Increased gene expression and copy number of mutated blaKPC lead to high-level ceftazidime/avibactam resistance in Klebsiella pneumoniae

Background Resistance to ceftazidime-avibactam was reported, and it is important to investigate the mechanisms of ceftazidime/avibactam resistance in K. pneumoniae with mutations in blaKPC. Results We report the mutated blaKPC is not the only mechanism related to CZA resistance, and investigate the role of outer porin defects, efflux pump, and relative gene expression and copy number of blaKPC and ompk35/36. Four ceftazidime/avibactam-sensitive isolates detected wild type blaKPC-2, while 4 ceftazidime/avibactam-resistant isolates detected mutated blaKPC (blaKPC-51, blaKPC-52, and blaKPC-33). Compared with other ceftazidime/avibactam-resistant isolates with the minimal inhibitory concentration of ceftazidime/avibactam ranging 128–256 mg/L, the relative gene expression and copy number of blaKPC was increased in the isolate which carried blaKPC-51 and also showed the highest minimal inhibitory concentration of ceftazidime/avibactam at 2048 mg/L. The truncated Ompk35 contributes rare to ceftazidime/avibactam resistance in our isolates. No significant difference in minimal inhibitory concentration of ceftazidime/avibactam was observed after the addition of PABN. Conclusions Increased gene expression and copy number of mutated blaKPC can cause high-level ceftazidime/avibactam resistance.


Background
Carbapenem-resistant Enterobacerales (CRE), especially carbapenem-resistant Klebsiella pneumonia (CRKP) have emerged as a major public health concern worldwide. In China, the production of K. pneumoniae carbapenemases (KPCs) is the predominant mechanism of carbapenem resistance and is frequently linked to a highly successful K. pneumoniae sequence type 11(ST11) clone [1]. The existing antibiotics treating infections caused by KPCproducing K. pneumoniae (KPC-Kp) have limited efficacy, and novel antibiotics are urgently needed.
CZA has been approved by the China State Drug Administration on May 21, 2019, for the treatment of complex intra-abdominal infections (cIAI), hospitalacquired pneumonia (HAP), and for the treatment of gram-negative bacterial infections in adults with limited therapeutic options: K. pneumoniae, E. cloacae, E. coli, Proteus mirabilis, and Pseudomonas aeruginosa. Before the wide use of CZA, CZA-resistant isolates carrying wild type KPC-2 have been reported in China [19,23]. Besides, we reported our experience in treating ten lung transplant recipients (9 with KPC-Kp infections, and 1 with Pseudomonas aeruginosa infection) with CZA at the China-Japan friendship hospital [24]. CZA resistant KPs with mutated bla KPC were recovered from 4 patients after CZA treatment [25]. Plasmid transfer and bla KPC cloning showed the mutated bla KPC in these isolates were associated with CZA resistance. We observed that the minimal inhibitory concentration (MIC) for CZA of KPC-51-producing K. pneumoniae clinical isolate was the highest (2048 mg/L) in our study, while the CZA MIC for the KPC-51-harbouring E. coli DH5a transformant was only 8 mg/L. We believed that the mutated bla KPC was only partly contributing to the resistance of CZA in this isolate, and other resistance mechanisms should be further investigated. Besides, though increased gene expression and copy number of bla KPC and/or porins defects were reported associated with CZA resistance, this finding was often reported in isolates with wild type bla KPC , and rarely reported in isolates with mutated bla KPC . The role of bla KPC expression and porins in CZA resistant isolates with mutated bla KPC is not clear, especially in isolates with different resistant levels.

Outer membrane porin gene sequence analysis
Compared with the sequences of wild type ompk35 (NCBI reference sequence WP_135730820.1) and ompk36 (GenBank accession number AEW62399.1) genes from the reference strain K. pneumoniae ATCC13883, the ompk35 sequence of all 8 isolates had a deletion after 85 bp, which caused a premature stop codon after amino acid position 62. And the Ompk36 in all 8 isolates had a glycine and aspartic acid duplication at amino acid 136 (136-137 GD insertion) ( Table 1). Different sequences of Ompk35 and Ompk36 were not found between the baseline isolates (A) and CZA resistant isolates (B) recovered from the same patient.

