In vitro evolution of cefepime/zidebactam (WCK 5222) resistance in Pseudomonas aeruginosa: dynamics, mechanisms, fitness trade-off and impact on in vivo efficacy

Abstract

Objectives

To study the dynamics, mechanisms and fitness cost of resistance selection to cefepime, zidebactam and cefepime/zidebactam in Pseudomonas aeruginosa.

Methods

WT P. aeruginosa PAO1 and its ΔmutS derivative (PAOMS) were exposed to stepwise increasing concentrations of cefepime, zidebactam and cefepime/zidebactam. Selected mutants were characterized for change in susceptibility profiles, acquired mutations, fitness, virulence and in vivo susceptibility to cefepime/zidebactam. Mutations were identified through WGS. In vitro fitness was assessed by measuring growth in minimal medium and human serum-supplemented Mueller–Hinton broth. Virulence was determined in Caenorhabditis elegans and neutropenic mice lung infection models. In vivo susceptibility to a human-simulated regimen (HSR) of cefepime/zidebactam was studied in neutropenic mice lung infection.

Results

Resistance development was lower for the cefepime/zidebactam combination than for the individual components and high-level resistance was only achieved for PAOMS. Cefepime resistance development was associated with mutations leading to the hyperexpression of AmpC or MexXY-OprM, combined with PBP3 mutations and/or large chromosomal deletions involving galU. Zidebactam resistance was mainly associated with mutations in PBP2. On the other hand, resistance to cefepime/zidebactam required multiple mutations in genes encoding MexAB-OprM and its regulators, as well as PBP2 and PBP3. Cumulatively, these mutations inflicted significant fitness cost and cefepime/zidebactam-resistant mutants (MIC = 16–64 mg/L) remained susceptible in vivo to the HSR.

Conclusions

Development of cefepime/zidebactam resistance in P. aeruginosa required multiple simultaneous mutations that were associated with a significant impairment of fitness and virulence.

Introduction

Pseudomonas aeruginosa exhibits a remarkable ability to acquire and express multiple resistance mechanisms, thus posing one of the highest therapeutic challenges.^1,2 In P. aeruginosa the acquisition of β-lactamases, such as MBLs and ESBLs, through horizontal gene transfer continue to signify spectrum gaps in current as well as newer β-lactam/β-lactamase inhibitor (BL/BLI) combinations—ceftolozane/tazobactam and ceftazidime/avibactam.^3–7 The traditional approach of using a BL/BLI is further challenged by the presence of non-enzymatic resistance mechanisms mediated by a vast repertoire of mutations in P. aeruginosa.^8–11

Zidebactam was the first described Gram-negative β-lactam enhancer belonging to the bicyclo-acyl hydrazide (BCH) series with standalone anti-pseudomonal activity; the combination of cefepime and zidebactam is currently under clinical development for infections caused by MDR/XDR Gram-negative bacteria, including P. aeruginosa. Although derived from a diazabicyclooctane (DBO) scaffold, BCHs were designed with an objective of augmenting PBP2 binding as opposed to the conventional approach of optimizing the β-lactamase-inhibitory activity.^12,13 The triple action (direct antibacterial activity, β-lactam enhancement and β-lactamase inhibition) provided by zidebactam offers a unique mechanistic opportunity to simultaneously overcome P. aeruginosa resistance mechanisms when combined with cefepime. However, potential mechanisms of resistance to zidebactam and cefepime/zidebactam in this organism remain unexplored. Thus, the objective of this work was to analyse the dynamics and mechanisms of in vitro development of resistance to cefepime, zidebactam and cefepime/zidebactam, through stepwise exposure to increasing drug concentrations, using a WT and a hypermutable (mutator) strain of P. aeruginosa. Selected mutants were analysed for (i) mutational changes through WGS; (ii) fitness cost of resistance; and (iii) in vivo susceptibility to cefepime/zidebactam.

Materials and methods

In vitro resistance development assay

A protocol previously described by our group was used.¹⁴ Briefly, at Day 0, 10 mL tubes of CAMHB containing 0.25–64 mg/L cefepime, 0.5–256 mg/L zidebactam or 0.25–64 mg/L cefepime/zidebactam at a 1:1 ratio or with zidebactam fixed at 4 mg/L were inoculated with P. aeruginosa PAO1 (WT) or PAOMS (mutS-deficient mutator derivative of PAO1)¹⁵ strains (10⁶ cfu/mL) in triplicate. The tubes were incubated at 37°C for 24 h with shaking at 180 rpm. Tubes from the highest antibiotic concentration showing visible growth were used to re-inoculate (1:1000 dilution) fresh medium containing the same concentration range of antibiotics. Cultures were also seeded in antibiotic-free Mueller–Hinton (MH) agar plates and two isolated colonies were randomly selected and stored frozen for further study. The same procedure was repeated for a total of seven consecutive days. Two colonies per strain, replicate experiments and antibiotic regimen were characterized as described below.

Susceptibility testing

The MICs of piperacillin/tazobactam, ceftazidime, ceftazidime/avibactam, ceftolozane/tazobactam, cefepime, aztreonam, imipenem, meropenem, ciprofloxacin, tobramycin, amikacin and colistin were determined by broth microdilution using customized Sensititre plates (Ref. FRCNRP, Thermo Fisher Scientific). MICs of cefepime, zidebactam, cefepime/zidebactam at a fixed concentration of 4 mg/L zidebactam and cefepime/zidebactam at a 1:1 ratio were further analysed through broth microdilution. MICs were interpreted using EUCAST 2020 breakpoints (www.eucast.org).

