Role of AmpG in the resistance to β-lactam agents, including cephalosporins and carbapenems: candidate for a novel antimicrobial target

Background A complex cascade of genes, enzymes, and transcription factors regulates AmpC β-lactamase overexpression. We investigated the network of AmpC β-lactamase overexpression in Klebsiella aerogenes and identified the role of AmpG in resistance to β-lactam agents, including cephalosporins and carbapenems. Methods A transposon mutant library was created for carbapenem-resistant K. aerogenes YMC2008-M09-943034 (KE-Y1) to screen for candidates with increased susceptibility to carbapenems, which identified the susceptible mutant derivatives KE-Y3 and KE-Y6. All the strains were subjected to highly contiguous de novo assemblies using PacBio sequencing to investigate the loss of resistance due to transposon insertion. Complementation and knock-out experiments using lambda Red-mediated homologous recombinase and CRISPR–Cas9 were performed to confirm the role of gene of interest. Results In-depth analysis of KE-Y3 and KE-Y6 revealed the insertion of a transposon at six positions in each strain, at which truncation of the AmpG permease gene was common in both. The disruption of the AmpG permease leads to carbapenem susceptibility, which was further confirmed by complementation. We generated an AmpG permease gene knockout using lambda Red-mediated recombineering in K. aerogenes KE-Y1 and a CRISPR–Cas9-mediated gene knockout in multidrug-resistant Klebsiella pneumoniae-YMC/2013/D to confer carbapenem susceptibility. Conclusions These findings suggest that inhibition of the AmpG is a potential strategy to increase the efficacy of β-lactam agents against Klebsiella aerogenes. Supplementary Information The online version contains supplementary material available at 10.1186/s12941-021-00446-7.


Background
Gram-negative bacteria are a major threat to hospitalized patients and are associated with high mortality rates [1][2][3][4]. Hospital-acquired infections such as pneumonia, bloodstream infections, urinary tract infections, wound or surgical site infections, and meningitis are of particular concern to the clinicians. In the presence of antibiotic selective pressure, bacteria are capable of acquiring or up-regulating genes that code for antibiotic drug resistance, eventually leading to the emergence and global dissemination of these pathogens [5]. Also, Gramnegative bacteria can have redundant resistance mechanisms, either using single mechanisms against multiple antibiotics or multiple mechanisms against a single antibiotic. The antimicrobial-resistance crisis has escalated for several reasons: the high cost associated with finding novel antibiotic targets and antibiotic discovery, the length of time needed to design the drug and evaluate the efficacy, and the increased frequency of emerging bacterial resistance.
Klebsiella aerogenes (previously known as Enterobacter aerogenes) is a clinically significant bacterium in the family Enterobacteriaceae. K. aerogenes are part of the normal microbiota of the gastrointestinal tract in 40 to 80% of the population [6]. Though not a primary human pathogen, it has been implicated in a variety of healthcareassociated conditions, such as systemic bacteremia and urinary and lower respiratory tract infections, which are intrinsically resistant to ampicillin and narrow-spectrum cephalosporins [7,8]. K. aerogenes possesses a chromosomal ampC β-lactamase gene that can be induced by antibiotic stress with various β-lactams. Mutations in the AmpC β-lactamase expression pathway can lead to resistance against extended-and broad-spectrum cephalosporins [9,10].
Carbapenems, one class of β-lactam antibiotics, have historically been successful for treating cephalosporinresistant K. aerogenes infections [11]. Alarmingly, K. aerogenes have recently emerged worldwide as carbapenem-resistant because of the high frequency of mutations in its ampR and ampD genes [12,13]. In this study, we examined the mechanism of carbapenem resistance in K. aerogenes using transposon mutagenesis. In addition, we incorporated the CRISPR-Cas9-mediated gene knockout system to delete ampG permease and understand its role in antimicrobial resistance.

