Emerging Microbes & Infections

2022, VOL. 11
https://doi.org/10.1080/22221751.2022.2065935

RESEARCH ARTICLE

Outer membrane vesicles-transmitted virulence genes mediate the emergence

of new antimicrobial-resistant hypervirulent Klebsiella pneumoniae
Yuneng Huaa,b*, Jingyu Wangb*, Mei Huanga, Yiyi Huangb, Ruyi Zhangb, Fan Bua, Biao Yanga,
Juanjiang Chena, Xiaomin Linb, Xiumei Hua,b, Lei Zhengb and Qian Wang a,b
a
Center for Clinical Laboratory, Zhujiang Hospital, Southern Medical University, Guangzhou, People’s Republic of China; bDepartment of
Laboratory Medicine, Nanfang Hospital, Southern Medical University, Guangzhou, People’s Republic of China

ABSTRACT
Hypervirulent Klebsiella pneumoniae (hvKp) is a notorious clinical pathogen that is more likely to cause severe primary
and metastatic abscesses. The dissemination of antimicrobial-resistant hvKp isolates has been reported worldwide,
posing a great challenge and severe clinical threat. However, the mechanisms of antimicrobial-resistant hvKp isolates
prevalent worldwide are not well precise. Outer membrane vesicles (OMVs) secreted from gram-negative bacteria are
an important vehicle for delivering effector molecules inter- and intra-species. To explore whether OMVs as the
vector of virulence genes horizontal transfer among Klebsiella pneumoniae and to explain the potential mechanism
for the development of antimicrobial-resistant hvKp isolates, we isolated OMVs from hvKp and classical Klebsiella
pneumoniae (cKp) by sequential differential centrifugation, respectively. Then, the characteristics and contents of
hvKp-OMVs and cKp-OMVs were analyzed. These hvKp-OMVs contain virulence genes, which could be transferred
from hvKp horizontally to extended-spectrum beta lactamase (ESBL)-producing cKp, leading to the production of
antimicrobial-resistant hypervirulent transformants. Further experiments confirmed the transformants exhibited
antimicrobial resistance and hypervirulent phenotypes in vitro and in vivo. In short, this work demonstrated that
hvKp-OMVs facilitated virulence genes transfer, allowing an increase in the virulence level of ESBL-producing cKp and
providing a new mechanism for the emergence of antimicrobial-resistant hvKp isolates.

ARTICLE HISTORY Received 27 January 2022; Revised 4 April 2022; Accepted 10 April 2022

KEYWORDS Outer membrane vesicles; hypervirulent klebsiella pneumoniae; horizontal gene transfer; virulence genes

Introduction
capsules’ polysaccharide of hvKp as a constraint on
Hypervirulent Klebsiella pneumoniae (hvKp) is a horizontal gene transfer, and chromosomal recombi-
notorious clinical pathogen with a hypermucoviscos- nation is rare in hvKp, with decreased plasmid diver-
ity phenotype, more likely to cause primary and meta- sity driving a reduction in pan-genome diversity.
static abscesses, clinically characterized by organ or These factors lead to the fact that the clonal diversity
life-threatening infections in community-dwelling of hvKp is much less than that of multidrug-resistant
healthy hosts [1,2]. hvKp contains a large virulence (MDR) counterparts, and the efficiency of MDR
plasmid that harbours many virulence-encoding clones in acquiring virulence plasmids far exceeds
genes, considered genetic factors that confer hvKp’s the efficiency of hvKp clones in developing drug-
hypervirulent phenotype [3,4]. These genetic determi- resistant plasmids [7,8]. However, antimicrobial-
nants of hypervirulence display phenotypes associated resistant hvKp isolates have now been reported world-
with increased capsule production, aerobactin pro- wide. The emergence of these strains may be because
duction and mucoviscosity [5]. The continuous evol- the classical Klebsiella pneumoniae (cKp) strains to
ution of Klebsiella pneumoniae (KP) depends on acquire pLVPK-like virulence plasmid or the acqui-
acquiring resistance-encoding and hypervirulence- sition of resistance determinants elements by hvKp
encoding mobile genetic elements [6]. For a long strains [9]. It is important to elucidate the underlying
period, hvKp strains were rarely resistant to com- mechanisms that contribute to the spread of virulence
monly used antimicrobials due to some limitation elements that promote the emergence of antimicro-
for gene acquisition. The significantly thickened bial-resistant hvKp.

