Virulence Characteristics and Molecular Typing of Carbapenem-Resistant ST15 Klebsiella pneumoniae Clinical Isolates, Possessing the K24 Capsular Type

Klebsiella pneumoniae is an opportunistic pathogen that frequently causes nosocomial and community-acquired (CA) infections. Until now, a limited number of studies has been focused on the analyses of changes affecting the virulence attributes. Genotypic and phenotypic methods were used to characterise the 39 clinical K. pneumoniae isolates; all belonged to the pan-drug resistant, widespread clone ST 15 and expressed the K24 capsule. PFGE has revealed that the isolates could be divided into three distinct genomic clusters. All isolates possessed allS and uge genes, known to contribute to the virulence of K. pneumoniae and 10.25% of the isolates showed hypermucoviscosity, 94.87% produced type 1 fimbriae, 92.3% produced type 3 fimbriae, and 92.3% were able to produce biofilm. In vivo persistence could be supported by serum resistance 46.15%, enterobactin (94.87%) and aerobactin (5.12%) production and invasion of the INT407 and T24 cell lines. Sequence analysis of the whole genomes of the four representative strains 11/3, 50/1, 53/2 and 53/3 has revealed high sequence homology to the reference K. pneumoniae strain HS11286. Our results represent the divergence of virulence attributes among the isolates derived from a common ancestor clone ST 15, in an evolutionary process that occurred both in the hospital and in the community.


Introduction
Klebsiella pneumoniae is one of the most important Gram-negative bacteria causing nosocomial and community-acquired (CA) infections, resulting in pneumonia, urinary tract infections (UTI), septicaemia, bronchitis and intra-abdominal infections [1]. K. pneumoniae is one of the most prevalent causes of catheter-associated urinary tract infections (CAUTIs), next to Escherichia coli. A high incidence of CAUTIs has substantial costs related to extended periods of hospital access. Furthermore, CAUTIs can affect the kidneys and enter the bloodstream causing systemic disease such as septicaemia [2]. Carbapenem-resistant Klebsiella pneumoniae (CR-Kp) has become a significant global public health challenge [3].
Severeness of the infection is basically determined by (i) physiological state of the host and (ii) virulence potential of the pathogenic agent. Hospitalization, antimicrobial therapy, prolonged use of invasive medical devices, and major surgeries are considered as an antibiotic. Mutations in ompK35 and ompK36 genes are the most known that are related to carbapenem resistance [29].
According to a recent study, infections caused by KPC-producing strains is more frequently associated with increased mortality compared with other mechanisms of resistance [30].
Recently, several studies have focused on the spread of antibiotic resistance and its evolutionary aspects in and outside of the hospital environment. However, knowledge of K. pneumoniae ecology, population structure or pathogenicity is relatively limited [17]. In this study appearance and emergence of CR-KP isolates with the same clonal origins and antibiograms in an English Teaching Hospital offered the possibility to analyse the evolutionary aspects affecting virulence attributes.

Bacterial Identification and Antimicrobial Susceptibility Testing
Isolates were analysed by MALDI-TOF MS and the results were compared with standard conventional identification. All the isolates (n = 39) were identified at the species level (log (score value) ≥ 2.0). The result of standard biochemical tests showed that, all isolates were negative for indole probe, methyl red test, ornithine-and arginine-decarboxylase test, motility test; and all isolates were positive for the adonite test, Voges-Proskauer test, citrate test, malonate test, urease test, lysine-decarboxylase test, saccharose and lactose test.
All isolates were resistant to penicillins, cephalosporins, carbapenems and fluoroquinolones. Antibiotic resistance profile and MIC values of the K. pneumoniae isolates is provided in Supplementary Table S2a and S2b, respectively. The minority of the isolates were susceptible to amikacin (48.71%) and gentamicin (30.76%). All isolates, except one (categorized as intermediate susceptible) were resistant to tobramycin (Supplementary Table S2a,b).
Of the 39 K. pneumoniae isolates included in the study, 100% (39/39) were shown to produce carbapenemase by the Carbapenem Inactivation Method (CIM).

Chromosomal Macro-Restriction Fragment Polymorphism Analysis with PFGE
Isolates were typed by PFGE of XbaI-digested total genomic DNA. Strains were considered to be the same clone if they showed ≥85% pattern similarity, or fewer than six fragment differences in the PFGE profile. Based on macrorestriction profile analysis by PFGE the isolates were grouped into three major clusters ( Figure 1).

Presence of Virulence-Associated Genes
The occurrence of virulence-associated genes was determined with PCR and results are summarized in Table 1. fimH-1 and mrkD genes encoding the type 1 and type 3 fimbrial adhesins, were present in 94.87% and 92.3% of the isolates, respectively. Phenotypic test ( Table 2) results were in agreement with the PCR data (Supplementary Table S3). The adhesion associated genes, mrkA, mrkJ and cf29a were detected at a prevalence of 92.3%, 92.3% and 10.25%, respectively. The activator of the allantoin regulator gene, allS was detected in 100% of isolates. Siderophore genes entB (enterobactin) and iutA (aerobactin) were detected at a prevalence of 94.87% and 5.13%, respectively. Siderophore phenotypic test results were in agreement with the PCR data. Serum resistance was observed in 46.15% of isolates at 3 h. traT gene, involved in resistance to serum, was detected in 46.15% of isolates. rmpA, the genetic determinant associated with the HMV phenotype was found in 10.25% of isolates. An enhancer of the colony mucoidity, was detected in 10.25%. uge and wziK24 genes encoding the uridine diphosphate galacturonate 4-epidermase gene and K24 capsular type gene, respectively, were detected in 100% of isolates. magA gene encoding mucoviscosity-associated gene A, causes hypermucoviscosity, as defined by positive results of the string test, were detected in 10.25% of isolates.

