Hypervirulent and hypermucoviscous extended-spectrum β-lactamase-producing Klebsiella pneumoniae and Klebsiella variicola in Chile

ABSTRACT Convergence of virulence and antibiotic-resistance has been reported in Klebsiella pneumoniae, but not in Klebsiella variicola. We, hereby, report the detection and genomic characterization of hypervirulent and hypermucoviscous K. pneumoniae and K.variicola recovered in Chile from health-care associated infections, which displayed resistance to broad-spectrum cephalosporins. One hundred forty-six K. pneumoniae complex isolates were screened by hypermucoviscosity by the “string test.” Two hypermucoid isolates, one hypermucoviscous K. pneumoniae (hmKp) and one K. variicola (hmKv), were further investigated by whole-genome sequencing. In vivo virulence was analyzed by the Galleria mellonella killing assay. In silico analysis of hmKp UCO-494 and hmKv UCO-495 revealed the presence of multiple antibiotic-resistance genes, such as blaCTX-M-1, blaDHA-1 and blaLEN-25 among others clinically relevant resistance determinants, including mutations in a two-component regulatory system related to colistin resistance. These genetic features confer a multidrug-resistant (MDR) phenotype in both strains. Moreover, virulome in silico analysis confirmed the presence of the aerobactin gene iutA, in addition to yersiniabactin and/or colicin V encoding genes, which are normally associated to high virulence in humans. Furthermore, both isolates were able to kill G. mellonella and displayed higher virulence in comparison with the control strain. In summary, the convergence of virulence and the MDR-phenotype in K. pneumoniae complex members is reported for the first time in Chile, denoting a clinical problem that deserves special attention and continuous surveillance in South America.


Introduction
Klebsiella pneumoniae complex includes K. pneumoniae sensu stricto, K. quasipneumoniae subsp. quasipneumoniae, K. quasipneumoniae subsp. similipneumoniae, K. variicola subsp. variicola, K. variicola subsp. tropica, K. quasivariico la, and K. africana, respectively [1]. Among members of this complex, K. pneumoniae and K. variicola have been widely recognized as important opportunistic human pathogens commonly involved in hospital-acquired infections (HAIs) [2,3]. The clinical importance of these species has been associated with multidrugresistance, mediated by the expression of extendedspectrum β-lactamases (ESBLs) and carbapenemases [4,5], and more recently with colistin resistance [6,7]. Lately, convergence of virulence and antibioticresistance has been reported in K. pneumoniae [8]. In this regard, hypervirulent K. pneumoniae (hvKp) isolates have been defined under the following criteria: i) occurrence of the hypermucoviscous (hmKp) phenotype, as det ermined by a positive "string test"; ii) presence of the rmpA gene, which regulates the capsule biosynthesis; and iii) pre sence of the aerobactin genes iucA/iutA [9,10]. Similarly to K. pneumoniae, K. variicola can also display the hypermucoviscous (hmKv) and/or hypervirulent (hvKv) phenotypes [1]. Currently, hvKp isolates have been reported mainly in Asia, Europe and North America, and more recently in South America [9], where sporadic reports have been restricted to Argentina and Brazil [10][11][12]. Hence, the aim of our study was to detect and characterize hypervirulent and hypermucoviscous ESBL-producing K. pneumoniae and K. variicola isolates recovered from Chilean hospitals.

K. pneumoniae complex isolates and antibiotic susceptibility testing
One hundred forty-six non-repetitive K. pneumoniae complex isolates collected between 2011 and 2018 in Chile, were investigated. All isolates were recovered from nosocomial infections and were initially identified by each hospital labo ratory as third-generation cephalosporin-resistant K. pneu moniae. Species identification was confirmed by conventional PCR according to previously described [13]. Antibio tic susceptibility testing to imipenem, ertapenem, meropenem, ceftriaxone, cefpodoxime, cefotaxime, ceftazidime, amoxicillin/clavulanic acid, amikacin, tobramycin, kanamycin, gentamicin, ciprofloxacin, levofloxacin and tetracycline was performed by the Kirby-Bauer method. ESBLproduction and colistin susceptibility were determined by the combined disc test and the broth microdilution method, respectively [14].

Phenotypic identification of hypermucoviscous isolates
The hypermucoviscous phenotype was determined by the "string test" [15]. In brief, when a bacteriological loop was able to generate a viscous filament ≥5 mm in length by stretching bacterial colonies growth at 37ºC by 18-24 h on a blood agar plate, the isolate was considered as positive, thus defined as hypermucoviscous. Two isolates resulted positive for the "string test," therefore, subsequent experiments included both strains.

