Involvement of Tn3 transposon in formation and transmission of hypervirulent and carbapenem-resistant Klebsiella pneumoniae

ABSTRACT Hypervirulent and carbapenem-resistant Klebsiella pneumoniae poses a severe threat to public health for its high pathogenicity, transmissibility, and drug resistance. This study aims to explore the evolutionary path and the role of Tn3 transposon in the virulence and carbapenem resistance transmission of K. pneumoniae. Bioinformatics analysis and some experimental tests such as plasmid conjugation experiments, antimicrobial susceptibility testing, and virulence-associated tests were performed to explore the virulence and drug resistance transmission. The complete genome sequencing, S1 nuclease pulsed-field gel electrophoresis, and bioinformatics analysis were used to investigate their transmission mechanism mediated by Tn3 transposons. Three hypervirulent and carbapenem-resistant K. pneumoniae isolates, which were obtained from different patients at different time points in the same ward, harbored a virulence plasmid and a carbapenem resistance plasmid. They were confirmed to have first evolved from hypervirulent isolates and then acquired a bla KPC-positive plasmid (CR-hvKp evolutionary pattern). The non-conjugative virulence plasmid pVir could be transferred to other bacterial strains via mobilization by the conjugative IncN/U-type plasmid pKPC, as well as by fusing with the conjugative pKPC plasmid (mediated by Tn3-based homologous recombination) to be self-transmissible, thus transferring drug resistance and virulence. The cointegration, pVir/KPC fusion plasmid, was further resolved into pVir and pKPC between the duplicated copies of the Tn3 transposon resolution site mediated by site-specific recombination. Therefore, Tn3 transposons can mediate hypervirulent and carbapenem-resistant K. pneumoniae strains transferring drug resistance and virulence to other bacteria. We must be vigilant to emerging transposon-mediated hypervirulent and carbapenem-resistant pathogens. IMPORTANCE Carbapenem-resistant Klebsiella pneumoniae (CRKP) is resistant to most common antibiotics, becoming the most important and prevalent nosocomial opportunity pathogen. Besides, K. pneumoniae can also cause severe community-acquired infections, such as primary liver abscess and endophthalmitis. These pathogens are commonly referred to as hvKp. CRKP and hvKp have evolved separately, each occupying its own clonal lineage and exhibiting a variety of properties. Our study provides important insights into the evolutionary events related to the arousal of virulence and drug resistance in K. pneumoniae through plasmid transmission, mediated by Tn3 transposon. Our study also provides evidence that multiple mechanisms contribute to the successful transfer of non-conjugative virulence plasmid, and the involvement of transposons enhances the efficiency. A good knowledge of its transmission mechanisms is fundamental to finding effective strategies to combat these threatening pathogens. Transposons are widely present in bacteria, spreading resistance and virulence genes between the environment and humans. Therefore, emerging transposon-mediated hypervirulent and carbapenem-resistant pathogens should be highly valued.


Molecular characteristics of K. pneumoniae isolates
Between January 2017 and February 2018, a total of 530 multicenter clinical K. pneumo niae isolates were collected, including 34 isolates characterized as carbapenem-resistant and hypervirulent (6).Among these, three isolates (KP130, KP131, and KP133) were classified into the ST268-K20 type.They were positive both for specific virulence factors (iucA, iroB, rmpA, and rmpA2) and for the bla KPC-2 gene.This finding indicates that they simultaneously harbored the pK2044-like virulence plasmid and the KPC plasmid conferring hypervirulence and carbapenem resistance, respectively.Complete genomes of three CR-hvKp isolates were sequenced.Their genome characteristics are summarized in Table 1.Notably, the pKPC and pVir/KPC plasmids had complete conjugative transfer elements, indicating that they could be transferred to other strains and could even mobilize non-conjugative plasmids.

