Antimicrobial Susceptibility, Virulence, and Genomic Features of a Hypervirulent Serotype K2, ST65 Klebsiella pneumoniae Causing Meningitis in Italy

The rise of a new hypervirulent variant of Klebsiella pneumoniae (hvKp) was recently reported, mainly linked to the ST23 lineage. The hvKp variants can cause severe infections, including hepatic abscesses, bacteremia, and meningitis, with a particularly disconcerting propensity to cause community-acquired, life-threatening infection among young and otherwise healthy individuals. The present study aimed to report the clinical characteristics of a hypermucoviscous K. pneumoniae strain isolated in Italy and sustaining recurrent meningitis in a patient of Peruvian origin. A further objective was to retrospectively investigate, by means of whole-genome sequencing (WGS) analysis, the genomic features of such an isolate. The hypermucoviscosity phenotype of the strain (sk205y205t) was determined using the string test. Genomic information was obtained by WGS (Illumina) and bioinformatic analysis. Strain sk205y205t was susceptible to most antibiotics, despite the presence of some resistance genes, including blaSHV-11, blaSHV-67, fosA, and acrR. The isolate belonged to ST65 and serotype K2, and exhibited several virulence factors related to the hvKp variant. Among these, were the siderophore genes entB, irp2, iroN, iroB, and iucA; the capsule-regulating genes rmpA and rmpA2; and the type 1 and 3 fimbriae fimH27 and mrkD, respectively. A further operon, encoding the genotoxin colibactin (clbA-Q), was also identified. The virulence plasmids pK2044, pRJA166b, and pNDM. MAR were also detected. Phylogenetic investigation showed that this Italian strain is highly similar to a Chinese isolate, suggesting a hidden circulation of this hvKp ST65 K2 lineage.


Introduction
The rise of a new hypervirulent variant of Klebsiella pneumoniae (hvKp) was recently reported. The hvKp variant was first described in Taiwan, causing invasive pyogenic liver abscess [1,2], but can also be responsible for other invasive diseases, including abscesses at other body sites (e.g., eyes, brain, prostate, and kidney), necrotizing fasciitis, severe pneumonia with bacteremia, and meningitis [3]. Although hvKp infection appears to often occur in diabetic patients, a particularly disconcerting problem is its ability to cause community-acquired, life-threatening infection among young and healthy individuals [4].
The hvKp strains often present colonies on agar plates with the peculiar aspect of hypermucoviscosity, detectable semi-quantitatively by a positive 'string test'; a widely Antibiotics 2022, 11, 261 2 of 14 used method for hvKp identification [5]. Usually, hvKp variants are susceptible to most antibiotics, except inherently resistant ampicillin, while several virulence factors have been identified [6]. The majority of hvKp show K1 or K2 capsular serotypes, even though non-K1/K2 serotype hvKp strains have also been reported and some K1/K2 strains are non-hvKp. Several sequence types (ST) have shown to be associated with hvKp, with ST23 being the most represented, related to the K1 serotype, followed by ST83 and ST65, associated with the K2 serotype [3].
An increasing number of studies are aiming at deciphering the virulence factor content and the specific feature of this K. pneumoniae variant. According to the genomic data, several virulence factors are tightly associated with the hvKp variant: the iron acquisition systems salmochelin (iroBCDN)/aerobactin (iucABCDiutA), the regulator of mucoid phenotype A gene (rmpA/A2), magA, kfu, fimH, wabG, uge, allS, a PhoPQ-activated integral membrane protein (pagO) and entB are the main virulence factors [7,8]. Usually, the virulence genes iroBCDN, iucABCDiutA, and rmpA/A2 are carried on the virulence plasmid pK2044. Their presence was significantly associated with K. pneumoniae causing invasive human infection [9].
The present study aimed to report the clinical characteristics of a hypermucoviscous K. pneumoniae strain causing meningitis, and to retrospectively investigate, by means of whole-genome sequencing (WGS) analysis, the genomic features of such an isolate.

