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

Background Klebsiella pneumoniae is an important opportunistic pathogen associated with nosocomial and community-acquired infections. A wide repertoire of virulence and antimicrobial resistance genes is present in K. pneumoniae genomes, which can constitute extra challenges in the treatment of infections caused by some strains. K. pneumoniae Kp13 is a multidrug-resistant strain responsible for causing a large nosocomial outbreak in a teaching hospital located in Southern Brazil. Kp13 produces K. pneumoniae carbapenemase (KPC-2) but is unrelated to isolates belonging to ST 258 and ST 11, the main clusters associated with the worldwide dissemination of KPC-producing K. pneumoniae. In this report, we perform a genomic comparison between Kp13 and each of the following three K. pneumoniae genomes: MGH 78578, NTUH-K2044 and 342. Results We have completely determined the genome of K. pneumoniae Kp13, which comprises one chromosome (5.3 Mbp) and six plasmids (0.43 Mbp). Several virulence and resistance determinants were identified in strain Kp13. Specifically, we detected genes coding for six beta-lactamases (SHV-12, OXA-9, TEM-1, CTX-M-2, SHV-110 and KPC-2), eight adhesin-related gene clusters, including regions coding for types 1 (fim) and 3 (mrk) fimbrial adhesins. The rmtG plasmidial 16S rRNA methyltransferase gene was also detected, as well as efflux pumps belonging to five different families. Mutations upstream the OmpK35 porin-encoding gene were evidenced, possibly affecting its expression. SNPs analysis relative to the compared strains revealed 141 mutations falling within CDSs related to drug resistance which could also influence the Kp13 lifestyle. Finally, the genetic apparatus for synthesis of the yersiniabactin siderophore was identified within a plasticity region. Chromosomal architectural analysis allowed for the detection of 13 regions of difference in Kp13 relative to the compared strains. Conclusions Our results indicate that the plasticity occurring at many hierarchical levels (from whole genomic segments to individual nucleotide bases) may play a role on the lifestyle of K. pneumoniae Kp13 and underlie the importance of whole-genome sequencing to study bacterial pathogens. The general chromosomal structure was somewhat conserved among the compared bacteria, and recombination events with consequent gain/loss of genomic segments appears to be driving the evolution of these strains.

This region is not found in K. pneumoniae NTUH-K2044, although there are evidences that it could have been lost during the evolutionary course of this genome. In NTUH-K2044, the CDS KP1_4647 is located adjacent to speB, the conserved gene that flanks the 5' terminal of this region in all strains compared. KP1_4647 is composed of 153 nt and was annotated as a possible membrane protein with the best hits in MGH 78578 and Kp13 being, respectively, KPN_03359 and KP02704, both having 1,269 nt in size. We hypothesize that this region suffered deletion in NTUH-K2044 and that this event led to the truncation of this CDS which resulted in its smaller size relative to the other studied K. pneumoniae. On the other hand, in the chromosome of strain 342 (where this RGP is also absent) the gene for this membrane protein is found intact and is 1,269 in length (KPK_0739).

RGP2
This plasticity segment contains transposases (KP02533, KP02534), a phage integrase (KP16268) as well as CDSs annotated as hypothetical. The conserved genes that flank this region at its 5' and 3' termini are, respectively, mutS and hmuV. In strain 342 and NTUH-K2044 they are contiguously localized and this region represents an unique insertion within the Kp13 chromosome. In the chromosome of MGH 78578 these genes are not found adjacent and there exists two CDSs between them, which were annotated as an aminoacyl-histidine dipeptidase (locus ID KPN_03092) and tagaturonate reductase (KPN_03093).