Whole-genome sequencing (WGS) and in silico analyses of hypermucoviscous isolates
Total DNA of both hypermucoviscous isolates was extracted for whole-genome sequencing (WGS) using the Wizard® Genomic DNA Purification kit (Promega, USA) following the manufacturer's protocol. Sequencing was performed by the Illumina MiSeq platform (2 × 250 bp paired end reads) with libraries prepared by the Next eraXT kit (Illumina), with a coverage of 30x.
We studied mutations in wzc, rcsAB, and lon genes in UCO-494 utilizing the K. pneumoniae (accession nu mbers LT174540 and JCMB01, respectively) genome as reference [16]. For all mutation, bioinformatic analysis was performed using the UGENE 1.32.0 Software.

Serum bactericidal assay and virulence behavior in the Galleria mellonella infection model
Serum bactericidal activity was analyzed according to previously described [17], with minor modifications. Briefly, 250 μL of a bacterial inoculum of 5 × 10 6 CFU/ ml were mixed with 750 μL of fresh human serum. Then, viable bacterial cell count was performed in tryptone soy agar (TSA) plates. A K. pneumoniae isolate that was previously characterized as hypervirulent in our laboratory was used as positive control, while serum inactivated at 56°C for 30 min was utilized as blank. All experiments were performed in triplicate. A bacterial survival of <1% after 3 h of incubation with serum was considered as susceptible. On the other hand, survival percentages of 1-90% or >90% were considered as intermediate and resistant, respectively [18]. Additionally, in order to compare the levels of virulence of hmKp UCO-494 and hmKv UCO-495, the G. mellonella infection model was utilized [19]. K. quasipneumoniae subsp. similipneumoniae ATCC700,603 and hvKp k1/ST23 UCO-448 [10] were used as negative and positive hypervirulent controls, respectively. Larvae survival was analyzed during 96 h, and Kaplan-Meier killing curves of G. mellonella were generated using the log rank test with p < 0.05. Each assay was performed in triplicate.

Capsular-polysaccharide (CPS) quantification and estimation of capsular size
Total capsular-polysaccharide (CPS) of hmKp UCO-494 and hmKv UCO-495 was estimated according to the phenol-sulfuric acid method, after extraction using zwittergent 3-14 [20], and incubated in tryptone soy broth (TSB) at 37ºC for 18 h with agitation. The estimation of capsular size was carried out by transmission electron microscopy (TEM) of a bacterial inoculum incubated at 37ºC for 24 h [21]. Prior to microscopy, the samples were centrifuged at 3,000 rpm for 5 min and washed once with PBS buffer.

Biofilm assay
Biofilms-quantification was performed as previously described [22]. In brief, a colony from each strain was grown overnight in TSB at 37°C. From this culture, 10 μL of a bacterial suspension was used to inoculate 96-well polystyrene plates containing 90 μL of TSB, and these plates were incubated at 37°C for 24 h. Subsequently, the medium was removed from the plates and each wells was washed three times with water. Immediately, the samples were stained with 125 μL of 0.1% crystal violet for 15 min. Excess dye was removed by rinse 4 times in water, and dried during 10 min at 65°C. Afterward, 125 μL of acetic acid solution (30% v/v) were added and then incubated for 15 min at room temperature. Then, 125 μL of the solubilized crystal violet were transferred to a new 96-well polystyrene plates and color intensity was determined at a 550 nm using a spectrophotometer. K. pneumoniae ATCC 700603 strain was used as positive control, and acetic acid solution (30% v/v) was used as a negative control. Biofilm-formation abilities were defined as follows: i) absorbance values between 0.084 and 0.168 (2x -4x blank absorbance) were considered as low biofilm-forming strains; ii) values ranging between 0.168 and 0.252 (4x -6x blank absorbance) were considered as medium biofilmforming strains, whereas iii) strains displayed absorbance values higher than 0.252 (>6x blank absorbance) were classified as high biofilm formers [23].

