Comparative analysis of the complete genome of KPC-2-producing Klebsiella pneumoniae Kp13 reveals remarkable genome plasticity and a wide repertoire of virulence and resistance mechanisms

General genomic features of K. pneumoniae isolate Kp13

MLST analysis classified Kp13 under the ST 442, sharing a single alelle with both ST 258 and ST 11, the main clusters associated with the worldwide dissemination of KPC-producing K. pneumoniae. Of notice, clinical isolates belonging to ST 442 have been only identified in Brazil so far [10, 11].

The complete genome of K. pneumoniae isolate Kp13 consists of 5,739,888 bp of which 5,307,003 bp form a single, circular chromosome (Figure 1) and 432,885 bp are split into six plasmids which range in size from 2,459 bp (pKP13a) to 294,493 bp (pKP13f) (Table 1). All the replicons were completely closed and the putative coding sequences (CDS) were manually validated. The reported sequence data can be obtained from the NCBI (http://www.ncbi.nlm.nih.gov/bioproject) under project accession ID PRJNA78291.

The G + C content of the Kp13 chromosome is 57.5%, which is in agreement with values previously reported for NTUH-K2044 (57.7%), MGH 78578 (57.5%) and 342 (57.3%) (Table 1). However, the plasmidial sequences present a much lower G + C content of 49.1 ± 4.4% (mean and standard deviation for the six replicons) suggesting the existence of DNA acquired during events of horizontal gene transfer (HGT), an observation that is corroborated by the number of mobile genetic elements (MGEs) such as transposons and phage sequences found in these plasmids. The number of transfer RNAs (tRNAs) identified in Kp13 was 86, which is identical to those of MGH 78578 and NTUH-K2044, but different than that reported in the strain 342 that carries 88 tRNAs genes (Table 1). Regarding the chromosome size, Kp13 lies between the human pathogens NTUH-K2044 and MGH 78578 [3], while K. pneumoniae 342 presents the largest chromosome size (Table 1). This is probably due to the accumulation of genes related to its adaptation to a different ecological niche since this strain is a N₂-fixing bacterium [4]. However, when considering only plasmidial sequences, the DNA content of Kp13 (432,885 bp, six plasmids) is markedly higher than other compared strains being followed by MGH 78578 (379,774 bp distributed among its five plasmids, Table 1). In fact, as it will be further discussed in this paper, a number of Kp13’s virulence- and resistance-related genes are plasmid-mediated and their respective proteins may contribute for evasion of host defenses and survival under antimicrobial selective pressure.

Comparison of the chromosomal architecture between K. pneumoniae

The overall architecture of the chromosomes of K. pneumoniae Kp13, NTUH-K2044, MGH 78578 and 342 were studied using the Mauve tool [13], and these multiple alignments are shown in Additional file 1. This analysis evidenced an overall structural conservation and colinearity among the chromosomes of the different K. pneumoniae strains compared. The multiple alignments showed the existence of nine large locally colinear blocks (LCBs); however, it was observed that strain-specific regions were also present within LCBs (white segments in Additional file 1), which may harbor specific adaptations of each bacterium, as it will be further discussed in the plasticity segment section. These regions most likely represent DNA acquired during events of HGT and may provide a greater metabolic versatility to Kp13 strain [14]. For instance, the chromosome of N₂-fixing strain 342 was the one that most accumulated differences relative to other human pathogens (Additional file 1), such as the nif cluster located in a plasticity region unique to that strain (KPK_1696-KPK_1715) [4].

Chromosomal inversions are thought to be one of the main events of genomic rearrangements in bacteria [15]. Of notice, no recombination events leading to inversions of DNA segments occurred within the larger LCBs (Additional file 1). Only LCB2 and LCB4 (both smaller than 10 kbp) were found inverted relative to the Kp13 chromosome (Additional file 2). This observation reinforces the conserved chromosomal architecture among the compared K. pneumoniae strains since these strains are not epidemiologycally related and, in the case of 342, can be considered ecologically distinct from one another.