Figure 1.
Relationships based on the PFGE profiles of K. pneumoniae isolates have revealed that apart from the minor clones they can divided into three major groups. In gel digestion of the chromosomal DNA was performed with XbaI. Cluster analysis was performed with Fingerprinting II Informatix (BioRad, Hercules, CA, USA). Sequenced isolates are indicated with asterisks (*).

Presence of Virulence-Associated Genes
The occurrence of virulence-associated genes was determined with PCR and results are summarized in Table 1. fimH-1 and mrkD genes encoding the type 1 and type 3 fimbrial adhesins, were present in 94.87% and 92.3% of the isolates, respectively. Phenotypic test ( Table 2) results were in agreement with the PCR data (Supplementary Table S3). The adhesion associated genes, mrkA, mrkJ and cf29a were detected at a prevalence of 92.3%, 92.3% and 10.25%, respectively. The activator of the allantoin regulator gene, allS was detected in 100% of isolates. Siderophore genes entB (enterobactin) and iutA (aerobactin) were detected at a prevalence of 94.87% and 5.13%, respectively. Siderophore phenotypic Figure 1. Relationships based on the PFGE profiles of K. pneumoniae isolates have revealed that apart from the minor clones they can divided into three major groups. In gel digestion of the chromosomal DNA was performed with XbaI. Cluster analysis was performed with Fingerprinting II Informatix (BioRad, Hercules, CA, USA). Sequenced isolates are indicated with asterisks (*).

Virulence Associated Phenotypic Assays
Presence of certain virulence attributes were revealed with phenotypic tests and results are summarized in Table 2. MSHA specific to type 1 fimbriae was detected in 94.87% of the isolates, while incidence of MRHA specific to type 3 fimbriae was 92.3%.
The ability to form biofilm was detected in 92.3% of the isolates, of which, 34 isolates were high producers, 2 isolates were medium producers, and 3 isolates were poor biofilm producers ( Table 2). All isolates that were able to form biofilm also produced type 3 fimbriae. No relationship was revealed between biofilm formation and isolate origin.
Phenotypic tests aiming to reveal the capacities of the CR-Kp isolates to produce siderophores showed that 94.87% of them produced enterobactin, while only 5.12% produced aerobactin ( Table 2). In two cases (53/1 and C8/15) the presence of both iron scavenging systems were revealed, while in one case (C1/16) none of the tested siderophore systems could be detected.
These isolates generally exhibited poor survival in human serum, with only 46.15% of isolates proving resistant to serum bactericidal activity after 3 h (Table 2). Similarly, HMV was found in only 10.25% of isolates.
Cell internalisation assays performed on two human epithelial cell lines (INT 407 and T24) have revealed that most of the isolates (31/39; 79.48%) were able to invade one or both of the cell lines to some extent during a 3 h co-incubation ( Figure 2). Fifteen isolates were internalized by both cell lines, while 16 isolates could only invade T24, but not INT407 to a detectable extent. Nine isolates were not able to invade any of the applied cell lines. No general relationship between the origin of isolation and the ability to interact with epithelial cells was established. The ability to form biofilm was detected in 92.3% of the isolates, of which, 34 isolates were high producers, 2 isolates were medium producers, and 3 isolates were poor biofilm producers ( Table 2). All isolates that were able to form biofilm also produced type 3 fimbriae. No relationship was revealed between biofilm formation and isolate origin.
Phenotypic tests aiming to reveal the capacities of the CR-Kp isolates to produce siderophores showed that 94.87% of them produced enterobactin, while only 5.12% produced aerobactin ( Table 2). In two cases (53/1 and C8/15) the presence of both iron scavenging systems were revealed, while in one case (C1/16) none of the tested siderophore systems could be detected.
These isolates generally exhibited poor survival in human serum, with only 46.15% of isolates proving resistant to serum bactericidal activity after 3 h (Table 2). Similarly, HMV was found in only 10.25% of isolates.
Cell internalisation assays performed on two human epithelial cell lines (INT 407 and T24) have revealed that most of the isolates (31/39; 79.48%) were able to invade one or both of the cell lines to some extent during a 3 h co-incubation ( Figure 2). Fifteen isolates were internalized by both cell lines, while 16 isolates could only invade T24, but not INT407 to a detectable extent. Nine isolates were not able to invade any of the applied cell lines. No general relationship between the origin of isolation and the ability to interact with epithelial cells was established.