Case Description
On June 2021, a 75-year-old man of Peruvian nationality was admitted at the Medical Department of the ASST of Lecco (Italy). Based on clinical and laboratory findings, including fever, headache, sleepy state, and cerebrospinal fluid analysis, the patient was suspected of meningitis and empirically treated with ceftriaxone (4 g daily, bid), ampicillin (12 g daily, tid), and vancomycin (2 g daily, bid). Microbiology cultures showed both urinary tract and bloodstream infections caused by K. pneumoniae. Thus, treatment was changed to meropenem (6 g, tid) and linezolid (1200 mg bid). On August, following complete relief of signs and symptoms, the patient was discharged with diagnosis of decapitated meningitis and hyposodemia. One month later, on September 2021, the patient was readmitted at the ASST of Lecco. On admission, he had fever (38.8 • C), headache, heart rate of 70 beats per minute, blood pressure 119/83 mm Hg, 95% oxygen saturation, and white blood cell count of 12.9 × 10 9 /L. A tomographic scan of the brain and a chest X-ray were not suggestive of a central nervous system infection. However, based on laboratory and clinical data the presence of sepsis and meningitis was again suspected. Empiric therapy was implemented with ceftriaxone (4 g daily, bid) and ampicillin (12 g daily, tid). Lumbar puncture showed a xanthochromic turbid cerebrospinal fluid (CSF) with meningitis features: pleocytosis with a white blood cell count of 8758/mm 3 (neutrophils, 97.4%). CSF also showed extreme hypoglycorrhachia (<1 mg/dL for a serum glucose level of 119 mg/dL) and high protein levels (1004 mg/dL). Given that a BioFire FilmArray ® Meningitis/Encephalitis panel (BioFire Diagnostics, Salt Lake City, UT, USA) was negative, meningitis caused by an uncommon bacterial etiologic agent was suspected. After a 10-h incubation period, an aerobic bottle from blood cultures performed at the emergency department was flagged positive by the instrument. Direct microscopic examination based on Gram staining evidenced the presence of Gram-negative rods. Empiric therapy was then changed to meropenem (6 g daily, tid), gentamicin (500 mg tid), and colistin (4.5 MU × 2, following a loading dose of 9 MU).
The patient had a Glasgow score of 7. Electroencephalography (EEG), brain nuclear magnetic resonance (NMR), and transesophageal echocardiography (TEE) were performed. EEG and NMR showed cerebral injury and hypodensity areas interpretable as vasculitic lesions, respectively, while TEE resulted negative. During the two weeks following the first CSF sample collection, blood cultures and a lumbar puncture were performed. Microbiology investigations resulted negative for these specimens. Nevertheless, this second CSF analysis again showed meningitis features: a white blood cell count of 80/mm 3 , a serum glucose level of 40 mg/dL, and 81 mg/dL protein levels.

Strain Identification and Antibiotic Susceptibility Test
Identification to the species level was performed by MALDI-TOF Mass Spectrometry (Vitek MS, bioMérieux, Marcy-l'Étoile, France), while susceptibility testing was routinely determined using a Vitek 2 system (bioMérieux) and Sensititre™ Gram Negative Plate (ThermoFisher, Waltham, MA, USA). Susceptibility results were interpreted according to current EUCAST criteria. Two sets of blood cultures and CSF were performed and sent to the laboratory. K. pneumoniae was isolated from both blood samples and CSF by culture in agar plates. The hypermucoviscosity phenotype of the two strains (sk205y205t and LC-1574/18) was determined using the string test. Hypermucoviscous K. pneumoniae  Table 1.