CONTACT Xiumei Hu [email protected]; Qian Wang [email protected] Center for Clinical Laboratory, Zhujiang Hospital,
Southern Medical University, Guangzhou, People’s Republic of China; Department of Laboratory Medicine, Nanfang Hospital, Southern Medical University,
Guangzhou, People’s Republic of China; Lei Zheng [email protected] Department of Laboratory Medicine, Nanfang Hospital, Southern
Medical University, Guangzhou, People’s Republic of China
*These authors contributed equally to this work.
Supplemental data for this article can be accessed online at https://doi.org/10.1080/22221751.2022.2065935.
© 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrest-
ricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1282 Y. Hua et al.

Bacterial outer membrane vesicles (OMVs) con- showed that hvKp-OMVs were larger (54-634 nm,
duct diverse functions, including the delivery of gen- median size 112 nm) (Figure 1A) than cKp-OMVs
etic materials or virulence factors, regulation of (17–523 nm, median 78 nm), which was consistent
intercellular communication, acting as a decoy for with what was described previously (Figure 1C). The
bacterial toxins and modulation of host immune OMVs were well purified, no bacteria were observed
responses [10]. Almost all gram-negative bacteria ubi- under the microscope, and the contaminated control
quitously produce and release OMVs that contain on the plates grew nothing.
numerous functional lipids, proteins, RNAs and
DNAs, which can be horizontally transferred to the
hvKp-OMVs are enriched in bacteria genetic
recipient bacteria or host cells [11]. OMVs represent
elements and functional proteins
a unique bacterial secretion pathway that selects and
protects its cargo, allowing bacteria to act on and The mean DNA concentrations (average of the three
interact with their environment over extended dis- experiments) of the hvKp-OMVs and cKp-OMVs
tances without the risk of direct contact [12]. As were 21.17 and 17.56 ng/µL, respectively (Figure
observed in previous studies, it is revealed that 2A). The mean protein concentrations (average of
OMVs protect luminal genes or plasmids against the three experiments) of the hvKp-OMVs and cKp-
DNases and function as an effective horizontal gene OMVs were 0.76 and 0.91 µg/µL, respectively (Figure
transfer system [13]. 2B). As demonstrated in Figure 2(C), DNA purified
Importantly, in Federica et al., genetically engin- from the OMVs gave specific amplified products for
eered Klebsiella pneumoniae-OMVs serve as vectors iroB and prmpA.
for antimicrobial resistance genes spread in the A total of 1177 proteins were detected by Nano LC-
microbial communities [14]. hvKp similarly secretes MS/MS analysis in hvKp-OMVs. Two hundred
outer membrane vesicles lumen-contained various twenty-three membrane-associated proteins
virulence factors and bioactive substances, stimulating (75.08%), 46 ribosomal proteins (15.49%), 22 cytosolic
the inflammatory response [15]. However, it is unclear proteins (7.41%), 3 cell wall proteins (1.01%) and 3
whether these hvKp-derived OMVs promote the hori- extracellular region proteins (1.01%) were found in
zontal transfer of mobile virulent elements between the vesicles (Figure 2D). Vesicle proteins were categor-
bacteria and thus produce the antimicrobial-resistant ized following subcellular localization site, biological
hvKp isolates. Our work first reported that hvKp- function, and molecular function (Figure 2E). Further
OMVs can transfer virulence genes to extended spec- bioinformatics analyses of the Kyoto Encyclopedia of
trum beta lactamase (ESBL)-producing cKp, resulting Genes and Genomes (KEGG) and Protein–Protein
in increased mucoviscosity and quantification of cap- Interaction Networks (PPI) were performed, which
sule, thus contributing to enhanced virulence levels of is shown in (Figure 2F, G).
the latter. ESBL-producing hvKp isolates exhibit
hypervirulent phenotype and simultaneously resistant
hvKp-OMVs mediate virulence genes transfer to
to almost all beta-lactam groups of antibiotics except
cKp strains
for carbapenems and cephamycins. In summary, our
data reveal the essential role of hvKp-OMVs in the To investigate whether hvKp-OMVs mediated viru-
transmission of virulence genes among Klebsiella lence genes transfer to cKp strains, we performed
pneumoniae strains, elucidating the potential mechan- transformation experiments to confirm the potential
isms for developing antimicrobial-resistant hvKp of hvKp-OMVs to deliver virulence genes to ESBL-
isolates. producing cKp. Exogenous DNA and protein were
digested by DNase and proteinase K. Treated OMVs
were used to transform ESBL-producing cKp. After
Results 24 h, treated cells were spread onto an LB agar plate
to identify successful transformants. Two strains of
Characterization of OMVs derived from hvKp
successful transformants, cKp-TC1 and cKp-TC2,
hvKp and cKp isolates were grown in Luria–Bertani were selected for further experiments.
(LB) broth, and the culture supernatants were col- Antibiotic sensitivity tests and REP-PCR assays
lected after 6 h of incubation. hvKp-OMVs and cKp- were established on donor, recipient, and transfor-
OMVs were isolated from the harvested broth and mants. The results showed that these transformants
characterized for morphology and size. A trans- exhibited minimal inhibitory concentration (MIC)
mission electron microscope (TEM) showed that and agarose gel electrophoresis images similar to
these purified OMVs were oval and spherical. There those of the recipient strains. There was no significant
was no significant difference between hvKp-OMVs difference in the band characteristics of the REP-PCR
(Figure 1B) and cKp-OMVs (Figure 1D) in the mor- amplified fragments between the recipient strains
phology. Nanoparticle tracking analysis (NTA) and transformants (Figure 3). Therefore, the
 Emerging Microbes & Infections 1283