Genome Sequencing and Bioinformatic Analysis of K. pneumoniae Isolates
A representative member of clinical K. pneumoniae strains, from four different isolation sites, were selected for whole-genome sequencing.
K. pneumoniae strain 11/3 was isolated from a faecal sample. De novo assembly generated 5,517,254 bp. The average contig length was 134,567 and the average GC content of the chromosome was 57.3%. The genome contained 5180 genes and 81 tRNA genes.
Analysis of the strain 11/3 sequence data revealed the presence of three types of plasmids, IncFIB (K), IncFII(K) and ColpVC. Based on ResFinder results, the presence of 21 genes related to antibiotics, including the previously characterized bla CTX-M-15 , bla SHV-106 and bla VIM-4 genes, were identified. Several genes and mutations associated with resistance to antimicrobials were detected. Mutations were identified in the ompK35 (p.A217S) and ompK36 (p.I128M, p.I70M) genes that are associated with carbapenem resistance. Seven fluoroquinolone resistance mutations (acrR p.F197I, p.K201M, p.L195V, p.G164A, p.R173G, p.F172S and p.P161R) and one tigecycline resistance mutation (ramR p.A19V) were also found in the genome. Based on VFDB results, 27 virulence genes were analysed (Supplementary Table S4), of which 3 virulence genes (manB, manC and wbbM) were only present in strain 11/3. The nucleotide sequence of strain 11/3 was deposited in the GenBank database under the accession number JAJTNS000000000. The general features of genomic analysis of K. pneumoniae strain 11/3 are summarized in Table 3 and are shown in Figure 3. K. pneumoniae strain 50/1 was isolated from blood culture. Shotgun sequences were assembled into one circular replicon measuring 5,379,211 bp. The average contig length was 298,845 and the average GC content of the chromosome was 57.4%. The genome contained 5022 genes and 73 tRNA genes. Analysis of the strain 50/1 sequence data has revealed the presence of one type of plasmid, IncFII(K). Based on ResFinder results, the presence of 14 genes related to antibiotics, including the previously characterized bla CTX-M-15 and bla SHV-106 genes were identified. Mutations were identified in the ompK35 (p.A217S) and ompK36 (p.I128M, p.I70M) genes that are associated with carbapenem resistance. Seven fluoroquinolone resistance mutations (acrR p.K201M, p.F172S, p.F197I, p.G164A, p.L195V, p.R173G and p.P161R) and one tigecycline resistance mutation (ramR p.A19V) were also found in the genome. Based on VFDB results, we analysed 29 virulence genes (Supplementary Table S4), of which 3 virulence genes (mrkB, mrkH and mrkJ) were only present in strain 50/1. The nucleotide sequence of strain 50/1 was deposited in the GenBank database under the accession number JAJTNT000000000. The general features of genomic analysis of K. pneumoniae strain 50/1 are summarized in Table 3 and are shown in Figure 3. . Genomic comparison map of the K. pneumoniae strain 53/3, strain 53/2, strain 50/1 and strain 11/3, with relative localization of the predicted virulence determinant genes. From the outside, the dark grey shows the predicted antimicrobial resistance (AMR) genes (black label), based on ResFinder results. Red labels show the predicted virulence genes, based on VFDB analysis. The second circle (grey) shows the genomic map of the K. pneumoniae strain 53/3 chromosome. The third circle (light blue) represents the genomic map of the K. pneumoniae strain 53/2 chromosome. The fourth circle (light green) shows the genomic map of the K. pneumoniae strain 50/1 chromosome, while the fifth circle (yellow) shows the genomic map of the K. pneumoniae strain 11/3 chromosome. The three innermost circles represent the chromosomes of the reference strains, like K. pneumoniae strain SH11286 (red), K. pneumoniae strain MGH78578 (pink) and K. pneumoniae strain NTUH-K2044 (purple). Circular genomic maps of the investigated K. pneumoniae isolates were obtained using the CGView Server.