Functional restoration of OmpK35 and Ompk36
The blunt vectors harboring functional wild-type Ompk35 or Ompk36 were transferred into selected isolates with different KPC variants by electroporation. The profiles of outer membrane proteins in all clinical isolates and transformants were analyzed by SDS-PAGE (Fig. 1). The truncated Ompk35 in clinical isolates were not shown, while the restoration of lost Ompk35 in corresponding transformants was confirmed in the profiles. The mutated Ompk36 did not influence the profiles of outer membrane proteins in clinical isolates. As a control, the empty blunt vector was also transferred into clinical isolates. The existence of wild type Ompk36 and empty blunt vector in corresponding transformants can not be reflected by SDS-PAGE, but it was confirmed using PCR and sequencing. Compared with clinical isolates with different KPC variants (8A, 1B, 3B, 7B, 8B), no significant reduction in CZA MICs was observed for selected isolates with the restoration of functional wild type Ompk35 or Ompk36 (Table 2). CZA MIC differences between clinical isolates, transformants with wild type Ompk35, transformants with wild type Ompk36, and transformants with original blunt vector were no more than 2-fold.
Gene expression and copy number of bla KPC , and ompk35/36 There was no significant difference in relative gene expression and copy number of bla KPC , or ompk35/36 between the baseline isolates (A) and CZA-resistant isolates (B) (Fig. 2a, c, Fig. 3a, c, Fig. 4a, c). When we compared the baseline isolate and the CZA-resistant isolate recovered from the same patient, the relative gene expression of bla KPC in 1B was higher than 1A (Fig. 2b), while no significant difference was observed in isolates of other patients. The relative copy number of bla KPC in 1B was higher than 1A, the relative copy number of bla KPC in 3B and 7B were lower than 3A and 7A, and no significant difference in relative copy number was observed between 8A and 8B (Fig. 2d).
The role of the AcrAB efflux pump After the addition of PABN, CZA MICs of all selected isolates were not decreased by more than 2fold (Table 1). This indicated that efflux is not a major mechanism for resistance to CZA.
Core genome phylogenetic analysis of isolated K. pneumoniae Core genome phylogenetic analysis was performed to compare the 8 isolates from this study based on the results of WGS analysis and 157 of K. pneumonia strains filtered by MLST from NCBI (until July 2019). As shown in Fig. 5, all of the 8 isolates were located in a large branch (light purple, A-class) from the phylogenetic tree as a whole, which was almost from China except for 2 Canadian sources. Ten isolated strains can be divided into three groups. Groups 1 (isolate 3A, 3B) was the closest phylogenetic tree branch (L122, L222, L211, and L124) which were all isolated from Hangzhou China.