Figure 1. Purification and characterization of OMVs derived from hvKp and cKp strains. (A) NTA of hvKp-OMVs and (C) cKp-OMVs
(n = 3). (B) TEM of hvKp-OMVs (arrows) and (D) cKp-OMVs (arrows), scale bar = 100 nm.

contamination of the donor strains was excluded. associated with hypervirulent phenotype, including
According to PCR results, the transformants contained regulator of the mucoid phenotype (prmpA/A2), sal-
a marker gene of the virulence plasmid (prmpA). Over- mochelin (iroBCDN) and aerobactin (iucABCD-
all, these results supported the idea that these transfor- iutA). String tests on blood agar plates showed a
mants had drug resistance-associated phenotype and marked increase in length by stretching colonies of
hypervirulence-associated phenotype (Table 2). These the transformants compared with the recipient strains
data indicated that the hvKp-OMVs could effectively (Figure 4A, B, C, and D). The results established that
transmit the virulence genes to ESBL-producing cKp. incubation of hvKp-OMVs with the ESBL-producing
cKp strains could increase the level of mucoviscosity
and capsular polysaccharide (uronic acid) in most of
hvKp-OMVs enhance the virulence level of cKp the transformants, which was closely related to the
strains virulence expression of hyper-virulent strains (Figure
To verify whether the transformants acquired viru- 4E, F, +OMVshvKp-100 µg). In our work, we found
lence genes, we conducted hypermucoviscosity assay that hvKp-OMVs at least 100 µg can cause an
to test for successful transformation. Virulence deter- increased level of mucoviscosity of ESBL-producing
minants carried by the pLVPK-like plasmid is cKp.
1284 Y. Hua et al.

Figure 2. Protein and DNA concentration of the OMVs. (A) DNA concentration of hvKp-OMVs and cKp-OMVs (n = 3). (B) Protein
concentration of hvKp-OMVs and cKp-OMVs (n = 3). (C) PCR profile of the hvKp-OMVs and hvKp strains used in this study. (D)
Cellular localization of protein from hvKp-OMVs. (E) Biological Process (left), Cellular Component (middle) and Molecular Function
(right) of protein from hvKp-OMVs. (F) Bioinformatics analyses of Kyoto Encyclopedia of Genes and Genomes (KEGG) and (G)
Protein-Protein Interaction Networks (PPI) were performed on protein from hvKp-OMVs.
 Emerging Microbes & Infections 1285

Figure 3. Agarose gel electrophoresis showed REP-PCR products with expected images. REP-PCR profile of the hvKp, cKp-TC1,
cKp-TC2 and cKp strains used in this study. The primers used to perform repetitive extragenic palindromic PCR (REP-PCR) are
REP (left) and ERIC (right), respectively.