WGS

Previously described protocols were used.¹⁶ Briefly, total DNA was isolated and indexed paired-end libraries were generated (Nextera XT DNA Library Preparation Kit, Illumina) and sequenced with a MiSeq desktop sequencer. The reads for each isolate were mapped against the genome of the P. aeruginosa reference strain PAO1 (RefSeq accession number NC_002516.2) using Bowtie 2 software, version 2.2.4.¹⁷ Pileups and raw files of the mapped reads were obtained using SAMtools, version 0.1.16¹⁸ and PicardTools, version 1.140. Read alignments surrounding all putative insertions/deletions (indels) were realigned using the Genome Analysis Toolkit (GATK), version 3.4-46.¹⁹

Gene expression

The expression of the genes encoding the chromosomal β-lactamase AmpC (ampC), MexAB-OprM (mexB) and MexXY-OprM (mexY) efflux pumps was determined from late-log-phase LB broth cultures at 37°C and 180 rpm shaking by real-time RT–PCR with a Bio-Rad system (CFX Connect Real-Time System), as previously described.²⁰

Caenorhabditis elegans killing assay

The assay for studying bacterial killing of C. elegans was performed as described previously.²¹ Briefly, a fresh culture of each bacterial strain was layered onto a 55 mm diameter plate containing 5 mL of potato dextrose agar and incubated at 37°C for 24 h to form bacterial lawns. Five worms per plate were then placed on top of the lawns. The plates were incubated at 24°C and scored for living worms at 0, 24, 48, 72 and 168 h. At least three independent experiments were performed and mean ± SD values were recorded. The Escherichia coli strain OP50, used to feed the nematodes, served as a non-pathogenic control. For each of the mutants, C. elegans virulence scores (CEVS) 1 to 5 were assigned following previously established definitions.²¹

In vitro growth in minimal medium and MH broth supplemented with human serum

Growth profiles of P. aeruginosa PAO1, PAOMS and their cefepime, zidebactam and cefepime/zidebactam-selected mutants were determined in M9 minimal medium (prepared by mixing 5 g/L NH₄Cl, 15 g/L KH₂PO₄, 64 g/L Na₂HPO₄·7H₂0, 2.5 g/L NaCl supplemented with 0.4% glucose, 0.1 mM CaCl₂ and 2 mM MgSO₄ at pH 7.2). Overnight bacterial cultures were diluted (1:10) in CAMHB and grown at 37°C and 180 rpm until the cultures reached mid-log phase. The cultures were then diluted in minimal medium (10⁴ cfu/mL) and the periodic growth at 37°C incubation in ambient air with 180 rpm shaking was monitored by measuring OD₆₀₀ till 28 h. Similarly, comparative growth assessment in MH broth supplemented with 50% pooled fresh human serum was also undertaken by measuring the changes in the cfu/mL (starting inoculum of 1 × 10⁶ cfu/mL) till 24 h.

In vivo growth and virulence assessment in the neutropenic mice lung infection model

All animal experiments were approved by the institutional animal ethics committee of Wockhardt. The in vivo fitness of parents and their representative mutants was assessed in a single experiment by monitoring (a) growth in the lungs of neutropenic mice over 24 h; and (b) mortality among infected mice (a measure of virulence) 24 h post-infection. Male/female Swiss albino mice (25–27 g) were rendered neutropenic by intraperitoneal injections of 150 and 100 mg/kg cyclophosphamide 4 days and 1 day prior to infection, respectively. Mice were transiently anaesthetized and intranasally inoculated with 0.05 mL of 10⁷–10⁸ cfu/mL of overnight bacterial cultures (5 × 10⁵–5 × 10⁶ cfu/mouse). In order to assess the in vivo growth, the bacterial counts in the lungs were enumerated at 2, 6 and 12 h and at the study endpoint of 24 h post-infection. For this purpose, five animals were randomly allocated at each timepoint of assessment. In the same experiment, the mortality was observed in the 24 h animal group. The lung bacterial count was enumerated as follows: mice were humanely euthanized, all lung lobes were aseptically harvested and homogenized and serial dilutions were plated on Trypticase soy agar for bacterial counts.

In vivo efficacy in the neutropenic mice lung infection model

This experiment (24 h treatment model) was conducted to assess whether the cefepime/zidebactam-selected mutants were susceptible in vivo to cefepime/zidebactam at its mice exposures that are equivalent to clinical exposures obtained from a 2 g/1 g, q8h dose regimen (proposed therapeutic dose of cefepime/zidebactam for Phase 3 clinical study). The efficacy of cefepime/zidebactam treatment was determined in comparison with standalone cefepime and zidebactam treatments.

Male/female Swiss albino mice (25–27 g) were rendered neutropenic as described earlier. Mice were treated with uranyl nitrate (5 mg/kg, intraperitoneal) 3 days before the infection to induce a predictable degree of renal impairment and facilitate simulation of clinical exposures.²² The human-simulated regimen (HSR) was identified by superimposing cefepime and zidebactam exposures (based on %fT_>MIC) in mice with human Phase 1 mean exposure profile. The mice doses of cefepime and zidebactam that generated human-simulated exposures are provided in Table S1 (available as Supplementary data at JAC Online). The plasma time–concentration profile of cefepime and zidebactam in human and mice is shown in Figure S1 and Table S2.

Two hours prior to the initiation of antibacterial therapy, each neutropenic mouse received an infective inoculum of 1 × 10⁸–5 × 10⁸ cfu/mL (5 × 10⁶–1 × 10⁷ cfu/mouse). Since these cefepime/zidebactam-selected mutants were inherently defective for in vivo growth, a relatively higher bacterial density was employed for infection to ascertain at least 2 log₁₀ growth after 24 h in lungs of untreated animals so that drug treatment effect could be judged reliably. The treatment of infected mice was initiated 2 h post-infection (0 h). The efficacy of the cefepime HSR, the zidebactam HSR and the cefepime/zidebactam HSR was judged by the change in the counts (cfu/lung) at 24 h-post initiation of treatment compared with 0 h. Each group (treatment or vehicle control) comprised five mice.

