
==== Front
eBioMedicine
EBioMedicine
eBioMedicine
2352-3964
Elsevier

S2352-3964(24)00362-1
10.1016/j.ebiom.2024.105326
105326
Articles
Decoding the origins, spread, and global risks of mcr-9 gene
Song Kaiwen abc
Jin Longyang ac
Cai Meng a
Wang Qi a
Wu Xingyu a
Wang Shuyi a
Sun Shijun a
Wang Ruobing a
Chen Fengning ab
Wang Hui whuibj@163.com
wanghui@pkuph.edu.cn
ab∗
a Department of Clinical Laboratory, Peking University People's Hospital, Beijing, China
b Institute of Medical Technology, Peking University Health Science Center, Beijing, China
∗ Corresponding author. Department of Clinical Laboratory, Peking University People's Hospital, Beijing 100044, China. whuibj@163.comwanghui@pkuph.edu.cn
c These authors contributed equally to this work and should be considered co-first authors.

10 9 2024
10 2024
10 9 2024
108 10532610 4 2024
23 8 2024
23 8 2024
© 2024 The Author(s)
2024
https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).
Summary

Background

The global spread of the plasmid-mediated mcr (mobilized colistin resistance) gene family presents a significant threat to the efficacy of colistin, a last-line defense against numerous Gram-negative pathogens. The mcr-9 is the second most prevalent variant after mcr-1.

Methods

A dataset of 698 mcr-9-positive isolates from 44 countries is compiled. The historical trajectory of the mcr-9 gene is reconstructed using Bayesian analysis. The effective reproduction number is used innovatively to study the transmission dynamics of this mobile-drug-resistant gene.

Findings

Our investigation traces the origins of mcr-9 back to the 1960s, revealing a subsequent expansion from Western Europe to the America and East Asia in the late 20th century. Currently, its transmissibility remains high in Western Europe. Intriguingly, mcr-9 likely emerged from human-associated Salmonella and exhibits a unique propensity for transmission within the Enterobacter. Our research provides a new perspective that this host preference may be driven by codon usage biases in plasmids. Specifically, mcr-9-carrying plasmids prefer the nucleotide C over T compared to mcr-1-carrying plasmids among synonymous codons. The same bias is seen in Enterobacter compared to Escherichia (respectively as their most dominant genus). Furthermore, we uncovered fascinating patterns of coexistence between different mcr-9 subtypes and other resistance genes. Characterized by its low colistin resistance, mcr-9 has used this seemingly benign feature to silently circumnavigate the globe, evading conventional detection methods. However, colistin-resistant Enterobacter strains with high mcr-9 expression have emerged clinically, implying a strong risk of mcr-9 evolving into a global “true-resistance-gene”.

Interpretation

This study explores the mcr-9 gene, emphasizing its origin, adaptability, and dissemination potential. Given the high mcr-9 expression colistin-resistant strains was observed in clinically the prevalence of mcr-9 poses a significant challenge to drug resistance prevention and control within the One Health framework.

Funding

This work was partially supported by the 10.13039/501100001809 National Natural Science Foundation of China (Grant No. 32141001 and 81991533).

Keywords

Colistin resistance gene
mcr-9
Phylogenomics
Bayesian analysis
Evolutionary adaptability
Codon usage bias
==== Body
pmc Research in context

Evidence before this study

In recent years, the mcr-9 gene has been widely discovered in different regions, genus, and sources. Currently, mcr-9 is not believed to confer resistance to colistin in bacteria. BLASTn have been used to compare sequences in the non-redundant nucleotide database. “mcr-9” was used as a keyword to search for corresponding strain sequences in the NCBI Pathogens database. The collected data spans from 1995 to 2023.

Added value of this study

We have constructed the most comprehensive dataset to date and analyzed the global evolution, spread, and origin of the mcr-9 gene. Our study also investigated genus-specific codon usage bias. Additionally, we have reported clinically colistin-resistant strains with high mcr-9 expression and confirmed the role of mcr-9.

Implications of all the available evidence

In this study, we systematically characterized the mcr-9 gene. For its low colistin resistance, mcr-9 has silently spread globally. Now, mcr-9 may becomes a significant threat to colistin, which poses a significant challenge to drug resistance prevention and control within the One Health framework. This study also defines new subtypes of mcr-9, providing a standardized basis for future research.

Introduction

The mcr-9 gene, discovered in recent years,1 belongs to the Mobile Colistin Resistance (mcr) gene family, which predominantly includes genes from mcr-1 to mcr-10. Colistin is one of the most important antibiotics globally and is considered the last line of defense against fatal infections caused by multidrug resistant bacteria.2 The phosphoethanolamine transferase encoded by the mcr family genes can catalyze the addition of phosphoethanolamine to the phosphate group of bacterial outer membrane lipid A, thereby reducing the negative charge of the bacterial outer membrane, diminishing its affinity with colistin, and leading to antibiotic resistance.2, 3, 4, 5, 6 The existence of this gene poses a major threat to public health because it leads to antimicrobial drug resistance,7,8 which is one of the “top ten public health threats globally” declared by the World Health Organization.9 In addition, most mcr genes are carried on plasmids that can undergo conjugative transmission, or on ColE-type plasmids that aid in plasmid transmission.10, 11, 12 Specifically, the structure of qseB-qseC-mcr-9 is similar to the structure of pmrC-pmrA-pmrB. The horizontal transmission ability of this gene family allows them to spread widely across the globe, leading to increased tolerance of various bacterial genus to colistin. These findings are of great significance for us to understand and cope with the challenge of antibiotic resistance, and remind us that we need to take more effective measures to prevent the continued propagation of these resistance genes.

Methods

Collection of new samples from China

Seventy mcr-9 positive isolates was selected from a large-scale screening, conducted through multi-center collection over the past decade.13, 14, 15, 16, 17 The collection of these samples spans from 2012 to 2022, encompassing 15 provinces and municipalities. This includes 17 samples from clustered outbreaks in Jilin province between 2017 and 2018. These 70 samples include Citrobactor, Klebsiella, Serratia, Enterobacter, and Escherichia (Supplemental Table S1).

In our study, we utilized both second-generation and third-generation sequencing technologies to sequence the bacterial strains. For the second-generation sequencing, genomic DNA from isolates was extracted with the DP302 Bacterial Genome DNA Kit (Tiangen Biotech, Beijing, China) and sequenced on the Illumina HiSeq 2500 platform, producing 150 bp paired-end reads with over 100× coverage. For the third-generation sequencing, we used the PacBio Sequel system. We constructed a library of approximately 30 Kb, and the sequencing depth reached more than 10×.