Discussion
Evolutionary aspects of antibiotic resistance spread and persistence and its impact on the clinical outcome in and out of the hospital environment has been the focus of recent studies [31][32][33][34][35][36]. It has been concluded that the CR-Kp infection-related mortality rate was higher than those of extended-spectrum β-lactamases (ESBL)-producing, and wild-type susceptible K. pneumoniae strains [37,38]. Moreover, infection with carbapenem-resistant strains is a risk factor for infection-related mortality [39][40][41][42][43]. Clones with common clonal origins, first identified in an English teaching hospital in 2010, provided a good opportunity to perform a comparative analysis focusing on virulence attributes.
In this study, the clones used were the first CR-Kps isolated in this hospital and originated from the first wave of the emergence (2010-2017). A common origin of the clones was confirmed as they all belonged to ST 15, possessed the moderately virulent capsule . Genomic comparison map of the K. pneumoniae strain 53/3, strain 53/2, strain 50/1 and strain 11/3, with relative localization of the predicted virulence determinant genes. From the outside, the dark grey shows the predicted antimicrobial resistance (AMR) genes (black label), based on ResFinder results. Red labels show the predicted virulence genes, based on VFDB analysis. The second circle (grey) shows the genomic map of the K. pneumoniae strain 53/3 chromosome. The third circle (light blue) represents the genomic map of the K. pneumoniae strain 53/2 chromosome. The fourth circle (light green) shows the genomic map of the K. pneumoniae strain 50/1 chromosome, while the fifth circle (yellow) shows the genomic map of the K. pneumoniae strain 11/3 chromosome. The three innermost circles represent the chromosomes of the reference strains, like K. pneumoniae strain SH11286 (red), K. pneumoniae strain MGH78578 (pink) and K. pneumoniae strain NTUH-K2044 (purple). Circular genomic maps of the investigated K. pneumoniae isolates were obtained using the CGView Server.
K. pneumoniae strain 53/2 was isolated from sputum. De novo assembly of the genome generated 5,269,114 bp. The average contig length was 292,728 and the average GC content of the chromosome was 57.4%. The genome contained 4908 genes and 81 tRNA genes. Analysis of the strain 53/2 sequence data revealed the presence of one type of plasmid, ColpVC. Based on ResFinder results, 20 genes related to antibiotics, including the previously characterized bla CTX-M-15 , bla SHV-106 and bla VIM-4 genes were identified. Several genes and mutations associated with resistance to antimicrobials were detected. Mutations were identified in the ompK35 (p.A217S) and ompK36 (p.I128M, p.I70M) genes that are associated with carbapenem resistance. Seven fluoroquinolone resistance mutations (acrR p.F172S, p.P161R, p.K201M, p.L195V, p.F197I, p.G164A and p.R173G) were found in the genome. Based on VFDB results, we analysed 18 virulence genes (Supplementary Table S4). The nucleotide sequence of strain 53/2 was deposited in the GenBank database under the accession number JAJTNR000000000. The general features of genomic analysis of K. pneumoniae strain 53/2 are summarized in Table 3 and are shown in Figure 3.
K. pneumoniae strain 53/3 was isolated from urine. De novo assembly generated 5,201,283 bp. The average contig length was 273,751 and the average GC content of the chromosome was 57.4%. The genome contained 4844 genes and 63 tRNA genes. Analysis of the strain 53/3 sequence data revealed the presence of one type of plasmid, IncFII(K). Based on ResFinder results, 21 genes were related to antibiotics, including the previously characterized bla CTX-M-15 and bla SHV-106 genes. Several genes and mutations associated with resistance to antimicrobials were detected. Mutations were identified in the ompK35 (p.A217S) and ompK36 (p.I128M, p.I70M) genes that are associated with carbapenem resistance. Seven fluoroquinolone resistance mutations (acrR p.R173G, p.F197I, p.G164A, p.P161R, p.K201M, p.F172S and p.L195V) and one tigecycline resistance mutation (ramR p.A19V) were also found in the genome. Based on VFDB results, we analysed 37 virulence genes (Supplementary Table S4), of which 5 virulence genes (cpsAPC, galF, irp1/ybt, irp2/ybt and wzi) were only in strain 53/3. The nucleotide sequence of strain 53/3 was deposited in the GenBank database under the accession number JACTNU010000000. The general features of genomic analysis of K. pneumoniae strain 53/3 are summarized in Table 3 and are shown in Figure 3.