Bacterial strains, plasmids, antibiotics and oligonucleotides
K. aerogenes YMC2008-M09-943034 (KE-Y1) was collected from a tertiary-care hospital in Korea in 2008, and non-carbapenemase-producing carbapenemresistant Klebsiella pneumoniae (YMC/2013/D) was collected from a different tertiary-care hospital in Korea in 2013. Bacterial identification was performed using the VITEK 32 GN system (BioMérieux, Marcy l'Etoile, France) and was confirmed using the direct colony method with MALDI-TOF MS (Bruker Daltonics, Bremen, Germany). The MICs of piperacillin, piperacillin-tazobactam, ampicillin, ampicillin-sulbactam, ceftazidime, ceftazidime-clavulanate, cefepime, imipenem, meropenem, ciprofloxacin, and aztreonam were determined by agar-dilution methods and E-test and interpreted according to the Clinical and Laboratory Standards Institute guidelines (2018). Lambda Red donor plasmid pKD46 was obtained from the E. coli Genetic Stock Center at Yale University (New Haven, CT, USA). Plasmid ZpUC-19 was a gift from Dr. Yo Suzuki at the J. Craig Venter Institute, La Jolla, CA, USA. Plasmids pCRISPR (Addgene plasmid # 42875) and pCas9 (Addgene plasmid #42876) were gifts from Luciano Marraffini. Plasmid pKDsgRNA-ack (Addgene plasmid #62654) was a gift from Kristala Prather. Recombinant strains were selected with spectinomycin, chloramphenicol, or kanamycin at a concentration of 50, 34, or 30 mg/l, respectively. Antibiotic concentrations were reduced by half for selecting strains harboring two or more plasmids. Oligonucleotides were synthesized by Macrogen, Inc., Korea.

DNA isolation, WGS, and analysis
The DNA of the wild-type and mutant strains were extracted using the Wizard genomic purification kit (Promega, WI, USA) with modifications to the manufacturer's protocol by adding 5 μl RNase solution during cell lysis and incubating the supernatant with DNA at − 20 °C for 1 h after addition of isopropanol. The WGS data were obtained by using the PacBio RS II sequencing system (Pacific Biosciences, Inc., Menlo Park, CA, USA) in a commercial laboratory (DNALink, Korea). To confirm the integrity of PacBio sequencing, all strains were sequenced on a 318 chip using the Ion Torrent PGM system and Ion Sequencing 200 kit (Life Technologies, Carlsbad, CA, USA). Annotations were performed using the RAST annotation pipeline with manual scrutiny [16]. All genomic analyses were performed using Geneious Pro 8.0 [17] (http:// www. genei ous. com).

Amino acid alignment and AmpG structure modeling
The amino acid sequences of AmpG protein from K. aerogenes KE-Y1 (CP045870), Klebsiella pneumoniae ATCC 13883 (KN046818), Escherichia coli O157:H7 (CP034384) and Pseudomonas aeruginosa PAO1 (CP053028) was aligned using Vector NTI align X. The structure model of AmpG was further established using I-TASSER according to homology modeling method [18]. The TOP 5 models predicted by I-TASSER and each model C-score are − 1.43, − 2.07, − 2.21, − 3.10, − 3.51 respectively. C-score is typically in the range of [− 5, 2], where a C-score of a higher value signifies a model with a higher confidence and vice-versa.

Allelic replacement mutagenesis using lambda Red recombineering
The ampG deletion mutant strain was constructed by gene replacement via double crossover recombination as described previously with a few modifications [20]. Briefly, three different fragments (including the upstream and downstream fragments of ampG) and the zeocin resistance cassette fragment were amplified using three different primer pairs i.e. Up_ampG_F/R, Zeo_F/R and Down_ampG_F/R (Table S3). The reverse primer of the upstream fragment and forward primer of the downstream fragment of ampG at their 5' ends included a 15-20-nt extension homologous to the primers used to amplify the zeocin marker gene. Nested overlap-extension PCR was performed using an equal concentration of the three fragments to generate a linear DNA template containing the zeocin marker gene flanked by both the upstream and downstream homologous regions. Plasmid pKD-sgRNA-ack was transformed into electrocompetent KE-Y1, and clones were screened on LB plates containing spectinomycin. Strain KE-Y1 containing pKD-sgRNA was grown in LB broth until reaching OD 600 = 0.2, and lambda Red recombinase was expressed by adding 0.2% arabinose and with an additional 1.  (Fig. 1). Average Nucleotide Identity (ANI) obtained with MUMmer and BLAST indicated more than 99.8% similarity among all three strains (Additional file 1: Table S1). Whole-genome alignment using Mauve [21] indicated the homology between the three strains with one locally collinear block reversed (Fig. 2).

Amino acid alignment and AmpG structure modeling
The amino acid sequence comparison of AmpG KE-Y1 with K. pneumoniae, E. coli and P. aeruginosa showed similarity of 96.33%, 93.08% and 41.58%, respectively. Also, all known activation motif residues (G25, A122, Q124, A181) were conserved (Fig. 3). Based on the homology modeling structure of AmpG KE-Y1, the activation motif residues are located inside the transmembrane. Therefore, it is expected to be similar to the previously known AmpG modeling structure [22].