Regions of genomic plasticity between Kp13 and other K. pneumoniae

RGP1 carries an insertion sequence at its 5′-end (IS1 family, KP05176) and harbors CDSs with predicted catalytic domains at the protein level including two possible dioxygenases (KP02716, KP02718) and a dehydrogenase (KP02715), which may be involved in the degradation of aromatic compounds [17], although further studies will be necessary to confirm this role. RGP2 contains genes for transposases and a possible phage integrase (KP16268) as well as hypothetical genes which do not allow for any inference on a possible metabolic role. RGP3 is 41 kbp in length and represents a phage insertion in the Kp13 chromosome, with the majority of its CDSs sharing greater identity to those of Salmonella phage SPN1S [GenBank:NC_016761]. A second, large bacteriophage-related RGP was detected in the Kp13 chromosome (RGP9, Table 2) and part of this region exhibits remarkable sequence similarity to that of K. oxytoca ϕKO2 prophage [18]. Apart from phage sequences, a possible antibiotic resistance cluster that contains genes coding for transcriptional regulators (KP04065, KP04061), an antibiotic biosynthesis monooxygenase (KP04063) and a glyoxalase/bleomicin dioxygenase (KP04064) were found, but further experimental studies are warranted to establish the relevance of this region to the MDR phenotype displayed by Kp13. RGP4 is one of three plasticity segments containing genes responsible for type VI secretion system (T6SS) formation. T6SS proteins are part of a recently described apparatus that secretes toxins using a needle-like mechanism [19]. They have been previously implicated in bacterial pathogenesis such as the HSI-I locus of Pseudomonas aeruginosa associated with persistent infections by this bacteria [20]. A CDS coding for the VgrG protein was identified in RGP4 (KP01061). This protein may play both the role of an effector (influencing the target cell physiology) as well as of structural component of the T6SS where it has been reported to form part of the puncturing needle [21]. Two other T6SS-related RGPs were detected in the Kp13 chromosome inside of the RGP6 and RGP8. RGP6 contains CDSs coding for bacteriocins, proteins that target other bacteria and may provide competitive advantage to Kp13. In contrast, the third T6SS locus found in RGP8 harbors the hcp gene (KP04341) whose product, Hcp, is also part of the needle structure along with VgrG [21]. It has been shown that Hcp inhibited phagocytosis by macrophages in Aeromonas hydrophila, thus acting as an effector in modulating the host’s innate immune response [22, 23]. A noteworthy plasticity region detected in the Kp13 chromosome was located in RGP5 (Figure 2), which has been previously established in other bacteria as being an ICE (Integrative and Conjugative Element) [24]. The integration of this region into the Kp13 chromosome could have been facilitated by tRNAs (for asparagine) located at its flanking termini as has been shown for other K. pneumoniae[24]. Genes usually involved in plasmidial mobilization (mobC, KP04798 and mobB, KP04799) as well as in T4SS (KP04803-KP04813) are also present and could have aided in the mobilization of this region (Figure 2). A great part of the gene composition of RGP5 is related to that of Yersinia pestis high-pathogenicity island (HPI) which is present in many Enterobacteriaceae genera [24]. The presence of genes coding for the yersiniabactin siderophores (irp1, KP04825 and irp2, KP04826) are characteristic of this HPI, and their occurrence in more virulent K. pneumoniae strains has been previously reported [24]. The presence of this region is indicative that Kp13 is well adapted as a pathogen, since iron scavenging is important during infection, as this nutrient is poorly available in the host in its free form. Alternative siderophore systems were also found in the Kp13 genome and will be further detailed. The ICE region is absent in the clinical strain MGH 78578 as well as in the N₂-fixing strain 342 that harbors in this region a different 115 kbp insertion containing genes related to its symbiotic lifestyle. Compared to NTUH-K2044 each strain carries unique segments within this RGP: in Kp13, this unique region is 12 kbp-long and contains several uncharacterized genes (KP04788-KP04791) (Figure 2). Since this ICE is usually found in more virulent strains, the search for this region in other Enterobacteriaceae strains isolated in Brazil may help to estimate its frequency and importance in clinically relevant bacteria. RGP7 represents an insertion found only in the chromosomes of Kp13 and MGH 78578 and contains CDSs with conserved domains at their predicted protein sequence such as glycosyltransferase (KP05060), acyltransferase (KP05301), two-component histidine kinase sensor (KP05061, KP05062) and peptidase (KP05059). The presence of kexD (KP05076) in this RGP called for special attention, since this gene codes for the recently characterized multidrug efflux pump KexD, an inner membrane protein belonging to the RND family [25]. It was found to be overexpressed in a number of multidrug-resistant K. pneumoniae isolates, and confered resistance to erythromycin, tetracycline, ethidium bromide and other drugs [25]. RGP12 contains Kp13-specific fimbriae and adhesins genes (KP02150-KP02154) and this insertion occurs after a tRNA gene for threonine. The aco operon (acoKABCD, KP32150; KP02898-KP02901) involved in acetoin catabolism [26] was also found in this region. It was detected in all compared strains, except for MGH 78578.

