Emergence of Resistance to Novel Cephalosporin–β-Lactamase Inhibitor Combinations through the Modification of the Pseudomonas aeruginosa MexCD-OprJ Efflux Pump

ABSTRACT

A ceftolozane-tazobactam- and ceftazime-avibactam-resistant Pseudomonas aeruginosa isolate was recovered after treatment (including azithromycin, meropenem, and ceftolozane-tazobactam) from a patient that had developed ventilator-associated pneumonia after COVID-19 infection. Whole-genome sequencing revealed that the strain, belonging to ST274, had acquired a nonsense mutation leading to truncated carbapenem porin OprD (W277X), a 7-bp deletion (nt213Δ7) in NfxB (negative regulator of the efflux pump MexCD-OprJ), and two missense mutations (Q178R and S133G) located within the first large periplasmic loop of MexD. Through the construction of mexD mutants and complementation assays with wild-type nfxB, it was evidenced that resistance to the novel cephalosporin–β-lactamase inhibitor combinations was caused by the modification of MexD substrate specificity.

INTRODUCTION

Pseudomonas aeruginosa is a major nosocomial pathogen characterized by an extraordinary capacity for developing resistance by the selection of mutations in chromosomal genes, frequently leading to the emergence of multidrug-resistant/extensively drug-resistant (MDR/XDR) phenotypes (1). Moreover, the growing prevalence of nosocomial infections produced by MDR/XDR P. aeruginosa strains is associated with significantly increased morbidity and mortality, since it compromises the selection of effective therapies (2). Over the past few years, some novel antipseudomonal agents, such as ceftolozane-tazobactam, have been introduced into clinical practice and helped to mitigate classical β-lactam resistance mechanisms in P. aeruginosa, including the overexpression of the chromosomal β-lactamase AmpC or efflux pumps, such as MexAB-OprM (3, 4). This enhanced activity, compared with that of other antipseudomonal β-lactams, relays into ceftolozane being a poorer substrate for AmpC hydrolysis and efflux pump extrusion. However, in vitro and in vivo emergence of resistance has been demonstrated to be frequently caused by the selection of mutations leading to increased catalytic efficiency of AmpC against ceftolozane (5–9). Here, we show that ceftolozane-tazobactam resistance may also emerge in vivo through the selection of mutations leading to the modification of the substrate specificity of one of the P. aeruginosa efflux pumps, MexCD-OprJ.

A 66-year-old female patient was admitted to the intensive care unit (ICU) of a Spanish hospital (H. Clinic, Barcelona) in March 2020 due to COVID-19 pneumonia and was initially treated with ceftriaxone, azithromycin, hydroxychloroquine, and lopinavir-ritonavir. Eleven days after admission, she developed ventilator-associated pneumonia (VAP) caused by P. aeruginosa, resistant only to ceftazidime and piperacillin-tazobactam, that was treated first with meropenem (2 g/8 h) during 8 days and ceftolozane-tazobactam (2 g/8 h + 1 g/8 h) for 2 additional weeks. A bronchial aspirate cultured at this time point was again positive for P. aeruginosa despite intermediate cultures the week before being already negative. Treatment was finally shifted to piperacillin-tazobactam plus amikacin, but the patient died of respiratory complications 2 weeks after. The P. aeruginosa strain isolated after ceftolozane-tazobactam treatment (HC-20-232) was documented to be resistant to meropenem and ceftolozane-tazobactam, along with other β-lactams and fluoroquinolones (Table 1). However, phenotypic and PCR assays (10) for acquired carbapenemases and extended-spectrum β-lactamases yielded negative results. Moreover, the quantification of ampC expression through previously described reverse transcription-PCR (RT-PCR) assays (22) ruled out AmpC hyperproduction as resistance mechanisms. Therefore, the strain was subjected to whole-genome sequencing as previously described (11). The presence of horizontally acquired resistance determinants and the sequence type (ST) were determined using online databases (https://cge.cbs.dtu.dk//services/). HC-20-232 was documented to belong to the ST274 clone, but no horizontally acquired resistance determinants were evidenced. Thus, the presence of mutations in a panel of 164 chromosomal genes related to antibiotic resistance (mutational resistome) was investigated, using as reference genomes those of wild-type PAO1 and a collection of ST274 isolates originating from difference sources (12). The analysis of the mutational resistome revealed the presence of four nonsynonymous mutations: a nonsense mutation leading to a truncated OprD (W277X), a 7-bp deletion leading to the inactivation of NfxB (nt213Δ7), and two missense mutations in the amino acid codons Q178R and S133G, located within the first large periplasmic loop (LPL) of MexD (13). While the oprD mutation did justify the carbapenem resistance exhibited by HC-20-232, ceftolozane-tazobactam, ceftazidime-avibactam, and even ceftazidime resistance was more puzzling. The inactivation of the negative regulator NfxB should lead to the overexpression of the efflux pump MexCD-OprJ, and that was indeed confirmed by real-time RT-PCR using previously described protocols (22) (Table 1). However, while MexCD-OprJ should explain the documented ciprofloxacin resistance, this efflux pump is known to extrude cefepime but not ceftazidime or ceftolozane. This phenotype was indeed confirmed using an nfxB mutant of PAO1, leading to increased MICs of ciprofloxacin and cefepime but without modifications of those of ceftazidime or ceftolozane (Table 1). Moreover, nfxB inactivation (using the cre-lox system as previously described [14]), and consequent MexCD-OprJ overexpression, did not increase the MICs of ceftolozane-tazobactam of a previously characterized (11) ST274 clinical isolate (PAMB148) that was already resistant to ceftazidime due to AmpC overexpression (Table 1). Since these assays confirmed that MexCD-OprJ overexpression was not responsible by itself for ceftolozane-tazobactam (and ceftazidime and ceftazidime-avibactam) resistance, we speculated that the underlying mechanism is the modification of the efflux pump substrate specificity through the modification of the MexD structure. To evaluate this hypothesis, we followed two parallel approaches: (i) we complemented nfxB deficiency in HC-20-232 with a wild-type copy of nfxB expressed from pUCP24 plasmid, as previously described (14), and (ii) we knocked out mexD in HC-20-232 using the cre-lox system as described previously (14). As shown in Table 1, both repression and inactivation of MexCD-OprJ restored wild-type ceftolozane-tazobactam (and ceftazidime and ceftazidime-avibactam) susceptibility, demonstrating the involvement of the expression of the modified MexCD-OprJ efflux pump in resistance.