Results

Dynamics of in vitro resistance development to cefepime, zidebactam and cefepime/zidebactam

Comparative analysis of stepwise resistance development to cefepime, zidebactam and cefepime/zidebactam (at a 1:1 ratio and with fixed 4 mg/L zidebactam) in WT PAO1 and its mutator derivative PAOMS is shown in Figure 1. As expected, emergence of resistance occurred more rapidly in the mutator strain than in the WT. Combination with zidebactam significantly reduced the emergence of resistance to cefepime in both strains. Remarkably, resistance development to the fixed 4 mg/L concentration of zidebactam did not occur for the WT strain PAO1, even with cefepime concentrations as low as ¼× MIC (0.25 mg/L).

Figure 1.

Dynamics of stepwise resistance development to cefepime, zidebactam and combinations in broth cultures in the WT reference strain PAO1 and its mutS-deficient (PAOMS) derivative mutant. Reference cefepime and zidebactam MICs were 1 and 4 mg/L, respectively, for both strains. Median values and ranges for three experiments are shown.

Mechanisms of in vitro resistance development to cefepime, zidebactam and combinations

Table S3 shows the complete list of non-synonymous mutations detected in each of the terminal mutants (Day 7; D7). The number of mutations acquired by PAOMS was higher than for PAO1 (28–62 versus 4–27 mutations). Interestingly, one cefepime-exposed PAO1 lineage showed a mutL loss-of-function mutation (nt772Δ1) indicating the acquisition of a hypermutator phenotype, thereby accumulating a higher number of mutations for this lineage compared with other PAO1 lineages (27 versus 4–8 mutations).

The characterization of the PAO1 and PAOMS terminal mutants is shown in Table 1 (susceptibility profiles) and Table 2 (resistance gene expression data and mutations detected in genes known to be involved in antibiotic resistance).

Table 1.

Susceptibility profiles of PAO1 and PAOMS mutant derivatives selected after 7 days of exposure to increasing concentrations of cefepime, zidebactam or cefepime/zidebactam combinations

IDa	MIC (mg/L)b
	TIC (S ≤ 4)	TZP (S ≤ 16)	ATM (S ≤ 16)	C/T (S ≤ 4)	CAZ (S ≤ 8)	CZA (S ≤ 8)	IPM (S ≤ 4)	MEM (S ≤ 2)	AMI (S ≤ 8)	TOB (S ≤ 4)	CST (S ≤ 2)	CIP (S ≤ 0.5)	FEP (S ≤ 8)	ZID	FEP/ ZID4c	FEP/ZID (1:1)d
PAO1	16	≤4	4	≤0.5	1	1	2	1	2	0.5	1	≤0.12	1	4	≤0.25	1
PAOMS	32	≤4	8	1	2	1	2	1	4	0.5	1	≤0.12	1	4	≤0.25	1
FEP-D7P1m1	256	64	32	2	16	16	4	8	≤2	≤0.25	1	0.5	8	4	4	4
FEP-D7P1m2	˃512	128	128	8	64	˃32	4	8	≤2	≤0.25	1	0.5	16	16	16	16
