and fingerprints) analyzed periodically throughout the batch, internal standards, and total ion chromatograms for each sample.Mixtures of pure standards containing over 200 metabolites were analyzed within the batch to aid in the identification of metabolites.

Metabolomics data processing and analysis
Metabolomics data analyses were performed using mzMatch72 and IDEOM, as we have previously described in greater detail (60)(61)(62)(63).Briefly, the quantification of each metabolite was conducted using the raw peak height.Univariate and multivariate analyses utilized MetaboAnalyst 3.074.Prior to analysis, relative peak intensity data were normalized by the median, log transformed, and scaled (by auto scale function) to reduce variance between the samples.The global metabolic profiles of samples with antibiotic treatments at each time point were analyzed using multivariate statistical analysis by the partial least square's discriminant analysis.Univariate analysis (t test; P < 0.05, fold change threshold = 1.5) was applied to identify significant metabolite changes between treated and untreated control samples at each time point.Metabolites that were detected as isomeric peaks with opposite abundance changes (increased and decreased levels) were excluded.Significant metabolites were filtered by selection of only those that showed a ≥0.59 log 2 -FC relative to the untreated control samples and an identification confidence score of 6 or more in IDEOM, i.e., removing likely LC-MS artifacts.Metabolic pathway analysis was performed for the statistically significant identified metabolites using KEGG Mapper (64).

FIG 1
FIG 1 Bar graphs show the differential expression of oqxB, acrA, acrB, ompC, ribC, and nfsB genes in six nitrofurantoin-resistant K. pneumouniae isolates compared with the wild-type isolates K. pneumouniae INF348.The data for each gene were derived from the average of three biological replicates, using the delta-delta Ct method.The grid line in the chart represents the log 2 FC cut-off of ≥0.59 or ≤−0.59.

FIG 2 (
FIG 2 (A) Interaction between amino acid residues and FMN plotted by LigPlot+: amino acid resi dues identified that are essential for stabilizing FMN of K. pneumouniae NfsB and (B) the mapping of critical mutations D198Y and W46C onto the dimeric structure of the enzyme utilizing a space-filling representation for the residues.NIC = nicotinic acid.

FIG 3
FIG 3 Schematic diagram shows the aminoacyl-tRNA biosynthesis and fold changes of significantly impacted metabolites K. pneumoniae INF348 following nitrofurantoin treatment at 1 and 4 h (log 2 FC ≥ 1.0, P ≤ 0.05).Data are mean ± SD for four biological replicates.The grid line in the chart represents the log 2 FC cut-off of ≥1 or ≤ −1.

FIG 4 (
FIG 4 (A) Schematic diagram depicting the significantly impacted purine metabolism of K. pneumoniae INF348 following nitrofurantoin treatment at 4 h.(B) Bar charts for the significantly impacted intermedi ates from purine metabolism of K. pneumoniae INF348 treated with nitrofurantoin at 1 and 4 h.(log 2 FC ≥ 1.0, P ≤ 0.05).Data are mean ± SD for four biological replicates.Blue rectangles: significantly decreased metabolites; red rectangles: significantly increased metabolites.The grid line in the chart represents the log 2 FC cut-off of ≥1 or ≤ −1.

FIG 6 (
FIG 6 (A) The significantly perturbed riboflavin metabolism of K. pneumoniae INF348 due to nitrofurantoin treatment.(B) Bar charts for the significantly perturbed intermediates from riboflavin metabolism at 1 and 4 h (log 2 FC ≥ 1.0, P ≤ 0.05).Data are mean ± SD for four biological replicates.Blue rectangles: significantly decreased metabolites; red rectangles: significantly increased metabolites.

TABLE 1
Mutations identified in resistant mutants derived from K. pneumoniae INF348 a

TABLE 2
Molsoft computation of the Gibbs free energy change in protein stability of a single-point mutation a The free energy change is displayed as ddG (kcal/mol).nfsB = nitroreductase enzyme.
of nitrofurantoin, using IDEOM (see Materials and Methods).A total of 1,007 putatively identified metabolites were obtained at 1 and 4 h post exposure, which included 73 metabolites involved in carbohydrate metabolism, 174 in amino acid metabolism; 326 in peptide metabolism, 49 in nucleotide metabolism, and 110 in lipid metabolism.Univariate analysis [t tests, False Discovery Rate adjusted p-value ≤0.05; log 2 -fold change (FC) ≥0.