Impact of transferable β-lactamases and intrinsic AmpC amino acid substitutions on the activity of cefiderocol against wild-type and iron uptake-deficient mutants of Pseudomonas aeruginosa

Abstract

Introduction

Cefiderocol is a newly FDA-approved siderophore-cephalosporin conjugate with potent activity against Pseudomonas aeruginosa, including strains with difficult-to-treat phenotypes.¹ The activity of cefiderocol against P. aeruginosa results from the following: (i) its remarkable stability against the intrinsic AmpC of P. aeruginosa (PDC) and transferable β-lactamases (including metallo-β-lactamases), owing to its hybrid cephalosporin structure that incorporates a pyrrolidinium group from cefepime and a carboxypropanoxyimino group from ceftazidime on the C-3 and C-7 side chains, respectively; (ii) the enhanced internalization in the bacterial cell via siderophore-mediated pathways resulting from a catechol moiety in the C-3 side chain, which leads to iron being chelated and transported to the periplasmic space via active iron uptake systems at higher rates than in any other β-lactam.²

Recently, analysis of in vitro evolution dynamics has shown that the selection of inactivating mutations in piuC (which encodes an iron-dependent oxygenase involved in the expression of the adjacent outer membrane receptor for iron transport designated piuA) is a key step in the acquisition of cefiderocol resistance in P. aeruginosa.³ Adding further concern, it has recently been demonstrated that the acquisition of some widespread transferable β-lactamases (e.g. NDM-1)⁴ and selection of specific amino acid substitutions in the P. aeruginosa PDC enzyme (e.g. E219K and L293P),³ are associated with increased cefiderocol MICs. However, the mechanistic basis underlying cefiderocol resistance in P. aeruginosa is not yet fully understood. More specifically, the contribution and interplay between impaired iron internalization and the expression of different β-lactamases have not yet been analysed in detail. Thus, to address these unresolved gaps and anticipate potential strategies to overcome cefiderocol resistance in P. aeruginosa, we tested the activity of this new cephalosporin against the most relevant transferable β-lactamases and intrinsic PDC variants under wild-type and iron uptake-deficient P. aeruginosa backgrounds.

Methods

Cloning of transferable β-lactamases and PDC mutants in isogenic wild-type and iron uptake-deficient P. aeruginosa backgrounds

Up to 31 of the most clinically relevant Ambler’s class A (bla_GES, bla_CTX-M, bla_SHV, bla_PER, bla_VEB and bla_KPC), B (bla_VIM, bla_IMP and bla_NDM), C (bla_CMY, bla_DHA and bla_FOX) and D (bla_OXA-2, bla_OXA-10 and bla_OXA-48) transferable β-lactamases and the intrinsic PDC-1 and 15 derivatives carrying specific amino acid substitutions previously associated with resistance to ceftazidime/avibactam or ceftolozane/tazobactam⁵ were selected. The β-lactamase genes were cloned in the pUCP24 plasmid following our previously described methodology.^6,7 The recombinant plasmids obtained were electrotransferred in parallel into the PAO1 reference strain and its iron uptake-deficient transposon mutant PAO ΔpiuC.⁸

Antimicrobial susceptibility testing

MICs of piperacillin/tazobactam, aztreonam, ceftazidime, ceftazidime/avibactam, ceftolozane/tazobactam, cefepime, imipenem, meropenem and cefiderocol were determined in triplicate, for all isolates, by reference broth microdilution assays. CAMHB was used in all experiments except for the cefiderocol assay, in which iron-depleted CAMHB (prepared according to CLSI M100 guidelines)⁹ was used. Avibactam and tazobactam were tested at a fixed concentration of 4 mg/L. EUCAST v 14.0 clinical breakpoints and guidelines (http://www.eucast.org/clinical_breakpoints/) were used for reference purposes. Reference strains Escherichia coli ATCC 25922 and P. aeruginosa ATCC 27853 were used as controls.

Molecular modelling

The molecular graphics program PyMOL was used to visualize and depict the substituted or deleted residues in the 3D structure of PDC associated with higher effects on cefiderocol resistance.¹⁰

Data availability

The nucleotide sequence data for the newly described PDC-577 variant (carrying the previously undescribed deletion ΔP215-G222) have been deposited in GenBank under accession number OR609871.1.