Evolutionary pathway of the ST268 CR-hvKp strains
According to the phylogenetic tree computed using the goeBURST software, ST268 belonged to the K. pneumoniae clonal group 815 (Group 2, Fig. S1A), adjacent to well-known MDR clones (Group 15; Fig. S1A), such as ST15 and ST14, and also adjacent to well-known hvKp clones (Group 35; Fig. S1A), such as CG23 and ST65.Group 1 had the highest resistance score, while Group 3 had the highest virulence score (Fig. S1B).The mean virulence and resistance scores of Group 2 were intermediate between Group 1 and Group 3 (Fig. S1B).From the bubble scatterplot data, it was evident that the ST268-adjacent genomes belonged to MDR clones (CG15) as well as hvKp clones (CG23 and ST65) (Fig. S1C).Therefore, the evolutionary pathway of the ST268 CR-hvKp strains is not straightforward to determine according to the goeBURST analysis because the ST268 clones were intermediate between hypervirulent and resistance clones.
However, the single nucleotide polymorphism (SNP)-based phylogenetic tree of 228 genomes in this cluster showed that ST268 was closely related to KL2 hvKp clones, such

Mobilization of virulence plasmids mediated by Tn3
Our previous study showed that non-conjugative virulence plasmid can be mobilized by the conjugative IncFII-type KPC plasmid because of the similar oriT site (6).Whether the IncN/IncU-type plasmid, with a complete conjugal transfer region, had the ability to mobilize the non-conjugative virulence plasmid needs to be further determined.We successfully conjugated KP130, KP131, and KP133 with sodium azide-resistant J53 and obtained the resulting transconjugants (Fig. 4A).Considering that the three transconju gants of each isolate had identical plasmid profiles [as determined from the S1 nucle ase pulsed-field gel electrophoresis (S1-PFGE) assays], one transconjugant from each isolate was chosen for further analysis.The complete genome analysis of the selected transconjugants confirmed that the pVir plasmid was co-transferred to the J53 strain (BioProject: PRJNA899795) with the pKPC plasmid.We observed that the nic sites of the pVir plasmids had 77.78% similarity with those from pKPC plasmids (Fig. 4B), suggesting that pVir plasmids could probably have been mobilized by pKPC plasmids.However, the virulence plasmid transfer frequencies have shown significant differences among the KP130, KP131, and KP133 strains (Fig. 4C).Thus, additional mechanisms may also be involved in the non-conjugative pVir plasmid mobilization.Moreover, the p131-1 plasmid of the J53-p131 strain was formed by the fusion of the pVir plasmid with partial sequences of the pKPC plasmid (which included its complete Type IV secretion system; Fig. S3).Interestingly, according to the S1-PFGE results, the pVir/KPC fusion plasmid 290 KB band was found not only in KP133 but also in KP130 and KP131 (Fig. 4A).Likewise, the KP133 strain also exhibited pVir (220 Kb) and pKPC (80 Kb) bands (Fig. 4A).However, these observations are not consistent with the complete genome sequencing results (Table 1).Next, as indicated in Fig. 5A, we designed four specific primer sets (V-K, K-V, V-V, and K-K) to target the sequences of the fusion region between the pVir and pKPC plasmids.If both V-K and K-V are positive, they will indicate the existence of pVir/KPC fusion plasmids in the tested strains.Additional explanations are given in Fig. 5B.Interestingly, the pVir/KPC fusion plasmid was found to be present in the KP130 and KP131 strains, which also carried the pVir and pKPC plasmids (Fig. 5C).Likewise, KP133 carried not only the pVir/KPC plasmid but also the pVir and pKPC plasmids (Fig. 5C).The same observation was obtained in three transconjugants (J53-p130, J53-p131, and J53-p133; Fig. 5C).Taken together, the pVir/KPC, pVir, and pKPC plasmids were simultaneously found in these isolates.However, the pVir/KPC, pVir, and pKPC plasmids could not maintain stable replication due to their incompatibility.We further dissected the sequences of the Tn3 transposon homologous region and found that the fusion sites could be targeted to 13 bp sequences (Fig. 5D).These sequences were also the resolution site (res) located inside the Tn3 transposon, responsible for the cointegration resolution of the transposition process.Based on these observations, we propose that pVir fuses with pKPC via Tn3-based homologous recombination to generate a new pVir/KPC fusion plasmid.Cointegration is further resolved between the duplicated copies of the transposon resolution site by resolvasemediated site-specific recombination (Fig. 5E).
In comparison with the parental J53 strain, the J53-p130, J53-p131, and J53-p133 transconjugants have shown clearly higher siderophore production levels, while their larvae-validated virulence levels were not increased (Fig. S4).Other virulence pheno types, such as serum resistance and viscosity, were also not significantly changed (Fig. S4).