Genomic dataset construction

To obtain sequences containing the mcr-9 gene, we utilized the NCBI's Basic Local Alignment Search Tool (BLASTn v.2.14.0+) (https://blast.ncbi.nlm.nih.gov/Blast.cgi) against the non-redundant nucleotide database.18 To ensure specificity, we set stringent thresholds for coverage and identity at 95% (Coverage was determined by the ratio of the aligned region to the query sequence length). Additionally, we searched the NCBI Pathogens database (https://www.ncbi.nlm.nih.gov/pathogens/) using “mcr-9” as a keyword. We downloaded sequences in batches using the NCBI's Batch Entrez tool19 based on assembly numbers obtained from the Pathogens database (https://www.ncbi.nlm.nih.gov/sites/batchentrez). This process yielded a total of 628 samples.

To retrieve metadata associated with the sequences, we employed a series of functions from the rentrez20 package in R to access the NCBI's EUtils API. By combining these sequences with newly collected laboratory data, we obtained a comprehensive dataset of 698 sequence samples spanning 38 countries across five continents (Supplemental Table S2).

Phylogenetic analysis

First, we preprocess the FASTA files that contain the base sequences. Depending on specific needs, we extract sequences of varying lengths that contain the mcr-9 gene. Then, we use the MAFFT v.7.49021 for multiple sequence alignment. To mitigate inference errors caused by special samples, we implement two measures: 1. We perform quality control on each site and eliminate those with fewer than two sequences. 2. We construct a preliminary tree for all sequences and compute the distance from each leaf node to the root node. We assume that the sampling follows a normal distribution, so we use the Median Absolute Deviation (MAD) algorithm to let the normal values fall in the middle 90% of the area and exclude outliers.

Next, after we conducted a Root-to-Tip analysis using TEMPEST to assess the temporal signal within the dataset, we use BEAUTi and BEAST2 v.2.7.522,23 software for phylogenetic molecular clock analysis. Since we believe that the evolution rate remains constant, we choose a strict molecular clock. As for the substitution model, we choose the most general GTR model, which assumes that all nucleotide substitution rates may be different. For each model, we run MCMC until convergence (We check the effective sample size and parameter value traces in the Tracer software to check for convergence, and the ESS of each parameter must be greater than 200). Finally, we use TreeAnnotator to best fit the posterior tree of the model to provide an annotated MCC tree. And, we run each model more than 3 times to ensure its repeatability. Evolutionary trees were visualized using FigTree v.1.4.4 or the ggtree24 package in R.

In addition, population dynamics will affect the shape of the tree, so we choose the Birth-Death Skyline Serial model (BDSKY) to analyze the evolutionary history and future trends of biological populations in a non-parametric way.25

As for the inference of the effective reproduction number, we obtain it by processing the log intermediate files generated by BEAST2 v.2.7.5 using packages such as coda,26 bdskytools,27 and beastio28 in R language.

Calculation of CUB

We consider the frequency of usage of each codon in its corresponding amino acid in all cds as the codon usage bias. The specific calculation is as follows.Xij=∑n=1mCodonln∗TPM

where Xij represents the occurrence of the jth codon of the ith amino acid in the genome, Codonln represents the occurrence of the lth codon in the nth gene, and m denotes the number of genes included in the analysis. TPM represents the expression level of the gene.CUBij=Xij∑j=1niXij

where ni indicates the number of codon types for the ith amino acid.

MIC determination and RT-PCR

We used the Broth Microdilution assay to determine the Minimum Inhibitory Concentration (MIC). In a 96-well plate, we introduced a colistin-containing medium with concentration gradients ranging from 1 to 1024 μg/mL, using a twofold dilution. For quality control, we employed the Escherichia coli ATCC 25922.

For RT-PCR, we chose the housekeeping gene rpoB as the reference gene. We designed specific primers (5’ to 3’) for mcr-9 and rpoB (rpoB_Enterobacter: Forward: TCAACTCCCTGTCCGTGTAC, Reverse: CAAAGTGGCCTTCGTCATCC; rpoB_Klebsiella: Forward: AGGATATGATCAACGCCAAG, Reverse:GGGTTGTTCTGGTCCATAA; mcr-9: Forward: GGCCGCAATAACTCGACATT, Reverse:GGAACATCAGCGTGGGTTTT).

Each sample underwent triplicate analysis, and Ct values were obtained. We applied the Livak method for relative quantification.

Gene sequence comparison

To enhance the accuracy of sequence similarity depiction, we devised a comprehensive scoring system. Each site is individually evaluated; successful alignment earns 2 points, while misalignment incurs a −1 point penalty. Gap opening and extension are assigned −0.5 and −1 points, respectively. To mitigate the impact of sequence length variations, we apply a final score correction by adding a value equivalent to the sequence length to the total score. Pairwise sequence comparisons generate a score matrix, which is then transformed into a distance matrix. Dimensionality reduction is achieved using the t-SNE algorithm.29

Structure prediction and comparison

We retrieved the 3D data structure of the MCR-1 from the RCSB (Research Collaboratory for Structural Bioinformatics) PDB database. We utilized alphafold2, based on the amino acid sequence of the MCR-9, to predict the three-dimensional structure of the protein transcribed by the mcr-9 gene. We employed PyMOL v.3.0 (https://pymol.org/2/) for visualization and structural alignment.

Visualization

To visually represent our data, we employed a carefully curated selection of tools and techniques. We utilized dedicated R language packages, FigTree v.1.4.4, and reputable online resources like Bioinformatics.com.cn. To construct the mcr gene and its upstream and downstream environments, we initially performed in-depth annotation analysis of the sequences using Prokka v.1.14.6.30 Next, we used ProgressiveMauve v.1.2.031 to conduct detailed comparisons of homologous regions across different sequences. Finally, we harnessed the R language to visualize the comparison results.

Cooperative analysis of resistance genes

To assess the co-occurrence of the mcr-9 gene with other resistance genes, we employed the ABRicate v.1.0.1, a tool that leverages multiple databases to provide comprehensive resistance gene information. Following analysis, we utilized R language for statistical analysis, presenting the results in an informative heatmap.

To further confirm the presence of these antimicrobial resistance (AMR) genes, we used BLASTn with stringent thresholds for coverage and identity set at 95%. Following these analyses, we utilized R language for statistical analysis, presenting the results in an informative heatmap.

Strain construction and gene Expression Analysis

The pBAD30 plasmid was used as the backbone. The synthesized mcr sequences, with manipulated codons, were inserted into the pBAD30 plasmid. The mcr-9 gene was tagged with a myc label for detection. The constructed plasmids were introduced into the appropriate bacterial strains via electroporation. Transformed cells were grown overnight in LB medium containing 0.2% arabinose to induce the expression of the mcr genes. The RNAP gene was used as an internal loading control. After induction, cells were harvested and lysed to extract proteins. The expression levels of the mcr genes were then analyzed using Western blotting. The Myc antibody is from ZSbio (catalog number TA-01). The RNAP antibody is from BioLegend (catalog number 4RA2). The secondary antibody is also from ZSbio (catalog number ZB-2305). The presence and intensity of bands corresponding to the MCR proteins confirmed their expression and allowed for semi-quantitative analysis. Western blot analyses were performed in triplicate to ensure the reproducibility of the results.