The large number of RGPs detected in Kp13 relative to the strains MGH 78578, NTUH-K2044 and 342 indicate a complex pattern of gain and loss of genetic material segments that occurred during the shaping of each of the compared chromosomes throughout their evolutionary history. It is thought that such events occur more frequently in pathogenic than in non-pathogenic bacteria, the genomes of the first group being regarded as more unstable and rearrangement-prone [27].

Unique and shared genes between the compared K. pneumoniae

We have analyzed the group of shared and unique genes between Kp13, MGH 78578, NTUH-K2044 and 342 to obtain insights that might explain their distinct virulence and pathogenic features.

The results from this analysis are shown in Figure 3 and in Additional file 4. As expected, the majority of the coding sequences from the compared organisms are part of a conserved genomic ‘core’ comprising 4,269 CDSs (Figure 3). Of these, 98.8% are chromosomally located in Kp13. As the chromosomes are generally more stable replicons, it is not surprising that most of the conserved genes were detected in this genetic replicon. Most of the so-called ‘accessory’ genes, which are present or not in bacterial strains, were found in plasmids of the compared K. pneumoniae. In Kp13, 40% of its unique genes (region XIV in Figure 3) were encoded in plasmids. The largest part of these unique genes (693 CDSs) could not have any functional clues attributed and were annotated as ‘hypothetical’ (Additional file 4), highlighting the need for follow-up studies focusing on whether these genes are in fact expressed and their possible roles.

Among the shared genes found only in clinical isolates (region III in Figure 3) we detected CDSs whose products are related to phosphonate utilization (phn cluster, KP00091 and adjacent genes in Kp13). These were previously studied in Klebsiella and may provide means to use alternative phosphorus sources such as aminoethylphosphonate, ethylphosphonate and methylphosphonate [28]. A CDS coding for D-aminoacylase (dan, KP32221) was also detected, and shared 49% identity with the recently characterized protein from Achromobacter xylosoxidans subsp. denitrificans (BLASTP against [Swiss-Prot:P72349]). This enzyme is involved in the conversion of N-acyl-D-amino acids to D-amino acids and fatty acids [29]. CDSs annotated as sulfatases were also found only in the clinical K. pneumoniae studied (Kp13 loci KP03498, KP03499), and their products may play a role in sulphate utilization.