FEP-D7P2m1	256	128	64	8	32	32	4	16	32	4	≤0.5	0.25	64	16	32	16
FEP-D7P2m2	512	˃256	64	16	˃64	˃32	4	8	16	2	1	0.5	64	8	32	8
FEP-D7P3m1	˃512	64	128	8	32	32	32	64	4	0.5	2	0.25	128	64	128	16
FEP-D7P3m2	˃512	64	˃128	16	˃64	˃32	32	˃64	8	1	1	1	128	64	128	32
FEP-D7M1m1	64	128	32	4	32	4	8	4	4	0.5	1	≤0.12	64	8	2	4
FEP-D7M1m2	256	˃256	64	8	˃64	32	16	8	4	0.5	1	≤0.12	64	16	8	8
FEP-D7M2m1	512	˃256	˃128	˃32	˃64	˃32	16	16	8	1	1	≤0.12	64	16	8	8
FEP-D7M2m2	512	˃256	128	8	˃64	8	1	1	≤2	≤0.25	1	≤0.12	64	˃128	8	8
FEP-D7M3m1	512	˃256	64	16	˃64	˃32	32	16	8	1	2	≤0.12	64	8	4	4
FEP-D7M3m2	256	˃256	64	16	˃64	˃32	16	16	8	1	2	≤0.12	64	8	8	8
ZID-D7P1m1	32	4	4	1	2	2	4	2	2	0.25	1	≤0.12	1	˃128	0.5	1
ZID-D7P1m2	16	4	4	≤0.5	2	2	4	2	2	0.25	1	≤0.12	1	˃128	0.5	1
ZID-D7P2m1	16	4	2	1	2	2	16	2	2	0.25	2	≤0.12	1	˃128	1	1
ZID-D7P2m2	16	4	2	≤0.5	2	2	8	2	2	0.25	2	≤0.12	1	˃128	1	1
ZID-D7P3m1	64	8	8	≤0.5	2	2	8	4	4	0.5	2	≤0.12	1	˃128	0.5	1
ZID-D7P3m2	32	8	4	1	2	2	8	4	4	0.5	2	≤0.12	1	˃128	0.5	1
ZID-D7M1m1	16	≤4	≤2	1	≤1	1	8	2	8	1	2	≤0.12	1	˃128	0.5	2
ZID-D7M1m2	32	16	8	≤0.5	4	2	8	8	4	0.5	2	≤0.12	2	˃128	0.5	2
ZID-D7M2m1	32	8	4	1	4	4	8	2	4	0.5	2	≤0.12	2	˃128	1	2
ZID-D7M2m2	128	32	32	2	8	8	16	16	4	0.5	2	0.25	4	˃128	4	4
ZID-D7M3m1	128	32	32	2	8	4	8	8	≤2	0.5	1	0.25	4	˃128	4	4
ZID-D7M3m2	64	16	16	1	4	1	16	16	4	0.5	2	0.25	2	˃128	2	2
FEP/ZID (1:1)-D7P1m1	256	8	4	2	8	8	8	16	8	1	1	0.5	16	64	16	16
FEP/ZID (1:1)-D7P1m2	256	8	8	2	8	8	4	16	8	1	1	0.5	16	64	16	16
FEP/ZID (1:1)-D7P2m1	128	16	16	2	8	8	8	8	4	0.5	2	1	8	32	8	8
FEP/ZID (1:1)-D7P2m2	256	32	32	2	16	16	8	16	4	0.5	1	0.5	16	32	16	8
FEP/ZID (1:1)-D7P3m1	256	32	16	2	8	8	16	16	4	0.5	1	0.25	8	32	8	8
FEP/ZID (1:1)-D7P3m2	256	32	16	2	8	8	8	8	4	0.5	1	0.25	8	32	8	8
FEP/ZID (1:1)-D7M1m1	512	32	12	4	˃64	˃32	64	˃64	4	0.5	2	0.5	64	˃128	32	32
FEP/ZID (1:1)-D7M1m2	˃512	32	128	4	64	˃32	16	˃64	4	0.5	1	0.25	64	˃128	32	32
FEP/ZID (1:1)-D7M2m1	512	64	˃128	4	64	˃32	8	˃64	4	0.5	2	0.25	64	˃128	64	64
FEP/ZID (1:1)-D7M2m2	512	32	128	4	64	32	8	˃64	4	0.5	1	0.25	64	˃128	64	64
FEP/ZID (1:1)-D7M3m1	˃512	32	˃128	16	˃64	˃32	16	˃64	2	0.5	2	0.25	128	˃128	128	64
FEP/ZID (1:1)-D7M3m2	˃512	64	˃128	8	˃64	˃32	8	˃64	4	0.5	1	0.25	128	˃128	128	64
FEP/ZID4-D7M1m1	512	64	32	8	64	32	16	16	≤2	0.5	1	0.25	32	128	64	64
FEP/ZID4-D7M1m2	512	32	32	8	64	˃32	16	32	≤2	0.5	1	≤0.12	64	64	32	32
FEP/ZID4-D7M3m1	512	64	128	4	64	32	4	8	4	1	1	≤0.12	64	128	64	64
FEP/ZID4-D7M3m2	˃512	64	128	4	64	32	8	16	4	1	1	0.25	64	32	64	64
FEP/ZID4-D7M6m1	512	32	128	4	32	32	8	4	4	0.5	1	0.5	128	32	32	32
FEP/ZID4-D7M6m2	512	64	128	4	64	32	16	32	4	1	1	0.5	128	32	32	32