DISCUSSION
An epidemic of hypervirulent and carbapenem-resistant K. pneumoniae has a higher mortality rate and is a greater public health challenge than an epidemic of the K. pneumoniae CRKP or hvKp strains (9,10,19).The mobilization of plasmids aggravates the emergence of such hypervirulent and carbapenem-resistant pathogens, regardless of the evolutionary pathways.Our previous studies simulated the evolutionary path of the CR-hvKp strains and found that the CR-hvKp strain could not simultaneously maintain hypervirulence and carbapenem resistance due to the rfaH mutation (6).K. pneumoniae in the CR-hvKp pattern have been reported, such as ST23-CR-hvKp (20)(21)(22), but little is known whether they exhibited both phenotypes: hypervirulence and carbapenem resistance.In this study, we collected three ST268-K20 K. pneumoniae strains (CR-hvKp pattern), which were successfully isolated from patients of the same hospital ward, and simultaneously carried the pLVPK-like virulence and bla KPC-2 -positive plasmids.
Most hvKP strains have either a K1 or K2 bacterial capsule type (23).Although the K20-type has been reported as hypervirulent (24), there are no previous reports on the evolution, virulence, and resistance of the ST268-CR-hvKp strain.We dissected the genetic relatedness of ST268 adjacent genomes and found that ST268 was located between CG15 and CG23.CG15 is always associated with antibiotic resistance, while CG23 is the most common hvKp type (12,25).The mean virulence and mean resistance scores were also intermediate between the two groups.The SNP-based phylogenetic tree showed clear evidence that ST268 had a close relationship with hypervirulent ST65 isolates rather than CG15 MDR isolates.This finding suggests that ST268-CR-hvKp isolates first became hypervirulent and then acquired the resistance plasmid.
HvKp isolates usually produce hyperviscous mucoid capsules on the bacterial surface; nevertheless, there is no clear understanding of their genetic basis and contribution to disease pathology (26).Although the KP130, KP131, and KP133 K. pneumoniae strains exhibited hypervirulence, they lacked a hyperviscous mucoid phenotype and capsule overproduction.This could be due to mutations in the rmpA and rmpA2 genes, which are involved in the biosynthesis of CPS.There is extensive evidence showing that K. pneumoniae strains carrying a virulence plasmid do not necessarily display a hypermucoviscous phenotype, and a hypermucoviscous phenotype may not also correlate with a hypervirulent phenotype of this organism (27,28).Among other factors, the ST268-CR-hvKp strain hypervirulence might be due to high levels of siderophore production.Siderophores, like aerobactin (encoded by iutA-iucABCD), are also critical virulence factors, enabling bacterial iron acquisition needed by hvKp strains, to proliferate infection (29).Moreover, ST268-CR-hvKp isolates also exhibited high resistance levels to many antibiotics.Therefore, ST268-CR-hvKp isolates were hypervirulent and carbapenem-resistant, while other K. pneumoniae strains in the CR-hvKp pattern may not exhibit carbapenem resistance and hypervirulence simultane ously.These findings indicate that ST268-K20 K. pneumoniae is more likely to evolve into hypervirulent and carbapenem-resistant strains compared with other types in the CR-hvKp pattern.
A non-conjugative plasmid harboring a mimic oriT sequence of conjugative plasmid can be mobilized by the in-trans activity of this conjugative plasmid (30).This in-trans relaxase mechanism of plasmid mobilization has also been confirmed in K. pneumoniae and Staphylococcus aureus (7,31).In this study, we provided evidence that the conjuga tive IncN/U plasmid may also mobilize the pLVPK-like non-conjugative virulence plasmid, enriching our understanding of the transmission and evolution of virulence plasmids.However, significant differences in transfer frequencies were observed among the KP130, KP131, and KP133 strains.The high transfer frequency of the pVir/KPC plasmid of the KP133 strain could be attributed to harboring a complete transfer element, conferring the self-transfer ability to the pKPC/Vir plasmid.We speculate that additional mecha nisms affecting pVir plasmid transfer into KP130 and KP131 strains may exist.Notably, complete genome sequencing and S1-PFGE have produced conflicting results.Thus, four primer sets have been further designed, and electrophoretic results confirmed the presence of the pVir, pKPC, and pVir/KPC plasmids in three ST268 CR-hvKp strains.Complete genome sequencing may have some constraints on low-copy plasmids.These plasmids were unstable in the host bacteria due to the plasmid incompatibility (32).Thus, genomic sequencing failed to identify sequence differences among the plasmids carried by the KP130, KP131, and KP133 strains.Taken together, pVir plasmids can be transferred by Tn3-mediated fusion with self-transferable pKPC plasmid medicated by Tn3, or can also be directly mobilized by pKPC plasmids.Moreover, fusions of the pVir plasmid with specific sequences of the pKPC plasmid (including the complete transfer region) in J53-p131 transconjugants may confer the higher transfer frequencies of pVir plasmids of KP131, in comparison with those of KP130 strains.
Plasmid conjugation is one of the basic methods of horizontal gene transfer, a process involved in the DNA exchange between neighboring bacteria.Although conjugation transfer provides a reasonable explanation for the acquisition of resistance and virulence, it can also cause gene rearrangements.The Tn3 transposon, which was found both in the pVir and the pKPC plasmids, mediates the formation of the pVir/KPC fusion plasmid by homologous recombination.This is the most common mechanism leading to the integration or fusion of two plasmids.It is becoming increasingly common for virulence plasmids to recombine with other plasmids.There can be mosaic virulence plasmids that encode both resistance and virulence genes, or conjugative virulence plasmids that enhance virulence gene transfer efficiency (32)(33)(34).The Tn3 transposon family encodes a DNA site-specific recombinase (or "resolvase") that resolves the cointegrate intermediate.Recombination occurs at specific transposon sites, the "internal recombination site" (IRS) or the "resolution site" (res) (15).Considering plasmid fusion at the res site and the simultaneous coexistence of pVir/KPC, pVir, and pKPC plasmids, it is reasonable to infer that pVir/KPC plasmid fusion and dissociation occur dynamically.The pVir plasmid could fuse with the pKPC plasmid by homologous recombination to generate a new fusion plasmid.The cointegrate is further resolved by resolvase-mediated site-spe cific recombination between the duplicated copies of the transposon resolution site.Recombination events facilitate plasmid evolution and confer adaptation advantages to complex and volatile environments.The transfer of resistance and virulence plasmids promotes the emergence and dissemination of CR-hvKp or hv-CRKP strains, posing challenging problems for public health and healthcare systems.
Our study had limitations that should be taken into account while interpreting these findings.First, there were a limited number of strains involved in this study.Second, the complexity of clinical strains cannot be avoided.Our study uncovers a novel mecha nism of CR-hvKp transmission and cannot be generalized for all CR-hvKp strains.These limitations could motivate future research on more potential mechanisms of CR-hvKp transmission.