Most of the unique genes found in Kp13 (region XIV in Figure 3) are related to MGEs such as transposases, insertion sequences and phage genes (Additional file 4). There are also CDSs whose products are involved in capsule formation (cps cluster) such as glycosyltransferases (KP03791, KP03802 and KP03803), which are unique to Kp13. The structure of this cluster was previously studied and belonged to a hitherto unreported capsular serotype [12]. CDSs whose predicted proteins have similarity to fructosamine deglycase (KP01294, 40% BLASTP identity to Bacillus subtilis [Swiss-Prot:O32157]) and fructosamine kinase (KP01295, 43% identity to [Swiss-Prot:O32153]) were also unique to Kp13. These enzymes are usually involved in the utilization of Amadori products, and fructosamine itself is produced via the condensation of glucose with amine compounds [30]. We formulated the hypothesis that, if indeed the products of these CDSs play a role in fructosamine metabolism as energy source, they could provide the Kp13 isolate with an adaptive advantage in the nosocomial setting, since Amadori products are intimately related to advanced glycation end-products (AGEs). In turn, AGEs are produced as result of poorly controlled chronic diseases such as diabetes [31], and from this perspective these compounds could enter the bacterial cell through an adjacent transporter coded in CDS KP01297 and be converted into glucose by the aforecited enzymes, providing Kp13 with an alternative energy source within the compromised host. In fact, Kp13 was recovered from a blood culture of a diabetic patient. However, additional studies, using animal models, are required to establish the occurrence of increased frequency and/or exacerbation of Kp13-infection in hyperglycemia. Another set of unique genes found in Kp13 included the cassettes sat (KP32246, located in plasmid pKP13d), coding for streptothricin acetyltransferase involved in the resistance to streptothricin antibiotics in the context of a class 2 integron, and a CDS whose product may be related to dihydrofolate reductase activity (KP31590, located in plasmid pKP13f within a class 1 integron), which may confer resistance to trimethoprim. The latter will be further discussed in the context of bla_CTX-M-2.

Virulence- and resistance-related ‘weapons’ identified in Kp13

Enzymatic inactivation of antimicrobials

The resistance determinant bla_SHV gene, which codes a narrow-spectrum class A beta-lactamase known to confer the intrinsic resistance to amoxicillin, ampicillin, ticarcillin and carbenicillin, was identified in Kp13 [41]. Indeed, both K. pneumoniae MGH 78578 and NTUH-K2044 genomes contain a bla_SHV-11 copy, encoding the narrow-spectrum SHV variant. Kp13 showed two copies of bla_SHV genes: bla_SHV-12 (found in the plasmid pKP13f) encoding an extended-spectrum beta-lactamase (ESBL); and a chromosomally-encoded bla_SHV-110 gene coding for an enzyme whose spectrum of activity has not been characterized yet (Table 3). The bla_SHV-12 copy was located downstream an IS26 element (KP05944), while bla_SHV-110 was found downstream the lac operon and was followed by an uncharacterized HTH-type transcriptional regulator CDS (KP01127).

Product	Gene	CDS	Genetic element	Best hit (% identity*, isolation country)
Beta-lactamase SHV-12	bla SHV12	KP32248	pKP13f	K. pneumoniae MGH 78578 (100%, USA)
Beta-lactamase OXA-9	bla OXA9	KP01389	pKP13f	K. pneumoniae FC1 (100%, Argentine)
Beta-lactamase TEM-1	bla TEM-1	KP01391	pKP13f	Neisseria gonorrhoeae (100%, nd)
Beta-lactamase CTX-M-2	bla CTX-M-2	KP03128	pKP13f	Salmonella typhimurium CAS-5 (100%, Argentine)
Beta-lactamase SHV-110	bla SHV110	KP31849	Chr	K. pneumoniae FSP 237/05 (100%, Brazil)
Beta-lactamase KPC-2	bla KPC-2	KP06703	pKP13d	K. pneumoniae 15 (100%, USA)

*,BLASTP against the NCBI database; Chr, chromosome; nd, not determined; pKP13[d/f], plasmid pKP13d/pKP13f.