Results

The use of cloning assays enabled us to construct a representative isogenic collection of up to 94 P. aeruginosa laboratory transformants, 47 of which were derived from PAO1 and 47 from its iron uptake-deficient mutant PAO ΔpiuC. Comparative antibiotic susceptibility for the 31 transformants expressing transferable β-lactamases is detailed in Table 1 (the MICs of conventional β-lactams only refer to PAO1 because, with the exception of cefiderocol, expression of the β-lactamases in PAO ΔpiuC had negligible effect on the activity). Overall, most of the transformants exhibited high levels of resistance towards classic antipseudomonals. Resistance to ceftazidime/avibactam (17/31, 54.8% resistant) and ceftolozane/tazobactam (24/31, 77.4% resistant) was frequent, with the MICs of both ranging from ≤1 to >512 mg/L. Cefiderocol was the most active β-lactam against PAO1 transformants (8/31, 25.8% resistant), with MICs ranging from 0.125 to 16 mg/L. In PAO1, cefiderocol resistance was caused by the production of SHV-12 (MIC = 16 mg/L), the ceftazidime/avibactam-resistant KPC variants KPC-35 and KPC-31 (MIC = 4–8 mg/L), NDM enzymes (MICs = 8 mg/L) and OXA-15 (MIC = 4 mg/L). Reduced susceptibility (MICs = 1–2 mg/L) was observed for PER-1, VEB-1, KPC-2, KPC-3, VIM-1, CMY-2, OXA-2 and OXA-14. The GES, CTX-M, VIM-2, VIM-20 (VIM-2-like), IMP, DHA, FOX, OXA-10 and OXA-48 enzymes had weaker effects on cefiderocol activity (≤4-fold increase; MICs = 0.125–0.5 mg/L). Expression of these same β-lactamases in PAO ΔpiuC (which yielded a 16-fold increase in the baseline cefiderocol MIC relative to PAO1) resulted in an additive effect that conferred high-level cefiderocol resistance (MICs = 16 to >256 mg/L) in all of the aforementioned enzymes associated with reduced susceptibility or resistance to cefiderocol in PAO1, with greater effects noted for SHV-12, KPC-35, KPC-31, NDMs and OXA-14. By contrast, in PAO ΔpiuC, production of GES-, CTX-M-, VIM-2-like enzymes, DHA-1, FOX-4, OXA-10 and OXA-48 conferred cefiderocol MICs which did not exceed 8 mg/L in any case (MICs = 2–8 mg/L).

Comparative MIC data of cefiderocol and the other β-lactams against the transformants expressing PDC-1 and variants with enhanced cephalosporinase activity are shown in Table 2. Overexpression of PDC-1 conferred resistance to the conventional antipseudomonals, but in all cases retained susceptibility to ceftazidime/avibactam (MIC = 2 mg/L), ceftolozane/tazobactam (MIC = 1 mg/L) and particularly cefiderocol (MIC = 0.25 mg/L). Relative to PDC-1, most variants were associated with decreased resistance to piperacillin/tazobactam (8/15, 53.3% resistant; MICs = 8–512 mg/L), with variable effects on the activity of aztreonam (6/15, 40.0% resistant; MICs = ≤4–128 mg/L) and cefepime (11/15, 73.3% resistant; MICs = 4–128 mg/L) and enhanced resistance to ceftazidime (15/15, 100% resistant; MICs = 16 to >256 mg/L), ceftazidime/avibactam (5/15, 33.3% resistant; MICs = ≤1–128 mg/L) and ceftolozane/tazobactam (14/15, 93.3% resistant; MICs = 1–256 mg/L). Cefiderocol was the most active β-lactam (3/15, 20.0% resistant; MICs = 0.125–4 mg/L) since most PDC variants only resulted in a 4-fold or less increase in the cefiderocol MIC compared to PDC-1. However, the allelic variants carrying the substitutions ΔP215-G222, E219K, L293P and F121L + E219K resulted in an 8- to 16-fold increase in MIC relative to PDC-1, leading in all cases to reduced susceptibility or cefiderocol resistance (MICs = 2–4 mg/L). As observed for transferable enzymes, expression of these PDC-derived β-lactamases in the PAO ΔpiuC strain again revealed an additive effect on cefiderocol resistance. The aforementioned PDCs associated with a weaker impact on the cefiderocol MIC generally did not yield MICs exceeding 8 mg/L. However, the variants carrying the ΔP215-G222, E219K, L293P or the dual F121L + E219K substitutions yielded in all cases high-level cefiderocol resistance (MICs = 32–64 mg/L). Molecular modelling of these PDC substitutions revealed that the ΔP215-G222 deletion and the E219K substitution were located in the Ω-loop of the PDC structure and that the L293P substitution was located in the H10 helix region (Figure 1).

