Evolution of ceftazidime-avibactam and cefiderocol resistance in ST131-H30R1-Escherichia coli isolates with KPC-3 mutants and application of FTIR biotyping

ABSTRACT Ceftazidime-avibactam and cefiderocol represent two of the few alternatives for infections by KPC-producing Enterobacterales. We reported the emergence of resistance to both ceftazidime-avibactam and cefiderocol in a KPC-producing ST131-Escherichia coli (KPC-ST131-Ec) clinical isolate. Antimicrobial susceptibility testing, Fourier-transform infrared (FTIR) spectroscopy, whole-genome sequencing, and cloning experiments were performed. A KPC-49-Ec isolate resistant to ceftazidime-avibactam (MICCZA > 16/4 mg/L) and susceptible to cefiderocol (MICFDC: 2 mg/L) was recovered in a blood sample from an oncologic patient hospitalized in the medical ICU (June 2019) during ceftazidime-avibactam treatment. After 44 days, a KPC-31-Ec resistant to both ceftazidime-avibactam and cefiderocol (MICCZA > 16/4 mg/L, MICFDC: 8 mg/L) was found in a rectal sample during a second cycle of ceftazidime-avibactam treatment. Both KPC-49 (R163S) and KPC-31 (D179Y) were detected in the epidemic ST131-H30R1-Ec high-risk clone and showed a phenotype resembling that of ESBL producers. FTIR spectroscopy managed to differentiate cefiderocol-susceptible and resistant ST131-Ec isolates, and these from others belonging to different clones. After cloning and transformation experiments, KPC-49 and KPC-31 were responsible for ceftazidime-avibactam resistance (MICCZA > 16/4 mg/L) and decreased carbapenem MICs (MICMER ≤ 0.12 mg/L, MICIMI ≤ 1 mg/L). KPC-31 was also shown to be associated with increased MICs of cefiderocol (twofold and threefold dilutions over KPC-3 and KPC-49, respectively). However, mutations in proteins participating in outer membrane stability and integrity, such as TolR, could have a more relevant role in cefiderocol resistance. The effects of ceftazidime-avibactam and cefiderocol co-resistance in clinical isolates of Enterobacterales producing KPC mutants make their identification challenging for clinical laboratories. IMPORTANCE Throughout four admissions in our hospital of a single patient, different KPC-3 variants (KPC-3, KPC-49, and KPC-31) were found in surveillance and clinical ST131-Escherichia coli isolates, after prolonged therapies with meropenem and ceftazidime-avibactam. Different patterns of resistance to cefiderocol and ceftazidime-avibactam emerged, accompanied by restored carbapenem susceptibility. The inability to detect these variants with some phenotypic methods, especially KPC-31 by immunochromatography, and the expression of a phenotype similar to that of ESBL producers, posed challenge to identify these variants in the clinical microbiology laboratory. Molecular methods and whole-genome sequencing are necessary and new techniques able to cluster or differentiate related isolates could also be helpful; this is the case of Fourier-transform infrared spectroscopy, which managed in our study to discriminate isolates by cefiderocol susceptibility within ST131, and those from the non-ST131 ones.