Plasmid knockout

The pCasKP-apr plasmid was introduced into C7811 via electroporation. Following transformation, cells were recovered in SOC medium and then plated onto LB agar plates containing 100 μg/mL of apramycin to select for transformants. Colonies appearing on the plates were indicative of successful transformation and plasmid knockout. To confirm the knockout of the mcr-9 plasmid, colonies from the plates were picked and cultured in LB broth. Plasmid DNA was then extracted and subjected to PCR amplification using primers specific for the mcr-9 gene. The absence of an amplification product confirmed the successful knockout of the mcr-9 plasmid.

Statistics

In this study, we employed hypothesis testing to determine the differences between various groups. All significance levels were set at 0.05. Assuming that the expression of mcr-9 in sensitive mcr-9-positive strains follows a normal distribution, we conducted a two-tailed z-test to assess whether there is a significant difference in overall mcr-9 expression between resistant and sensitive strains. When calculating the difference in codon preference, the Wilcoxon-Mann-Whitney test is used.

Ethics

Researchers do not require separate ethical clearance.

Role of funders

The funders did not participate in study design, data collection, data analyses, interpretation, or writing of report.

Results

The specific distribution characteristics of the mcr-9 gene

According to clinical analysis and observation, we found that the positive rate of the mcr-9 gene in the clinical CRE (Carbapenem-Resistant Enterobacteriaceae) strains we collected is high, with the positivity rate approaching nearly 1%. In order to study the characteristics of mcr-9 gene-carrying strains, we have collected 70 mcr-9 gene-positive strains in China. Among these 70 isolates, mcr-9 is located on the plasmids in 66 of the isolates, and on the chromosomes in the remaining 4 isolates. We also included the sequence data and whole genome sequence data of mcr-9 positive in the NCBI public database, totaling to 698 samples. At the same time, we collected time, geographical information, genus, etc. of this collection for future analysis.

Based on the division of mcr-9 subtypes on NCBI, we analyzed the geographical distribution characteristics of different subtypes (Fig. 1a). We have identified new variants in the collected mcr-9 sequences that have not yet been included in the NCBI database. We have assigned numbers and names to these new variants, ranging from mcr-9.4 to mcr-9.27 (Supplemental Table S3). It shows that mcr-9.1 and mcr-9.2 were the most prevalent. In Asia, America, and most regions, mcr-9.1 has the largest proportion; but when it comes to Europe (especially Western Europe), mcr-9.2 has become the dominant subtype. This intriguing difference raises questions about the underlying evolutionary forces driving this divergence.Fig. 1 Comprehensive Overview of the mcr-9 Dataset. (a) Geographic distribution of mcr-9-positive assemblies. Circle sizes reflect strain counts, and colors indicate subtypes. Subtypes with higher counts are placed below those with lower counts. (b) The stacked graph shows the changes in the number of strains carrying the mcr-9 gene in various bacterial genus over time. (c) The stacked graph shows the changes in the number of strains carrying the mcr-9 gene in various sources over time. (d) The stacked graph shows the changes in the number of strains carrying the mcr-9 gene in various regions over time. (e) This network graph shows the types of plasmid replicons on which the mcr-9 gene is located (plasmids or chromosomes), with each combination of plasmid replicons represented by a pie chart, and the size of the pie chart indicating the quantity. Combinations of plasmid replicons with a common one are connected by a line (if there is no common one, two similar ones are found and so on).

The collection time of the samples spans 28 years (from 1995 to 2023). From the perspective of genus, the collected mcr-9 is most widely distributed in the genus Enterobacter, followed by the genus Salmonella. Around 2012, mcr-9 had a brief outbreak in the genus Escherichia (Fig. 1b). In addition, we also classified the source of the strains. The majority of mcr-9-harboring strains are human-derived, indicating a predominant prevalence of the mcr-9 gene among human isolates (Fig. 1c). The prevalence of the mcr-9 gene varies in different countries/regions. Among the samples collected, the United States and China have the most mcr-9 positive strains, followed by Australia and the United Kingdom (Fig. 1d). In addition, because the position of the mcr-9 gene in the genome is mostly on the plasmid, we use plasmidfinder to identify the “plasmid type” of the plasmid or chromosome where mcr-9 is located.32 Although there are various types, IncHI2 and IncHI2A types constitute the majority.33 As is widely known, IncHI2 type plasmids have significant transmission capabilities.34, 35, 36 From another perspective, although the IncFII type does not account for much in the overall, it occupies a significant proportion on the chromosome carrying mcr-9 (Fig. 1e).

Differences in sequences carrying the mcr-9 gene

To elucidate the relationship between the mcr-9 gene and its upstream and downstream environment, we compared 5000 bp sequences flanking the mcr-9 gene and visualized the pairwise similarity matrix using t-SNE.29

First, we added time information to the scatter plot. In the two-dimensional plane, we used the interpolation method to fill the area with colors corresponding to time. In the variable of mcr-9 subtype, different subtypes of mcr-9 exhibit distinct clustering. Within variables such as genus, source, plasmidtype, position of mcr-9 in the genome, region, and area, we can observe that there is a degree of overlap and intersection among the different categories within each variable. This actually reflects the diversity of the mcr-9 gene during its transmission process (Supplementary Figure S1).

Additionally, a Most Recent Common Ancestor (MRCA) analysis of the mcr-9 gene, based on time signals, was performed. In this process, we selected the sequences of the mcr-9 gene, along with 5 bases upstream and downstream of it, and after classification, we obtained 66 sequence types. Temporal signals were found based on regression of root-to-tip genetic distance against sampling time (R2 = 0.25, Supplementary Figure S2a). Then, we used BEAST2 software to infer the common ancestor from the existing variations of the mcr-9 gene. As is known to all, most strains carrying the mcr-9 gene are sensitive to colistin, so we chose the strict molecular clock model (This model assumes that each branch evolves at the same evolutionary rate in the evolutionary tree), we assume that the mcr-9 gene is evolving at a relatively constant rate. The results show that the earliest collected mcr-9 gene started to differentiate from 1961. There is a certain clustering phenomenon in the three variables of genus, source, and environment, and from the evolutionary tree (Fig. 2a).Fig. 2 The phylogenetic tree of the mcr-9 gene and the genetic environment. (a) We conducted a time-based evolutionary analysis of all mcr-9 genes (n = 698). The circles on the leaf nodes in the figure represent the number of strains (size, logarithm to base ten plus one) and mcr-9 subtypes (color) at that node. The three circles outside the tree represent the genus, source, and continent where the strains were found. (b) From each major branch of the evolutionary tree, one to two samples are extracted to display the mcr-9 gene and its upstream and downstream gene environment. The delta symbol (∗) indicates that the gene is incomplete. Scale bar: 2 kb bases.

