Pseudomonas aeruginosa is one of the most frequent nosocomial pathogens, particularly as a source of acquired pneumonia in intensive care units with a tendency towards multidrug resistance [1]. Cefiderocol (FDC) is a novel siderophore cephalosporin that shows activity against most multidrug-resistant P. aeruginosa strains, including carbapenem-resistant P. aeruginosa [2]. Its broad and excellent activity is explained by its unique and so-called Trojan horse strategy relying on the active penetration into Gram-negative bacterial cells using its iron transport system [3]. There is some evidence of acquired resistance to FDC. Nevertheless, it seems that the potential for resistance acquisition remains low [4, 5]. The mechanisms underlying this resistance remain poorly understood. Resistance to FDC in P. aeruginosa has been demonstrated to be associated with alterations of iron uptake pathways [6], structural modification of the natural AmpC β-lactamase [7], and modification of the expression of efflux systems. The aim of this study was to decipher the mechanisms and associated genetic determinants responsible for increased resistance pattern to FDC among four sequential P. aeruginosa isolates recovered from a single patient who had never been treated with FDC.

Case history

A 65-year-old patient was hospitalized in the surgical intensive care unit of the Lille University Hospital for acute respiratory distress syndrome following pancreaticoduodenectomy. His medical history included fibrosing interstitial lung disease and rheumatoid arthritis. This patient developed ventilator-acquired pneumonia, for which four sequential clinical isolates of P. aeruginosa (P-01, P-02, P-03, and P-04) were recovered over a period of 3 weeks. The first and second isolates (P-01 and P-02) were from tracheal aspirate samples during the first week of ventilator-associated pneumonia. The second isolate (P-02) was already resistant to all tested β-lactams except ceftolozane-tazobactam (C/T). The patient was treated with meropenem (1 g q8h) IV associated with colistin IV (9 MUI q12h) because C/T was unavailable due to long-term drug shortage. The third and fourth isolates (P-03 and P-04) were recovered from tracheal aspirate samples during the second and third week of treatment, respectively.

Methods

Antimicrobial susceptibility testing was performed by using the disk diffusion method on Mueller–Hinton agar plates for selected antibiotics. Minimum inhibitory concentrations (MICs) were then determined using Etest strips (bioMérieux, La Balme-les-Grottes, France) on Mueller–Hinton agar plates at 37 °C for all antibiotics or antibiotic combinations except for FDC and colistin. MIC values of FDC were determined with the reference BMD method using iron-depleted cation-adjusted Mueller–Hinton (ID-CAMH) broth prepared following the protocol described by Hackel et al. [8]. MICs of colistin were determined using broth microdilution in cation-adjusted Mueller–Hinton broth (Bio-Rad). The results were interpreted according to the latest EUCAST breakpoints (https://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Breakpoint_tables/v_12.0_Breakpoint_Tables.pdf) [9]. The reference strain P. aeruginosa ATCC 27,853 was used as quality control for all testing.

Whole-genome sequencing (WGS) was performed for the four isolates with the ultimate goal to investigate the molecular mechanisms underlying such resistance pattern. The entire genome was sequenced using a MiSeq Illumina platform (Illumina, San Diego, CA, USA) using the Nextera sample preparation method with 2 × 150 bp paired end reads. Illumina short reads were assembled using the CLC Genomic Workbench (version 20.0.4; CLC Bio, Aarhus, Denmark), and contigs with a minimum contig length of 800 nucleotides (nt) were generated. The resulting assembled sequences were analyzed using ResFinder 4.1 software (for antimicrobial resistance genes) and MLST 2.0 software (for Multilocus sequence typing (MLST) analysis on the Center for Genomic Epidemiology server (http://www.genomicepidemiology.org/). The raw sequence data project had been deposited at GenBank under accession no PRJNA856996.

