Molecular Epidemiology of Hypervirulent Carbapenemase-Producing Klebsiella pneumoniae

Objective To investigate the overall distributions of key virulence genes in Klebsiella pneumoniae, especially the hypervirulent bla KPC-positive K. pneumoniae (Hv-bla KPC(+)-KP). Methods A total of 521 complete genomes of K. pneumoniae from GenBank were collected and analyzed. Multilocus sequence typing, molecular serotyping, antibiotic-resistance, virulence genes and plasmid replicon typing were investigated. Results Positive rates of virulence genes highly varied, ranging from 2.9 (c-rmpA/A2) to 99.6% (entB). Totally 207 strains presented positive fimH, mrkD, entB and wzi and 190 showed positive fimH, mrkD, entB, irp2 and wzi, which were the two primary modes. A total of 94, 165 and 29 strains were denoted as hypervirulent K. pneumoniae (HvKP), bla KPC(+)-KP and Hv-bla KPC(+)-KP. ST11 accounted for 17 among the 29 Hv-bla KPC(+)-KP strains; Genes iucA, p-rmpA2 and p-rmpA were positive in 28, 26 and 18 Hv-bla KPC(+)-KP strains respectively. Among the 29 Hv-bla KPC(+)-KP strains exhibiting four super clusters from GenBank, IncHI1B plasmids carrying virulence genes and IncFII ones with bla KPC were responsible for both 23 strains respectively. Conclusions Positive rates of virulence genes vary remarkably in K. pneumoniae. Genes iucA, p-rmpA2 and p-rmpA were primary ones inducing Hv-bla KPC(+)-KP. IncHI1B plasmids carrying virulence genes and IncFII ones with bla KPC constitute the primary combination responsible for Hv-bla KPC(+)-KP. The making of Hv-bla KPC(+)-KP is mostly via bla KPC(+)-KP acquiring another plasmid harboring virulence genes.

With years passing, K. pneumoniae, regardless of cKP or HvKP, becomes more and more drug-resistant, among which carbapenem-resistance is of great concern. Carbapenem-resistance is mostly conferred by carbapenemase gene (bla KPC ), New Delhi metallo-b-lactamase gene (bla NDM ), and oxacillinases-48 gene (bla , which are predominantly carried on the mobile genetic elements (Zhang et al., 2015;Lee et al., 2016). Among them, bla KPC , particularly bla KPC-2/3 is predominant (Kopotsa et al., 2019). Carbapenem-resistant K. pneumoniae (CRKP) has now become a great public health threat worldwide (Lee et al., 2016;Niu and Li, 2019), due to its causing high mortality and medical burden.
In the past decades, hypervirulence and drug-resistance advance separately in K. pneumoniae. CRKP was not usually considered hypervirulent (Zhang et al., 2017a). However, their convergence was found in recent years worldwide (Zhang et al., 2015;Lam et al., 2019;Wozniak et al., 2019). Not surprisingly, such K. pneumoniae strains could induce an overwhelming mortality (Gu et al., 2018). Due to the mobility of elements carrying virulence and drug-resistance genes, hypervirulent carbapenem-resistant K. pneumoniae (Hv-CRKP) gained more and more prevalence with its positive rate reaching 7.4-15.0% among CRKP in recent years (Lee et al., 2017). To date, the overall distribution of key virulence genes in K. pneumoniae strains, in particular hypervirulent bla KPC -positive K. pneumoniae (Hv-bla KPC (+)-KP), was rarely reported. Here, we collected 521 K. pneumoniae strains from GenBank. Upon the yielded data, we could get insight into the distributions of key virulence genes in K. pneumoniae, particularly Hv-bla KPC (+)-KP.

Multilocus Sequence Typing (MLST)
The DNA fasta sequences of the 521 genomes were compared with the K. pneumoniae MLST database (Larsen et al., 2012) containing the seven housekeeping genes (gapA, infB, mdh, pgi, phoE, rpoB and tonB) and the STs were yielded.

Determination of Serotypes, Antibiotic-Resistance and Virulence Genes
For the genomes of K. pneumoniae from GenBank, the accession numbers were directly used to determine the capsular types via the database of Institute Pasteur (https://bigsdb.pasteur.fr/ klebsiella/klebsiella.html). The potential beta-lactamase genes were determined using the Resfinder software version 3.2 (https://cge.cbs.dtu.dk/services/ResFinder/) (Zankari et al., 2012) with the minimum coverage of 60% and minimum identity of 90%, and the virulence genes were predicted using NCBI_BLAST (megablast) searches against the virulence genes of K. pneumoniae with experimental supports (Table S2) with the cut-off coverage of 80% and cut-off identity of 80%.