TIC, ticarcillin; TZP, piperacillin/tazobactam; ATM, aztreonam; C/T, ceftolozane/tazobactam; CAZ, ceftazidime; CZA, ceftazidime/avibactam; FEP, cefepime; IMP, imipenem; MEM, meropenem; AMI, amikacin; TOB, tobramycin: CST, colistin; CIP, ciprofloxacin; ZID, zidebactam.

a

ID is the strain name comprising treatment [FEP, cefepime; ZID, zidebactam; FEP/ZID (1:1), cefepime + zidebactam (1:1); FEP/ZID4, cefepime + zidebactam (4 mg/L)]-days (D) of exposure, parental strain (P, PAO1; M, PAOMS) and mutant (m1 or m2).

b

MICs were interpreted using EUCAST 2020 breakpoints (www.eucast.org).

c

ZID concentration was fixed at 4 mg/L.

Table 2.

Characterization of resistance mechanisms for PAO1 and PAOMS mutant derivatives selected after 7 days of exposure to increasing concentrations of cefepime, zidebactam or cefepime/zidebactam combinations

IDa	Mutations in resistance genesb																			Gene expressionc			Large chromosomal deletion
	PA0424 (mexR)	PA0426 (mexB)	PA1180 (phoQ)	PA1777 (oprF)	PA2020 (mexZ)	PA2491 (mexS)	PA2492 (mexT)	PA2494 (mexF)	PA3047 (dacB)	PA3168 (gyrA)	PA3574 (nalD)	PA3721 (nalC)	PA4003 (pbpA)	PA4020 (mpl)	PA4109 (ampR)	PA4266 (fusA1)	PA4418 (ftsI)	PA5045 (ponA)	PA5235 (glpT)	↑ampC	↑mexB	↑mexY	Length (kb)	Loci
FEP-D7P1m1	T130P																			–	+	–	350	PA1905– PA2207
FEP-D7P1m2	T130P																			–	+	–	350	PA1905– PA2207
FEP-D7P2m1			nt870Δ1		E120K												N283H			–	–	+	2	PA2020– PA2023
FEP-D7P2m2			nt870Δ1		E120K									nt1180InsC			N283H			+	–	+	2	PA2020– PA2023
FEP-D7P3m1											T158P					N351S	R504C			–	+	+	217	PA2020– PA2211
FEP-D7P3m2											T158P					N351S	H394R; R504C			–	+	+	217	PA2020– PA2211
FEP-D7M1m1									G368R					nt112InsC						+	–	–
FEP-D7M1m2									G368R					nt112InsC						+	–	–
FEP-D7M2m1															D135N					+	–	–
FEP-D7M2m2															D135N				V327M	+	–	–
FEP-D7M3m1									G244R					M315V			Y268C			+	–	–
FEP-D7M3m2									G244R					M315V			Y268C			+	–	–
ZID-D7P1m1				nt121Δ11																–	–	–
ZID-D7P1m2				nt121Δ11																–	–	–
ZID-D7P2m1		nt196Δ3											V516M							–	–	–
ZID-D7P2m2		nt196Δ3											V516M							–	–	–
ZID-D7P3m1																				–	–	–
ZID-D7P3m2																				–	–	–
ZID-D7M1m1							N326Y													–	–	–
ZID-D7M1m2																				–	–	–
ZID-D7M2m1													V516M							–	–	–
ZID-D7M2m2											L193P		V516M							–	+	–
ZID-D7M3m1											W205R		V516M							–	+	–
ZID-D7M3m2						V125I					L171P		V516M							–	+	–
FEP/ZID (1:1)- D7P1m1	T130P	F136L																		–	+	+
FEP/ZID (1:1)- D7P1m2	T130P	F136L																		–	+	+
FEP/ZID (1:1)- D7P2m1		V45L									T43I									–	+	+
FEP/ZID (1:1)- D7P2m2		V45L									T43I									–	+	+
FEP/ZID (1:1)- D7P3m1		W634G									W49R									–	+	+
FEP/ZID (1:1)- D7P3m2		W634G									W49R									–	+	+
FEP/ZID (1:1)- D7M1m1		T329A									T43I		T328M				R504C			–	+	–
FEP/ZID (1:1)- D7M1m2		T329A									T43I		T328M				R504C			–	+	–
FEP/ZID (1:1)- D7M2m1	L45P	T329A											A596V				R504C			–	+	–
FEP/ZID (1:1)- D7M2m2	L45P	T329A											A596V				R504C			–	+	–
FEP/ZID (1:1)- D7M3m1	R32X	T329A						Y845H									H394R; R504C			–	+	–
FEP/ZID (1:1)- D7M3m2	R32X	T329A						Y845H									H394R; R504C			–	+	–
FEP/ZID4- D7M1m1		R360C										R48Q					R504C	S702P		–	+	–
FEP/ZID4- D7M1m2		R360C										R48Q					R504C			–	+	–
FEP/ZID4- D7M3m1	L57P	V612M															R504C			–	+	–
FEP/ZID4- D7M3m2	L57P	V612M								R226C							R504C			–	+	–
FEP/ZID4- D7M6m1	G101E	M630T															S505P			–	+	–
FEP/ZID4- D7M6m2	G101E	M630T															S505P			–	+	–

a

ID is the strain name comprising treatment [FEP, cefepime; ZID, zidebactam; FEP/ZID (1:1), cefepime + zidebactam (1:1); FEP/ZID4, cefepime + zidebactam (4 mg/L)]-days (D) of exposure, parental strain (P, PAO1; M, PAOMS) and mutant (m1 or m2).

b

Mutations detected in a panel of 64 well-known genes related to resistance development. Ins indicates insertion and Δ indicates deletion.

c

Relative gene expression with respect to PAO1 control strain: ampC (–), <5-fold; ampC (+), >10-fold; mexB (–), <2-fold; mexB (+), >3-fold; mexY (–); <5-fold; mexY (+), >10-fold.²¹

Mutants selected upon cefepime exposure

All cefepime-exposed PAO1 lineages showed large deletions of specific chromosomal regions, always including mexZ and galU, together with mutations leading to β-lactam target [ftsI (PBP3)] modification and/or mutations leading to the overexpression of MexAB-OprM [mexR (nalB) or nalD] or MexXY-OprM (mexZ) efflux pumps. In contrast, all PAOMS lineages selected by cefepime exposure showed ampC overexpression due to mutations in PBP4 (dacB) or ampR regulators, but none of them presented any large deletions. These differential mechanisms led to differences in cefepime/zidebactam MICs, these being significantly lower for PAOMS mutants. Likewise, differential mechanisms also impacted non-β-lactam MICs, such as those of ciprofloxacin and aminoglycosides, mostly linked to efflux pump overexpression.

Mutants selected upon zidebactam exposure

Zidebactam resistance development was associated with specific pbpA (PBP2) mutations. Mutations in the efflux pump MexAB-OprM and its regulators were also documented in some lineages. Cefepime/zidebactam MICs ranged from 0.5 to 1 and from 1 to 4 mg/L for zidebactam-selected PAO1 and PAOMS lineages, respectively.

Mutants selected upon cefepime/zidebactam exposure

Development of resistance to cefepime/zidebactam was associated with mutations in the MexAB-OprM efflux pump (mexB gene) and its regulators (MexR, NalC or NalD), along with specific mutations in PBP2 and PBP3. The terminal cefepime/zidebactam MICs ranged from 8 to 16 and from 32 to 64 mg/L for PAO1 and PAOMS lineages, respectively.

Virulence of mutants exposed to cefepime, zidebactam or cefepime/zidebactam in the C. elegans model

A major loss of virulence-associated lethality was evidenced for cefepime and cefepime/zidebactam-selected mutants, whereas such an effect was lower for zidebactam-selected mutants (Table 3). To substantiate this observation, in vitro and in vivo studies in clinically relevant animal models were performed using one representative mutant from each strain–antibiotic pair.

Table 3.