Results

According to results of antimicrobial susceptibility testing and considering the EUCAST resistance breakpoint for FDC at > 2 μg/ml, the first P. aeruginosa isolate recovered (P-01) was susceptible to FDC (2 μg/ml), albeit this MIC value was higher than that of wild-type P. aeruginosa (0.125–0.25 μg/ml). The subsequent isolates (P-02, P-03, P-04) displayed gradual elevation of FDC MICs (8, 16, and 64 μg/ml, respectively) (Table 1). Those P. aeruginosa isolates showed also variable and increasing levels of resistance to most β-lactams including meropenem (8 to > 32 μg/ml), meropenem/vaborbactam (4 to > 64 μg/ml), imipenem (> 32 μg/ml), imipenem/relebactam (3 to > 32 μg/ml), ceftazidime (16 to > 256 μg/ml), ceftazidime/avibactam (1.5 to > 256 μg/ml), ceftolozane/tazobactam (0.75 to > 2 μg/ml), cefepime (8 to 48 μg/ml), aztreonam (8 to > 256 μg/ml), tetracycline (2 to > 256 μg/ml), and chloramphenicol (> 256 μg/ml). The MIC of colistin remained unchanged and was found to be 1–2 μg/ml for all those isolates.

Table 1 MIC determination and genomic features of P. aeruginosa isolates with progressive in vivo development of resistance to cefiderocol
Full size table

All isolates possessed identical intrinsic β-lactamase genes (blaOXA-847 [encoding a narrow-spectrum oxacillinase], blaPDC-1 [encoding an AmpC-type cephalosporinase]) but no acquired β-lactamase were identified. All four P. aeruginosa strains were found to belong to sequence-type ST244, one of the most widely distributed clones worldwide. WGS data of the four strains were analyzed compared to the reference strain P. aeruginosa PAO1 (GenBank no. AE004091.2).

Considering that no acquired β-lactamases were found in any of those isolates, efforts were directed towards examining P. aeruginosa targets potentially involved in other known resistance to FDC. These included insertions, deletions, and mutations in pirS, piR, pirA, piuA, piuB, exbB-exbD-tonB3 or mutations in the pvdS, fecI, fecR, fecA, fpvA, fpvB, and fiuA—all components of the bacterial iron transport system. Also, proteins associated with impaired permeability were also assessed [7, 12]. Several identical mutations were identified in various TonB-dependent receptor proteins in all four strains including pirA (A370T), piuA (Q38H), piuB (I343L, AN573-574TD), TonB3 (F35L, V122A), fvpB (T67K, A125T, E262D), fecA (T339A, H363R) and fiuA (M670I, A677T, T771A) (Table 1). A single amino acid substitution (D89E) was identified in MexR, the negative regulator of the MexAB-OprM efflux pump, in the three FDC-resistant strains (P-02, P-03, P-04) but not in the FDC-susceptible strain P-01.

To evaluate whether this reduced susceptibility to FDC could be related to increased expression of the ampC gene, MIC values for FDC were determined in combination with cloxacillin, an inhibitor of AmpC activity (Table 1). Hence, MICs of FDC dropped to 0.5, 2, 2, and 2 μg/ml for P-01, P-02, P-03, and P-04, respectively, when combined with cloxacillin at a fixed concentration of 2000 μg/ml. Further analysis of the WGS data identified the absence of the ampD gene in all isolates, this feature being known to induce upregulation of the ampC gene. Moreover, a single amino acid substitution (Ala219Thr) was identified in AmpDh3 of the four isolates analyzed. By analyzing the outer membrane proteins of P. aeruginosa, the OprD outer membrane porin, which is involved in carbapenem uptake, was found to be truncated in all four strains that were actually resistant to carbapenem.

