A 66-year-old female patient was hospitalized, in a tertiary hospital in Hangzhou, Zhejiang province, China, in 2019, with symptoms of pyosepticemia, pancreatic malignancy, leukopenia, thrombocytopenic purpura, hepatic failure, and renal insufficiency. Initially, the patient received tigecycline by intravenous injection. Isolates were cultured from the inpatient during her hospitalization. The first isolate KP1-1, cultured form the blood sample of the inpatient within 24 h after admission, displayed extensively antibiotic resistance (including tigecycline) but susceptible only to colistin. The antibiotic therapeutic strategy was then adjusted to that of intravenous colistin (500,000 Unit every 8 h for the first 4 days and every 12 h for the following days). After received colistin treatment for 9 days, the second isolate KP1-2 was cultured from the stool sample with a colistin MIC of 32 mg/L. At the end of the treatment period, the inpatient expired from pyosepticemia and pancreatic malignancy. The isolates were identified by VITEK 2 (bioMérieux, Marcy-l’Étoile, France) and Matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF-MS, Bruker, Billerica, MA, United States).

Antimicrobial susceptibility testing was performed by the microdilution broth method for the following antimicrobial agents: aztreonam, fosfomycin, ertapenem, ceftazidime, cefepime, cefoperazone-sulbactam, cefoxitin, levofloxacin, ciprofloxacin, amikacin, tetracycline, minocycline, tigecycline, colistin, cefotaxime, meropenem, imipenem, gentamicin, trimethoprim-sulfamethoxazole, piperacillin, and piperacillin-tazobactam. The results were interpreted according to Clinical and Laboratory Standards Institute (CLSI) guidelines, except for tigecycline and colistin, which were interpreted according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines. Escherichia coli (ATCC 25922) were used as quality control strains for antimicrobial susceptibility testing.

Genomic DNA was extracted using a QIAamp DNA MiniKit (Qiagen, Valencia, CA, United States) following the manufacturer’s instructions. The bacterial genome was fragmented by sonication using a Covaris M220 sonicator (Covaris, Woburn, MA, United States) and the sheared DNA fragments were then used to prepare a shotgun paired-end library with an average insert size of 350 bp via a TruSeq DNA Sample Prep kit (Illumina, San Diego, CA, United States). The prepared library was sequenced using the Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, United States) through the 150 bp paired-end protocol. The short reads were assembled using Unicycler v0.4.8 software (Wick et al., 2017).

The genome annotation was conducted using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al., 2016). Antibiotic resistance genes, virulence genes, and plasmid replicons were queried using ABRicate 1.0.1 in tandem with ResFinder 4.1, CARD 2020, VFDB 2019, and PlasmidFinder 2.1 databases, with a 90% threshold for gene identification and a 60% minimum length to respective database entries. With the genomic sequence of the first isolate KP1-1 as the reference, the reads of the second isolate KP1-2 was mapped against that of KP1-1 using CLC Genomics Workbench 12. The gene variations, including single nucleotide polymorphisms (SNPs) and insertion and deletion mutations were predicted, and the variations were verified by PCR and Sanger sequencing. In silico multilocus sequence typing (MLST) analysis and bacterial source tracking using core genome MLST (cgMLST) strategy were performed using BacWGSTdb 2.0 server (Ruan and Feng, 2016; Ruan et al., 2020; Feng et al., 2021). The phylogenetic relationship between K. pneumoniae KP1-1, KP1-2 and a total of 631 publicly available ST11 K. pneumoniae isolates recovered from China were analyzed.

The high-copy plasmid pCR2.1-Hyg, constructed by inserting a hygromycin-resistant gene into the HindIII site of pCR2.1, was used as a genetic vector. Escherichia coli DH5α was used as the host for recombinant plasmids. The differential genes were amplified from the colistin-susceptible isolates KP1-1 using the primers shown in Table 1. The purified amplified fragments were, respectively, cloned into the plasmid pCR2.1-Hyg using the ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). Then, the recombinant plasmids were separately transformed into the E. coli DH5α for amplification, and eventually introduced into the colistin-resistant isolates KP1-2 by electroporation. The plasmid pCR2.1-Hyg was also transformed into KP1-2 as blank control. Electro-transformants were selected on Mueller-Hinton agar supplemented with 40 g/mL of hygromycin.

Expression levels of phoP, phoQ, pmrA, pmrB, pmrC, pmrD, pmrE, pmrK, and mgrB genes were determined by qRT-PCR with the primers listed in Table 1. Total RNA was extracted from the mid-log phase bacterial culture using the PureLink^TM RNA Mini Kit (Thermo Fisher Scientific, Waltham, MA, United States) and treated with RNase-free DNase I (Takara Biotechnology, Dalian, China) to remove genomic DNA contamination. Then, the corresponding cDNAs were generated from 500 ng of RNA using the HiFiScript cDNA Synthesis Kit (CWBio, Beijing, China) according to the manufacturer’s instructions. qRT-PCR was performed using MagicSYBR Mixture (CWBio, Beijing, China) on the Applied Biosystems^TM QuantStudio^TM 1 (Thermo Fisher Scientific, Waltham, MA, United States). Thermal cycling conditions were as follows: 95°C for 30 s for enzyme activation, followed by 40 cycles of denaturation at 95°C for 5 s and annealing at 60°C for 30 s. The relative gene expression levels were determined using the comparative threshold cycle (ΔΔC_t) method with rpsL gene normalization. The experiments were performed in triplicate and repeated three times.