In order to understand the differences in the gene environment of different subtypes clearly and intuitively, we selected six main clusters in the evolutionary tree. We chose one to two representative sequences in each cluster to illustrate the genetic context of the mcr-9 gene within a 5000 bp region upstream and downstream. We can clearly find that there are significant differences in the insertion sequences before and after the different mcr-9 subtypes. The side of the mcr-9.2 tends to directly connect the two insertion sequences IS1R and IS1D, while the mcr-9.1 prefer IS26 (Fig. 2b).

We sought to determine which gene structure, between the two distinct ones, is more advantageous for the spread or survival of mcr-9. Our analysis revealed that in Western Europe, the mcr-9.2 initially dominated, but its prevalence declined upon the emergence of the mcr-9.1. However, in the America and East Asia, mcr-9.1 has consistently been the predominant subtype, even after the introduction of mcr-9.2, suggesting that the gene structure of mcr-9.1 may confer greater adaptability (Supplementary Figure S3).

At the base level, gene mutations are also noteworthy. We comprehensively analyzed all mutation types and discovered the diversity of mcr-9 mutations. Notably, T to C mutations were the most prevalent, with 11 occurrences, surpassing the second-place A to G and A to T mutations (5 each) (Supplementary Figure S4a). Interestingly, these mutations were not evenly distributed across the gene, showing a tendency to cluster on one side (Supplementary Figure S4b). Furthermore, analyzing new amino acid mutations across different regions revealed that the United States had the highest number of mutations, followed by France and Canada (Supplementary Figure S4c).

Geographical spread inference

In order to deeply understand the trajectory of the spread of the mcr-9 gene in geographical distribution, we cluster it into five regions according to the sampling latitude and longitude of each sample, namely East Asia, Western Europe, America, South Asia, and the Middle East. Based on the regression analysis, temporal signals were detected in the genetic data (59 representative samples were selected) over time (R2 = 0.32, Supplementary Figure S2b). The results of Bayesian system geography inference based on ancestor state reconstruction show that the mcr-9 gene may initially spread from Western Europe to America and East Asia, and then the strains in South Asia and the Middle East gradually acquired this gene.

Since the data newly included in this study are all from China, we further analyzed the spread of mcr-9 within China. A total of 94 samples were collected from China, of which 58 were newly sequenced in this study, accounting for 61.7%. We also divided the geographical location into five major blocks, time-related patterns (Supplementary Figure S2c, 32 representative samples were selected) were identified in the genetic data, and found that mcr-9 first appeared in East China and then spread to the inland areas. From a time perspective, the first wave of spread within China (2007–2012) occurred after Western Europe was introduced into East Asia; from the direction of spread, the direction of spread from the coast to the inland is consistent with the direction of spread from East Asia to South Asia. This finding indicates that there is no conflict between the spread within the global scope and the spread within China (Fig. 3).Fig. 3 Geographical transmission of the mcr-9 gene. Based on the Bayesian model, we inferred the geographic origin of mcr-9. First, we divided the geographical locations into five categories based on latitude and longitude information: East Asia, South Asia, Western Europe, the America, and the Middle East. According to the inference results, the origin of mcr-9 is Western Europe, and it first spread to the America and East Asia, while South Asia and the Middle East obtained the gene later. Specifically in China, we found that the spread path was from the coast to the inland. The size of the circles on the Chinese map represents the number of samples from each city.

In addition, within China, we found that there are short-term outbreaks in specific areas. What is more alarming is that based on our multi-center data mining, we found its wide geographical distribution in china, which may imply that the mcr-9 gene has a wide distribution worldwide.

Differences in the spread of mcr-9 genes in different regions

The mcr family has received much attention due to its transferable characteristics, which has led to an increase in the global risk of colistin resistance. Therefore, it is particularly important to trace back the changes in the ability of gene transmission in the past. For this reason, we chose the effective reproduction number as the quantification standard. The effective reproduction number (Re) is an indicator used to measure the transmission ability of infectious diseases within a certain period of time. The mcr-9 gene has similar characteristics to infectious diseases. This gene can be transmitted between strains. Therefore, it is reasonable to use a birth-death model to infer the effective reproduction number. Time-related patterns in America (Supplementary Figure S2d), East Asia (Supplementary Figure S2e) and Western Europe (Supplementary Figure S2f) were identified in the genetic data.

For each region, we used the bdsky model based on BEAST2 for analysis, which is a birth-death skyline model. Here, we extracted 10,000 samples from the posterior distribution to draw a grid-like skyline plot. The results show that the effective reproduction numbers of the three regions show different trends over time.

In East Asia, the effective reproduction number exhibited a sudden increase around 2005, followed by a plateau and a subsequent sharp decline, eventually returning to a low level. In contrast, the America experienced oscillating effective reproduction numbers around 1, with a notable decline in 2015 that resulted in values consistently below 1. Western Europe, however, displayed a different pattern, characterized by a low effective reproduction number in the early stages, followed by a rise after 2010 and a more pronounced increase after a decline in 2015, eventually stabilizing at a high level (Fig. 4).Fig. 4 Effective Reproduction Number in East Asia, the America, and Western Europe. “Re” refers to the effective reproduction number, which represents the average number of different microorganisms that a mobile resistance gene can transfer to in a given time. That is to say, if Re is greater than 1, the transmission ability of the resistance gene is higher; if Re is less than 1, the prevalence of the resistance gene will be more likely to decline. We inferred the effective reproduction numbers for the three regions (East Asia (a-b), the America (c-d), and Western Europe (e-f)) with the largest sample sizes and found that East Asia and the America have stabilized after experiencing peaks, while the infectivity of mcr-9 in Western Europe has reached another peak after a brief decline. The shaded area represents the Highest Posterior Density (HPD) at each grid point.

The higher effective reproduction number observed in Western Europe suggests a potential gain in transmission capability due to evolutionary changes. Furthermore, the “further rise after decline” phenomenon observed in this region highlights the risk of resurgence in other regions where the effective reproduction number is currently low, emphasizing the need for continued vigilance. These findings provide valuable insights into the spread of the mcr-9 gene and underscore the importance of ongoing surveillance and targeted interventions to mitigate its impact.

Origin of genus and source

The genus Enterobacter was the most prevalent among the mcr-9 positive samples, with a majority originating from humans. This observation raises questions about whether this distribution is a result of evolutionary processes or a historical artifact. To investigate this, we conducted a source tracing analysis.