Conclusions
The pVir plasmid can be transferred via mobilization by the conjugative IncN/U-type pKPC plasmid, as well as by fusing with the conjugative pKPC plasmid (mediated by the Tn3 transposon) to be self-transmissible.The involvement of transposons greatly enhances the efficiency of non-conjugative virulence plasmids transmission.We must be vigilant to emerging transposon-mediated hypervirulent and carbapenem-resistant pathogens, as transposons are ubiquitous in Enterobacteriaceae and contribute to the spread of resistance and virulence genes.

Bacterial strains and microbiological characteristics
The K. pneumoniae KP130, KP131, and KP133 strains were isolated from clinical speci mens from different patients in the same ward at a hospital in Zhejiang involved in our previous retrospective multicenter study (6).Virulence-associated genes (iucA, iroB, rmpA, and rmpA2) and carbapenem resistance genes were first amplified and sequenced.The antimicrobial susceptibilities to common antibiotics were determined by the microdilu tion method.The results were interpreted according to the Clinical and Laboratory Standards Institute (CLSI M100) (35).

Pulsed-field gel electrophoresis and S1 nuclease pulsed-field gel electropho resis
The PFGE methodology was used to analyze the genetic relationship among the KP130, KP131, KP133 K. pneumoniae strains, recipient Escherichia coli J53, and their transconju gants.These isolates were digested by XbaI endonuclease (Takara, Dalian, China) and analyzed by PFGE according to a previous study (36).The S1-PFGE methodology was used to determine the numbers and molecular weight of the plasmids carried by the isolates involved in this study.The relevant isolates digested with S1 nuclease (Takara, Dalian, China) were subjected to PFGE as described above.