TABLE 1.

Susceptibilty profiles and characterization of the studied strains

Strain	Description/relevant genotype	mexDa	MIC (μg/ml)b
			TIC (R,  >16)	ATM (R,  >16)	PTZ (R,  >16)	TAZ (R,  >8)	FEP (R,  >8)	IMP (R,  >4)	MER (R,  >8)	CTZ (R,  >4)	CZA (R,  >8)	IPR (R,  >2)	CIP (R,  >0.5)	AMI (R,  >16)	TOB (R,  >2)	COL (R,  >2)
PAO1	Wild-type reference strain	1	16	4	≤4	2	≤1	2	1	0.5	1	0.25	≤0.12	4	0.5	2
PAONB	nfxB knockout mutant of PAO1	744 ± 80	≤8	≤2	≤4	≤1	4	≤0.5	≤0.5	0.25	≤0.5	≤0.12	2	≤2	≤0.25	2
HC-20-232	Clinical isolate belonging to ST274 OprD (W277X), nfxB (nt213Δ7), mexD (Q178R, S133G)	241 ± 12	32	8	8	64	64	32	16	16	32	1	1	8	2	2
HC-20-232MxD	mexD knock out mutant of HC-20-232	ND	32	4	8	4	2	32	8	1	2	1	≤0.12	4	1	2
HC-20-232 (pUCP nfxB)	HC-20-232 strain harboring plasmid pUCP24 with cloned wild-type nfxB	4.9 ± 0.2	32	8	8	4	4	32	8	1	2	1	≤0.12	8	2	2
PAMB148	Clinical isolate belonging to ST274 ampD P41L (AmpC overexpression)	1.3 ± 0.1	256	64	>256	64	32	4	1	4	8	0.25	≤0.12	8	2	2
PAMB148NB	nfxB knockout mutant of PAMB148	76 ± 17	128	32	256	64	32	2	1	2	4	≤0.12	2	4	1	2

^aRelative level of mexD mRNA with respect to that of wild-type PAO1 according to previously described protocols (22). ND, not done.

^bBroth microdilution MIC results obtained in triplicate experiments. EUCAST v 11.0 resistance breakpoints are indicated. TIC, ticarcillin; ATM, aztreonam; PTZ, piperacillin-tazobactam; TAZ, ceftazidime; FEP, cefepime; IMP, imipenem; MER, meropenem; CTZ, ceftolozane-tazobactam; CZA, ceftazidime-avibactam; IPR, imipenem/relebactam; CIP, ciprofloxacin; AMI, amikacin; TOB, tobramycin; COL, colistin.

This work is the first to demonstrate the potential involvement of efflux pumps in ceftolozane-tazobactam resistance. Moreover, together with recent evidence on MexAB-OprM (15), results from our work highlight the potential relevance of missense mutations within efflux pump structural components in substrate specificity and antimicrobial resistance evolution. Future experiments should be directed toward a detailed analysis of the impact of the specific mutations in the structure and function of the MexCD-OprJ efflux pump. However, our results are consistent with a previous in vitro work that showed that other mutations located within the first LPL of MexD were able to modify its β-lactam substrate specificity (16). Likewise, from the epidemiological and biological perspectives, it is also of interest to understand which are the driving forces leading to the selection of such a complex resistance mechanism. While the reconstruction of the sequence of genetic events that had occurred during treatment was not possible due to the lack of availability of the preceding susceptible isolates, we can speculate that azithromycin treatment played a role. Indeed, despite not showing significant antipseudomonal activity, azithromycin is known to be active against P. aeruginosa biofilms (17, 18), such as those occurring in cystic fibrosis and intubated patients (19), and it is associated with the frequent selection of mutations leading to the overexpression of MexCD-OprJ, since it is a major substrate of this efflux pump (14, 20). Thus, azithromycin could have selected the nfxB mutation, leading to the overexpression of the efflux pump in the first place, and the subsequent ceftolozane-tazobactam treatment could have selected the mutations leading to the structural modification of MexD. Finally, from the clinical perspective, we aimed to evaluate whether the documented resistance mechanism could be overcome with the recently introduced imipenem-relebactam combination. Since HC-20-232 was documented to be fully susceptible to imipenem-relebactam (Table 1), it was concluded that relebactam, through AmpC inhibition, efficiently overcame OprD-mediated resistance, as previously noted (21), and that the documented MexD mutations failed to confer significant imipenem resistance.

Data availability.

The complete sequence of HC-20-232 has been deposited in the European Nucleotide Archive under accession number ERS6196999.