Discussion
Evolutionary aspects of antibiotic resistance spread and persistence and its impact on the clinical outcome in and out of the hospital environment has been the focus of recent studies [31][32][33][34][35][36]. It has been concluded that the CR-Kp infection-related mortality rate was higher than those of extended-spectrum β-lactamases (ESBL)-producing, and wild-type susceptible K. pneumoniae strains [37,38]. Moreover, infection with carbapenem-resistant strains is a risk factor for infection-related mortality [39][40][41][42][43]. Clones with common clonal origins, first identified in an English teaching hospital in 2010, provided a good opportunity to perform a comparative analysis focusing on virulence attributes.
In this study, the clones used were the first CR-Kps isolated in this hospital and originated from the first wave of the emergence (2010-2017). A common origin of the clones was confirmed as they all belonged to ST 15, possessed the moderately virulent capsule type K24, had the same antibiogram (Supplementary Table S2a,b), and showed high sequence homology based on PFGE and partially by whole genome sequence analysis. Well documented controls were used, including the hypervirulent K. pneumoniae strain NTUH K-2044 [44] and the moderately virulent reference strain MGH78578 [45]. Many attributes were common, but the slight differences among them indicated a divergence or evolutionary separation via an evolutionary process.
The common clonal origins of the K. pneumoniae isolates were supported by the fact that all isolates (n = 39) belonged to ST15 and possessed the K24 capsular type and their antibiograms showed high similarities to each other (Supplementary Table S2a,b). Furthermore, identified point mutations (location inside the gene is labelled with: p.XY) in the resistance genes of the four sequenced isolates 11/3, 50/1, 53/2 and 53/3 were localized at the same positions affecting the carbapenem resistance in the ompK35 (p.A217S) and ompK36 (p.I128M, p.I70M) [46]. This was also the case in other two inactive resistance genes, namely the acrR and ramR. The acrR gene (1492-4449 bp) confers resistance to fluoroquinolones [47,48], is located on a plasmid, and was affected with the same seven point mutations (p.F197I, p.K201M, p.L195V, p.G164A, p.R173G, p.F172S and p.P161R). Only one mutation (p.A19V) was identified in all sequenced isolates in the ramR gene, that confers resistance to tigecycline [47][48][49].
Expression the type 1 (92.3%) and type 3 (94.87%) fimbriae (Table 2) showed the capacity of the clones to colonize abiotic and biotic surfaces, and this contributed not only to survival but also to pathogenesis. These findings are in agreement with earlier studies which show that the adhesive subunit FimH in particular, of type 1 fimbriae, plays an important role in UTIs caused by K. pneumoniae [50]. Type 3 fimbriae assist adhesion to human tissue structures (e.g., lung, kidney) and promote biofilm formation on abiotic surfaces. As such, they may play a role in biofilm-associated infections in catheterized patients [2,[51][52][53], but are also present in UTI E.coli isolates [54]. Correlation between the presence of mrkD, the expression of type 3 fimbriae, and strong biofilm formation was also confirmed by our study (Tables 1 and 2), indicating its pivotal role in survival and persistence of the isolated clones. Incidence of mrkD genes (92.3%) among our CR-Kp isolates correlate with recent findings (94%) [55]. Results of our and this latter recent study are also comparable based on the incidence of magG as being 10.25% and 11%, respectively.
The high rate of enterobactin expression (94.87%, Table 2) was consistent with previous studies [56][57][58][59] and further supported the pathogenic potential of the isolates. Furthermore, aerobactin was also reported to be an important determinant of virulence of K. pneumoniae [16] if present, although it was more rarely found in Klebsiella. Its reported rates range from 3-6% [21], similar to our findings (5.12%). Although in our study no serious outcomes were registered among the patients infected by the isolates, evidence in the literature is mounting that K. pneumoniae strains carrying acquired siderophores have enhanced capacities to cause invasive diseases [17,[59][60][61][62].
A high rate of siderophore production could also support cell internalization and invasion. The heterogeneous picture of the invasion capacities of the isolates with common clonal origin testify to unknown molecular biological changes that could affect the invasion capacities of these bacteria. Cell internalization in most bacteria is a still an unsolved multifactorial process, that in case of K. pneumoniae was first reported more than two decades ago [63]. Because of its shielding effect, internalization is a bacterial survival strategy that could also spoil the efficacy of targeted therapies presenting a challenge to the clinician. Despite these therapeutic consequences, cell internalization is still not in the focus of K. pneumoniae research [64][65][66]. In our study, internalization experiments ( Figure 2) not only revealed the differences among the isolates, but also revealed that most isolates showed preference for the T24 bladder carcinoma cells, instead of INT 407.
In contrast, past studies have focused on the HMV phenotype of K. pneumoniae and its contribution to hypervirulence [61,67,68]. Our data confirmed the position that HMV and hypervirulence are different phenotypes that should not be described synonymously. This finding is consistent with recent opinions [69,70]. Although HMV positive clones (53/3, 53/11, 50/3, 11/1) can form biofilm, produce enterobactin and resist degradation in human serum, they were isolated either from mild UTIs or from faecal samples. Furthermore, our results confirmed earlier findings that the presence of the rmpA gene, encoding a positive regulator of mucopolysaccharides expression, is necessary for the HMV phenotype, since we detected this regulator only in the 4 HMV positive clones. To best of our knowledge this is the first study where the rmpA gene has been detected in a CR-Kp strain belonging to the ST 15 clone.
The described differences among the isolates with common clonal origins could be a result of an evolutionary process happening in the community and subsequently brought into the hospital. Based on the cluster analysis, we hypothesize that members of cluster I may originate from these two epidemiological processes. However, based on the isolation years and genomic similarities between members of cluster II and cluster III, revealed with PFGE, we postulate that two internal hospital epidemics could have occurred in 2011 and 2015, respectively. We have no available community data from the Chester region from this period, but based on the hospital data, the dominance of the ST 15 clone could be speculated to have emerged from 2010-2017.
Despite these similarities, it is interesting to note that no serious infections were detected among the patients. One explanation for this is the genetic background of the clones. Most clones (53.85%) were not resistant to the bactericidal effect of human serum. On the other hand, almost all clones possessed the K24 capsule, which in contrast to K1, K2, and K5 [21,71,72] is not considered as an important capsule type in the pathogenic process. Nevertheless, the ability of all clones to form a firm biofilm with the help of the K24 capsule maintains that this capsule type, together with type 3 fimbria, supports the survival of bacteria on abiotic and potentially on biotic surfaces. This survival potential of the K24 capsule type might explain why this is the most frequent capsule type linked to ST 15 [18], an otherwise pan-drug resistant widespread clone recently identified worldwide [18,19,60,[72][73][74].
Although we do not know what attributes the original ST 15 clone had, we hypothesize that after its appearance this clone underwent, and is still currently undergoing a homing process, when the bacterium adapts to survival in the hospital environment by losing some of its virulence attributes, such as the ability to survive in serum and to invade eukaryotic cells. Such a process in the hospital, community, or in the environment is an important aspect of bacterial evolution with relevance to human pathogenesis and was also formerly outlined in cases of S. aureus and L. pneumophilia [75,76]. Till now no, relevant data were found in case of Klebsiella pneumoniae. This was the reason that our study mainly focused on the divergence of the virulence attributes among isolates with common clonal origin. In addition, sequence analysis could also provide an insight into the contributing mechanisms of the carbapenem resistance in the isolates i.e., mutations in ompK35 and ompK36. Isolates were resistant to cefoxitin (Supplementary Table S2a,b), indicating that the mutations detected by WGS analysis in the ompK35 and ompK36 truly manifested.