Only the OMVs secreted by hvKp could induce the Transformants exhibited stable
increased level of mucoviscosity and capsular polysac- hypermucoviscosity phenotype
charide (uronic acid) in the ESBL-producing cKp
To determine if the transformants exhibited stable
strains, as OMVs isolated from cKp or E.coli strains
hypermucoviscosity phenotype, all transformants
could not increase the level of mucoviscosity in the
were passaged for five generations, and then the
ESBL-producing cKp strains (Figure 4G, +OMVscKp-
expression of the hyper-mucoviscosity phenotype of
500μg, +OMVsE.coli-500μg). Furthermore, cKp-
these five generations was measured. String tests on
OMVs could not cause increased mucoviscosity of
blood agar plates showed a marked increase in length
the hype-virulent strains (Figure 4H, +OMVscKp-
by stretching these strains (Figure 6A). The results
500 µg), and hvKp strains acquisition of its OMVs
demonstrated that the level of mucoviscosity was sig-
could not enhance the level of mucoviscosity even at
nificantly increased in these five generations and the
a higher concentration (Figure 4H, +OMVshvKp-
expression level of hypermucoviscosity phenotype
500 µg). The presence of plasmid-borne genes
without the impact from the passage. (Figure 6B, C)
prmpA and iroB as a marker of the virulence plasmid
Plasmid-borne gene prmpA of these five gener-
in the transformants was determined by PCR (Figure
ations were analyzed using PCR assay, showing that
4I, J).
the transformants’ acquisition of virulence genes still
contained prmpA after several generations (Figure
6D, E).
The virulence level of transformants was tested
by the mouse lethality assay Discussion
The virulence level of these two transformants was Klebsiella pneumoniae, a concerning global pathogen,
further detected by the mouse lethality assay. At is widely recognized as a cause of hospital- and com-
5.0 × 105 colony-forming units (CFU) of bacteria munity-acquired infections posing a mounting threat.
in infected mice, cKp-TC1 and cKp-TC2 resulted With its evolution, it has gradually evolved into classi-
in 80% and 60% mortality at 120 h, respectively, a cal Klebsiella pneumoniae (cKp) and hypervirulent
level slightly lower than that of hvKp but substan- Klebsiella pneumoniae (hvKp) [16]. However, hvKp
tially higher than that of ESBL-producing cKp. Fur- is more virulent than cKp, and can cause nosocomial
thermore, when infected with the same dose of and community-acquired infections in healthy indi-
ESBL-producing cKp or phosphate buffer saline viduals [3]. cKp often carries antimicrobial-resistant
(PBS), the post-infection mortality rate was 0% at genetic elements, such as β-lactamase and carbapenem
120 h (Figure 5). These data confirm that the acqui- genes. Unlike cKp, hvKp harbours pLVPK-like viru-
sition of virulence plasmids by ESBL-producing cKp lence plasmids encoding important virulence factors
strains resulted in an enhanced level of virulence such as capsular polysaccharides regulator (prmpA/
comparable to that of the hvKp strains, as demon- A2) and siderophores (iroBCD/iucABCD). hvKp
strated by the ability to cause a higher mortality strains acquire antimicrobial-resistance genetic
rate in the mouse lethality assay. elements or drug-resistant cKp strains obtain
1286 Y. Hua et al.

Figure 4. hvKp-OMVs enhanced mucoviscosity and capsule production of cKp strains. (A) String tests on blood agar plates of
hvKp, (B) cKp-TC1, (C) cKp-TC2, (D) cKp strains. (E) Mucoviscosity and (F) uronic acid production of hvKp, cKp-TC1, cKp-TC2
and cKp strains (+OMVshvKp-100 μg). Results are presented as mean ± SEM, ***P < 0.001. (G) hvKp-OMVs could induce an
increased level of mucoviscosity in cKp strains (+OMVshvKp-500 μg), while the level of mucoviscosity in cKp strains is not enhanced
by E.coli-OMVs (+OMVsE.coli-500 μg) and cKp-OMVs (+OMVscKp-500 μg). Wilcoxon paired t-test, ***P < 0.001, significant difference.
Wilcoxon paired t-test, ns P > 0.05, no significant difference. (H) cKp-OMVs (+OMVscKp-500 μg) could not induce the increased
level of mucoviscosity in hvKp strains, while the level of mucoviscosity in hvKp strains is not enhanced by its own OMVs even
at higher concentrations (+OMVshvKp-500 μg). Wilcoxon paired t-test, ns P > 0.05, no significant difference. (I)(J) PCR amplified
fragments of the hvKp, cKp-TC1, cKp-TC2 and cKp strains, respectively.
 Emerging Microbes & Infections 1287