Case report
A 48-year-old female diagnosed with biliary cholangiocarcinoma was admitted to the Oncology ward of our hospital four times during 2019, always due to fever or bad general condition, and diagnosed later with acute cholangitis.Five different episodes of bacteremia were documented and Ec was isolated in all of the blood samples.During the first two stays, meropenem (1 g/8 h iv) was administrated for 14 days and 4 days, respectively, to treat two episodes of Ec bacteremia (Ec-S1 and Ec-S2, MIC MER ≤ 0.12 mg/L, MIC IMI ≤ 1 mg/L; MIC CZA ≤ 0.25/4 mg/L and MIC FDC ≤ 0.03 mg/L) (Fig. 1; Table 1).In the third admission, a KPC-producing Ec (KPC-Ec) was detected in a blood culture (Ec-R1; MIC MER = 8 mg/L, MIC IMI = 4 mg/L; MIC CZA = 0.5/4 mg/L and MIC FDC = 0.25 mg/L) and ceftazidime-avibactam therapy was implemented (2 g/0.5 g/8 h iv).After 10 days, a second KPC-Ec isolate (Ec-R2) with an ESBL phenotype but resistant to ceftazidime-avi bactam was isolated in another blood sample (MIC MER = 0.5 mg/L, MIC IMI = 2 mg/L; MIC CZA = 16/4 mg/L and MIC FDC = 2 mg/L).Due to that, amikacin was added for 7 days (1 g/24 h iv) (Fig. 1; Table 1).KPC-3 and KPC-49 (R163S mutant of KPC-3) were confirmed by PCR and Sanger sequencing in Ec-R1 and Ec-R2, respectively.Both Ec-R1 (KPC-3) and Ec-R2 (KPC-49) were identified as KPC producers by the immunochromatography test and the rapid molecular methods (Eazyplex-Superbug-CRE system and Xpert Carba-R assay) (Table 2).Surveillance cultures were all negative for E. coli until this moment.
All but one of the cefiderocol susceptibility testing results had similar interpretations between disk diffusion and broth microdilution techniques (EUMDROXF and ComASP) (Table 1).The antibiotics and dosages received during the admissions are summarized in Table S1.

ST131 screening and Biotyper analysis
PCR screening for the ST131 high-risk clone was positive for all Ec isolates recovered from clinical and surveillance samples during the third and fourth admissions, all being KPC producers.The susceptible Ec isolates (Ec-S1 and Ec-S2) were typed as non-ST131 (Table 2).
Using the cut-off proposed by the FTIR or Biotyper software (0.165), the cefiderocolresistant isolates (KPC-31-Ec-R3 and KPC-31-Ec-R4) were classified very close, whereas susceptible isolates (KPC-3-Ec-R1 and KPC-49-Ec-R2) were separated from these two isolates in the same cluster, belonging all of them to ST131 clone.Apart from that,  .
non-ST-131 susceptible E. coli (Ec-S1 and Ec-S2) were also differentiated from the previous ones.The distribution is represented in a dendrogram (Fig. 2A) and a scatter plot (Fig. 2B).