Bacterial Isolates, Growth Conditions
Thirty-nine K. pneumoniae isolates (one isolate per patient) collected from the Microbiology Laboratories of University Hospital in Chester, England were studied (Table 4). They were recovered from different specimens: faecal (n = 21), urine (n = 11), sputum (n = 4), and blood cultures (n = 3). Identification of the isolates was performed by Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS, Microflex, Bruker Daltonics, Billerica, MA, USA) and was confirmed with the standard biochemical tests (indole, adonite, Voges-Proskauer, methyl red, citrate, malonate, urease, lysine-, ornithine-, arginine-decarboxylase, saccharose, lactose and motility tests) prior to the study.  Bacteria were routinely grown on 37 • C in Luria-Bertani (LB) broth or on eosinmethylene blue (EMB) agar (Difco, Fisher Scientific, Leicestershire, UK) plates. Special media and control strains for fimbria, biofilm, and siderophore production tests, are indicated at relevant sections. K. pneumoniae strains NTUH-K2044 [44] and MGH 78578 [45] were used as controls for each experiment. Other control strains are designated in the relevant assays.

Antimicrobial Susceptibility Testing
Antimicrobial susceptibility was determined with the disc diffusion method on Mueller-Hinton (MH) agar (Oxoid, Basingstoke, UK). The inoculum was adjusted to an optical density of 0. Antimicrobial disks were purchased from Oxoid, Hungary. E. coli ATCC 25922 was used as a control. The results were interpreted according to the guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) [77].
A new phenotypic test, called the Carbapenem Inactivation Method (CIM), was used to detect carbapenemase activity [78]. To perform the CIM, a suspension was made by suspending a full 10 µL inoculation loop of culture, taken from a Muller-Hinton (MH) agar (Oxoid, UK) plate in 400 µL Tryptic Soy Broth (TSB). Subsequently, a susceptibilitytesting disk containing 10 µg meropenem (Oxoid, UK) was immersed in the suspension and incubated for 4 h at 35 • C. After incubation, the disk was removed from the suspension using tweezers, placed on a MH agar plate inoculated with a susceptible E. coli ATCC 29522 indicator strain and incubated for 18 h at 35 • C. Inoculation of the MH agar plate with the indicator strain was done with suspension of OD 595 1.25, streaked in three directions using a sterile cotton swab. If the bacterial isolate produced carbapenemase, the meropenem in the susceptibility disk was inactivated allowing uninhibited growth of the susceptible indicator strain. Disks incubated in suspensions that do not contain carbapenemases yielded a clear inhibition zone. Each isolate was tested two times.

Chromosomal Macro-Restriction Fragment Polymorphism Analysis with Pulsed-Field Gel Electrophoresis (PFGE)
Pulsed-field gel electrophoresis (PFGE) was used to reveal the clonal relationship among the K. pneumoniae isolates. Strains were grown overnight (10 h) on LB agar at 37 • C. The ODs were adjusted to 1.3 to 1.4 (~10 8 CFU/mL) at 540 nm. Genomic DNA was prepared in low-gelling point agarose (BioRad, USA) by a procedure developed at the Centre for Disease Control (CDC) [79]. DNAs were in cube digested with XbaI (New England Biolabs, San Diego, CA, USA) restriction endonuclease for 12 h, with 10 units/mL. Separation of the fragments was performed by using the CHEF-DR II system (BioRad, Hercules, CA, USA). DNA was electrophoresed for 24 h at 14 • C in a 1.2% agarose gel (Sigma-Aldrich, St. Louis, MO, USA) at 6 V/cm with a linear gradient pulse time of 54 s. Interpretation of PFGE patterns was based on the criteria of Tenover et al. [80]. After photographing, gels were analysed and interpreted with Fingerprinting II Informatix Software (BioRad, USA). Levels of similarities were calculated with the Dice coefficient, and unweighted pair group method with arithmetic averages (UPGMA) was used for the cluster analysis of the PFGE patterns. Pulsotypes (PTs) were defined at 85% similarity between macrorestriction patterns [20].
Genetic determinants of the K24 capsule type were revealed after whole-genome sequencing of four representative strains and analysis of these sequences. Based on the wzi cluster analysis, the K24 specific primer pair (Fw: 5 -AGATAATAGG CAACAGCGTTCT-3 and Rev: 5 -GATACGTTAAA CGCCTCAAGTA-3 ) was designed by Primer-BLAST software [82].
For DNA extraction, 1.5 mL of overnight (10 h) broth culture was centrifuged at 12,000 rpm for 2 min. The pellet was resuspended in 100 µL sterile deionized water and boiled for 10 min. After a final centrifugation at 12,000 rpm for 10 min, the supernatant containing template DNA was recovered, and used for the analysis. Conditions of the amplification reactions were as follows: 95 • C for 2 min for initial denaturation, which was followed by 34 cycles consisting of 95 • C denaturation for 30 s, annealing temperatures for 30 s, and a 72 • C elongation for 1 min. A one step termination was carried out at 72 • C for 10 min. Samples were electrophoresed in 1% agarose (Invitrogen, Waltham, MA, USA) gels, stained with ethidium bromide (Acros Organics, Antwerp, Belgium), and visualized under UV light. The traditional S. cerevisiae agglutination test was used to detect the expression of the type 1 fimbriae-Mannose Sensitive Haemagglutinins (MSHA)-on the surface of the K. pneumoniae isolates. A cell suspension (37 µL; 1 × 10 9 CFU/mL) of the standard Sacharomyces cerevisiae W303 was mixed with the bacterial dilution (1 × 10 8 CFU/mL) on a glass slide, and gently rotated. To confirm the specificity, tests were also performed in the presence of α-methyl-D-mannoside (5%) (Fluka, Buchs, Switzerland) as it inhibited the specific agglutination [83]. All agglutination assays were carried out three times. K. pneumoniae strain 71 and K. pneumoniae strain 39 were used as positive controls, and K. pneumoniae strain 130 and K. pneumoniae strain 131 were used as negative controls.