To accurately depict the situation, we selected the America, East Asia, and Western Europe for the source tracing inference of ancestral genus. We used 43 samples from America, 151 samples from East Asia, and 38 samples from Western Europe. The results revealed that in the America (Fig. 5a) and Western Europe (Fig. 5c), the ancestral bacterial genus was Salmonella, while in East Asia (Fig. 5b), it was the genus Enterobacter. Notably, the proportion of Salmonella in East Asia was relatively low.Fig. 5 The cross-genus transmission process of the mcr-9 gene in the America, East Asia, and Western Europe over time. We analyzed the cross-genus transmission process of the mcr-9 gene in the three regions with the largest sample size (America, East Asia, and Western Europe), with different colors representing different bacterial genera of the mcr-9 positive strains in each branch. Over time, the transmission of mcr-9 between genus is depicted in (a), (b), and (c) respectively. This suggests that the transmission of mcr-9 in Western Europe and the America originated from Salmonella, while the origin of mcr-9 in East Asia came from Enterobacter. The numbers in brackets represent the 95% highest density interval.

Previous reports have revealed the potential transmission of the mcr-9 gene between animals and humans.34 We further explored the transmission dynamics of mcr-9 positive strains among humans, animals, and the environment. The evolutionary tree with time information indicates that the earliest mcr-9 gene likely originated in humans and subsequently spread to animals and the environment (Fig. 6).Fig. 6 The spread of the mcr-9 gene among sources. (a) Evolutionary tree of sources based on time inference, with different colors representing different source types (human, animal, and environment). (b) Schematic diagram of the spread of the gene among sources. The direction of transmission is from human sources to animal sources and from human sources to environmental sources.

Genus bias and codon usage bias

Owing to the variability in the supply of tRNA within host cells, a specific amino acid's codon is often utilized at varying frequencies, a phenomenon referred to as “Codon Usage Bias” (CUB). Most organisms exhibit CUB to varying extents.37

Mcr-9 and mcr-1, despite both belonging to the mcr family, exhibit significant differences in terms of function, dominant genus, and hosts. To elucidate this, we took into account the CUB of mcr-1 and mcr-9 within their respective plasmids and strains. We employed the Principal Component Analysis (PCA) method to condense the usage frequency of 64 codons into two dimensions, with the first two principal components accounting for 70.486% and 9.554% of the variance, respectively. Upon dimension reduction, it can be observed that the codon usage of mcr-positive bacterial genomes and mcr-carrying plasmids forms two parallel lines with slopes approximately equal to 1. While the codon usage preferences of different genus vary significantly across the entire genome, this variation is less pronounced in plasmids due to their exogenous nature (Fig. 7a).Fig. 7 Differences in codon usage bias. (a–b) Use PCA to perform dimensionality reduction visualization of codon usage preferences, and display differences from genus and gene environment respectively. (c) The difference in the codon frequency used by mcr-1 and mcr-9 positive plasmids in the six amino acids N, H, D, Y, C, and F, where AAC_N represents the frequency of codons AAC in amino acid N. (d) Differences in codon usage bias among Escherichia and Enterobacter in these amino acids. The symbol ‘∗∗∗’ is used to denote statistical significance (Wilcoxon-Mann-Whitney test). Specifically, ‘∗∗∗’ represents a P-value less than 0.001.

Taking into account the differences between mcr-1 and mcr-9, we discovered that, regardless of whether in plasmids or the whole genome, mcr-9 generally trends in the positive slope direction of mcr-1 (Fig. 7b). What accounts for this? Upon analysis (the Wilcoxon test was used), we uncovered an intriguing pattern: for synonymous codons (considering only amino acids with two distinct codons, XXY type), mcr-9 tends to favor C over mcr-1, and mcr-1 tends to favor T over mcr-9 (Fig. 7c).

Interestingly, these trends align with the codon usage tendencies of the genera in which mcr-1 and mcr-9 are predominantly found: Escherichia and Enterobacter, respectively. Escherichia tends to favor T over Enterobacter, and Enterobacter tends to favor C over Escherichia (Fig. 7d). This suggests that the preference of mcr-9 for Enterobacter as its most prevalent genus rather than Escherichia may be intimately linked to the inherent codon usage bias among different genera.

We experimentally validated the impact of codon bias on gene expression across different species. We selected representative clinical strains–E. coli ATCC 25922 and Enterobacter cloacae ATCC 13047, and targeted the mcr-1 and mcr-9 genes. For each gene, two variants were designed, with the last position of the codons expressing the six amino acids N, H, D, Y, C, F changed to C (denoted as C-type) or T (denoted as T-type). These two different genes and the wild-type mcr (denoted as WT) were introduced into the strains via the pBAD30 plasmid and induced by arabinose. Western blot results showed that, whether it was mcr-1 or mcr-9, when the last position of the six amino acid codons changed from C to T (i.e., C-type became T-type), the increase in gene expression in ATCC 25922 was significantly higher than in ATCC 13047 (Fig. 8). This implies that E. coli is more inclined to use T in these six codons compared to E. cloacae.Fig. 8 Impact of Six Amino Acid Codon Changes on Expression. (a) Western Blot Analysis: This examines the expression of the mcr-1 gene and mcr-9 gene in ATCC 25922 (Escherichia coli) and ATCC 13047 (Enterobacter cloacae). The wild-type (WT), C-type (C), and T-type (T) are selected separately. The C/T type refers to all six amino acid codons N, H, D, Y, C, and F ending in C/T. RNAP is selected as the loading control and Myc as the tag protein. (b) Calculation of Expression Level Change: This calculates the change in expression level of the six amino acid codons from T to C at the end, and the standardized expression level of the T type is divided by the expression level of the C type (Three biological replicates).

Coexistence of mcr-9 gene and other drug-resistant genes

Multi-drug resistance among bacteria is a growing concern due to factors like antibiotic misuse. The mcr-9 gene, associated with colistin resistance, has been reported to coexist with other drug-resistant genes,38, 39, 40, 41, 42, 43 necessitating attention to potential prevention and control challenges.