Several acquired beta-lactamase-encoding genes were identified in plasmids pKP13f and pKP13d of Kp13, namely bla_TEM-1, bla_OXA-9, bla_CTX-M-2, and bla_KPC-2 (Table 3), genes that if properly expressed are enough to confer resistance to all beta-lactam agents. bla_TEM-1 encodes a class-A beta-lactamase with limited spectrum of activity, that confers resistance against penicillins and first-generation cephalosporins and whose activity is inhibited by clavulanic acid [42]. The bla_OXA-9 gene probably encodes a class-D beta-lactamase with narrow spectrum of activity, although its complete biochemical characterization remains to be determined. OXA-9 is inhibited by clavulanic acid and cloxacillin, but differently from most OXA enzymes is not inhibited by NaCl, conferring a phenotype typical of class A enzymes production [43]. The bla_CTX-M-2 gene encodes an ESBL commonly observed in Enterobacteriaceae isolates from humans, swine and poultry manure, although this gene has also been identified in P. aeruginosa, Vibrio cholerae and Acinetobacter baumannii clinical isolates [44]. In Kp13, bla_CTX-M-2 is found downstream a class 1 integrase CDS (KP03119) and adjacent to the Orf513-coding gene (KP03126), also named ISCR1, which possibly mediates the dissemination of this gene [45]. As previously described [45], this integron contains a duplication of its 3′ segment where truncated qacE (between CDSs KP18328 and KP03130) and sul1 (KP03130 and KP03134) were found. The overall architecture of the bla_CTX-M-2 region in Kp13 is closer to that described for the class 1 integron harboring bla_CTX-M-59 (termed In0506) recovered from patients in São Paulo, Brazil [46], with which the Kp13 sequence shares 99% identity (BLASTN against GenBank accession no. EU622856), and the difference relies on the non-synonymous change in bla_CTX-M-2. This region also carries resistance determinants to other antimicrobial classes, including the genes dfrXVb (KP31590), encoding an alternative dihydrofolate reductase that is associated with resistance to trimethoprim and a cmlA- like (KP18323), which codes for an efflux pump related to chloramphenicol resistance [47]. Of special interest in Kp13 was the detection of bla_KPC-2 (KP06703, found in pKP13d) gene coding for the K. pneumoniae class A carbapenemase (KPC-2). The expression of KPC-2, especially when associated with porin alterations, can give rise to resistance against all beta-lactams. The genetic context of bla_KPC-2 includes an upstream insertion sequence (ISKpn6, KP06702), as well as a resolvase CDS (Figure 4). It is of notice that no inverted repeats related to Tn4401 were detected, since this transposon is frequently associated with the mobilization of the carbapenemase gene [48]. The absence of such signals can be regarded as evidence that the spread of bla_KPC-2 also occurs solely by recombination events involving ISKpn6 and the flanking Tn3-family sequences.

Interstrain variation at the nucleotide level may relate to lifestyle-specific adaptations

SNPs were determined for three comparisons using the complete chromosome sequence of Kp13 as reference aligned against: i) the complete chromosome sequence of strain MGH 78578, ii) complete chromosome sequence of strain NTUH-K2044 and iii) complete chromosome sequence of strain 342. The number of SNPs falling within non-coding and coding regions (non-synonymous, synonymous and non-sense) between each comparison are listed in Figure 5A. It can be seen that strain 342 accumulated a total number of mutations much larger than the other two strains in comparison to Kp13 (data not shown). This result is in accordance with the proposed hypothesis that this strain should be reclassified as K. variicola[62]. Then, we investigated the common SNPs identified in all CDSs of Kp13 (grouped according to their COG functional category) against both pathogenic strains, MGH 78578 and NTUH-K2044 (Figure 5B and Additional file 9). These SNPs represented unique Kp13 alleles relative to the latter two human pathogenic strains. Disregarding uninformative categories such as general function prediction and unknown function, the ones which accumulated more SNPs were E (amino acid transport/metabolism) and G (carbohydrate transport/metabolism). Also, 124 SNPs were identified within CDSs related to category V (defense mechanisms). Subsequently, in order to understand the significance of SNPs in relation to the Kp13 lifestyle, we investigated all SNPs falling within CDSs related to virulence and resistance. Notably, most SNPs (141 in total) were identified within CDSs involved in drug resistance (MDR/DR) (Figure 5C, Additional file 7). They are followed by CDSs whose products function as OM proteins, accounting for 42 mutations, such as the CDS for an OmpA domain-containing protein, which carries four non-synonymous mutations (out of a total of 9) (Additional file 7). However, the role of non-synonymous mutations found among CDSs involved in drug resistance in K. pneumoniae remains unknown.