RGP3 and RGP9 (bacteriophage-related)
These RGPs represent probable phage insertions in the Kp13 chromosome. In RGP3, identical att sites were detected flanking the termini of the region (attL/attR, ATTGAGTGGGAAT) where recombination between phage and bacterial DNA probably occurred. The guaA gene (locus KP00870 in Kp13) is conserved among the compared strains and flanks the 5' region of this RGP, and it has been previously reported that the attachment site that follows this gene functions as recombination spot between Salmonella species and ε15 phages [3] and the same mechanism might have occured in K. pneumoniae Kp13 and related strains. There are no insertions in this region in the chromosomes of strains NTUH-K2044 and MGH 78578, while the analog region in K. pneumoniae 342 harbors a distinct 12.9 kbp insertion that may represent an integrated plasmid carrying genes related to plasmidial replication and mobilization [4].
The CDSs predicted in this region code for many hypothetical proteins as well as methylases (KP00873, KP00885) which could have a role in protecting the inserted DNA against the action of bacterial restriction enzymes. The phage genes identified in RGP3 using the PHAST webserver code for multiple bacteriophage-related functions, including tail and capsid formation, fibers, proteases and terminases.
The second bacteriophage-related RGP was designated RGP9. At its initial portion a variety of CDSs unrelated to phage metabolism are found, and their predicted domains reveal a possible signal transduction role (KP04084) as well as enzymatic activities such as oxidoreductase (KP04069), phosphohydrolase (KP04085) and asparaginase (KP04086). Also in this region are two CDSs displaying high amino acid identity to CioA and CioB from Pseudomonas aeruginosa, KP04075 (70.95% BLASTP identity to UniProtKB accession O07440) and KP04076 (66.07% identity to accession O07441). In P. aeruginosa, these proteins are cyanide-insensitive terminal oxidases which could provide a way to sustain aerobic respiration in the presence of cyanogenic compounds [5]. If these proteins are indeed expressed by Kp13 and related K. pneumoniae it would offer a mechanism to compete against cyanide-producing organisms. The central and 3' terminal part of RGP9 contain phage-related genes and their insertion occurs next to the conserved icd gene (KP04861) which lies outside the boundaries of this plasticity segment. Previous reports indicate that the icd gene in Escherichia coli harbors an integration spot at its 3' terminal which facilitates prophage insertion through homologous recombination [6,7]. Most CDSs in this region display similarity to the ones found in plasmidial prophage ϕKO2 found in K. oxytoca. However, the coverage of the Kp13 region to this prophage is limited to the segments responsible for bacteriophage head and tail formation, as the initial portion of ϕKO2 contains genes involved in plasmid mobilization. In strains 342 and NTUH-K2044 non-identical phage insertions also took place, and while the regions present in 342 and Kp13 share limited similarity, the inserted elements at each one of them have distinct organization and are probably the result of independent recombination events. The NTUH-K2044 region also lacks the initial segment that is found in Kp13, 342 and MGH 78578 containing CDSs unrelated to phage metabolism.

RGP4, RGP6 and RGP8 (T6SS-related)
The DNA segment comprised within RGP4 is the first of three identified T6SS in the Kp13 chromosome. RGP4 is flanked in its 3' terminal by an Arg-tRNA (located between coordinates 1,481,583-1,481,657) and it contains multiple CDSs with predicted T6SS-related domains, such as IcmF (KP01067) and DotU (IcmH, KP03057). Although both proteins have been initially classified as belonging to the T4SS, being homologous to the Icm/Dot archetype of Legionella pneumophila, they are both equally found in t6ss [8,9]. Only two (ClpV and FHA domain-containing protein [COG3456]) from a total of 14 proteins described by Bingle et al. [9] as the most conserved in this secretion system could not be identified in this region. ClpV is the energizing component of this system, while FHA-domain homologs regulate protein secretion via phosphorylation of its threonine residues [10]. However, a ClpV homolog (KP04342) was identified at a distinct T6SS region in Kp13 and could possibly be used to complement the expression of this system in RGP4. The insertion of this region may have been facilitated by the action of transposases which were found in this region (KP32241, KP32243).
The presence of this T6SS locus in all compared strains serves as evidence that this region might have been transferred early during the K. pneumoniae evolution and have since suffered reorganization within the host chromosomes. Similarly to the T6SS locus-III in strain MGH 78578, this T6SS segment also harbors bacteriocin genes whose predicted protein sequence contain pyocin/cloacin domains (IPR016128) such as KP01496 and KP01501/KP01504, although the latter are found truncated due to the insertion of an IS family 3 belonging transposase. CDSs with a possible immunity role against self bacteriocins are also found in this RGP (KP01494, KP01495, KP01500).
Thus, the T6SS found in RGP6 could act conferring Kp13 an advantage on competing with related bacteria through the synthesis and export of bacteriocins. An HTH-type regulator (KP04175) located in the opposite strand to the genes supposedly responsible for the enzymatic activities is present, as well as a possible transporter (serine transporter, KP04173). The presence of these genes might indicate a possible orchestrated role for this region that could provide Kp13 and related K. pneumoniae (but not strain 342) with metabolic distinctness. Adjacent to the transposase is an CDS (KP04178) annotated as related to β-lactamases due to a conserved domain, although direct evidence is needed in order to confirm whether in fact this gene is expressed and the properties of its product.