Open reading frames deactivated by transposon insertions
Whole-genome sequencing (WGS) analysis of KE-Y1, KE-Y3, and KE-Y6 revealed six insertions of gentamicinencoded transposons. The transposon-mediated inactivated genes are listed in Additional file 1: Table S2. AmpG permease gene insertions were seen in both carbapenem-susceptible transposon insertion strains KE-Y3 and KE-Y6. In strain KE-Y3, the AmpG permease gene was truncated in the middle by the transposon insertion. However, in strain KE-Y6, the transposon inserted immediately upstream of the gene, leading to the loss of the promoter required for AmpG permease gene expression (Fig. 4). The genome position of the genes responsible for induced β-lactamase expression is illustrated in Fig. 5, along with its comparison to that of the wild-type and its transposon mutant derivatives. AmpC-AmpR and AmpD-AmpE are closely linked with each other. The localization of the AmpG, AmpC-AmpR, and AmpD-AmpE genes are similar and identical in all the three strains.

CRISPR-Cas9 mediated AmpG permease knockout
The clinical isolate K. pneumoniae-YMC/2013/D was included for CRISPR-Cas9-mediated gene knockout studies to evaluate the role of AmpG permease in another Gram-negative multidrug resistant pathogen. The CRISPR/Cas9 knockout system, coupled with lambda Red recombineering, has been used to  to 8 μg/ml) and twofold (8 to 4 μg/ml) decrease in resistance to imipenem and meropenem, respectively, compared to the wild-type isolate.

Discussion
Our results demonstrate that carbapenemases do not mediate carbapenem resistance in K. aerogenes. Thus, control and management of carbapenem resistance should not be focused solely on the use of carbapenemases. Carbapenem resistance can also be due to overexpression of AmpC, efflux pumps, and porin loss (or a combination of these). To the best of our knowledge, this is the first report of K. aerogenes and the role of AmpG permease on carbapenem resistance following induction of ampC during antibiotic-induced stress. This stress may lead to many other mutations in the bacterial genome that cause resistance to a wide variety of non-carbapenem antibiotics such as cefotaxime, ceftazidime and cefepime [10]. AmpC is a chromosomally encoded group I, class C cephalosporinase produced by K. aerogenes at basal levels. The presence of β-lactams, such as cefoxitin and imipenem, highly induce AmpC expression [23], which involves a complex network of regulatory genes closely linked with peptidoglycan recycling [24]. During antibiotic treatment, the balance of peptidoglycan synthesis is compromised, releasing GlcNAc-anhydro-MurNAcoligopeptides into the periplasm. Resulting murapeptides are transported into the cytosol by an AmpG transporter encoded by ampG [25]. However, AmpG has no influence on ampC induction, nor does it show a gene dosage effect [26]. Upon entry into the cytosol, the GlcNAc sugar residue is removed by β-N-acetylglucosaminidase (NagZ) to generate 1,6-anhydromuropeptide, which is processed by N-acetyl-anhydromuramyl-L-alanine amidase (AmpD) during the non-induced state [27,28]. However, growth in the presence of β-lactams leads to increased breakdown of peptidoglycan or mutations in the ampD gene and may eventually lead to an increased intracellular concentration of murapeptides [29]. Another gene, ampE, located near ampD; AmpD modulates the response exerted on β-lactamase expression by AmpE [30].
In the carbapenem-resistant KE-Y1 wild-type strain, this increased concentration of intracellular murapeptides might have induced the AmpC production by interacting with the LysR-type transcriptional regulator AmpR, making the strain resistant to carbapenems [31]. The ampR gene is located immediately upstream of ampC and encodes a DNA-binding protein that activates ampC [32,33].
Consistent with the above observations and data obtained in this study, the ampG knockout implicates this protein as a potential pharmaceutical target for controlling ampC hyper-expression. Studies indicate that cells lacking AmpD or AmpG lose 40% of the peptidoglycan layer per generation [31]. Both carbapenem-susceptible transposon insertion mutants KE-Y3 and KE-Y6 contained an ampD gene; therefore, the susceptibility to the carbapenem was due to the loss of functional ampG. Similar studies previously suggested the role of AmpG in the antimicrobial resistance of P. aeruginosa and K. cloacae [26,34].

Conclusions
To our knowledge, this is the first report illustrating the role of AmpG in carbapenem resistance in K. aerogenes. We used knockout studies with transposon mutagenesis in K. aerogenes KE-1 and the CRISPR-Cas9 gene