Lethality in the C. elegans infection model of PAO1 and PAOMS mutants evolved in the presence of increasing concentrations of cefepime, zidebactam and combinations for 7 days

IDa	Surviving nematodes atb
	0 h	24 h	72 h	168 h
E. coli OP50	5.0 ± 0.0	5.0 ± 0.0	8.0 ± 2.0	93.3 ± 11.6
PAO1	5.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0
FEP-D7P1m1	5.0 ± 0.0	4.3 ± 0.6	9.0 ± 3.0	≥100
FEP-D7P2m1	5.0 ± 0.0	4.7 ± 1.2	9.7 ± 5.7	6.3 ± 3.5
FEP-D7P3m1	5.0 ± 0.0	1.3 ± 0.6	1.7 ± 0.6	0.0 ± 0.0
ZID-D7P1m1	5.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0
ZID-D7P2m1	5.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0
ZID-D7P3m1	5.0 ± 0.0	1.3 ± 0.6	0.0 ± 0.0	0.0 ± 0.0
FEP/ZID (1:1)-D7P1m1	5.0 ± 0.0	3.3 ± 1.5	0.3 ± 0.6	0.0 ± 0.0
FEP/ZID (1:1)-D7P2m1	5.0 ± 0.0	2.0 ± 1.0	0.7 ± 0.6	0.0 ± 0.0
FEP/ZID (1:1)-D7P3m1	5.0 ± 0.0	2.3 ± 1.2	0.3 ± 0.6	0.0 ± 0.0
PAOMS	5.0 ± 0.0	0.3 ± 0.6	0.0 ± 0.0	0.0 ± 0.0
FEP-D7M1m1	5.0 ± 0.0	1.0 ± 1.7	0.3 ± 0.6	0.3 ± 0.6
FEP-D7M2m1	5.0 ± 0.0	1.3 ± 0.6	0.0 ± 0.0	0.0 ± 0.0
FEP-D7M3m1	5.0 ± 0.0	2.3 ± 0.6	3.7 ± 1.5	15.0 ± 5.3
ZID-D7M1m1	5.0 ± 0.0	1.7 ± 0.6	1.3 ± 1.5	0.0 ± 0.0
ZID-D7M2m1	5.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0
ZID-D7M3m1	5.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0
FEP/ZID (1:1)-D7M1m1	5.0 ± 0.0	2.7 ± 0.6	0.7 ± 1.2	0.0 ± 0.0
FEP/ZID (1:1)-D7M2m1	5.0 ± 0.0	3.7 ± 1.5	15.0 ± 8.5	44.0 ± 48.7
FEP/ZID (1:1)-D7M3m1	5.0 ± 0.0	1.7 ± 1.5	0.0 ± 0.0	0.0 ± 0.0
FEP/ZID4-D7M1m1	5.0 ± 0.0	5.0 ± 1.0	24.0 ± 16.5	≥100
FEP/ZID4-D7M2m1	5.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0
FEP/ZID4-D7M3m1	5.0 ± 0.0	2.0 ± 1.0	0.7 ± 1.2	0.0 ± 0.0

a

ID is the strain name comprising treatment [FEP, cefepime; ZID, zidebactam; FEP/ZID (1:1), cefepime + zidebactam (1:1); FEP/ZID4, cefepime + zidebactam (4 mg/L)]-days (D) of exposure, parental strain (P, PAO1; M, PAOMS) and mutant (m1 or m2).

b

Surviving nematodes (mean ± SD from three independent experiments) at 0, 24, 72 and 168 h are shown. Values for E. coli OP50, used to feed the nematodes and as a non-pathogenic control, and WT PAO1 and PAOMS strains are included for comparative purposes. Data for each of three experiments per strain [D7 mutants of PAO1 (P1–3) or PAOMS (M1–3)] and compound [FEP, cefepime; ZID, zidebactam; FEP/ZID (1:1), combination at 1:1 ratio; FEP/ZID4, combination with fixed ZID concentration of 4 mg/L] are shown.

In vitro growth of mutants exposed to cefepime, zidebactam and cefepime/zidebactam in minimal medium and human serum

Representative PAO1 mutants selected on cefepime (mutant FEP-D7P2m2) and cefepime/zidebactam (1:1) [mutant FEP/ZID (1:1)-D7P1m1] showed retarded growth patterns in M9 minimal medium compared with the parent PAO1 strain (Figure 2a). For instance, an extended lag phase (∼20 h) was observed with the cefepime/zidebactam (1:1) mutant and the maximum OD₆₀₀ attained was <0.2 after 28 h (compared with the OD₆₀₀ of >0.45 for the parent). The zidebactam-selected representative mutant (mutant ZID-D7P2m1), however, did not show significant growth impairment. All four representative mutants from the PAOMS lineage [FEP-D7M3m1, ZID-D7M2m1, FEP/ZID (1:1)-D7M2m1 and FEP/ZID4-D7M3m1] also displayed significantly reduced growth rates in M9 minimal medium (OD₆₀₀ of <0.3 at 28 h) compared with the parent strain (Figure 2b).

Figure 2.

Growth profiles of WT P. aeruginosa PAO1 (a, c), hypermutator PAOMS (b, d) and their respective mutants grown in either M9 minimal medium (a, b) or CAMHB with 50% serum (c, d). Growth was monitored till 28 h by measuring OD₆₀₀ (a, b) or till 24 h by determining cfu/mL and represented as log₁₀ cfu/mL (c, d). FEP, cefepime; ZID, zidebactam; FEP/ZID, cefepime/zidebactam.

