Objective

The hypermucoviscous phenotype and hypercapsule production are driven by the rmpADC operon identified in Klebsiella pneumoniae [1]. There are several reports of hypermucoviscous KpSC isolates that do not carry the rmpADC operon; hereafter this phenotype is refer to as hypermucoviscous-like. This phenotype has been  described in K. pneumoniae [2], K. quasipneumoniae subsp. similipneumoniae [3] and K. variicola [4] but rarely seen outside the KpSC. It has been demonstrated that the hypermucoviscous-like phenotype promotes virulence, so those isolates displaying this phenotype are likely to increase its virulence [4]. The Klebsiella oxytoca complex (KoxSC) comprise nine species in which K. michiganensis is a member. Precise species identification requires genome-based analysis thus misidentification is likely to occur by phenotypic, biochemical test or MALDI-TOF [5].

The prevalence and type of infections caused by K. michiganensis is largely unknown but the species of the KoxSC are opportunistic pathogens that are associated with hemorrhagic colitis, urinary tract infections and bacteremia, and occasionally outbreaks in immunocompromised patients or those requiring intensive care [5]. K. michiganensis may acquire extended-spectrum β-lactamases (ESBL), carbapenemases and colistin resistance [6, 7]. To date, no hypervirulent or hypermucoviscous-like K. michiganensis strains has been described. In this study, the semi-quantitative string test [8] was applied to a collection of 30 presumptive K. oxytoca isolates resulting one isolate positive to this test which was further characterized.

Data description

The K. michiganensis 9273 isolate was obtained from urinary tract infection at the Hospital Civil de Guadalajara, Mexico in 2015. First, the species identification was performed at the hospital using MicroScan and resulted in K. oxytoca. The strain was subjected to disk-diffusion-method, accordingly to CLSI (2022) guidelines [9], to determine its antimicrobial susceptibility profile. It showed resistance to ampicillin (256 mg/ml) and susceptibility to ceftazidime, cefotaxime, imipenem, amikacin, gentamicin, nalidixic acid, ciprofloxacin, gentamicin, sulfamethoxazole plus trimethoprim and tetracycline, which characterize a non-MDR profile. The string test was evaluated in Mac Conkey agar plates and revealed a 4 cm string long. The plasmid profile was investigated by alkaline lysis protocol using the E. coli NCTC 50192 strain, which contains plasmids of 154-,66-, 48-, and 7-kb as a molecular size marker [10]. The K. michiganensis 9273 isolate possess three plasmids of ~ 190-, 160-, and 90-kb (Additional file 1: Fig. S1).

Total genomic DNA from the 9273-isolate was extracted using 5 ml of an overnight culture then purified using the DNeasy Kit (Qiagen, Germany). In 2015, the whole-genome sequence was generated using pyrosequencing methodology with the 454 Roche FLX Titanium platform. Reads (99.9% above Q40) longer than 500 bp were used for de novo assembly with the CLC Genomics Workbench version 4.0 (CLC bio). The total sequence data are 295,456 reads with 30- to 943-bp length range and a total of 145 contigs, with an estimated genome size of 6,632,597 bp with 20X coverage. The genome sequence was subjected to average nucleotide identity (ANI) analysis against reference genomes of the nine members of the KoxSC. The ANI values of K. oxytoca (GCA_003812925.1), K. grimontii (GCA_900200035), K. huaxiensis (GCA_003261575), K. pasteurii (GCA_018139045.1), K. spallanzanii (GCA_901563875.1), taxon 1 (QJJG00000000), taxon 2 (CP046115) and taxon 3 (CP055481) were below the cutoff point for species distinction (87.3% to 91.7%) but 99.24% ANI value with K. michiganensis (GCA_015139575.1) was observed. The Whole Genome Sequencing project was deposited at DDBJ/ENA/GenBank under the accession number JAVFHI000000000.

MLST ty** tool determined the sequence type (ST) 50 (https://cge.cbs.dtu.dk/services/MLST/). This ST50 did not correspond to globally expanding beta-lactam resistant sequence type [11]. Carriage of antimicrobial resistance genes was determined by ResFinder-4.1 (https://cge.cbs.dtu.dk/services/ResFinder/). The OXY-1 β-lactam resistance gene was identified which represents the phylogenetic group Ko1. In addition, the aph(3')-Ia gene for aminoglycoside resistance and multidrug efflux pumps (acrD, acrB, mdtB, mdtC, bepE, msbA, and emrB genes) were identified. Lastly, virulence genes mrk, kfu, fyuA (siderophore receptor of yersiniabactin) were found.