Figure 1.

Crystal structure of PDC-1 from P. aeruginosa in the wild-type form (PDB ID 4GZB, 1.79 Å).¹² The enzyme sites (Ω-loop and H10 helix) and the deleted or replaced amino acids (sticks) in the variant enzymes identified are highlighted in the overall (left) and close-up (right) views. The position of key residues in the enzyme active site involving the SVSK and KTG motifs are also shown (yellow). Note how the amino acid sequence modifications identified in this study are located in the Ω-loop [E219K (orange) and ΔP215-G222 (green)] and in H10 helix (L293P, pink). These sites are involved in the R1 and R2 binding sites, respectively, of cephalosporines. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Open in new tabDownload slide

Discussion

Cefiderocol activity rates of almost 100% have been reported for representative collections of clinical isolates of P. aeruginosa recovered as part of large-scale antimicrobial surveillance initiatives.^1,13,14 Cefiderocol resistance in P. aeruginosa has been associated with the production of β-lactamases with increased cephalosporinase activity or truncated iron internalization systems. However, the specific contribution and cooperation of these mechanisms to cefiderocol resistance have scarcely been explored. In agreement with previous findings,³ a first look at the baseline cefiderocol MIC of the wild-type PAO1 strain and its transposon mutant derivative PAO ΔpiuC revealed a 16-fold increase in the cefiderocol MIC (from 0.125 to 2 mg/L, respectively). This demonstrates that blocking the ability of cefiderocol to additionally penetrate the P. aeruginosa cell via this iron transport system relegates the antipseudomonal activity to levels comparable with those of conventional cephalosporins such as ceftazidime and cefepime, against which PAO1 usually yields MICs of 1–2 mg/L. Among the different PAO1-derived transposon mutants available in the PAOUW library⁸ with inactivated iron uptake systems (pirA, pirR, etc.), the PAO ΔpiuC mutant was selected as host in the present study because it has recently been identified as the most important factor for cefiderocol uptake in P. aeruginosa (it was recently found to be mutated in up to 21 of 30 P. aeruginosa mutants obtained after in vitro cefiderocol exposure).³

Regarding the effects of the interplay between inactivated piuC and transferable β-lactamases on cefiderocol activity, our results provide new evidence regarding how high-level resistance (MICs of 16 to >256 mg/L) to this agent may be achieved when impaired iron uptake is combined with the production of SHV-12, KPC enzymes (particularly those carrying Ω-loop mutations), NDM variants, OXA-2, OXA-14, OXA-15 and, to a lesser extent, PER-1, VEB-1, VIM-1 and CMY-2. While some of the enzymes considered here are still an infrequent cause of β-lactam resistance in P. aeruginosa (SHV-12, CMY-2, etc.), other variants are already widespread (e.g. VIM-1, OXA-14, OXA-15 or PER-1) or are globally expanding (e.g. NDMs, KPC Ω-loop mutants) among P. aeruginosa strains worldwide.^15,16 Our findings indicate the potential association between isolates carrying such β-lactamases and the emergence of high-level resistance, thus arguing for close monitoring of their spread. Moreover, and probably more importantly, some of the enzymes included here and of well-known clinical relevance had minimal effects on the cefiderocol activity, even when expressed in PAO ΔpiuC. This applied to many GES variants, CTX-M-9, CTX-M-15, VIM-2-like enzymes, IMPs, DHA-1, FOX-4, OXA-10 and OXA-48, for which our findings reinforce the positioning of cefiderocol as an attractive and highly stable β-lactam.