Complete genome sequencing and bioinformatic analysis
After bacterial DNA isolation, the genomes of the KP130, KP131, and KP133 K. pneumo niae strains, as well as their transconjugants, were sequenced using the Illumina NovaSeq and MinION platforms.De novo assembly was performed using HGAP and CANU using default settings (37,38).Assembly errors were corrected using Pilon (39), prior to obtaining the complete genome.To screen ST268-related genome sequences, we collected 1,165 publicly available K. pneumoniae genomes from the GenBank database (Supplementary Data 1).Kleborate was used to determine the sequence type, K_locus, antimicrobial resistance genes, and virulence genes of the strains (http://github.com/katholt/Kleborate).The circular maps for plasmid genome comparison were generated using the BLAST Ring Image Generator (BRIG) software (http://sourceforge.net/projects/ brig) (40).A comparison of the fusion regions of related plasmid genomes was per formed using Easyfig software (https://mjsull.github.io/Easyfig/)(41).oriT sequences and transfer elements were predicted using oriTfinder's default parameters (42).

Phylogenic analysis
The clusters of ST268 adjacent genomes (Supplementary Data 2) were evaluated using the goeBURST software.The phylogenic analysis of related genomes was performed by Parsnp and visualized in iTOL.The virulence scores and resistance scores were calculated according to the algorithm described in a previous study (25).

Confirmation of virulence phenotypes
A series of virulence phenotypic assays were conducted as described in a previous study with some modifications (6,8).The serum resistance assay was performed by mixing mid-log phase bacteria with serum from healthy people (at a 1:3 ratio), followed by a 3-h incubation at 37°C.The survival percentage of each strain was plotted against the incubation period to determine the serum resistance.The hypermucoviscosity was determined by measuring the absorbance at 600 nm of supernatants after centrifugation at 10,000 g for 30 s.The siderophore production was confirmed by two methods: agar plates containing Chrome Azurol S and quantification of the siderophore production by Siderophore units (Su).The uronic acid content was quantified to evaluate CPS production.The virulence levels were assayed both in the mice and Galleria mellonella larvae infection models.The bacterial inoculum employed in mice was adjusted to 5 × 10 6 colony-forming units (CFU).The bacterial inoculum of the KP130, KP131, and KP133 strains employed in larvae was adjusted to 1 × 10 5 CFU.In transconjugants, it was increased to 1 × 10 6 CFU in mice and 1 × 10 7 CFU in G. mellonella.All experiments were repeated at least three times except the mice infection experiments, which were repeated twice.

Plasmid conjugation assays
The plasmid conjugation assays were conducted to determine whether the virulence plasmids of the K. pneumoniae KP130, KP131, and KP133 strains could be transferred to other strains, conferring virulence and resistance.KP130, KP131, and KP133 with tellurite resistance were used as donors, while sodium azide-resistant J53 was used as the recipient.We cultured both donors and recipients to the logarithmic phase at 37°C.After mixing 100 µL of donor cells and 400 µL of recipient cells, an LB agar plate was inoculated and incubated at 37°C for 16-18 h.Transconjugants were selected with potassium tellurite (3 µg/mL) and sodium azide (150 µg/mL).The transconjugants were further amplified by PCR using iucA and bla KPC-2 primers.Three transconjugants of each strain were chosen for PFGE and S1-PFGE analysis, as described above.Next, the complete genome of these transconjugants was sequenced to determine the numbers and molecular weights of the plasmids, as well as to detect any recombination events during the transmission process.

Statistical analysis
GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA) was used to assess the statistical significance of the data with the Student's t-test and the log-rank test.

FIG 5
FIG 5 The formation and resolution of pVir/KPC plasmids mediated by Tn3.(A) Primers designed for detecting pVir/KPC plasmids.(B) Interpretations for the PCR results.V-K: V-F/K-R; K-V: K-F/V-R; V-V: V-F/V-R; K-K: K-F/K-R.(C)DNA electrophoretograms for four genes specific for pVir, pKPC, and pVir/KPC plasmids.(D) 13 bp sequences involved in fusing with pKPC and pVir plasmids.(E) The formation and resolution of pVir/KPC plasmids.pVir fuses with pKPC via Tn3 transposon-based homologous recombination to generate a new pVir/KPC fusion plasmid, and the cointegrate is further resolved by resolvase-mediated site-specific recombination between the duplicated copies of the transposon resolution site.

TABLE 1
The genomic characteristics of ST268-CR-hvKp strains in this study a ORF, open reading frame.b +, have such information.c -, no such information.