Type 3 Fimbriae Assay (Mannose-Resistant Haemagglutination, MRHA)
Presence of the type 3 fimbriae mediated mannose-resistant haemagglutination on K. pneumoniae isolates was revealed by using the classical method of Podschun and Sahly [84]. Tannic acid (Fluka, Hungary) treated bovine erythrocyte suspension (Culex, Budapest, Hungary) was used. Bacteria were grown at 37 • C on brain heart infusion (BHI) agar plates (Oxoid, UK) for 24 h. Cells were collected, resuspended in physiological saline and their cell counts were adjusted to~10 8 cells/mL in physiological saline. Thirty-two µL of bacterial suspensions and 38 µL of erythrocytes were mixed on glass slide, manually rotated, and observed for 10 min at room temperature (22 • C). Agglutination was finally read after further incubation for 5 min at 4 • C [39]. All agglutination assays were carried out three times. K. pneumoniae strain 130 and K. pneumoniae strain 131 were used as positive controls, and K. pneumoniae strain 71 and K. pneumoniae strain 79 were used as negative controls.

Biofilm Assay
The biofilm forming capacities of the K. pneumoniae isolates were tested by the slightly modified crystal violet binding plate assay [85]. Bacteria were subcultured in LB broth for 12 h at 37 • C, in a shaking incubator (120 RPM). The ODs were adjusted to 0.9 to 1.0 (~10 8 CFU/Ml) at 540 nm. Twenty µL of the adjusted bacterial cultures and 180 µL LB broth were transferred to 96-well polystyrene microtiter plates (Sarstedt, Nümbrecht, Germany) and incubated for 24 h at 37 • C. Planktonic bacteria were removed with gentle washing and fixed with 2% formalin-phosphate buffered saline (PBS; Sigma-Aldrich, USA). Intensities of biofilm formations were revealed by 1% crystal violet (Sigma-Aldrich, USA) staining for 20 min at room temperature (22 • C), and by solubilizing the layer with 1% SDS (Sigma-Aldrich, USA). Extinction of the solubilized crystal violet in each well was measured at 595 nm with a FLUOstar Optima Microplate Reader (BMG Labtech, Ortenberg, Germany). Controls were performed with crystal violet binding to the wells exposed only to the culture medium without bacteria. Biofilm assays were repeated three times in three independent experiments and in each assay, quantification was performed in four separate wells. K. pneumoniae strain 703+ was used as a positive control and K. pneumoniae strain 446-was used as a negative control. Isolates were classified as high biofilm-producers (OD > 3.0), medium producers (OD 1.0-3.0) or poor producers (OD < 1.0).

Serum Bactericidal Assay
Normal human serum (NHS), pooled from healthy volunteers, was divided into equal volumes and stored at −20 • C before use. The serum bactericidal activity was measured using the method described by Podschun et al. [56], with slight modification. Bacteria were grown at 37 • C in LB broth for 24 h, with shaking at 100 rpm in an incubator. After washing with physiological saline solution, OD 600 was adjusted to 0.4. Bacteria were diluted to 2-3 × 10 6 cell/mL in physiological saline. Then, 25 µL from the bacterial suspensions and 75 µL from the undiluted NHS were mixed in the 96 well polystyrene microtiter plates (Sarstedt, Germany), and incubated at 37 • C. Samples were taken immediately after mixing and after incubation for 1 and 3 h, and serial dilutions were plated on LB agar for colony forming unit (CFU) determination. Resistance was graded by the mean 3 h survival ratio (ratio of colony count after serum treatment for 3 h compared with baseline). The highly serum resistant K. pneumoniae strain NTUH-K2044 was used as a positive control [44]. Each strain was tested three times.