When PAO1 and its mutants were grown in CAMHB supplemented with human serum (to mimic the relevant in vivo conditions), cefepime and cefepime/zidebactam-selected mutants showed substantial serum susceptibility, as evidenced by an ∼2 log₁₀ reduction in the bacterial count during the initial 4 h (Figure 2c). After the initial drop in bacterial density, even though growth resumed, it remained slow compared with the parent strain. PAO1 parent and the zidebactam-selected mutant retained optimal growth features in serum-supplemented medium. In the case of PAOMS, all the mutants showed significant loss in viability during the initial 4–6 h and continued to show retarded growth till 24 h (Figure 2d). In summary, except the PAO1 zidebactam-selected mutant, all other PAO1 and PAOMS mutants had significant in vitro growth defects.

In vivo fitness of cefepime, zidebactam and cefepime/zidebactam mutants in the neutropenic mice lung infection model

Post intranasal infection, PAO1 showed a time-dependent increase in count leading to 100% mortality at 24 h, an indication of optimal expression of virulence (Figure 3a). Consistent with the in vitro growth profile, the zidebactam-selected PAO1 mutant (ZID-D7P2m1) showed parent-comparable growth but did not result in mortality, suggesting virulence loss. On the other hand, cefepime and cefepime/zidebactam (1:1)-selected mutants [FEP-D7P2m2 and FEP/ZID (1:1)-D7P1m1] showed hampered in vivo growth with the former was unable to cause mortality at 24 h, while the latter was able to do so (Figure 3a). Thus, taking into account the growth as well as the mortality, none of the mutants showed parent-comparable in vivo fitness. All PAOMS mutants showed a decreased rate of growth as well as the inability to trigger mortality, pointing towards extensive fitness cost being incurred while acquiring cefepime, zidebactam or cefepime/zidebactam resistance (Figure 3b).

Figure 3.

Growth assessment of WT P. aeruginosa PAO1 (a), hypermutator PAOMS (b) and their respective mutants in the neutropenic animal model. After intranasal administration of fixed counts, bacterial growth in animals was monitored till 24 h by determining cfu/lung and represented as log₁₀ cfu/lung. In the case of parents (PAO1, PAOMS) and the PAO1 cefepime/zidebactam (1:1)-selected mutant, none of the animals survived, resulting in 100% mortality at 24 h and hence cfu counts could not be obtained. FEP, cefepime; ZID, zidebactam.

In vivo efficacy of cefepime/zidebactam against cefepime/zidebactam-selected mutants

In vivo efficacy studies were undertaken with a higher infecting load to compensate for the suboptimal in vivo growth of high cefepime/zidebactam MIC mutants, selected from cefepime/zidebactam exposures. The resultant bacterial density at 2 h post-infection was 5.8 ± 0.35 log₁₀ cfu/lung for PAO1 FEP/ZID (1:1) [D7P1m1], 6.0 ± 0.70 log₁₀ cfu/lung for PAOMS FEP/ZID (1:1) [D7M2m1] and 5.3 ± 0.20 log₁₀ cfu/lung for PAOMS FEP/ZID4 [D7M3m1]. Mortality of 100% was observed in the untreated group of PAO1 mutants while the other two mutants showed net growth (≥2 log₁₀ cfu/lung) at 24 h. Standalone cefepime or zidebactam HSRs resulted in net stasis or net growth, while MIC-dependent killing was observed with the cefepime/zidebactam HSR. For the cefepime/zidebactam PAO1-selected mutant (MIC = 16 mg/L) a 2 log₁₀ kill was observed. Likewise, for the cefepime/zidebactam 4 mg/L PAOMS-selected mutant (MIC = 64 mg/L) and the cefepime/zidebactam (1:1) PAOMS-selected mutant (MIC = 64 mg/L), the cefepime/zidebactam HSR exerted a 1.1 and 0.5 log₁₀ bacterial kill, respectively (Figure 4b). As anticipated, the cefepime/zidebactam HSR was efficacious (>2 log₁₀ kill) against both parent strains (Figure 4a).

Figure 4.

In vivo efficacy study for parent strains (a) and cefepime/zidebactam-selected mutants (b) in the neutropenic mice lung model. The efficacy was assessed as change in the lung bacterial count at 24 h compared with 0 h (initiation of antibiotic treatment). Cefepime/zidebactam (1:1) MICs were 1, 4, 16, 64 and 64 mg/L for PAO1, PAOMS, PAO1 cefepime/zidebactam (1:1)-selected mutant (D7P1m1), PAOMS cefepime/zidebactam (1:1)-selected mutant (D7M2m1) and PAOMS cefepime/zidebactam-4 mg/L-selected mutant (D7M3m1), respectively. FEP, cefepime; ZID, zidebactam.

Discussion

The triple action of zidebactam imparts a unique mechanistic contribution to its partner cefepime to overcome multiplicity of resistance mechanisms in P. aeruginosa. Zidebactam significantly retarded resistance development in both the WT and its hypermutable derivative. Moreover, the development of high-level cefepime/zidebactam resistance was only achieved when using the DNA mismatch repair-deficient strain known to show spontaneous mutation rates up to 1000-fold higher than the WT. This type of strain, frequently encountered in chronic infections rather than in acute episodes,²³ is useful for analysing the emergence of resistance phenotypes during long-term treatment or in infection sites known to show fluctuations in drug exposures, a condition suitable for resistance selection.^24,25

The WGS analysis of the selected mutants provided interesting results. Cefepime resistance development in PAO1 invariably included large genomic deletions in galU, which encodes for a UDP-glucose pyrophosphorylase required for LPS core synthesis. This type of deletion has been previously documented upon ceftazidime and meropenem exposure.¹⁵ One difference, however, is that in two of the three experiments, the deletion included, in addition to galU, the negative regulator of MeXY (mexZ) but not the efflux pump components. Thus, the deletion determined a double resistance mechanism: galU deletion and MexXY overexpression. This differential aspect of cefepime-selected deletions is in agreement with the reported major role of the MexXY efflux pump in cefepime resistance.^26,27 In contrast, large deletions were not detected when the PAOMS strain was exposed to cefepime, which is not surprising since this type of mutation is not amplified in DNA mismatch repair-deficient strains.²⁸ Exposure to cefepime in PAOMS resulted in the selection of AmpC-hyperproducing mutants as previously documented for ceftazidime.¹⁵ It is noteworthy that these cefepime-resistant mutants were susceptible to cefepime/zidebactam.

