In vivo Emergence of Colistin Resistance in Carbapenem-Resistant Klebsiella pneumoniae Mediated by Premature Termination of the mgrB Gene Regulator

Multidrug-resistant (MDR) Klebsiella pneumoniae is a severe threat to public health worldwide. Worryingly, colistin resistance, one of the last-line antibiotics for the treatment of MDR K. pneumoniae infection, has been increasingly reported. This study aims to investigate the emergence of evolved colistin resistance in a carbapenem-resistant K. pneumoniae isolate during colistin treatment. In this study, a pair of sequential carbapenem-resistant K. pneumoniae isolates were recovered from the same patient before and after colistin treatment, named KP1-1 and KP1-2, respectively. Antibiotic susceptibility testing was performed by the microdilution broth method. Whole genome sequencing was performed, and putative gene variations were analyzed in comparison of the genome sequence of both isolates. The bacterial whole genome sequence typing and source tracking analysis were performed by BacWGSTdb 2.0 server. Validation of the role of these variations in colistin resistance was examined by complementation experiments. The association between colistin resistance and the expression level of PhoP/PhoQ signaling system and its regulated genes was evaluated by quantitative real-time PCR (qRT-PCR) assay. Our study indicated that KP1-1 displayed extensively antibiotic resistant trait, but only susceptible to colistin. KP1-2 showed additional resistance to colistin. Both isolates belonged to Sequence Type 11 (ST11). The whole genome sequence analysis uncovered multiple resistance genes and virulence genes in both isolates. No plasmid-mediated mcr genes were found, but genetic variations in five chromosomal genes, especially the Gln30∗ alteration in MgrB, were detected in colistin-resistant isolate KP1-2. Moreover, only complementation with wild-type mgrB gene restored colistin susceptibility, with colistin MIC decreased from 32 to 1 mg/L. Expression assays revealed an overexpression of the phoP, phoQ, and pmrD genes in the mgrB-mutated isolate KP1-2 compared to the wild-type isolate KP1-1, confirming the MgrB alterations was responsible for increased expression levels of those genes. This study provides direct in vivo evidence that Gln30∗ alteration of MgrB is a critical region responsible for colistin resistance in K. pneumoniae clinical strains.

In this study, we investigated the genomic variations between a paired colistin-susceptible and -resistant K. pneumoniae isolates consecutively recovered from a single patient, and demonstrated the mechanism responsible for the emergence of high-level resistance to colistin during in vivo treatment.

The Patient and Isolates
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).

Whole-Genome Sequencing
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 pairedend 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.

Complementation Assays
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.

Accession Number
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.

Isolate Characterizations
The minimum inhibitory concentrations (MICs) of the two isolates to 21 antibiotics were shown in Table 2.

Genetic Variations in Colistin-Susceptible and -Resistant Isolates
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 colistinsusceptible 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 Frontiers in Microbiology | www.frontiersin.org FIGURE 1 | Phylogenetic relationship between K. pneumoniae KP1-1, KP1-2 and a total of 631 ST11 K. pneumoniae strains retrieved from NCBI GenBank. The lines connecting the circles indicate the clonal relationship between different isolates. The scale bar represents a pairwise allelic difference of core genome multilocus sequence typing (cgMLST) loci. The number of isolates from each province is given in square brackets. The solid line indicates the pairwise allelic differences of two isolates is less than 50 alleles. Branches longer than 50 allelic differences are proportionally shortened and are represented by dashed lines. The two yellow circles indicate K. pneumoniae KP1-1, KP1-2. 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).

Complementation Experiments
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).

Gln30 * Substitution in MgrB Associated With phoPQ and pmrD Overexpression
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 wildtype 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).

DISCUSSION
Over the past decade, increased rates of resistance to antibiotics in K. pneumoniae has been reported worldwide (Olaitan et al., 2014a;Giamarellou, 2016;van Duin and Doi, 2017). Colistin has regained a significant part of the therapeutic regimen for treatment of infection caused by carbapenemresistant bacteria. However, it rapidly developed resistance due to the frequent use in clinical settings, becoming a major public health concern (Olaitan et al., 2014a;Giamarellou, 2016). In the present study, we monitored the evolved colistin resistance in a 66-year-old female inpatient infected with carbapenem-resistant K. pneumoniae during colistin treatment. Our data provide the direct evidence that genetic evolution in the mgrB gene can lead to a high level of colistin resistance and cause treatment failure.