Hypermucoviscosity (HMV) Testing
Single colonies obtained after overnight culture on blood agar plates were tested for their ability to form viscous strings when a standard inoculation loop was touched onto their surface and slowly raised. The formation of string greater than 5 mm in length is indicative of the hypermucoviscosity (HMV) positive phenotype [87]. K. pneumoniae strain NTUH-K2044 [44] was used as a positive control. Each isolate was tested twice.

Whole-Genome Sequencing
In order to get a detailed view of the genome organization of the ST 15 K. pneumoniae lineage, sequencing of four representative members from different specimens was carried out: K. pneumoniae strain 11/3 (isolated from faeces), strain 50/1 (isolated from blood culture), strain 53/2 (isolated from sputum) and strain 53/3 (isolated from urine). DNA was extracted from K. pneumoniae strains 11/3, 50/1, 53/2 and 53/3. DNA for whole-genome sequencing were extracted from cultures grown overnight (10 h) in LB agar, using DNA extraction kit (PureLink Genomic DNA Mini Kit, Thermo Fischer Scientific, Waltham, MA, USA) following the manufacturer's protocol. At the end of the extraction process, DNA samples were dissolved in 100 µL of sterile nuclease free water. Genomic DNA sequencing libraries were prepared using the Nextera XT Library Preparation kit (Illumina, San Diego, CA, USA). Sequencing was performed using MiSeq Reagent Kit v3 (600 cycles) on an Illumina MiSeq (Illumina, San Diego, CA, USA). Assembly of the pure sequence was performed with the MyPro pipeline [88]. Open reding frames were predicted and annotated with the Rapid Annotation using Subsystem Technology [89]. Homology searches were conducted with the BLAST tools available at the NCBI website [90].
Analysis of antibiotic resistance was performed with ResFinder 4.1 [91]. Additional software, including Pathogen Wach [92], was used for analysis of specific plasmid genetic features. Nucleotide sequences for validated Klebsiella genus virulence genes were downloaded from the Virulence Factor Database (VFDB) [93]. Nucleotide-nucleotide alignment was run using BLASTN v2.11.0 on macOS 11.6 for our Klebsiella sp. contig sequences against the downloaded virulence factor database. Our annotated sequences along with three Klebsiella pneumoniae subsp. pneumoniae reference genomes (strain HS11286, SAMN02602959; MGH78578, SAMN02603941; NTUH-K2044, SAMD00060934) were visualized using the CGView Server Beta web tool [94]. BLAST alignments for all sequences were run against the nucleotide sequence of K. pneumoniae strain 53/3 and visualized. Virulence genes for strain 11/3, 50/1, 53/2 and 53/3 identified previously using the VFDB were marked on the CGView map.

Cell Internalization Assay
Two human cell lines, the human intestinal cell line INT 407 and the bladder carcinoma cell line T24 were used to reveal the abilities of the investigated CR-Kp isolates to invade cultured cells. Invasion assays were performed essentially as described previously [95]. Briefly, semiconfluent cell monolayers were prepared (3 × 10 5 cells/well) in RPMI 1640 medium (Lonza, Verviers, Belgium) supplemented with 10% heat-inactivated (30 min for 56 • C) calf bovine serum (Sigma-Aldrich, USA), 10,000 U/mL of penicillin, 10 µg/mL of streptomycin and 0.5 mg/mL of neomycin and incubated overnight (10 h) at 37 • C in a humidified 5% CO 2 incubator. On the following day, cells were washed with PBS (pH 7) and to each well 1 mL RPMI 1640 medium (Lonza, Belgium) and bacterial suspensions (OD 600 = 1,~1 × 10 8 CFU/mL) were added to reach 10× dilution [96]. Plates were incubated at 37 • C in a humidified, 5% CO 2 incubator for 3 h. The plates were then washed three times with PBS (pH 7) to remove unbound bacteria. Fresh cell culture medium containing 100 µg/mL Polymixin B was then added to kill all extracellular bacteria and incubated for 1 h at 37 • C. Then wells were washed three times with PBS (pH 7) and lysed with 0.05% Triton X-100 (Sigma-Aldrich, Budapest, Hungary). The intracellularly surviving K. pneumoniae cell counts were determined by out-plating. All assays were performed in triplicate and were repeated independently twice. Salmonella enterica serotype Typhimurium strain ATCC14028, K. pneumoniae strain 3091 (accession number: SAMEA8948279) [97] and K. pneumoniae strain NTUH-K2044 (accession number: SAMD00060934) were used as positive controls and K. pneumoniae strain MGH78578 (accession number: SAMN02603941) was used as a negative control.

Conclusions
In this study an evolutionary process could be outlined for a CR-Kp ST 15 clone that first appeared in an English Teaching Hospital in 2010. We conclude that during this process, which likely occurred in parallel both in the hospital and the community, divergence of virulence attributes could be observed that support persistence of the original clone rather than virulence. Long term studies in hospital environments, supplemented with community data, could reveal, in the future, the changes in the virulence potentials of emerging clones, similar to changes that are studied for antibiotic resistance. To the best of our knowledge this is the first study where the rmpA gene has been detected in a CR-Kp strain belonging to the ST 15 clone.