Regarding zidebactam resistance, mutants from three of six experiments showed a mutation (V516M) within a conserved residue of PBP2. Thus, these results are consistent with a PBP2 inhibition-based mechanism of action of zidebactam.¹² The impact of the V516M mutation on PBP2 functionality requires further investigation. Expectedly, all zidebactam mutants were susceptible to cefepime/zidebactam, indicating that sole PBP2 mutations do not impact activity of the combination due to concurrent multiple PBP inactivation.

Finally, the development of cefepime/zidebactam resistance required the combination of individual mutations leading to cefepime and zidebactam resistance, mostly including PBP2 and PBP3 mutations, along with mutations leading to efflux pump (MexAB-OprM) overexpression. Thus, in addition to the range of mutations encountered in cefepime- or zidebactam-selected mutants, all cefepime/zidebactam-selected mutants showed mexB mutations affecting efflux pump structure. Selection of mexB mutations upon ceftazidime and imipenem/relebactam exposure has also been previously noted^14,29 and therefore one could presume that these types of mutations could facilitate the extrusion of both cefepime and zidebactam. On the other hand, unlike previously shown for ceftolozane/tazobactam and ceftazidime/avibactam, cefepime/zidebactam exposure did not select mutations in AmpC or in its regulators. This indicates that although zidebactam is a potent inhibitor of AmpC, the antibacterial activity of cefepime/zidebactam against P. aeruginosa is primarily driven by multiple PBP binding, thus preventing the combination from selection of AmpC mutations, which otherwise occurs with ease in the case of classical BL/BLIs. Thus, while resistance to other available combinations such as ceftolozane/tazobactam and ceftazidime/avibactam can emerge with a single mutation in AmpC structure, resistance to cefepime/zidebactam requires multiple concurrent mutations in the efflux system and PBPs, the frequency of which is expected to remain much lower.³⁰

In addition to requiring multiple mutations, cefepime/zidebactam resistance was associated with a major fitness cost. These results are consistent with previous findings showing that several of the involved mechanisms, such as large genomic deletions and PBP3 mutations, are associated with a significant fitness cost in P. aeruginosa.^15,30 Earlier, we reported that large deletions and PBP3 mutations in PAO1 upon meropenem exposure were linked to fitness cost.¹⁵ Likewise, mecillinam, which also binds to PBP2, albeit in E. coli, has also been demonstrated to select mutations in PBP2 as well as genes regulating intracellular ppGpp (stringent response molecule) concentrations, resulting in mecillinam resistance and significant fitness cost.³¹ Thus, cost of fitness associated with PBP mutations probably explains infrequent involvement of such mutations in clinical resistance to P. aeruginosa. This is in contrast to the frequently implicated role of β-lactamases, efflux and porin down-regulation in imposing clinical resistance, indicating that these mechanisms are tolerated well without inflicting fitness cost.

Apart from the fitness cost, we also investigated the implication of acquired mutations on the in vivo susceptibilities to cefepime/zidebactam employing the neutropenic mice lung infection model, particularly against high cefepime/zidebactam MIC (16–64 mg/L) mutants. Consistent in vivo efficacy of the cefepime/zidebactam HSR for the cefepime/zidebactam-selected high-MIC mutants was observed. The coverage of mutants with cefepime/zidebactam MICs in the range of 16–64 mg/L observed in this study is in accordance with the pharmacokinetic/pharmacodynamic breakpoint of ≤64 mg/L for cefepime/zidebactam identified through Monte Carlo simulation and PTA analyses.^32–34 This shows that exposure to cefepime/zidebactam selects high-MIC mutants with fitness cost trade-off and such mutants largely remain susceptible to cefepime/zidebactam in vivo. Independent investigators in the past have reported a remarkable bactericidal effect of a cefepime/zidebactam humanized regimen in neutropenic mouse lung or thigh against Acinetobacter baumannii and P. aeruginosa with cefepime/zidebactam MICs up to 64 and 32 mg/L, respectively.^22,35–37 The β-lactam enhancer action of zidebactam, unlike BLIs, causes a substantial lowering of cefepime %fT_>MIC requirement, enabling the coverage of high-MIC strains.³⁴

In summary, our results show that zidebactam significantly reduces resistance development to cefepime in vitro and high-level cefepime/zidebactam resistance selection requires multiple simultaneous mutations, the frequency of which is expected to remain low. Moreover, due to loss of fitness and infectivity, such a mutant population may be less successful in disseminating the acquired resistance. Importantly, the high cefepime/zidebactam-MIC mutants encountered in this study remained susceptible in vivo to the cefepime/zidebactam combination.

Funding

This work was supported by the Wockhardt Research Centre (India) and by the Ministerio de Economía y Competitividad of Spain, Instituto de Salud Carlos III—co-financed by the European Regional Development Fund ‘A way to achieve Europe’ ERDF, through the Spanish Network for the Research in Infectious Diseases (RD16/0016).

Transparency declarations

Snehal Palwe, Prashant Joshi, Swapna Takalkar, Hariharan Periasamy, Sachin Bhagwat and Mahesh Patel are employees of Wockhardt. All other authors: none to declare.

Supplementary data

Tables S1 and S2, Figure S1 and Table S3 are available as Supplementary data at JAC Online.