mechanism remained to be clarified. The mass spec-

trometric results suggested that a total of 1177 pro-
teins in hvKp-OMVs, including outer and inner
membrane proteins.
Recent work confirmed that Klebsiella pneumoniae-
OMVs could carry drug-resistant genetic elements for
horizontal gene transfer (HGT) between bacteria, and
the DNA wrapped in them was not affected by
DNases, giving this new HGT mechanism an
additional advantage [21]. We proved that hvKp
could release OMVs containing the vital virulence
Figure 5. Virulence potential of hvKp, cKp-TC1, cKp-TC2 and genes, including the regulator of mucoid phenotype
cKp strains with an inoculum of 5 × 105 CFU at 120 h after
infection in a mouse lethality assay (n = 5). The log-rank Man-
(prmpA) and siderophores (iroB). After hvKp-OMVs
tel–Cox test was used for the analysis of survival curves. carried virulence genes were transferred to ESBL-pro-
ducing cKp, the transformants showed enhanced
hypermucoviscosity phenotype. Key genes related to
pLVPK-like virulence plasmids, leading to the emer- the virulent plasmids were detectable by PCR, such
gence of multidrug-resistant hvKp isolates. However, as iroB, prmpA, and iutA. When the hvKp, ESBL-pro-
at least two reasons limit the emergence of those ducing cKp, and the transformants were injected
strains. One thought is that pLVPK-like virulence intravenously into mice, the mouse survival rate of
plasmids are non-conjugative. In addition, hvKp has the transformants was significantly decreased com-
some sort of genetic limitation and the constraint of pared with that of the ESBL-producing cKp strains,
significantly thickening the capsule, making it challen- closer to the high virulence level, proving that the
ging to obtain mobile genetic elements, which is one of transformants displayed hypervirulent and multi-
the prominent features of hvKp [7]. The virulent plas- drug-resistant phenotypes.
mid, once confined to hvKp, has gradually spread to To verify our work was not only applied to a
cKp [17]. The emergence and dissemination of multi- specific strain, but OMVs-mediated transformation
drug-resistant hvKp isolates undoubtedly pose an experiments were conducted on new Klebsiella pneu-
alarming challenge to antimicrobial therapy and pub- moniae clinical isolates. PCR-based assays were per-
lic health [18]. formed on peg-344, iroB, iucA, prmpA and prmpA2
Outer membrane vesicles (OMVs) are nano-sized genes. Results suggested that iroB, iucA, prmpA and
vesicles with various biological properties secreted by prmpA2 genes of this clinical isolates are positive
various gram-negative bacteria under physiological (Supplementary Fig 2). Two strains of successful
and pathological conditions [19]. OMVs play a critical transformants, cKp-TC3 and cKp-TC4, were selected
role in cell–cell communication, directing the contents for further experiments. The viscous string showed a
to be nearby or distant cells or tissues and assisting marked increase in length by stretching bacterial colo-
cells in exchanging proteins, lipids, nucleic acids and nies of the transformants compared with the recipient
virulence factors [20]. Work described OMVs as a strains on blood agar plates (Supplementary Fig 3A).
carrier for horizontal gene transfer (HGT) is known OMVs-mediated transformation experiments demon-
about A. baumannii OMVs facilitate intra- or inter- strated that OMVs derived from this new hyperviru-
species transfer of plasmids containing OXA-24 Car- lent Klebsiella pneumoniae clinical isolates enhanced
bapenemase Gene and blaNDM-1 gene to surround- the level of the mucoviscosity of cKp strains. (Sup-
ing bacterial [21,22]. Moreover, recent studies plementary Fig3B,+OMVshvKp-100 µg). prmpA and
demonstrate that OMVs mediate the transfer of viru- iroB genes as a molecular marker of the virulence plas-
lence genes from Escherichia coli O157:H7 to other mid in the successful transformants were determined
enteric bacteria [23]. by the PCR assay (Supplementary Fig 3C).
In this work, the OMVs were extracted from hvKp Antimicrobial-resistant hvKp infections have been
and cKp, respectively. The results of TEM and NTA widely disseminated during the past years, which exhi-
showed that OMVs have typical structures. There bit severe challenges to the prevention and treatment
was no significant difference in the morphology of of these strains. The main limitation of this study is
hvKp-OMVs and cKp-OMVs, but the particle size of that the molecular mechanism underlying the viru-
hvKp-OMVs was larger than that of cKp-OMVs. lence genes transfer event, which enables the emer-
This particle size difference may be due to the signifi- gence of antimicrobial-resistant hvKp isolates is not
cantly thickened capsule of hvKp isolates. According completely clear. Further explorations are required
to previous studies [24], the size of vesicles of capsu- to find out the specific mechanism of DNA wrapping
lated bacterial or fungal strains was generally larger into the vesicles and to clarify how the vesicles enter
than that of non-capsulated strains, and the specific the recipient bacterial strains. In addition, although
1288 Y. Hua et al.