Analysis of the impact of PDC substitutions also revealed useful clues to understanding the evolutionary potential of this enzyme towards cefiderocol resistance. Interestingly, most of the commonly described ceftolozane/tazobactam resistance substitutions were not able to increase the cefiderocol MIC further than 0.5 mg/L in PAO1 or 8 mg/L (interestingly, the clinical breakpoint for ceftazidime and cefepime) when expressed in PAO ΔpiuC, highlighting the stability of cefiderocol against most PDC variants. These results are consistent with our recent findings on ceftolozane/tazobactam- and ceftazidime/avibactam-resistant P. aeruginosa strains and position cefiderocol as a useful therapeutic alternative in such cases.⁵ The exceptions are those variants carrying the E219K and the previously undescribed 8 amino acid deletion ΔP215-G222, which conferred cross-resistance to ceftolozane/tazobactam, ceftazidime/avibactam and cefiderocol. Interestingly, both of these substitutions are located in the Ω-loop of the PDC structure, which is key to accommodating the R1 side chain of cephalosporins.¹⁷ On the other hand, the L293P was included due to its recent association with in vitro resistance development to cefiderocol.³ Fortunately, this amino acid change, which is located in the H10 helix (thus suggesting a major impact on the accommodation of substrates with a bulky R2 side chain)¹⁷ and has been previously encountered in up to 9 PDC variants found in strains of clinical origin,¹⁸ does not confer cross-resistance to ceftolozane/tazobactam or ceftazidime/avibactam. Intriguingly, these three types of amino acid alterations (E219K, ΔP215-G222 and L293P) have a strong impact on cefiderocol resistance in both normal and impaired iron uptake conditions, and thus their potential emergence during cefiderocol treatment should be closely monitored.

Altogether, the study findings conclusively demonstrate that high-level cefiderocol resistance in P. aeruginosa results from the combination of specific transferable β-lactamases or PDC substitutions and decreased iron uptake. Managing the information about the underlying β-lactamase genotype is of critical importance for optimizing cefiderocol treatments against P. aeruginosa.

Acknowledgements

We gratefully acknowledge Shionogi for providing us with cefiderocol powder.

Funding

This work was supported by the Instituto de Salud Carlos III (ISCIII) through the projects PI20/01212, PI21/00017, PI21/00704, PI22/01212 and PI23/00851, and co-funded by the European Union. The research was also funded by Centro de Investigación Biomédica en Red de Enfermedades Infecciosas (CIBERINFEC, CB21/13/00055 and CB21/13/00099), integrated in the National Plan for Scientific Research, Development and Technological Innovation and funded by the ISCIII–General Subdirection of Assessment and Promotion of the Research–European Regional Development Fund (ERDF) ‘A way of making Europe’. The study was also funded by the Axencia Galega de Innovación (GAIN), Consellería de Innovación and Consellería de Economía, Emprego e Industria, Xunta de Galicia (IN607D 2021/12 to A.B., IN607A 2016/22 to G.B. and IN607D 2024/008 to J.A.-S.). Financial support from the Spanish State Agency of Research (PID2022-136963OB-I00/AEI/10.13039/501100011033 to C.G.-B.), the Xunta de Galicia [ED431C 2021/29 and Centro Singular de Investigación de Galicia accreditation 2019–2022 (ED431G 2019/03), C.G.-B.], and the European Regional Development Fund (ERDF) is gratefully acknowledged. L.G.-P. was financially supported by the ISCIII project PI21/00704 and the ISCIII PFIS program (FI23/00074). T.B.-M. was financially supported by the ISCIII project PI20/00686 and the ISCIII Río Hortega program (CM23/00095). I.A.-G. was financially supported by the ISCIII Juan Rodés program (JR23/00036). S.R.-P. was financially supported by the ISCIII Río Hortega program (CM23/00104). M.O.-G. was financially supported by GAIN-Xunta de Galicia (IN606A 2023/023). M.A.G.-F. was financially supported by the ISCIII PFIS program (FI22/00039). J.C.V.-U. was financially supported by GAIN-Xunta de Galicia (IN606B 2022/009). J.A.-S. was financially supported by the ISCIII Juan Rodés program (JR21/00026).

Transparency declarations

Shionogi provided cefiderocol powder and did not exercise any control over the conduct or reporting of the research. J.C.V.-U. has received honoraria for lectures and/or presentations from Merck Sharp & Dohme (MSD). A.O. has received grants or contracts from MSD, Wockhardt and Shionogi, consulting fees and honoraria for lectures and/or presentations from MSD, Pfizer and Shionogi. G.B. has received funding and study materials from MSD, grants or contracts from MSD, Pfizer, ABAC Therapeutics and Roche, consulting fees and honoraria for lectures and/or presentations from MSD, Shionogi Pfizer, Roche and Menarini, and support for attending meetings and/or travels from Pfizer. J.A.-S. has received honoraria for lectures and/or presentations from MSD, Shionogi and Advanz Therapeutics, and support for attending meetings and/or travels from Pfizer. All other authors: none to declare.

References

Author notes