RGP11
As some of the previously identified RGPs, this region also has a syntenic break relative to the K. pneumoniae 342 chromosome. An insertion sequence is found in this region (KP31763), as well as CDSs with a possible role in sugar transport possibly belonging to the phosphotransferase system family (KP03465-KP03468). Adjacent to these CDSs two sequences with sugar isomerase domains were detected (KP03469, KP03470). It is known that the use of alternative carbon sources represents a selective survival advantage to bacteria, but domain analyses solely do not allow for inferences on the exact carbohydrate that could be transported by these proteins.

RGP12
The insertion of this region occurs near a tRNA for threonine which is immediately followed by an CDS coding for a phage integrase family protein (KP02143) having 91% identity with a sequence from E. coli SMS-3-5 (BLASTN, [GenBank:ACB17891]). Multiple transposases are present throughout the region and reveal a complex mosaic structure with Kp13-specific segments as well as shared regions between the compared strains. CDSs coding for fimbrial proteins and adhesins, specific to Kp13, are found in RGP12 (KP02150-KP02154). During the infection process, adhesins bind to specific molecules of the host and thus have been proposed as vaccine targets [12]. The region where RGP12 was inserted may serve as recombination spot in K. pneumoniae since in the other compared bacteria the region following the flanking tRNA contain different insertions having only few conserved segments. For instance, in MGH 78578 there is a 25 kbp insertion, while in NTUH-K2044 and 342 the length of these insertions are, respectively, of 23 and 60 kbp (counting from the conserved flanking genes proA up to csiD). At the 3' terminal of RGP12 we found the aco operon (acoKABCD, KP31250; KP02898-KP02901) involved in acetoin catabolism, this compound being the product of the fermentative metabolism of many microorganisms [13]. These genes are found in Kp13, NTUH-K2044 and 342, but not in MGH 78578, which raises the question of whether the specific segment where aco is located was lost in that strain or simply not acquired through HGT events.

RGP13
Similar to the previously described region, the insertion of RGP13 occurs after a tRNA (for leucine), also followed by an CDS with similarity to phage integrases.(KP32364).
Most of the CDSs identified in this region are unique to Kp13 relative to the compared strains, and include an acid phosphatase (KP01195) followed by a membrane protein (KP01196) which carries a porin domain (InterPro accession IPR001702). This protein shares 35% identity with the outer-membrane cytotoxic protein ComP from Pleiomonas shigelloides (BLASTP, [SwissProt:A0JCJ5]), which was described as able to induce apoptosis in human cells [14]. Multiple ABC-related transporter proteins are found in this RGP, and their coding genes are KP01205, KP32165, KP01200 and KP01201. Also detected were many hypothetical genes, restriction enzymes, insertion sequences and transposases, and it is probable that this region suffered extensive structural reorganization.