Phylogenetic Analysis and Plasmid Replicon Analysis
The phylogenetic tree of K. pneumoniae strains was generated using kSNP3 (Gardner et al., 2015) software for K. pneumoniae chromosomes and displayed by iTOL (Letunic and Bork, 2016) with midpoint rooting. For the plasmids, the phylogenetic patterns were based on the presence/absence of orthologous gene families of all the plasmids under analysis. A binary gene presence/absence matrix was created using OrthoFinder (Emms and Kelly, 2019) with default settings and a hierarchical cluster result was shown by iTOL (Letunic and Bork, 2016).
Plasmid replicon typing was determined using the PlasmidFinder software version 2.0.1 with the minimum coverage of 60% and minimum identity of 95%.

Distributions of Virulence Genes in Hv-bla KPC (+)-KP
A total of nine modes of virulence genes were found among the 29 Hv-bla KPC (+)-KP strains, as shown in Figure 5. And the first four modes consisted of eight (27.6%), seven (24.1%), five (17.2%) and three (10.3%) strains, which constituted the majority.

DISCUSSION
This study investigated the general distributions of key virulence genes in K. pneumoniae, in particular Hv-bla KPC (+)-KP.
Among the 521 strains, 65 were positive for peg-344, of which 63 were denoted as HvKP. A sensitivity of 96.9% was therefore yielded, similar as the report (p = 0.5791) (Russo et al., 2018). Gene allS was not restricted to K1 and K2 strains, different from the document (Yu et al., 2008). The reason may lie in the different specimen types of analyzed strains. Gene wzy-K1 (formerly designated magA), corresponding to K1 serotype, vice versa, could help K. pneumoniae yield macromolecular EPS, which confers hypervirulence (Fang et al., 2004). Wzi is a protein riveting capsular polysaccharides, loss of which K. pneumoniae should be acapsular (Rahn et al., 2003). Acapsule was found in 13 (2.5%) strains, which means low virulence. A total of four kinds of siderophores were found in K. pneumoniae strains: enterobactin, salmochelin, yersiniabactin, and aerobactin (Russo and Marr, 2019). Intriguingly, one strain (strain AR_0096, accession number: CP027612.1) was found for none siderophore, indicating other ferric uptake systems than siderophores may also provide a certain amount of iron for growth and reproductivity (Hsieh et al., 2008).
Except for macromolecular EPS and excessive siderophores, hypercapsule could also contribute to hypervirulence (Russo and Marr, 2019), which is typically conferred by p-rmpA, p-rmpA2 or c-rmpA/A2 genes. Hypercapsule played an equal role with excessive siderophores (15.4 vs 16.3%, p = 0.6714) in hypervirulence of K. pneumoniae. The reason lies in the same pLVPK-like plasmids harboring rmpAs and siderophore genes concurrently.
Gene bla KPC was first reported from USA in 1996 (Yigit et al., 2001). Then, the first bla KPC-2 (+)-KP strain was reported in mainland China in 2007 (Wei et al., 2007). CRKP has now shared 70 -90% of carbapenem-resistant Enterobacteriaceae in the European Union and China (Grundmann et al., 2017;Zhang et al., 2017b). To date, bla KPC consists of more than 50 subtypes, among which bla KPC-2 is the most successful one and predominates CRKP worldwide. bla KPC-2 was positive in 132 (25.3%) strains while bla KPC-3 was found in 30 (5.8%) strains. Our study also showed clonal group 258 but not ST11 made up the majority of bla KPC (+)-KP Fu et al., 2019); The reason comes from the global distribution of the 521 strains.
The first Hv-CRKP, belonging to K2 and ST65, was unveiled in mainland China in 2015, which was isolated from blood in Wuhan FIGURE 7 | Characteristics of the 29 Hv-bla KPC (+)-KP strains. The phylogenetic relationship of the 29 Hv-bla KPC (+)-KP strains was analyzed by kSNP 3.0 Based on the predicted results, the binary gene presence/absence matrix was created reflecting the collection year, collection region, bla KPC-2 gene and core virulence genes. The STs of Hv-bla KPC (+)-KP strains were marked on the right of the phylogenetic tree. ND: not defined. The presence of genes, etc. is represented by a solid box. and the absence of others is represented by a white box.