Genomic Characterization
The strains (isolated by culture in agar plates from CSF and positive blood culture) were subjected to whole-genome sequencing, as follows: genomic DNA was extracted with a QIAamp DNA minikit (Qiagen, Düsseldorf, Germany), following the manufacturer's instructions, and sequenced on the Illumina Miseq platform with a 2 × 250 paired-end run, after a Nextera XT library preparation step (Illumina Inc., San Diego, CA, USA).
The reads quality was checked using FastqC software (https://www.bioinformatics. babraham.ac.uk/projects/fastqc/, accessed on 12 July 2020), and the low quality terminal residues were trimmed using Trimmomatic software (V0.36) (MAXINFO: targetLength = 50, strictness = 0.7) [10]. Genomic reads were assembled using SPAdes software (k-mer sizes: 21, 33, 55, 77, 99, 127 with "-careful" settings) [11]. The genome assembly was submitted to the European Nucleotide Archive (ENA), with the codes pending. The genomic distance among the study strain and the K. pneumoniae genomes retrieved from PATRIC database [12] was estimated using Mash software [13]. The 59 genomes most similar to the study strain were selected to perform the remaining analyses. The genomes were aligned against the reference K. pneumoniae genome 'NTUH_K2044' using progressive Mauve [14]. The core SNPs were named as described by Gona and colleagues [15]. The alignments of the core SNP sequences were subjected to phylogenetic analysis using RAxML software, with a 100 pseudo-bootstrap, after best model selection using ModelTest-NG [16].
The multilocus sequence type (MLST), the virulence gene content, the resistance genes, and the plasmid incompatibility groups of the strain were assessed using the Kleborate tool [17]. The variants of the genes fimH and mrkD, and the presence/absence of the gene magA were assessed using Blastn search (E-value threshold: 0.00001).
Several virulence genes were detected, both specific and nonspecific for hvKp. At least two fimbriae-related genes were found: the mrk operon, which encodes type 3 fimbriae (mrkABCDFHJ), and the type 1 fimbriae fim operon (fimH27). The wzi gene, involved in the capsule attachment to the host cell surface and used for the prediction of capsular (K) antigen type, was found. Notably, the strain carried the wzi-72 allele.
Both isolates also contained the rmpA and rmpA2 genes (regulators of the mucoid phenotype that increase capsule production), highly predictive of an hvKp strain. Virulence factors such as wabG (responsible for the biosynthesis of core lipopolysaccharide), and uge (uridine diphosphate galacturonate 4-epimerase), involved in capsule production were detected; thus, confirming the pivotal role of capsule-associated genes in the establishment of a hypervirulent behavior. Although less commons in hvKp, genes encoding colibactin (clbA-Q), a genotoxic metabolite, were also identified in sk205y205t and LC-1574/18 isolates.
It has already been established that serotype K2 isolates lack the kfu iron uptake gene and allS, a gene for allantoin metabolism; another gene required for hypermucoviscosity and virulence is magA, which appears to be restricted to K1 isolates. To confirm these features, the Klebsiella ferrous iron uptake system (kfuA, kfuB, and kfuC genes), allS and magA were not present in the two isolates. Furthermore, the strains also lacked the twocomponent system kvgAS, the microcin E495 and peg-344 genes. Virulence factors and genomic features are shown in Table 2.
Plasmids harbored by the sk205y205t and LC-1574/18 strains were determined based on the assembled genome using the BLASTn program, which was able to match with plasmid sequences recorded in plasmid banks; as a result, plasmids pK2044, and pRJA166b were identified with 100% and 99.1% identity, respectively. A further identified plasmid was an IncHI1B plasmid, showing 99.3% identity with pNDM.MAR_JN420336.
In addition to virulence plasmids, hvKp strains have acquired other virulence genes through mobile genetic elements in the chromosome. The ICEKp10 encoding genes for colibactin synthesis had already been identified in the two isolates studied here. The resistance mechanisms identified included the bla SHV-11 , bla SHV-67 , oqxA, and oqxB responsible for the resistance to beta-lactams; fosA that may be responsible for fosfomycin resistance; and acrR, which may confer resistance to fluoroquinolones. Nine different point substitutions were detected in the OmpK36 porin. Of these, N49S, L59V, L191S, F207W, D224E, L228V, and E232R have been previously associated to cephalosporins resistance; while A217S and N218H were associated to carbapenem resistance [18]. Three different point substitutions were found in the OmpK37 porin: I70M, I128M, and N230G, previously associated with carbapenem resistance [19]. Last, the following seven substitutions, observed on the AcrR sequence, P161R, G164A, F197I, F172S, R173G, L195V, and K201M had previously been associated with fluoroquinolone resistance [20].
A comparison of the sk205y205t isolate with other K. pneumoniae genomes deposited in the PATRIC database is shown in Figure 1. Notably, sk205y205t showed a major similarity with a ST65 strain collected in China. The Chinese isolate showed the same capsular serotype and virulence genes content, but a greater number of resistance genes, which also include qnrS1, ermB, bla TEM-1d , bla CTX-M-14 , and bla CTX-M-3 .
A comparison of the sk205y205t isolate with other K. pneumoniae genomes deposited in the PATRIC database is shown in Figure 1. Notably, sk205y205t showed a major similarity with a ST65 strain collected in China. The Chinese isolate showed the same capsular serotype and virulence genes content, but a greater number of resistance genes, which also include qnrS1, ermB, blaTEM-1d, blaCTX-M-14, and blaCTX-M-3. Figure 1. Maximum likelihood phylogenetic tree including the sk205y205t K. pneumoniae isolate (in red and bold type) and background strains retrieved from the PATRIC database. The geographic origin of all the strains is reported on the first level colored ring. The sequence type of the strains is indicated on the second colored ring.

Discussion
Hypervirulent K. pneumoniae (hvKp) is an emerging and evolving pathotype. hvKp shows an increased ability to cause severe infections, including hepatic abscesses, central nervous system infections, and endophthalmitis, with a particularly disconcerting propensity to cause community-acquired, life-threatening infection among young and otherwise healthy individuals [21]. Incidence of infections due to hvKp has been steadily increasing over the last three decades, and although, for the moment, they are mainly susceptible to antibiotics, hvKp strains are becoming increasingly resistant to antimicrobials via acquisition of mobile elements carrying resistance determinants [2].
Several genetic factors that confer the hypervirulent phenotype have been identified. In particular, those involved in increased capsule production and iron uptake systems are established hvKp-specific virulence factors. In addition, this also includes high mucus, lipopolysaccharide, fimbriae-related (type I fimbriae and type III pili), and biofilm formation [22].
In this study we analyzed by WGS and bioinformatic analysis the genomic features of a hmKp isolate (sk205y205t) from a case of meningitis. We also studied a subsequent K. pneumoniae isolate, LC-1574/18, collected from a blood culture of the same patient, in order to assess if one strain was responsible for those infections. For both strains we identified multiple virulence genes, associated with hypervirulence, and few antibiotic re-