Discussion
In our report, we demonstrate a new mechanism of high-level CZA resistance in a KPC-producing K. pneumoniae strain in a lung transplant recipient, which is that high-level resistance to CZA is due to increased gene expression and copy number of the mutated bla KPC . Though increased bla KPC expression and copy number and/or Ompk defects were reported associated with increased CZA MICs [6,21,26], this result was often reported in isolates with wild type KPC, and the role of bla KPC expression and porins in isolates with mutated bla KPC is not clear. Our results have supplemented this evidence. In isolates with mutated bla KPC , the mutated bla KPC may play a major role in CZA resistance, and the increased gene expression and copy number of the mutated bla KPC could cause high-level CZA resistance, while the truncated Ompk35 may rarely contribute to the increased MIC of CZA.
Resistance to CZA has been reported in KPCproducing K. pneumoniae following treatment of CZA. Shields et al. first reported the evolution of CZA resistance during the treatment of K. pneumoniae infections in three patients from the United States in 2016 [3]. The treatment-emergent CZA resistance was subsequently described in Greece [27], Italy [9,15,16], Finland [10], Germany [12], and Spain [28] since then. And we recently reported four CZA-resistant KPC-KP recovered from lung transplant recipients after 13-22 days of CZA treatment in China [25].
The role of mutated porins is not clear for CZA resistance. Modifications of Ompk35 and Ompk36, accompanied with various beta-lactamases, lead to carbapenem resistance, while do not influence CZA [29,30]. However, several reports showed Ompk35/36 defects were   [33]. Similarly, we observed that a truncated Ompk35 and a GD insertion at amino acid position 136-137 in Ompk36 in all our isolates, including 4 CZA-resistant isolates and 4 CZA-susceptible isolates.
We hypothesize that the different KPC variants combined with Ompk35/36 defects could lead to the different levels of CZA MIC, and the KPC-51 coupled with porins defects may lead to the highest MIC of CZA. However, after the restoration of functional Ompk35 or Ompk36, no significantly decreased CZA MICs were observed in CZA-resistant isolates. We believed that the mutated bla KPC was the most important mechanism in our isolates, and the mutations of Ompk35 or Ompk36 contribute no or rare to CZA resistance. The role of mutated porins for CZA should be further investigated.
The expression and copy number of bla KPC were often associated with reduced susceptibility to CZA. Overexpression of the bla KPC gene is a potential mechanism of CZA resistance in wild type bla KPC isolates. Nelson et al. reported that porins alteration combined with increased bla KPC-3 gene copy number and gene expression can cause CZA resistance [22]. And the relative bla KPC-2 copy numbers and relative expression of bla KPC-2 in the reduced susceptibility group were significantly higher than those in the susceptibility group [19,23,33]. For mutated bla KPC , the report of KPC variants combined with gene expression and copy number is rare. As we writing this article, one study [34] showed the increased bla KPC-53 gene dosage (two copies) coupled with porins alterations may lead to high-level CZA resistance (MIC = 64 mg/L). But there was only one strain with increased bla KPC-53 gene dosage was reported in the study, and the comparison between isolates with different KPC variants was not investigated.
Our results showed that the relative gene expression and copy number of bla KPC in isolate with the highest CZA MIC was higher than baseline isolate carrying wild type bla KPC-2 , while this phenomenon did not appear in other CZA-resistant isolates. These results indicate that the mutated bla KPC was the dominant mechanism of CZA resistance in our isolates, and when combined with increased gene expression and copy number of bla KPC could lead to the higher level MIC of CZA.
This study is limited by its small patient population. CZA resistance was discovered before the widespread use of CZA in China. More research needs to be done, especially in data collected after the CZA treatment.

Conclusions
In summary, we found that mutated bla KPC is not the only mechanism related to CZA resistance, the increased gene expression and copy number of mutated bla KPC can cause high-level CZA resistance. With the use of CZA, more CZA resistant isolates have been reported worldwide, and more mechanisms of CZA resistance need to be explored. In consideration of the rapid acquisition of CZA resistance after therapy, our founding suggests that it is crucial to monitor the MIC of CZA in KPC-KP.

Isolates
Eight previously described isolates [25], including 4 baseline isolates recovered before CZA treatment and 4 corresponding CZA-resistant isolates recovered after CZA treatment, were analyzed in the present study. K. pneumoniae ATCC13883, K. pneumoniae ATCC700603, E.coli ATCC25922, and Salmonella enteric serotype Braenderup H9812 were used as reference isolates.