Figure 6. Transformants exhibited stable hypermucoviscosity phenotype after several generations. hvKp-OMVs enhanced mucov-
iscosity of cKp strains. (A) String tests on blood agar plates of five generations of cKp-TC1 and cKp-TC2. (B)Mucoviscosity of five
generations of cKp-TC1 and (C)cKp-TC2. (+OMVShvKp-100 μg). Results are presented as mean ± SEM, ***P <0.001. (D) PCR amplified
fragments of five generations of cKp-TC1 and (E) cKp-TC2 (prmpA).
 Emerging Microbes & Infections 1289

Table 1. Oligonucleotides used in this work. Whole Genome sequencing

Primer Sequence (5′ to 3′ ) Length (bp) Use(s)
prmpA Forw ACTGGGCTACCTCTGCTTCA 536 PCR
Genomic DNA was extracted from K. pneumoniae
prmpA Rev CTTGCATGAGCCATCTTTCA clinical isolates using the bacterial genomic DNA
iroB Forw CCCTGGCATATCAAAGGCGT 534 PCR extraction kit (TIANGEN Biotech, China) and
iroB Rev GACAACAACGCGGGCATTTA
peg-344 TGGGGTTATTCTTTCGCT 508 PCR sequenced by the MinION (Oxford Nanopore Tech-
Forw nologies) platform. The strain had a 5.45-Mb chromo-
peg-344 Rev TTTCCAAGCTTACTGCAATT
prmpA2 Forw GTGCAATAAGGATGTTACATTA 430 PCR some and four main plasmids with sizes of
prmpA2 Rev GGATGCCCTCCTCCTG 177,011base pairs (bp), 113,492 (bp), 10,455 (bp)
iucA Forw AATCAATGGCTATTCCCGCTG 239 PCR
iucA Rev CGCTTCACTTCTTTCACTGACAGG
and 9,319 (bp), respectively (Supplementary Table
REP Forw NNNNCGNCGNCATCNGGC – REP- 1). The whole sequence of K. pneumoniae clinical iso-
REP Rev NCGNCTTATCNGGCCTAC PCR lates had been submitted to the GenBank databases
ERIC Forw ATGTAAGCTCCTGGGGATTCAC – REP-
ERIC Rev AAGTAAGTGACTGGGGTGAGCG PCR (accession number JAKFHZ000000000).