For the different sample data types we have, we separately considered the presence of other drug-resistant genes in the same chromosome (Supplementary Figure S5a), the same plasmid (Supplementary Figure S5b), and the same strain (Supplementary Figure S5c) with the mcr-9 gene. We constructed a heatmap based on whether each sample can identify a certain drug-resistant gene, where blue represents the absence of the gene, and pink represents the presence of the gene. Then, we perform clustering on rows and columns with similar properties. We found that there is a very obvious clustering of mcr subtypes, that is, the coexistence patterns of mcr-9.1 and mcr-9.2 with other drug-resistant genes are different. We specifically look at the three drug-resistant genes with the greatest difference between mcr-9.1 and mcr-9.2: they are ant(2″)-la, aadA2, and blaCTX-M-9. Compared to mcr-9.1, these three genes are more inclined to coexist with mcr-9.2 (Supplementary Figure S5d). Ant(2″)-la and aadA2 confer resistance to aminoglycoside antibiotics, while blaCTX-M-9 is a variant of extended-spectrum β-lactamases that can be plasmid-mediated, leading to cephalosporin resistance.44,45

Clinical drug-resistant strains with high expression of mcr-9 have emerged

The sensitivity of the most mcr-9-positive strains to colistin has been a significant factor contributing to its lack of attention. Over the past decade, our sampling efforts have revealed several mcr-9-positive strains that exhibit resistance to colistin. After excluding other known resistance mutations, one particular strain captured our interest: Enterobacter asburiae (C7811), which was isolated from a sputum sample of a 22-year-old woman. Using the liquid medium dilution method, we determined that C7811 has a colistin MIC of over 1024 μg/mL. Notably, C7811 not only carries the mcr-9 gene but also harbors the blaNDM-1 gene on the same plasmid, rendering it resistant to multiple drugs. We eliminated the plasmid carrying the mcr-9 gene and used antimicrobial susceptibility testing to test the MIC of the strains before and after knock out. It was found that the MIC of C7811 to colistin decreased from >1024 μg/mL to 32 μg/mL, a reduction of more than 32 times. This implies that the plasmid carrying the mcr-9 gene plays a key role in colistin resistance.

Furthermore, we conducted a comparison of mcr-9 expression levels between C7811 and six randomly selected mcr-9-positive Enterobacter strains that were sensitive to colistin using RT-PCR. The results revealed that the mcr-9 expression level of C7811 was significantly higher than that of the sensitive strain (with a P-value under the Livak method of 0.005). Interestingly, when analyzing other mcr-9-positive resistant bacteria with different colistin resistance mechanisms, we found that their mcr-9 expression levels did not significantly differ compared to sensitive strains (with a P-value of 0.66). This suggests that certain unknown colistin resistance pathways may inhibit or not significantly influence mcr-9 expression. The emergence of high mcr-9 expression in clinically resistant strains signals a dangerous trend toward widespread colistin resistance worldwide.

Discussion

Our laboratory's previous investigations have comprehensively analyzed the global prevalence and dissemination dynamics of the colistin resistance gene, mcr-1.46 While both mcr-9 and mcr-1 belong to the mcr family, strains harboring mcr-9 exhibit greater sensitivity to colistin compared to those carrying mcr-1.47 On the one hand, this is due to the weak activity of mcr-9 itself; on the other hand, it may be because the expression level of mcr-9 is relatively low. Moreover, mcr-9 is one of the most prevalent subtypes of the mcr family worldwide.10,48,49 This raises concerns that if mcr-9 acquires higher colistin resistance, it could have unpredictable consequences. Despite its significance, research on the evolutionary transmission of mcr-9 remains limited. Our study aims to address this gap, providing valuable insights for future prevention and control efforts.

We present a comprehensive and systematic analysis of the mcr-9 gene, constructing a global dataset of 698 mcr-9-positive sequencing isolates from six continents and 38 countries/regions. This dataset is the most extensive to date, providing a valuable resource for understanding the prevalence and diversity of mcr-9.

However, we acknowledge the limitations of our study, particularly the limited number of samples, especially early samples, and uneven sampling distribution due to inadequate microbial genomic monitoring in low- and middle-income countries.50 To mitigate the potential bias, we performed separate analyses for the three regions with the most samples (Western Europe, America, and East Asia) and conducted a separate transmission analysis for the Chinese region, strengthening the statistical reliability of our results and establishing a complete evidence loop.

The transmission of mcr-9 positive strains between sources is also noteworthy. Previous studies suggest that mcr-1 originated from livestock in China,46 whereas mcr-9 first appeared in humans and then spread to animals and the environment. This contrasting “origin” may influence its resistance evolution.

Furthermore, mcr-9 has three subtypes defined on NCBI. We have supplemented this with new subtypes. Notably, mcr-9.1 differs from mcr-9.2 by a single tryptophan residue (W) deletion. This seemingly minor variation leads to significant differences in upstream and downstream genes, including drug-resistance genes. Before our study, mcr-9 subtype analysis was virtually absent.

A critical innovation of this study is the integration of the concept of the effective reproduction number, widely used in epidemiology, into the analysis of transferable drug-resistant genes, a unique approach not employed in previous research. By conceptualizing bacteria as microbial communities, we can analogize the spread of drug-resistant genes among various strains to the transmission of viruses among individuals. This concept allows us to speculate on the transmissibility of mobile drug-resistant genes over time and identify specific events that may influence their spread, thereby aiding policy formulation. For instance, in East Asia, the effective reproduction number of the mcr-9 gene exhibited a steep decline around 2017, coinciding with China's prohibition of colistin as an animal growth promoter from April 30, 2017. Although the resistance of mcr-9 to colistin is not high, and there are reports indicating that mcr-9 has been detected in the E. coli found in pigs, which is unrelated to the use of colistin in pig farms.51 It is difficult for us to dismiss the potential connection, which warrants verification by further experiments.

Our study unveiled some key limitations that warrant future investigation. Notably, mcr-9 accumulates multiple mutations over time, which raises questions about which mutations increase mcr-9's resistance activity and which may weaken it. To delve deeper, we identified 30 nonsynonymous mutations in mcr-9 compared to the current most prevalent sequence. Remarkably, among these mutations, seven result in leucine substitutions. Moreover, irrespective of the reference sequence, the top four frequently observed amino acid changes (leucine, isoleucine, valine, and proline) collectively constitute nearly half (44.8%) of all substitutions. Intriguingly, these four amino acids are hydrophobic and include three branched-chain amino acids (BCAAs: leucine, isoleucine, valine). BCAAs, notably leucine, are crucial components of the mcr-9 gene product, comprising 0.087, 0.058, and 0.059 of the total amino acid composition, respectively. Proline, although present at a lower proportion (0.039), also plays a significant role in microbial physiology, as evidenced by studies linking proline degradation metabolism to the pathogenicity of Candida albicans.52 These observations suggest that mcr-9 mutations could potentially impact its physiological function. Furthermore, similarity clustering analysis of the composition of each mcr subtype revealed a clustering pattern akin to that observed in phylogenetic tree construction, with mcr-1, mcr-2, and mcr-6 forming one cluster, and mcr-7, mcr-8, mcr-9, mcr-10, and mcr-4 forming another (Supplementary Figure S6).

Furthermore, we compared mcr-1 and mcr-9 directly. Despite their low sequence similarity, these genes share a high structural resemblance. Extensive research has been conducted on the activity of mcr-1, providing valuable insights into its mechanism. According to its crystal structure, MCR-1 possesses a conserved catalytic center, which is centered around Zn2+ and consists of five residues,8,53 in addition to a similar phosphatidylethanolamine (PE) substrate-loading tunnel that includes six PE-binding residues.54,55 Based on this structural alignment, we analyzed the corresponding sites in MCR-9. For instance, in MCR-1, a mutation at position 465 from aspartic acid (D) to alanine (A) reduces the strain's resistance to colistin.54,56 Interestingly, at the corresponding position 450 in MCR-9 the amino acid can be either D (aspartic acid) or N (asparagine). The question arises: could mcr-9 evolve to exhibit activity comparable to mcr-1, posing a greater risk of colistin resistance worldwide? These are pressing challenges that require further investigation.