Strain
Colistin MIC (mg/L)  Colistin resistance is mainly mediated by chromosome or horizontal gene transfer. Chromosomal mutations in twocomponent systems (PmrA/PmrB and PhoP/PhoQ) and genes regulating these systems can lead to colistin resistance in K. pneumoniae (Olaitan et al., 2014a,b;Poirel et al., 2017). Moreover, plasmid-mediated mcr genes, which encodes a phosphoethanolamine transfer enzyme, have been identified to confer resistance to colistin via horizontal gene transfer (Liu et al., 2016;Caniaux et al., 2017;Ga et al., 2019). None of the plasmid encoded mcr-1 to mcr-10 genes were detected in KP1-2, which demonstrating that the colistin resistance is mediated by chromosomally encoded mechanisms.
By mapping the whole genome sequences of KP1-1 and KP1-2, five genes with non-silent alterations were exhibited in colistinresistant KP1-2, including inactivated mgrB gene mediated by premature termination. MgrB, a small transmembrane protein with 47 amino acids, mediates potent negative feedback on the PhoQ/PhoP regulatory system, which regulates genes implicated in the LPS modifications and colistin resistance (Lippa and Goulian, 2009;Olaitan et al., 2014b;Poirel et al., 2017). Until now, the insertion of IS elements (especially the IS5-like element), non-sense mutations, and missense mutations have recently been reported in colistin resistance in K. pneumoniae isolates in diverse clinical and non-clinical isolates (Cannatelli et al., 2013(Cannatelli et al., , 2014Olaitan et al., 2014a;Poirel et al., 2015;Aires et al., 2016;Haeili et al., 2017;Hamel et al., 2020). Among the above genetic variations, insertional inactivation of mgrB by IS elements, especially IS5-like elements, seemingly to be the most common mechanism of mgrB variation. Regarding the missense mutations, genetic alterations in MgrB, including Q30stop and C28stop, have been identified to be responsible for colistin resistance in K. pneumoniae isolates (Olaitan et al., 2014a;Aires et al., 2016). We suppose that premature termination within mgrB, found in the present study, result in MgrB inactivation and therefore lead to PhoP/PhoQ activation which in turn activates the PmrA/PmrB response regulator.
Complementation experiments showed that only transformation of wild mgrB gene had the ability to restore colistin susceptibility in KP1-2. This result agrees with that mgrB disruptions and mutations represent a strong association with colistin resistance mechanism in K. pneumoniae (Cannatelli et al., 2013(Cannatelli et al., , 2014Olaitan et al., 2014a;Poirel et al., 2015;Aires et al., 2016;Haeili et al., 2017;Hamel et al., 2020). The Gln30 * substitution, has been reported in several studies in different countries, found in the KP1-2 reinforced the hypothesis that position C88 in the mgrB (codon 30 in protein) is a critical region, which is prone to mutate upon colistin treatment (Olaitan et al., 2014a;Poirel et al., 2015;Aires et al., 2016;Haeili et al., 2017). Compared with the colistin resistance resulting from the mcr genes and two-component systems, the inactivation of MgrB leads to a higher level of colistin resistance (Olaitan et al., 2014a;Liu et al., 2016;Ga et al., 2019). In the current study, MgrB variation conferring colistin resistance occurred in a successful pandemic clone ST11, which will likely cause global presence of pan-drug-resistant K. pneumoniae and need continuous monitor.
The disruption of mgrB results in the activation of PhoP/PhoQ signaling system, which is known to indirectly activate the PmrA/PmrB via PmrD (Lippa and Goulian, 2009;Olaitan et al., 2014b). The activation of the PmrA/PmrB leads to the upregulation of pmrCAB and pmrHFIJKLM-pmrE operons that transfer of PEtN and L-Ara4N cationic groups to the LPS, which is responsible for the acquisition of colistin resistance in K. pneumoniae (Olaitan et al., 2014b;Poirel et al., 2017). Thus, it is generally accepted that loss of MgrB activates the cross-regulation of PhoPQ-PmrD-PmrAB signal transduction pathway in K. pneumoniae. However, activation of PhoP/PhoQ through mgrB mutation dose not significantly activate the production of PmrA/PmrB and confers colistin resistance. Therefore, PhoP/PhoQ activation alone is able to confer colistin resistance even without any additional effects caused by PmrA/PmrB activation (Cheung et al., 2020). In our study, we observed a signification association between colistin resistance, attributed to Gln30 * substitution in MgrB, and upregulation of phoPQ operon and pmrD gene despite there being no upregulation of pmrHFIJKLM and pmrCAB operons. This can be explained by the fact that some unexplained mechanisms other than pmrHFIJKLM and pmrCAB might be involved in mediating colistin resistance in K. pneumoniae, which warrants further investigation.
In conclusion, our findings identified that Gln30 * substitution in MgrB is responsible for the upregulation of PhoP/PhoQ signaling system and of the pmrD gene that confers colistin resistance in K. pneumoniae. To the best of our knowledge, this is the first report to provide direct in vivo evidence that the alteration of MgrB confers colistin resistance in a carbapenemresistant K. pneumoniae isolate in China.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ supplementary material.