The SNPs analysis performed in Kp13 relative to the two other human pathogenic strains has shown that there are Kp13-specific alleles occurring in many functional COGs, suggestive that these SNPs could result in a variety of diverse phenotypic effects. Moreover, the amount of SNPs found in CDSs associated with drug resistance may give additional clues on the distinctness of Kp13 compared to MGH 78578 and NTUH-K2044. An interesting follow-up for this study would be to analyze the relationship between the amino acid sequence substitutions in these proteins and their corresponding structural alterations, which could relate to the resistance phenotype of Kp13. As an example, non-synonymous mutations in possible MDR-related genes were detected, such as in mdtB (respective changes at the protein level Ile93 → Met, Ala352 → Thr, Arg526 → His, considering the Kp13 CDS KP04638 as reference) and an acrB homolog (Thr511 → Ala, Val542 → Ala) (Additional file 7).

Finally, the results from the SNPs analysis shed light on the plasticity of the Kp13 genome at the nucleotide level, which complement the observed plasticity at the broader level of genomic segments, discussed within the context of RGPs.

Ethics statement

The Ethics Committee of the Universidade Estadual de Londrina (UEL) approved the present study under reference number CAAE:3356.0.000.268-09. Clinical evaluation and blood sampling were performed after diagnostic routine procedures in the intensive care unit of the Hospital Universitário (UEL) with written informed consent of the patient.

Bacterial strain

Klebsiella pneumoniae isolate Kp13 was obtained from the blood culture of a patient admitted to the intensive care unit with diabetes mellitus and cranial encephalic trauma [12]. This bacterium was isolated during an event of nosocomial outbreak due to KPC-2-producing K. pneumoniae that occurred at the Hospital Universitário (UEL) between February and May 2009. The antimicrobial susceptibility profile of Kp13 was confirmed by broth microdilution. Antimicrobials tested included ampicillin, amoxicillin/clavulanic acid, piperacillin/tazobactam, cefoxitin, ceftriaxone, ceftazidime, cefepime, imipenem, meropenem, ertapenem, gentamicin, amikacin, levofloxacin, polymyxin B and minocycline, and tigecycline. MICs were determined and interpreted following the CLSI guidelines for Enterobacteriaceae[63, 64] except for tigecycline and polymyxin B where FDA [65] and EUCAST [66] breakpoints were applied, respectively. Kp13 was susceptible to tigecycline (MIC, 0.25 mg/L) and minocycline (MIC, 2 mg/L), and displayed resistance to ampicillin (MIC, >32 mg/L); amoxicillin/clavulanic acid (MIC, >32/16 mg/L), piperacillin/tazobactam (MIC, >128/4 mg/L), cefoxitin (MIC, >32 mg/L), ceftriaxone (MIC, >64 mg/L), ceftazidime (MIC, >32 mg/L), cefepime (MIC, >32 mg/L), ertapenem (MIC, ≥1024 mg/L), imipenem (MIC, 64 mg/L); meropenem (MIC, 64 mg/L), amicacin (MIC, >64 mg/L), gentamicin (MIC, 256 mg/L) and polymyxin B (32 mg/L). In addition, this isolate showed the carbapenemase producer phenotype and was confirmed to carry the gene encoding KPC-2 [12].

DNA sequencing, genome assembly and annotation

Both shotgun and 3 kb paired-end library were constructed and the genome sequencing of K. pneumoniae isolate Kp13 was carried out using the Genome FLX sequencer (454 Life Sciences/Roche), as previously described [12]. Genome assembly was performed using Newbler v 2.6 (Roche) and Celera genome assembly v 6.1 (JCV Institute). Gaps within scaffolds resulting from repetitive sequences were resolved by in silico gap filling. We achieved mean sequence coverage of 111× for this genome. The SABIA pipeline [67] was used for gene prediction and automatic annotation followed by manual validation of each predicted CDS.