Antimicrobial susceptibility testing
Antimicrobial susceptibility testing performed using the VITEK-2 compact system (bioMerieux, Marcy-l'Etoile, France). Broth microdilution susceptibility testing of CZA was performed, and the result was interpreted according to the guidelines established by the Clinical and Laboratory Standards Institute (CLSI, M100, 2019). Avibactam was tested at a fixed concentration of 4 μg/ ml in combination with increasing concentrations of ceftazidime. The reference strains E. coli ATCC25922, and K. pneumoniae ATCC700603 were used as controls.
Detection of genes encoding β-lactamases, and outer membrane proteins PCR detection for the presence of beta-lactamase genes encoding carbapenemases (bla KPC , bla NDM-1 , bla VIM , bla IMP , and bla OXA-48 ), ESBL associated genes (bla CTX-M , bla SHV , and bla TEM ), and plasmid-borne AmpC betalactamases (bla ACC , bla DHA , and bla CMY ) were performed as described previously [35]. Outer membrane protein genes were amplified by PCR as described previously [36]. PCR amplicons were sequenced and compared with sequences available in the GenBank database using BLAST searches.
Quantitative real-time PCR (qRT-PCR) determinating gene expression and copy number of bla KPC and ompk35/36 The gene expression and copy number of bla KPC and ompk35/36 of 4 baseline isolates (A) and 4 CZAresistant isolates (B) were examined as described previously [36][37][38], and the reference genes were listed in Table 3. Total RNA at mid-logarithmic growth phase bacterial cultures were obtained using RNeasy Mini Kit (Qiagen, Germany) and treated with RNase-Free DNase Set (Qiagen) in accordance with the manufacturer's protocol. RT-PCR was performed using the QuantiTect SYBR Green RT-PCR Kit (Qiagen) on a QuantiStudio 12 K Flex system (Thermo Fisher Scientific). DNA was extracted by QIAamp DNA Mini Kit (Qiagen), and copy numbers were measured using the QuantiFast SYBR Green PCR Kit (Qiagen) on a QuantiStudio 12 K Flex system. K. pneumoniae house-keeping gene rpoB was used to normalize the gene expression and copy number of bla KPC and ompk35/36. Statistical analyses were performed using GraphPad Prism 9.

Preparation of competent cells and electroporation
The competent cells of K. pneumoniae clinical isolates were prepared using 10% glycerol as previously described [39]. The mixture of 50ul electrocompetent cells and 5ul plasmid was transferred into a 2 mm electroporation cuvette and electroporated using MicroPuler System (Bio-Rad) at 2.5 kV. The cells were plated onto Luria-Bertani (LB) agar containing apramycin at 50 mg/ L. The plates were incubated at 37°C overnight, and the successful clone was identified using PCR and sequencing. The MICs of the recombinant strains were determined in the presence of IPTG at 100 μM [22].

Isolation and analysis of outer membrane components
Outer membrane proteins were isolated according to Carlone's rapid procedure [40], and analyzed with SDS-PAGE.

Efflux pump inhibitor tests
MICs of CZA in combination with PABN (phenylalanine-arginine beta-naphthylamide) (25 mg/mL) [19,41], a pump inhibitor, were determined. A fourfold decrease in MIC after the addition of PABN was considered significant.
Core genome phylogenetic analysis WGS of all 8 isolates was carried out using the Illumina NovaSeq system in our previous study [25], and the assembled genome sequence has been deposited on NCBI with BioProject 'PRJNA588110'. SNP analysis was performed using snippy software and the JM45 sequence was used for reference sequences. 8313 SNV sites were identified and the Modelfinder was used to find the best model. A phylogenetic tree was built by Iqtree [42,43] with the best model (HKY + F + ASC + R4) and the number of bootstraps was set to 1000 times.  KPC-R GCCGCCCAACTCCTTCA [38] RPOB-F CTGATGCCTCAGGATATGATCAAC [38] RPOB-R CTGGCTGGAACCAAAGAACTCT [38] OmpK35-F TCCCTGCCCTGCTGGTAG [36] OmpK35-R CTGGTGTCGCCATTGGTGG [36] OmpK36-F GCGACCAGACCTACATGCGT [36] OmpK36-R AGTCGAAAGAGCCCGCGTC [36]