Discussion

The siderophore cephalosporin cefiderocol is one of the most promising commercialized agents against carbapenem-resistant Gram negatives including carbapenem-resistant P. aeruginosa. Although still uncommon, reduced susceptibility or resistance to FDC is being reported in that species. In this study, four sequential P. aeruginosa isolates were recovered from a single patient and analyzed using WGS. These isolates showed an increasing trend of resistance to broad-spectrum spectrum cephalosporins, carbapenems, and FDC over the time, without history of FDC exposure. Using the complete genome sequence data of the four P. aeruginosa isolates, all strains were shown to belong to the same sequence type, namely, ST244, a globally disseminated high-risk clone [10, 11].

The same mutations were detected in various iron transporters (pirA, piuA, piuB, TonB3, fvpB, fecA and fiuA), responsible for the active transport of FDC, for the four strains. These mutations may play an essential role in the basic increase of FDC MICs in all four strains and could be a potential mechanism of FDC resistance development in P. aeruginosa as previously reported [13,14,15,16]. However, those mutations in corresponding genes cannot explain by themselves the difference of susceptibility to FDC between those four strains.

The amino acid substitution found in MexR in the three FDC-resistant strains (P-02, P-03, P-04) was likely resulting in the overexpression of the mexAB-oprM multidrug efflux pump genes that may contribute to additional FDC resistance in those three strains as previously reported [17]. No mutation in the genes encoding TonB-dependent receptors (TBDRs) was identified in those isolates [17]. Very recently, Ikawa et al. indicated that the MexAB-OprM drug efflux system contributes to the intrinsic resistance to FDC in P. aeruginosa and found that the mexAB-oprM-deficient P. aeruginosa mutant displayed increased FDC susceptibility compared to the wild-type strain. On the other hand, the overexpression of mexAB-oprM in the mexAB-oprM-deficient mutant increased the MIC value of FDC [18]. In another study, MIC of FDC was slightly affected by the oprD gene deficiency or the overproduction of MexAB-OprM multidrug efflux pump due to a defect in the mexR or nalD regulatory genes in P. aeruginosa [4]. Although the association between MexAB-OprM overproduction and reduced FDC activity is still poorly understood, the role of mexAB-OprM overexpression in reducing FDC activity needs to be further explored as FDC could be a substrate of this efflux pump. Interestingly, truncation of the OprD-encoding gene had been previously reported among in vivo selected FDC-resistant clinical P. aeruginosa isolates (after FDC treatment) [14].

MICs of FDC were significantly decreased in the presence of cloxacillin, suggesting that overexpression of ampC gene was involved in the increased MICs of FDC in those four P. aeruginosa isolates. This increased expression of the ampC gene could be explained by the inactivation of ampD gene in the four isolates as previously described [19,20,21,22]. In a study, mutations in the ampD gene were associated with modest increases in FDC MICs to 2 μg/ml, remaining in the susceptible range in that latter case [17]. Moreover, the amino-acid substitution (Ala219Thr) identified in AmpDh3 of the four isolates might play an important role in overproduction of cephalosporinase leading to ceftazidime resistance, as previously reported [23], which likely cause collateral resistance to FDC since FDC combines chemical moieties of ceftazidime and cefepime. We noticed that the FDC-resistant P. aeruginosa strains (P-02, P-03, and P-04) showed a 4- to tenfold increase in FDC MICs compared to the FDC susceptible isolate (P-01). This increase in FDC MICs could be attributed to strong collateral-resistance with ceftazidime for which MICs were elevated by 4- to ≥ eightfold in those FDC-resistant P. aeruginosa strains compared to the FDC-susceptible isolate (P-01). Similar results were observed for cefepime and ceftazidime/avibactam (Table 1). Very recently, it has been shown that development of FDC resistance was often associated with collateral susceptibility changes towards other β-lactams such as increased resistance to ceftazidime and ceftazidime/avibactam [24].

In conclusion, we report here in vivo development of FDC resistance in P. aeruginosa clinical isolates without exposure to FDC is raising concerns about the effectiveness of this promising drug. Whole-genome sequence analysis suggests that this resistance was driven by multifactorial mechanisms including changes in iron transporter proteins, ampC gene overproduction, and mexAB-oprM overexpression.