The draft genome sequence of K. pneumoniae KP1-1 and KP1-2 were deposited in the NCBI GenBank database under the accession numbers JAERIE000000000 and JAERIF000000000.

The minimum inhibitory concentrations (MICs) of the two isolates to 21 antibiotics were shown in Table 2. Antimicrobial susceptibility testing showed that the first isolate KP1-1 was extensively drug-resistant, including aztreonam, ertapenem, ceftazidime, cefepime, cefoperazone-sulbactam, cefoxitin, levofloxacin, ciprofloxacin, amikacin, tetracycline, minocycline, tigecycline, cefotaxime, meropenem, imipenem, gentamicin, trimethoprim-sulfamethoxazole, piperacillin, and piperacillin-tazobactam. However, it remains susceptible to colistin (MIC < 0.03 mg/L). After the colistin therapy for 9 days, the second isolate KP1-2 was recovered with an increased colistin MIC of 32 mg/L.

The whole genome sequence data analysis classified isolates KP1-1 and KP1-2 to the same Sequence Type 11 (ST 11). Both KP1-1 and KP1-2 harbored multiple antimicrobial resistance genes, including aminoglycosides (aadA2 and rmtB), β-lactams (bla_CTX–M–65, bla_KPC–2 and bla_TEM–1B), fluoroquinolones (qnrS1), fosfomycin (fosA), phenicols (catA2), sulfonamides (sul2), tetracyclines [tet(A)], and trimethoprim (dfrA14). Moreover, we also identified several virulence genes, including aerobactin (iutA, iucA, iucB, iucC, and iucD), hypermucoviscosity (rmpA and rmpA2), and yersiniabactin (ybtA, ybtE, ybtP, ybtQ, ybtU, ybtT, and ybtX). Phylogenetic analysis indicated that the majority of ST11 K. pneumoniae isolates in NCBI GenBank database were recovered from Sichuan and Hangzhou, and the most closely related strain to K. pneumoniae KP1-1 and KP1-2 is L20, another ST11 strain previously recovered from a human feces sample in Hangzhou in the year 2016, which differed by only 15 cgMLST loci (Figure 1).

Regarding to the colistin resistance, no plasmid-mediated genes (mcr-1 to mcr-10) were found in isolate KP1-2. To explain the mechanism of colistin resistance, the raw sequence reads of KP1-1 and KP1-2 were mapped and putative variations were identified in seven genes (Table 3). Compared with the colistin-susceptible KP1-1, a premature stop codon was detected in the mgrB gene at the position 88 (C88T) in isolate KP1-2, leading to a non-sense mutation in the amino acid sequence for glutamine to stop at position 30 of the protein (Gln30^∗). The full-length MgrB protein, which is 47 amino acids in wild type isolates, was therefore only 29 amino acids long in isolate KP1-2.

Moreover, Missense variants were detected in aromatic amino acid transporter AroP (A689T, Glu230Val), PTS N,N’-diacetylchitobiose transporter subunit ChbC (C413T, Ala138Val), transcriptional activator protein LysR (G770A, Cys257Tyr), and a frameshift variant was detected in aminoimidazole riboside kinase (714dupC, Ala239fs). Synonymous variants were found in fimbrial biogenesis outer membrane usher protein (T1641C) and Methyl viologen resistance protein SmvA (A711>C).

The significance of the non-silent alterations of the five genes detected in colistin-resistant KP1-2 was determined by complementation experiments. The introduction of PCR2.1-Hyg-mgrB, which carries a cloned copy of the KP1-1 mgrB gene along with part of its flanking sequence, in isolate KP1-2 was able to restore susceptibility to colistin, with MIC decreased from 32 to 1 mg/L (Table 4). However, complementation with the rest four functional genes did not alter colistin susceptibility in KP1-2 isolates. This indicates that colistin resistance is not related to these alterations in the other four genes. On the contrary, transformed with the PCR2.1-Hyg, minor change of colistin MIC was found in KP1-2 (MIC decreased to 16 mg/L).

Expression levels of the phoPQ operon and pmr genes were analyzed to assess the impact of the Gln30^∗ substitution in MgrB. Analysis of phoP, phoQ, and pmrD transcription by qRT-PCR revealed a two- to three- fold increase in KP1-2 carrying an inactivated mgrB allele in comparison with KP1-1 and with the mgrB mutants complemented with a cloned copy of wild-type mgrB (Table 5). However, there was no upregulation in the expression of pmrA, pmrB, pmrC, pmrE, and pmrK genes. Interestingly, significant upregulation of mgrB was observed for KP1-2 (5-fold) and KP1-2 complemented with wild-type mgrB (10-fold).