Plasmids with incompatibility groups repA, IncFIBK and IncFII were detected by PlasmidFinder-V2.1 (https://cge.food.dtu.dk/services/PlasmidFinder/). The hypermucoviscous-like phenotype in K. variicola has been linked with the acquisition of an IncFIBK-plasmid (pKV8917) which conferred the highly viscous phenotype. The pKV8917 self-transmissible plasmid has the potential to disseminate to other Klebsiella species [4]. Thus, additional studies are required to determine whether there is a plasmid-borne or chromosomal mechanism in K. michiganensis 9273.

The relationship between resistance genes and mobile genetic element were predicted using Mobile Element Finder (https://cge.food.dtu.dk/services/MobileElementFinder/) showing that aph(3′)-Ia and fyuA belonged to the same contig, suggesting a putative co-harboring location.

Quantification of capsule by means of uronic acid measurement [4] revealed that the hypermucoviscous-like K. michiganensis isolate 9273 produced more capsule (148 µg/109 CFUs) than non-hypermucoviscous (24 µg/109 CFUs) and hypervirulent K. pneumoniae (63.6 µg/109 CFUs) isolates reported in previous studies (Fig. 1A) [12]. Phagocytosis assay [4] showed that the 9273-isolate was less phagocyted in comparison to a non-hypermucoviscous isolate (Fig. 1B). In addition, we performed serum killing assay using human serum obtained from healthy volunteers [4] and was used to challenge ~ 106 CFUs of the 9273 isolate. This assay was performed in triplicate and revealed a serum resistant type.

Fig. 1
figure 1

Assays for determining the capsule-associated virulence phenotype in K. michiganensis 9273. A Quantification of uronic acid and B Phagocytosis assay. We included one hypervirulent K. pneumoniae (14660, ST86-K2) and one classical and non-hypermucoviscous K. pneumoniae (9468) as reference strains for capsule production and phagocytosis resistance assays. C Kaplan–Meier survival curves of infected G. mellonella larvae with K. michiganensis  isolate 9273 at three doses 1 × 104, 1 × 105 and 1 × 106 CFUs. K. variicola F2R9 (106 CFUs) [12], K. pneumoniae 9468 (106 CFUs) and hypervirulent K. pneumoniae 14660 (104 CFUs) were used respectively as nonlethal and lethal doses control. Statistical analysis was carried out based on one way ANOVA (A and B), long rank (Mantel-Cox) and Chi-square test (C); p < 0.0001. Uronic acid quantification and phagocytosis assay are presented as the mean ± standard deviation of three independent experiments. Kmg K. michiganensis, hvKpn hypervirulent K. pneumoniae, non-hmv Kpn non-hypermucoviscous K. pneumoniae, PBS Phosphate-buffered saline

Full size image

Finally, infection assays were realized according to Sugeçti [13]. G. mellonella larvae were acquired from Petmmal company (México). Briefly, bacterial cultures of K. michiganensis 9273 were prepared with 104, 105 and 106 CFU per 20 µl in PBS 1X. Seven-instar larvae of G. mellonella were chilled on ice for 5 min and surface sterilized in 95% ethanol. Then, 20 µl of each culture was injected into the hemocoel of each G. mellonella larvae with a hamilton syringe. Larvae were examined every 24 h and were scored as dead when they were melanized or unresponsive to touch.

We inoculated 106 CFUs of classical K. variicola F2R9 [14] and K. pneumoniae 9468 as nonlethal controls and 104 CFUs of the hypervirulent K. pneumoniae 14660 as lethal control. All experiments used 40 larvae per treatment. After 5 days post bacterial challenge, similar survival percentages were obtained for K. michiganensis 9273 (84%, 104 CFUs), K. variicola F2R9 (76.63%, 106 CFUs) and K. pneumoniae 9468 (74.45%, 106 CFUs). However, inoculation of 105 and 106 CFUs of K. michiganensis 9273 resulted in 50% and 0% survival, respectively (Fig. 1C). Similarly, 0% survival for hypervirulent K. pneumoniae 14660 (104) was observed after 3 days (Fig. 1C). Taken together, these results prove that K. michiganensis 9273 isolate is virulent, and this effect may be linked with its capsule-associated phenotype.

Limitations

Unlike the fact that the hypermucoviscous-like phenotype in K. variicola was determined to be plasmid-mediated, the present work did not address the genetic basis for this phenotype.