To sum up, through comprehensive analyses of sequencing data from positive isolates, we constructed an unprecedented global multicenter mcr-9 positive sequencing isolate dataset, the largest to date. Utilizing bioinformatics and phylogenetic tools, we unearthed profound insights into the evolutionary dynamics and transmission patterns of the mcr-9 gene across the world. The emergence of clinically resistant strains with high expression of mcr-9 serves as strong evidence of extremely high resistance risk. The “future” may already be on its way.

Contributors

HW conceived and designed the study. QW, SS, XW collected the bacteria isolates and sequenced them. XW, LJ and KS conducted microbial susceptibility testing as well as reverse transcription-polymerase chain reaction (RT-PCR) analysis on the bacterial strains. LJ and KS constructed expression vectors and strains, performed Western blot, and conducted plasmid elimination experiments. Bioinformatics analyses were performed by KS. KS wrote the draft. HW and LJ performed revisions. During this process, HW, MC, SW, SS, RW, FC provided a lot of guidance for background research, article writing, image production, and other items. LJ and KS have verified the underlying data. The authors have read and approved the final manuscript.

Data sharing statement

All new assembly data used for analysis have been submitted to NCBI, and the corresponding BioProject accession number is PRJNA1064478. Any additional information required in this paper is available from the corresponding author on reasonable request.

Declaration of interests

The authors declare that they have no competing interests.

Appendix A Supplementary data

Supplemental Table S1

Supplemental Table S2

Supplemental Table S3

Supplemental Figures

Supplemental Western blots

Acknowledgements

This work was supported by the 10.13039/501100001809 National Natural Science Foundation of China (Grant No. 32141001 and 81991533). The authors also thank all the partners in the CRE network for their contribution to this study.

Statement: During the preparation of this work the authors used “Bing Chat” and “Gemini” in order to improve readability. After using these services, the authors reviewed and edited the content as needed and takes full responsibility for the content of the publication.

Appendix A Supplementary data related to this article can be found at https://doi.org/10.1016/j.ebiom.2024.105326.
==== Refs
References