City in March 2013 (Zhang et al., 2015). Armed with hypervirulence and extreme drug-resistance, Hv-CRKP causes greater mortality and becomes notorious (Gu et al., 2018). Our study showed a positive rate of 5.6% for Hv-bla KPC (+)-KP worldwide. Different prevalence of iucA, p-rmpA2 and p-rmpA in Hv-bla KPC (+)-KP strains suggested their different roles in hypervirulence. The modes of virulence genes were rather diverse in Hv-bla KPC (+)-KP. Similar prevalence of ≥ 3 siderophores and p-rmpA/p-rmpA2 (p > 0.9999) indicated their equal roles in hypervirulence of Hv-bla KPC (+)-KP strains, which also originated from the same pLVPK-like plasmids harboring rmpAs and siderophore genes simultaneously. The proportion of K64 was (11, 37.9%), lower than another report (Zhang et al., 2020) (p < 0.0001). Further, IncHI1B plasmids carrying virulence genes and IncFII ones with bla KPC were responsible for both 23 strains, suggesting IncHI1B and IncFII plasmids jointly constitute the most successful combination. Furthermore, the phylogenetic trees revealed that the 29 Hv-bla KPC (+)-KP strains belonged to four super clusters although three clusters all possessed ST11 strains.
Hv-bla KPC (+)-KP evolution may occur through two mechanisms. The first pathway is via HvKP acquiring a plasmid carrying drug-resistance determinants (Wei et al., 2016;Feng et al., 2018) or by the insertion of resistance genes into virulence plasmid or chromosome harbored by HvKP (Zhang et al., 2016;Fu et al., 2018). The second pathway is via multidrug-resistant/extreme drug-resistant cKP acquiring a pK2044-or pLVPK-like virulence plasmid or integrated virulence genes into drug-resistance plasmids (Gu et al., 2018). Our data showed it was most likely that Hv-bla KPC (+)-KP mainly evolved through the second pathway, i.e. via bla KPC (+)-KP acquiring another plasmid harboring virulence genes. Zhou et al. (2020) and Tang et al. (2020) preached that CRISPR-Cas system deficiency in ST11 may play a vital role. However, the two papers elucidated only bla KPC entering ST11 strains; IncHI1B plasmids are different from IncFII ones: rare protospacers were FIGURE 8 | Details for the characteristics of the 57 antibiotic-resistance or virulence plasmids carried by the 29 Hv-bla KPC (+)-KP strains. The phylogenetic patterns were based on the presence/absence of orthologous gene families of 57 plasmids under analysis. Seven categories of information were presented in this figure, including the phylogenetic tree of 57 plasmids, STs of host strains, collection year, collection region, bla KPC-2 gene, core virulence genes and replicon types of plasmids. ND, not defined. The presence of genes, etc. is represented by a solid box and the absence of others is represented by a white box.
found and they lacked Type IV secretion systems, e.g. traM gene. Therefore, the mechanisms behind IncHI1B plasmids entering ST11 strains would be sophisticated and intriguing.
This study has some limitations. First, the specimen types of 521 K. pneumoniae strains are not well known. Second, some positive virulence genes do not inevitably mean "exact" hypervirulence.
Taken together, positive rates of virulence genes vary overwhelmingly in K. pneumoniae. Hypercapsule plays an equal proportion with excessive siderophores in hypervirulence of K. pneumoniae. Virulence genes iucA, p-rmpA2 and p-rmpA are primary ones inducing Hv-bla KPC (+)-KP. IncHI1B plasmids carrying virulence genes and IncFII ones with bla KPC constitute the primary combination responsible for Hv-bla KPC (+)-KP. Hvbla KPC (+)-KP urges more insightful investigations.

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. This data can be found here: https://pan.baidu.com/s/1sbsl_phsx8 IRoQeeY87e-w (password: xf5l).

AUTHOR CONTRIBUTIONS
DH, YL and PR conceived the study. DT, WC, PF, WW and XJ collected the 521 genomes. DH, YL, PR and XL did bioinformation analysis. DH and YL wrote the manuscript, which was revised by XL and XJ. All authors contributed to the article and approved the submitted version.