MLST analysis

Multilocus sequence typing was performed as previously described [68]. Briefly, the trimmed nucleotide sequences of seven housekeeping genes namely rpoB, gapA, mdh, pgi, phoE, infB and tonB were analyzed using the Klebsiella pneumoniae MLST Database, available at http://www.pasteur.fr/recherche/genopole/PF8/mlst/Kpneumoniae.html. For each allele was given a number that combined yielded the sequence type (ST). The goeBURST algorithm was employed to generate a minimum spanning tree using the information of isolates available and the PHYLOViZ tool [69]. Clonal complexes included STs sharing at least 5 identical alleles between their representatives.

Comparative genomic analyses

The MicroScope platform [70, 71] provided some of the tools used for the comparative genomic analyses, such as the RGPFinder and Phyloprofile modules for determination of regions of genomic plasticity (RGPs) and unique/shared gene content identification, respectively. RGPs are defined as DNA segments over 5 kbp possibly related to events of horizontal exchanges, and they are identified using a series of constraints such as G + C% deviation, compositional biases as measured by the tools Alien_Hunter [72] and SIGI-HMM [73], synteny breaks and proximity to tRNAs. In these analyses, the Kp13 chromosome was set as the reference and RGPs were searched relative to those of strains 342, MGH 78578 and NTUH-K2044. Each identified RGP was manually inspected based on its genomic context and conserved genes located at the flanks of the region. Elements characteristic of horizontal transference events such as tRNAs, transposases and prophage sequences were also searched for within each region. EasyFig [74] was used to generate comparative figures of specific regions, while the BLAST Ring Image Generator 0.95 [75] was used to plot the chromosome map as well as arrange and visualize the RGPs within the overall Kp13 chromosome context.

In order to identify unique and shared gene content, the PhyloProfile Exploration tool within MaGe was used, and the following criteria from Jenssen et al. [76] were adopted to define orthology between two translated open reading frames: (i) the smaller protein covering at least 80% the length of the larger (minLrap ≥ 0.8); (ii) protein identity ≥ 35%. A Venn diagram of the unique/shared gene content was generated with a custom R script using the package VennDiagram [77].

Chromosomal architecture comparison between K. pneumoniae strains were carried out using the progressiveMauve algorithm implemented in Mauve 2.3.1 [78] at its default parameters. For this analysis, sequences from the chromosome of strains 342 [GenBank:CP000964], MGH 78578 [GenBank:CP000647] and NTUH-K2044 [GenBank:AP006725] were downloaded from the NCBI website and used as input to Mauve.

Phage-related sequences were searched using PHAST [79], while T4SS-related sequences were scanned for using the AtlasT4SS database [80].

The investigation of the resistance and virulence repertoire was facilitated by BLASTX sequence comparison to the Antibiotic Resistance Genes Database [81], as well as throughout the manual annotation of the Kp13 genome, when CDSs related to these functions were flagged. Insertion sequences were characterized using the ISFinder database (http://www-is.biotoul.fr/) [82].

SNP analysis was carried out as previously described [83]. Firstly, three primary SNP-call sets were generated from comparisons between Kp13 and the strains MGH 78578, NTUH-K2044 and 342. Then, scripts in PERL were written to function as quality filters to acquire the most reliable SNP-call sets for the comparisons. Alignments were made using the NUCmer algorithm from the MUMmer v.3.0 package [84]. SNPs were also classified regarding their location on intergenic region or CDS (with its corresponding annotated function and related COG category) [83].

Nucleotide sequence accession numbers

The NCBI BioProject accession of K. pneumoniae Kp13 is PRJNA78291. The sequences reported here have been deposited in the GenBank database (http://www.ncbi.nlm.nih.gov/Genbank) under accession number [GenBank: CP003999] for the chromosome. The six plasmids are available under accession numbers [GenBank:CP003996] (pKP13a), [GenBank: CP003994] (pKP13b), [GenBank: CP003995], (pKP13c), [GenBank: CP003997] (pKP13d), [GenBank: CP003998] (pKP13e) and [GenBank: CP004000] (pKP13f).