DISCUSSION
To the best of our knowledge, this is the first description of a KPC-31-producing ST131-H30R1-Ec clinical isolate resistant to both ceftazidime-avibactam and cefiderocol.Ceftazidime-avibactam-resistant KPC enzymes usually derive from point muta tions in bla KPC-2 and bla KPC-3 genes, frequently after the antibiotic exposure (18).Recent publications have demonstrated that the ceftazidime-avibactam and cefiderocol resistance phenotypes and the corresponding resistance mechanisms are commonly correlated, mainly in K. pneumoniae (8,19).However, variable percentages of resistance to both ceftazidime-avibactam (0.5-18%) and cefiderocol (0.9-8%) have been reported in the ST131-Ec population (14,20).
In our study, we reported the emergence of resistance to ceftazidime-avibactam in the ST131-H30R1-Ec, first producing the KPC-49 (R163S) and then the KPC-31 (D179Y), with a difference in time of 44 days and each of them during a different treatment cycle with ceftazidime-avibactam.Although further studies would be needed to demonstrate this fact, our results suggest that the production of KPC-49 and KPC-31 enzymes in the ST131-Ec high-risk clone has possibly occurred through two different routes, since we observed that the genetic distance between KPC-3-Ec and KPC-31-Ec isolates was smaller than the one between KPC-49-Ec and both KPC-31-Ec.
On the other hand, co-resistance to cefiderocol was only detected in KPC-31-ST131-Ec isolates, even though the MIC value for cefiderocol in the KPC-49-ST131-Ec (MIC FDC = 2 mg/L) was threefold higher than that in the KPC-3-ST131-Ec isolate (MIC FDC = 0.25 mg/L) and at least sixfold higher than that in the susceptible Ec isolates (MIC FDC ≤ 0.03 mg/L), recovered before the initiation of the ceftazidime-avibactam therapy.Hobson et al. showed that KPC-31-producing isolates show cross-resistance to both ceftazidimeavibactam (MIC CZA > 32 mg/L) and cefiderocol (MIC FDC = 4 mg/L) (19).Our results demonstrated that both KPC-31 and KPC-49 were responsible for the ceftazidime-avibac tam resistance (MIC CZA > 16 mg/L) in our ST131-H30R1-Ec.Apart from that, the MIC value for cefiderocol was threefold dilutions higher in the KPC-31-Ec transformant (MIC FDC = 1 mg/L) than in the corresponding KPC-3 (MIC FD = 0.25 mg/L) and twofold dilutions higher than in the KPC-49-transformant (MIC FDC = 0.125 mg/L), but still not considered resistant any of them.
In addition to the changes in β-lactamases, a mutation leading an amino acid change in tolR (A112V) was identified in both KPC-31-ST131-Ec.TolR (homolog of ExbD in the TonB-ExbB-ExbD System), is part of the Tol-Pal system and is required for the main tenance of the outer membrane stability and integrity (21).TonB-ExbB-ExbD system is involved in the active transport of iron siderophores and facilitates the entry of cefiderocol into the bacterial periplasmic space (22,23).Modifications in TonB-depend ent receptors have been demonstrated to be related to cefiderocol resistance in Pseudomonas aeruginosa and several Enterobacterales species (24,25).According to our results, mutations in TolA-Q-R proteins may also inhibit the siderophore-drug uptake and contribute to cefiderocol resistance.Mutations associated with deficiency in iron transporters (cirA and fiu) and the PBP-3 have also been showed to be implicated in increased cefiderocol MIC values in Ec isolates, but identical genes were found in our KPC-3, KPC-49, and KPC-31-Ec isolates (5,6,9).
Co-colonization with a KPC-31-K.pneumoniae isolates resistant to ceftazidime-avibac tam and cefiderocol was also detected during the second cycle of treatment with ceftazidime-avibactam.This KPC-31-Kp isolate was typed in a previous study as the ST307 high-risk clone, widely disseminated in our hospital since 2018 and linked to the production of other KPC variants conferring resistance to ceftazidime-avibactam (7,26).Note that co-resistance to ceftazidime-avibactam and cefiderocol has previously been reported in our hospital in KPC-62-producing ST307-K.pneumoniae isolates with permeability defects during an outbreak in the medical ICU in 2020 (7).The KPC-31-ST307-Kp isolate carried a modification of the PBP2 protein (A523T), possibly also involved in increasing the cefiderocol MIC value, which was not found in any of the E. coli isolates (7).
An IncF [F1:A2:B20] plasmid closely related to that found in the ST307-K.pneumoniae isolates carrying novel KPC variants previously detected in our hospital were identified in our KPC-3, KPC-49, and KPC-31-Ec isolates (7,26).The presence in circulating K. pneumoniae hospital clones of KPC variants conferring resistance to ceftazidime-avibac tam and possibly with an impact on the cefiderocol susceptibility, could facilitate the persistence of these resistance mechanisms and their dissemination to other bacterial species such as Ec.
Note also that the infection and colonization by ceftazidime-avibactam and cefiderocol-resistant KPC-31-Ec and KPC-31-K.pneumoniae isolates occurred in this patient before the approval and the use of cefiderocol in clinical practice in our hospital.Besides this, despite of resembling an ESBL phenotype, the immunochromatography test was able to detect the KPC-49 enzymes, but not KPC-31.The detection of cefiderocol resistance in microbiology laboratories represents a considerable challenge due to the technical issues of the phenotypic methods used for the antimicrobial susceptibility testing (27).These limitations, in addition to the successful adaptability of these new KPC variants in high-risk clones with a high capacity for dissemination in the hospital environment, such as the ST131-Ec, makes the impact of these findings even more important.
Interestingly, FTIR biotyping managed to differentiate not only the two Ec belonging to clones other than ST131 from this one, but also those resistant and susceptible to cefiderocol among ST131 isolates.Changes in the bacterial membrane due to mutations affecting the Tol-Pal system that are believed to contribute to cefiderocol resistance could be the explanation for the ability of this technique to discriminate among them.It should be noted that Ec-R3 was obtained from a surveillance rectal swab, whereas the rest of the Ec isolates were from blood samples, which might have had an impact on FTIR due to other biochemical features different from KPC changes.FTIR was originally designed to be used as an early warning tool for outbreaks by generating a fingerprint of the bacterial composition and clustering isolates that belong to the same clone; however, more evaluations are needed to fully understand the application in these situations involving similar carbapenemases with different susceptibility results.
In conclusion, our results indicate that mutated KPC β-lactamases are not the only factor responsible for the increased cefiderocol MICs and that mutations affecting genes involved in the energy transduction for the outer membrane integrity, such as the Tol-Pal system, could also contribute to cefiderocol resistance in Ec.New biotyping technologies, such as FTIR, might be promising in discriminating mechanisms of resistance involving biochemical changes in the membrane.The production of mutated KPCs affecting cefiderocol susceptibility during treatment with ceftazidime-avibactam in successful high-risk clones such as the ST131-Ec, is a public health problem that needs to be monitored and poses a challenge for clinical microbiology laboratories.