cKp-OMVs could not cause increased mucoviscosity String test

of the hype-virulent strains, and hvKp strains acqui- The strain was then inoculated onto a blood agar plate
sition of its own OMVs could not enhance the level and showed a positive string test, in which the viscous
of mucoviscosity even though it is possible at higher string was greater than 5 mm in length when stretch-
concentration, the potential roles of cKp-OMVs in ing bacterial colonies grown overnight at 37°C using a
mediating resistance-encoding mobile genetic bacteriology inoculation loop on an agar plate, indi-
elements transfer to hvKp strains remain unclear. cating that it showed hypermucoviscosity phenotype.
Similarly, we wonder whether drug-resistant hvKp
simultaneously disseminates drug-resistance and
hypervirulence genetic factors. The role of OMVs Polymerase chain reaction
from the drug-resistant hvKp isolates in delivering
mobile genetic elements among Klebsiella pneumoniae The extraction of genomic DNA from all
deserves further exploration. K. pneumoniae isolates was performed with a bacterial
In summary, we demonstrated for the first time that genomic DNA extraction kit (TIANGEN Biotech,
the hvKp-OMVs could carry key virulence genes, China). PCR-based assays were performed on peg-
leading to the increased level of mucoviscosity and 344, iroB, iucA, prmpA and prmpA2 genes of these
virulence expression in drug-resistant strains’ acqui- strains for the presence of virulence-encoding plas-
sition of virulence determinants. First this work eluci- mid, as previously reported (Supplementary Figure
dates a new mechanism governing the emergence and 1) [25]. Not all hvKp are hypermucoviscous and the
dissemination of antimicrobial-resistant hypervirulent detection of five virulence genes markers (peg-344,
klebsiella pneumoniae. iroB, iucA, prmpA, and prmpA2) provides >95% diag-
nostic accuracy for distinguishing hvKp from cKp.
The five markers were detected, and the positive rate
Materials and methods of those ≥2 combined with the clinical features was
preliminarily determined to be a hypervirulent Kleb-
Klebsiella strain siella pneumoniae. Table 1 lists the specific primers,
A Klebsiella pneumoniae clinical isolate were detected PCR conditions and product size.
in the pharyngeal swab of a 63-year-old man in an
otorhinolaryngology clinic of Nanfang Hospital in
Purification of OMVs
Guangdong Province in 2019, which was identified
by VITEK MS system (BioMérieux, China) or BD hvKp was incubated in Luria–bertani (LB) medium
Phoenix 100 automatic identification and suscepti- (Hope Bio-Technology Co., China) at 37°C, for the
bility testing system (Becton Dickinson, USA).The preparation of OMVs [26]. To remove dead cells,
patient was referred to an otorhinolaryngology clinic foreign proteins, and vesicles larger than 1 µm from
due to parotid abscess. the bacterial solution, differential centrifugations

Table 2. Phenotypic and genotypic characteristics of donor, transformant, and recipient strains.
MIC(μg ml−1)
Strain Cefazolin Cefotaxime Imipenem Meropenem Ampicillin Aztreonam Tetracycline prmpA
cKp 16 2 2 <1 >16 >16 >8 –
cKp-TC1 16 2 2 <1 >16 >16 >8 +
cKp-TC2 16 2 2 <1 >16 >16 >8 +
hvKp ≤4 ≤1 ≤1 <1 16 ≤2 ≤2 +
1290 Y. Hua et al.