Discussion
Traditionally, K. variicola has been considered as susceptible to most antibiotic classes, but this description has change over time, due to an increase in the MDR-K. variicola reports [1]. In South America, there is a single report in Colombia describing a KPC-2-producing K. variicola strain, which was resistant to all βlactams [5].
Worryingly, it is the emergence of hypervirulent-MDR phenotype, especially in K. variicola isolated. In this regard, Farzana R et al. describe a fatal MDR-hvKv outbreak in neonates in Bangladesh. The isolates contained a bla CTM-M-15 and bla NDM-1 genes, among others, in addition to several virulence genes like siderophore (kfuABC) and Enterobactin (entABCDEFHIJ) associated with hypervirulent phenotype [6]. On the other hand, Lu et al. described the first hvKv isolated from blood from a patient with cholangitis in China, which was resistant to colistin (MIC = 8 ug/mL) [28]. These are concordant with our study since we identified an MDR K. variicola that was resistant to colistin. In the case of K. pneumoniae, colistinresistant hvKp isolates has been reported previously. Specifi cally, Lu et al. reported five colistin-resistant hmKp strains recovered from blood samples in China [29]. Similar to our findings, these isolates were colistin-resistant and carbapenems-susceptible. Moreover, Huang et al. characterized diverse colistin-resistant hmKp isolates that were also resistant to carbapenems, since they produced the KPC-2 carbapenemase [30].
Our findings described the convergent hypervirulent phenotype and colistin-resistance in K. pneumoniae and K. variicola MDR strains. In this sense, the mutations in genes involved in colistin-resistance might be mediating this phenotype. As described previously, point mutations or deletions in pmrA or pmrB genes result in the addition of phosphoethanolamine to the lipid A [31]. Moreover, it has been demonstrated in vivo the role of PmrAB system, in which it has been associated to intra-macrophage survival and virulence in K. pneumoniae [32]. In case of hvKp UCO-494, we identified a point mutation in pmrB, similarly to the description of Lagerbäck et al., where a NDM-1-producing K. pneumoniae isolate presented an amino acid substitution in G256R in the pmrB gen [33], which was related with colistin-resistance K. pneumoniae [34]. Furthermore, it is important to highlight that the mechanism of colistinresistance in hvKv UCO-495 was mediated by chromosomal mutations in the two-component system PhoPQ, especially in the D150G substitution in PhoP. Even though mutations in these systems are associated to colistinresistance [30], general data of molecular mechanisms of colistin-resistance in K. variicola are scarce; therefore, our results describe a non-classical pmrAB and phoQ mutations in this species [7]. In this regard, we determined that these mutations are neutral according to in silico models, in consequence, in vivo studies should be performed in order to determine if they have an impact on colistinresistance [16,35].
WGS analyses reflect a widely diverse resistome. In this sense, the bla LEN-25 gene was detected in the K. variicola UCO-495 genome, which corresponds to an intrinsic-chro mosomal β-lactamase. Furthermore, we found that hvKv UCO-495 strain was resistant to cephalosporins, which might be mediated by bla SHV-12, while hvKp UCO-494 resistance was mediated by bla CTX-M-1. In this case, there are some reports of convergence of hypervirulent phenoty pe and ESBL genes in K. pneumoniae. For instance, hypervirulent and ESBL-producing have been linked to several ESBLs genes, such as bla CTX-M-14 , bla CTX-M-18 , bla CTX-M-3 and bla SHV-12 [3637,38].
In case of heavy-metal resistance genes, we found in hvKv UCO-495, the tellurium resistance genes terW and terZCD, which are related to the plasmid pKV8917 [39] in hvKp and hvKv strain [1,40]. These genes were not detected in hvKp UCO-494. Relevantly, we identified the presence of the quaternary-ammonium resistance gene emrD in K. pne umoniae UCO-494. As note, these compounds have been heavily used during the SARS-CoV-2 pandemic as disinfectants, which could have an important ecological impact on selecting MDR-bacterial isolates due to selective pressure [41].
Furthermore, Moura et al. identified a K. pneumoniae serotype K19 isolate in Brazil [10]. In this study, the authors determined that this serotype has a similar killing ability compared to hypervirulent K1-isolates [10]. Moreover, the Brazilian isolate produced the ESBL CTX-M-15, which belongs to the same group of the ESBL detected in hvKp UCO-494 isolate (CTX-M-1) [42]. These findings suggest that this serotype could be endemic to South America, where could being disseminated through the region. In addition, molecular epidemiology determined by MLST revealed that hvKp UCO-494 belonged to the ST1161, which is apparently endemic to Chile since it has been detected previously in the country [43]. In the case of hmKv UCO-495, it was designated as ST173, which corresponds to a new ST that could be endemic to this geographical area. In consequence, further epidemiological studies are needed, in order to understand their prevalence and epidemiology in South America.