Discussion
Hypervirulent K. pneumoniae (hvKp) is an emerging and evolving pathotype. hvKp shows an increased ability to cause severe infections, including hepatic abscesses, central nervous system infections, and endophthalmitis, with a particularly disconcerting propensity to cause community-acquired, life-threatening infection among young and otherwise healthy individuals [21]. Incidence of infections due to hvKp has been steadily increasing over the last three decades, and although, for the moment, they are mainly susceptible to antibiotics, hvKp strains are becoming increasingly resistant to antimicrobials via acquisition of mobile elements carrying resistance determinants [2].
Several genetic factors that confer the hypervirulent phenotype have been identified. In particular, those involved in increased capsule production and iron uptake systems are established hvKp-specific virulence factors. In addition, this also includes high mucus, lipopolysaccharide, fimbriae-related (type I fimbriae and type III pili), and biofilm formation [22].
In this study we analyzed by WGS and bioinformatic analysis the genomic features of a hmKp isolate (sk205y205t) from a case of meningitis. We also studied a subsequent K. pneumoniae isolate, LC-1574/18, collected from a blood culture of the same patient, in order to assess if one strain was responsible for those infections. For both strains we identified multiple virulence genes, associated with hypervirulence, and few antibiotic resistance genes. In our case, the production of SHV-11 and SHV-67 determinants did not confer resistance to third generation cephalosporins; hence, showing a non-ESBL phenotype. The putative virulence plasmids harbored by the sk205y205t and LC-1574/18 isolates highly resembled the hvKp-specific pLVPKlike plasmid.
More than 50% of meningitis presented abscesses (mostly hepatic) as first localization. Strains causing meningitis mostly showed an antibiotic multi-susceptible phenotype. Death occurred in more than 50% of cases.
Our case represents the first description of meningitis sustained by hmKp/hvKp in Italy. In particular, the patient had a Peruvian origin. Notably, three cases (3/16) of meningitis from the scientific literature presented origins from Central-South America. The patient reported no medical history; hence, suggesting a community-acquired infection. No other specific risk factors were reported. Moreover, no abscess was revealed in our patient; thus, suggesting that meningitis occurred after sepsis (probably starting from urinary tract infection) sustained by hvKp. The hvKp strain showed an antibiotic multi-susceptible phenotype, as well as other strains causing meningitis from literature cases. Moreover, antimicrobial treatment improved the patient's clinical condition, probably due to the absence of important risk factors or comorbidities.
The isolation of hmKp/hvKp strains represents an emergent cause of concern, often associated with invasive infections. This case highlights the severity of hypervirulent hypermucoviscous Kp with multiple and serious localizations in a few hours. The various virulence factors produced by this strain contributed to the invasive form of the infection. Moreover, the hypermucoviscous nature of the strain make very difficult a complete clinical resolution of the infection, despite the antibiotic multi-susceptible profiles. This aspect is often associated with recurrent infections, as occurred in our patient. This case also highlights the major contribution of the bacteriology laboratory, which allowed a rapid diagnosis by blood and CSF culture in association with a string test. This permitted the early detection of the hypermucoviscous phenotype and an optimal antibiotic treatment.
The string test method could be very useful for the rapid detection of the hypermucoviscous phenotype. However, clinical laboratories are unable to differentiate cKp from hvKp. The definition of biomarkers that can accurately predict hvKp strains (i.e., virulence factors associated to invasive infections) by WGS and the development of a more specific diagnostic test seem to be necessary for optimal patient care and epidemiologic surveillance of the dissemination of such strains.

Conclusions
The hypermucoviscous and hypervirulent phenotypes in K. pneumoniae represent a serious cause of concern, mostly causing severe invasive and recurrent infections. In particular, the meningitis sustained by hmKp/hvKp are associated with a mortality rate >50%. Fortunately, these strains are commonly associated with antibiotic multi-susceptible profiles, but resistant variants are emerging. The rapid detection of hmKp/hvKp phenotype, together with an optimal antibiotic treatment, could be of utmost importance for the clinical resolution of severe invasive infections sustained by these strains.
Author Contributions: A.P. performed microbiological investigations and sequencing, analyzed and interpreted the data, and wrote the paper; F.C. and M.P. performed bioinformatic analyses; C.M. and E.M. performed microbiological investigations, provided the clinical data, contributed to the revision of the manuscript draft; F.L. supervised the activities and revised the manuscript; L.P. conceived the study, interpreted the data, and contributed to draft the paper. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.

Institutional Review Board Statement:
The study was conducted in the context of normal clinical routine. All information and metadata about patients have been anonymized. Samples were coded and analyses were performed with an anonymized database.