Bacterial isolates and patient data
We conducted a retrospective screening of cefiderocol-resistant isolates among the ceftazidime-avibactam-resistant KPC-producing Enterobacterales isolates recovered in our hospital between 2018 and 2020.A KPC-Ec resistant to both ceftazidime-avibactam and cefiderocol was found in a surveillance rectal sample from an oncologic patient admitted to our hospital in June 2019.One KPC-Kp isolate with the same phenotype was also recovered in a rectal sample in the same period of time in this patient.Clinical and epidemiological data were retrospectively reviewed.All Ec and Kp isolates that could be recovered in surveillance and clinical samples from this patient were included in the study.
The patient had previously been admitted to our hospital (April 2019) with a bacteremia by a KPC-3-producing Ec susceptible to ceftazidime-avibactam (MIC CZA = 0.5/4 mg/L), and a KPC-49-producing Ec isolate with a ceftazidime-avibactam resistant phenotype (MIC CZA = 16/4 mg/L), following 10 days of standard-dose ceftazidime-avi bactam treatment.The susceptibility testing and molecular characterization of these two KPC-3-and KPC-49-Ec isolates were reported in a previous study (17).

Molecular typing
ST131 screening and subtyping to detect the H30R1 (clade 1) and H30Rx (clade 2) subclones were performed by SNP-based PCR as previously descri bed (28).Bruker's IR-Biotyper (Bruker-Daltonics, Germany) was also used as biotyping method (https://www.bruker.com/en/products-and-solutions/microbiologyand-diagnostics/microbial-strain-typing.html).Ec samples were prepared according to the manufacturer's instructions, performing five replicates per sample with default settings.The cutoff value proposed by the software was used to determine the cluster aggrupation (16,29).
bla KPC cloning bla KPC-3 , bla KPC-49 , and bla KPC-31 genes were amplified by PCR using primers previously described by Shields et al. (18).Amplicons were cloned into the pCR-Blunt TM II-TOPO TM vector following the manufacturer's instructions (Zero Blunt TOPO PCR cloning kit; Invitrogen, Cergy-Pontoise, France).The recombinant plasmids were transformed by heat shock into competent Ec cells (One Shot TOP10 Chemically Competent E. coli cells, ThermoFisher) and then selected on Luria broth agar medium supplemented with ampicillin (30 mg/L), kanamycin (50 mg/L), and IPTG-Xgal (80 mg/L), as previously described (17).Successful transfer of bla KPC genes was confirmed by PCR and Sanger sequencing.MIC values in KPC-3, KPC-49, and KPC-31 transformants were also measured by standard broth microdilution.

FIG 1 TABLE 1 a
FIG 1 Timeline of isolates and antibiotic treatments* received by the patient during the study period, showing the observed phenotype and the KPC variant.Horizontal bars represent the duration of treatments and all isolates recovered for the analysis are marked with capital letters, indicating the origin of the sample with the shape of the mark (square or circle).Isolates C, D, E, F and H were analysed by WGS.*Only antibiotics with activity against Escherichia coli are represented in the figure.mer: meropenem, imi: imipenem, cza: ceftazidime-avibactam, fdc: cefiderocol.

TABLE 2
Sample data and results of phenotypic methods for KPC detection a