In the case of siderophore production, it has been demo nstrated that yersiniabactin, salmochelin, and aerobactin are the most predominant in K. pneumoniae and K. variico la [44]. Specifically, the aerobactin system has four biosynthetic enzymes, iucABCD, and an outer membrane transporter, iutA [44]. Interestingly, epidemiological studies have shown a significant relationship between iucABCD-iutA with the hmKp phenotype; therefore, aerobactin is con sidered a substantive virulence factor in hvKp isolates [45]. However, the occurrence of multiple siderophore systems in hvKp strains suggests that siderophore systems in addition to Iuc-system play important roles in the pathogenesis of these microorganisms during either colonization or invasive processes [46].
Although all Klebsiella pneumoniae complex species could form mucoid colonies, it is well recognized the existence of two well-defined phenotypes. The classical (cKp/cKv) and hypermucoviscous (hmKp/Kv) phenotypes, both differentiated by their ability of forming a viscous and adhesive mucous string in solid media. Because of this, it is important to elucidate the mechanisms of CPS-production in hypermucoviscous K. pneu moniae strains that lack the rmpA/rmpA2 genes and do not belong to the predominant K1 or K2 serotypes [47]. In this sense, Ernst et al. studied the impact of singlenucleotide polymorphisms of the wzc gene in the capsule biosynthesis, which could confer a hypercapsule production phenotype, enhancing virulence [16]; and additionally, contribute to the resistance to polycationic peptides, such as colistin [48]. On the other hand, diverse mechanisms are related with hypercapsule production, such as mutation in wzc, rcsAB and lon protease genes [49]. Our results showed a mutation in all of this gen in hvKp UCO-494. In this sense, some authors suggest that a single amino acid substitution in wzc, rcsA or lon protease genes could increase capsule production [16], and this mechanism could be related to the hypermucoviscous phenotype in K. pneumoniae UCO-494; however, this phenomenon has not been studied in K. variicola.
In the case of virulence, the irp1 (polyketide synthetase) and irp2 (iron acquisition yersiniabactin synthesis enzyme) encode for iron-repressible high molecular weight proteins that are involved in yersiniabactin production [4]. This siderophore system was first described for Yersinia species; however, they could be also present in other Enterobactera les [50]. It is believed that its dissemination occurred via horizontal gene transfer events since the responsible genes have been identified within pathogenicity islands, such as ICEKp, which is frequently identified in K. pneumoniae [2]. The mannose-sensitive type 1 fimbriae are common in K. pneumoniae. These fimbriae are encoded by fim-like genes, in which the major components are fimA and fimH that confer its ability to adhere to human mucosal or epithelial surfaces [51]. Furthermore, other important adh esin in K. pneumoniae is the mannose-resistant Klebsiellalike (type III) codified in the fimbriae cluster mrkABCDF HIJ [52]. This is considered as a virulence factor and contri butor to mucous adherence, tissue colonization, and biofilm [53]. In our case, only UCO-494 irp1 and irp2 genes.
Importantly, biofilm-formation ability of hmKp contributes to hypervirulence, since hypervirulent strains generate more biofilms in comparison with less virulent isolates [54]. Specifically, biofilms provide protection against environmental conditions, such as desiccation, and also protect bacteria from the immune system acti on [46]. Accordingly, diverse studies associate biofilm phenotype to capsule, and/or fimbriae; however, it has been also demonstrated that the lack of capsule enhan ces biofilm-formation in K. pneumoniae [46]. Our resul ts revealed that K. pneumoniae UCO-494 presented a low biofilm-formation ability, and at the same time displayed a lower G. mellonella killing ability in comp arison to K. variicola UCO-495. Moreover, hvKp UCO-494 was susceptible to the serum activity, in contrary to hvKv UCO-495 that was resistant. However, hvKp UCO-494 produced more CPS in comparison with hvKv UCO-495, which is concordant with the bacterial-size capsule, in which hvKp UCO-494 has a thicker capsule than hvKv UCO-495. These discordant results suggest that more research is needed in order to establish the specific role of biofilm-formation and virulence in Klebsiella species. In this regard, some studies have demonstrated no significant differences in biofilm-form ation ability between invasive (more virulent) and noninvasive (less virulent) K. pneumoniae isolates [55]. In another study, K. pneumoniae mutant strains with dec reased biofilm production ability did not show any diff erence in their ability to survive serum activity, which reaffirms the need for further studies in this regard.
In conclusion, we identified the convergence of hypermucoviscous phenotype and MDR K. pneumoniae and K. variicola isolates in Chile. It is important to consider the relevance of these phenotypes since they are not normally screened by a routine laboratory. Moreover, our results demonstrate the relevance of K. variicola as pathogen, due to its antibiotic-resistance and virulence features. Moreover, our results suggest that the hypermucoviscous/hypervirulent phenotype of K. pneumoniae-complex isolates is the results of multiple mechanisms, including siderophores and biofilm-production, which have not been well elucidated yet. Our results remark the need for more detailed research of the mechanisms and epidemiology of hypervirulent strains, in order to elucidate the role of high-risk K. pneumoniae-complex lineages.