1 Carroll L.M. Gaballa A. Guldimann C. Identification of novel mobilized colistin resistance gene mcr-9 in a multidrug-resistant, colistin-susceptible Salmonella enterica serotype typhimurium isolate mBio 10 3 2019
2 Liu J.H. Liu Y.Y. Shen Y.B. Plasmid-mediated colistin-resistance genes: mcr Trends Microbiol 32 4 2024 365 378 38008597
3 Poirel L. Jayol A. Nordmann P. Polymyxins: antibacterial activity, susceptibility testing, and resistance mechanisms encoded by plasmids or chromosomes Clin Microbiol Rev 30 2 2017 557 596 28275006
4 Abavisani M. Bostanghadiri N. Ghahramanpour H. Colistin resistance mechanisms in Gram-negative bacteria: a Focus on Escherichia coli Lett Appl Microbiol 76 2 2023
5 Gogry F.A. Siddiqui M.T. Sultan I. Current update on intrinsic and acquired colistin resistance mechanisms in bacteria Front Med 8 2021 677720
6 Grégoire N. Aranzana-Climent V. Magréault S. Clinical pharmacokinetics and pharmacodynamics of colistin Clin Pharmacokinet 56 12 2017 1441 1460 28550595
7 Sun J. Zhang H. Liu Y.H. Towards understanding mcr-like colistin resistance Trends Microbiol 26 9 2018 794 808 29525421
8 Zhang H. Srinivas S. Xu Y. Genetic and biochemical mechanisms for bacterial lipid A modifiers associated with polymyxin resistance Trends Biochem Sci 44 11 2019 973 988 31279652
9 EclinicalMedicine Antimicrobial resistance: a top ten global public health threat eClinicalMedicine 41 2021 101221
10 Ling Z. Yin W. Shen Z. Epidemiology of mobile colistin resistance genes mcr-1 to mcr-9 J Antimicrob Chemother 75 11 2020 3087 3095 32514524
11 Hussein N.H. Al-Kadmy I.M.S. Taha B.M. Mobilized colistin resistance (mcr) genes from 1 to 10: a comprehensive review Mol Biol Rep 48 3 2021 2897 2907 33839987
12 Wang Y. Liu F. Hu Y. Detection of mobile colistin resistance gene mcr-9 in carbapenem-resistant Klebsiella pneumoniae strains of human origin in Europe J Infect 80 5 2020 578 606
13 Wang Q. Wang X. Wang J. Phenotypic and genotypic characterization of carbapenem-resistant Enterobacteriaceae: data from a longitudinal large-scale CRE study in China (2012-2016) Clin Infect Dis 67 suppl_2 2018 S196 S205 30423057
14 Zhang Y. Jin L. Ouyang P. Evolution of hypervirulence in carbapenem-resistant Klebsiella pneumoniae in China: a multicentre, molecular epidemiological analysis J Antimicrob Chemother 75 2 2020 327 336 31713615
15 Jin L. Wang R. Wang X. Emergence of mcr-1 and carbapenemase genes in hospital sewage water in Beijing, China J Antimicrob Chemother 73 1 2018 84 87 29040585
16 Wang X. Liu Y. Qi X. Molecular epidemiology of colistin-resistant Enterobacteriaceae in inpatient and avian isolates from China: high prevalence of mcr-negative Klebsiella pneumoniae Int J Antimicrob Agents 50 4 2017 536 541 28668693
17 Wang R. Liu Y. Zhang Q. The prevalence of colistin resistance in Escherichia coli and Klebsiella pneumoniae isolated from food animals in China: coexistence of mcr-1 and blaNDM with low fitness cost Int J Antimicrob Agents 51 5 2018 739 744 29409993
18 Camacho C. Coulouris G. Avagyan V. BLAST+: architecture and applications BMC Bioinf 10 2009 421
19 Benson D.A. Cavanaugh M. Clark K. GenBank Nucleic Acids Res 41 Database issue 2013 D36 D42 23193287
20 Winter D.J. rentrez: an R package for the NCBI eUtils API R J 9 2017 520
21 Katoh K. Standley D.M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability Mol Biol Evol 30 4 2013 772 780 23329690
22 Bouckaert R. Heled J. Kühnert D. Beast 2: a software platform for Bayesian evolutionary analysis PLoS Comput Biol 10 4 2014 e1003537
23 Bouckaert R. Vaughan T.G. Barido-Sottani J. Beast 2.5: an advanced software platform for Bayesian evolutionary analysis PLoS Comput Biol 15 4 2019 e1006650
24 Yu G. Smith D. Zhu H. ggtree : an R package for visualization and annotation of phylogenetic trees with their covariates and other associated data Methods Ecol Evol 8 2016
25 Stadler T. Kühnert D. Bonhoeffer S. Birth-death skyline plot reveals temporal changes of epidemic spread in HIV and hepatitis C virus (HCV) Proc Natl Acad Sci U S A 110 1 2013 228 233 23248286
26 Plummer M. Best N. Cowles K. CODA: convergence diagnosis and output analysis for MCMC R News 6 2005
27 Plessis Ld bdskytools: BDSKY Tools 2016
28 Plessis Ld beastio: functions for pre- and post-processing of BEAST and BEAST2 files 2024
29 Kobak D. Berens P. The art of using t-SNE for single-cell transcriptomics Nat Commun 10 1 2019 5416 31780648
30 Seemann T. Prokka: rapid prokaryotic genome annotation Bioinformatics 30 14 2014 2068 2069 24642063
31 Darling A.E. Mau B. Perna N.T. progressiveMauve: multiple genome alignment with gene gain, loss and rearrangement PLoS One 5 6 2010 e11147
32 Carattoli A. Zankari E. García-Fernández A. In silico detection and typing of plasmids using PlasmidFinder and plasmid multilocus sequence typing Antimicrob Agents Chemother 58 7 2014 3895 3903 24777092
33 Li Y. Dai X. Zeng J. Characterization of the global distribution and diversified plasmid reservoirs of the colistin resistance gene mcr-9 Sci Rep 10 1 2020 8113 32415232
34 Khalifa H.O. Oreiby A.F. Abd El-Hafeez A.A. First report of multidrug-resistant carbapenemase-producing bacteria coharboring mcr-9 associated with respiratory disease complex in pets: potential of animal-human transmission Antimicrob Agents Chemother 65 1 2020
35 Liang Q. Yin Z. Zhao Y. Sequencing and comparative genomics analysis of the IncHI 2 plasmids pT5282-mphA and p112298-catA and the IncHI 5 plasmid pYNKP001-dfrA Int J Antimicrob Agents 49 6 2017 709 718 28390961
36 Khalifa H.O. Soliman A.M. Saito T. First report of foodborne Klebsiella pneumoniae coharboring blaVIM-1, blaNDM-1, and mcr-9 Antimicrob Agents Chemother 64 9 2020
37 Wu X. Shan K.J. Zan F. Optimization and deoptimization of codons in SARS-CoV-2 and related implications for vaccine development Adv Sci 10 23 2023 e2205445
38 Macesic N. Blakeway L.V. Stewart J.D. Silent spread of mobile colistin resistance gene mcr-9.1 on IncHI 2 'superplasmids' in clinical carbapenem-resistant Enterobacterales Clin Microbiol Infect 27 12 2021 1856.e7 1856.e13
39 Chavda K.D. Westblade L.F. Satlin M.J. First report of blaVIM-4- and mcr-9-coharboring Enterobacter species isolated from a pediatric patient mSphere 4 5 2019
40 Lin M. Yang Y. Yang Y. Co-occurrence of mcr-9 and blaNDM-1 in Enterobacter cloacae isolated from a patient with bloodstream infection Infect Drug Resist 13 2020 1397 1402 32494170
41 Simoni S. Mingoia M. Brenciani A. First IncHI 2 plasmid carrying mcr-9.1, blaVIM-1, and double copies of blaKPC-3 in a multidrug-resistant Escherichia coli human isolate mSphere 6 3 2021 e0030221
42 Umeda K. Nakamura H. Fukuda A. Genomic characterization of clinical Enterobacter roggenkampii co-harbouring blaIMP-1- and blaGES-5-encoding IncP6 and mcr-9-encoding IncHI 2 plasmids isolated in Japan J Glob Antimicrob Resist 24 2021 220 227 33385587
43 Xu T. Xue C.X. Huang J. Emergence of an epidemic hypervirulent clone of Enterobacter hormaechei coproducing mcr-9 and carbapenemases Lancet Microbe 3 7 2022 e474 e475 35659344
44 Lahlaoui H. Ben Haj Khalifa A. Ben Moussa M. Epidemiology of Enterobacteriaceae producing CTX-M type extended spectrum β-lactamase (ESBL) Med Mal Infect 44 9 2014 400 404 25234380
45 Cantón R. Coque T.M. The CTX-M beta-lactamase pandemic Curr Opin Microbiol 9 5 2006 466 475 16942899
46 Wang R. van Dorp L. Shaw L.P. The global distribution and spread of the mobilized colistin resistance gene mcr-1 Nat Commun 9 1 2018 1179 29563494
47 Xu T. Xue C.X. Chen Y. Frequent convergence of mcr-9 and carbapenemase genes in Enterobacter cloacae complex driven by epidemic plasmids and host incompatibility Emerg Microbes Infect 11 1 2022 1959 1972 35848148
48 Liao W. Cui Y. Quan J. High prevalence of colistin resistance and mcr-9/10 genes in Enterobacter spp. in a tertiary hospital over a decade Int J Antimicrob Agents 59 5 2022 106573
49 Kawamoto Y. Kaku N. Akamatsu N. The surveillance of colistin resistance and mobilized colistin resistance genes in multidrug-resistant Enterobacteriaceae isolated in Japan Int J Antimicrob Agents 59 1 2022 106480
50 Acharya K.P. Subramanya S.H. Pitout J.D.D. Inclusion of next-generation leaders and cost-effective precision diagnostic techniques are vital in combatting antimicrobial resistance in low- and middle-income countries JAC Antimicrob Resist 2 2 2020 dlaa032
51 Guarneri F. Bertasio C. Romeo C. First detection of mcr-9 in a multidrug-resistant Escherichia coli of animal origin in Italy is not related to colistin usage on a pig farm Antibiotics 12 4 2023
52 Silao F.G.S. Jiang T. Bereczky-Veress B. Proline catabolism is a key factor facilitating Candida albicans pathogenicity PLoS Pathog 19 11 2023 e1011677
53 Ma G. Zhu Y. Yu Z. High resolution crystal structure of the catalytic domain of mcr-1 Sci Rep 6 2016 39540
54 Xu Y. Lin J. Cui T. Mechanistic insights into transferable polymyxin resistance among gut bacteria J Biol Chem 293 12 2018 4350 4365 29462787
55 Xu Y. Chen H. Zhang H. The mcr-3 inside linker appears as a facilitator of colistin resistance Cell Rep 35 7 2021 109135
56 Gao R. Hu Y. Li Z. Dissemination and mechanism for the mcr-1 colistin resistance PLoS Pathog 12 11 2016 e1005957