(5,000 g, 15 min, 4°C; 10,000 g, 30 min, 4°C) were mass-spectrometric data were analyzed via MaxQuant
used. The collected supernatant was then filtered 1.6.1.0. Protein database used: uniprot-Klebsiella
through a 0.22-µM membrane filter (Merck Millipore, pneumoniae [573]-397877-202102. fasta.
Germany). After ultracentrifugation at 100,000 g of Intravesicular DNA was quantified using the
the residual supernatant at 4°C for 70 min in an method, as previously described [21]. 50μg of
SW32Ti rotor (Beckman Coulter, USA), the pellet OMVs was treated with 1 U/μL DNase I (Invitrogen,
was resuspended in 500 µL PBS (Gibco Company, USA) and 50 mg/L proteinase K (Beyotime Biotech-
USA). OptiPrep (60% iodixanol; Axis-Shield, Norway) nology, China) according to the manufacturer’s pro-
density gradient solutions with 5%, 10%, 20%, and tocols. Following DNase and proteinase K treatment,
30% density gradients were prepared and distributed OMVs were lysed for 30 min at 37°C with 0.125%
into discrete 5%-30% density gradient layers. A Triton X-100 solution (Sigma-Aldrich, USA). The
500μL of resuspended OMVs sample from the pre- DNA concentration and purity were then quantified
vious step was deposited on top of the density gradient using a Nanodrop (Thermo Scientific, USA); a mini-
and centrifuged at 100,000 g for 18 h at 4°C with a mum requirement is that the ratio of OD260/280 ≥
SW40 Ti rotor. After centrifugation, the OMVs layer 1.8. Purified DNA was used for subsequent PCR
was resuspended in phosphate-buffered saline (PBS). analyses.
The OMVs suspension was then filtered again through
a 0.22-µM membrane filter and cultured on blood agar
OMVs-mediated transformation
plates to ensure procedures were sterile. (Merck Milli-
pore, Germany). OMVs were treated with 1 U/μL OMVs-mediated transformation experiments were
DNase I (Invitrogen, USA) according to the manufac- conducted, as previously described [22]; an ESBL-pro-
turer’s protocols and stored at −80°C for further ducing K. pneumoniae ATCC 700603, which was used
analysis. as a recipient strain, was incubated in LB broth at 37°C
under orbital shaking (150 rpm) for 6 h until the cul-
ture reached an OD600 of 0.4. Then recipient strains
Transmission electron microscope (TEM)
were resuspended in cold LB broth and adjusted to a
Purified OMVs were visualized by a transmission elec- concentration of 107 CFU/mL. Purified OMVs were
tron microscope (H-7650, Hitachi, Ltd., Japan). A isolated from the donor strains. Following that,
20 µL of OMVs was dropwise added into the copper 100 µg of hvKp-OMVs was mixed with 200 µl of reci-
mesh and then stained with 2% uranyl acetate for pient cells statically at 37°C for 4 h and then for
5 min. The morphology of OMVs was observed another 4 h under shaking (160 rpm) at 37°C. Fresh
under TEM, operating at 80 kV. LB-broth was added to the mixture, and the culture
was incubated overnight under orbital shaking
(160 rpm) at 37°C. The collected suspensions were
Nanoparticle Tracking analysis (NTA)
carefully inoculated onto LB agar plates. The success-
The size distribution of OMVs was determined by ful transformation was preliminary confirmed using
NanoSight NS300 instruments (Malvern, UK). the string test. PCR assay was used to determine the
OMVs were suspended in sterilized PBS and measured presence of prmpA and iroB as key molecular markers
within the optimal dilution. Samples were then in successful transformants.
injected into the measuring chamber and analyzed
by NTA software.
Mucoviscosity assay
A sedimentation assay [27] was used to determine the
Quantification of OMVs
mucoviscosity of hvKp, cKp-TC, and cKp. In short,
BCA Protein Assay Kit (Beyotime Biotechnology, cultures incubated at 37°C overnight in LB broth
China) was used to determine the protein concen- were subcultured in a fresh medium to an OD600 of
tration of OMVs samples according to the manufac- 0.2. The cultures were normalized to OD of 1.0 ml−1
turer’s protocols. Simultaneously, the total proteins after 6 h and centrifuged at 1,000 g for 5 min. The
from OMVs samples were extracted and analyzed supernatant was gently removed without disturbing
using the LC–tandem mass spectrometry (LC-MS/ the pellet to determine the OD600. Stretching bacterial
MS) analysis. The peptides were extracted from each colonies on sheep blood agar plates with an inoculum
sample by a gradient elution at a nanoliter flow rate ring were used to perform a string test.
using the Easy nLC 1200 system (Thermo Scientific,
USA). The peptide fragments were separated and ana-
Extraction and quantification of capsule
lyzed using a Q-Exactive HF-X mass spectrometer
(Thermo Scientific, USA) for data-dependent acqui- Uronic acid was extracted and quantified in the same
sition (DDA) mass spectrometry for 120 min. The way previously described [28]. A 500 µl of bacterial
 Emerging Microbes & Infections 1291

cultures grown for 6 h was combined with 100 µl of Author contributions

1% Zwitterion 3–14 detergent in 100 mM citric acid
YH, JW, XH, LZ, and QW: conceived the study, super-
(pH 2.0) and incubated at 50°C for 20 min. The
vised the whole project and wrote the manuscript. YH
cells were then centrifuged, and 250 µl of the super-
and JW conceptualized the work, performed most of
natant was added to 1.2 ml of 100% ethanol, incu-
the experiments and drafted the manuscript. MH par-
bated for 20 min at 4°C, then centrifuged for 5 min
ticipated in study design, strains collection and trans-
at maximum speed. Dried and resuspended the pellet
formation assay. YYH and RZ did the whole-genome
in 200 µl distilled water, then 1.2 ml of 12.5 mM
sequencing . FB and BY did the mouse infection
sodium tetraborate sulfate was added, incubated for
experiments and participated in data interpretation.
5 min at 100°C, and cooled for 10 min. Then, 20 µl
JC and XL participated in mucoviscosity assays.
of 0.15% 3-phenylphenol in 0.5% NaOH was added.
The absorbance at 520 nm after 5 min of incubation
at room temperature was measured. The glucuronic
acid content, expressed as micrograms OD unit −1, Disclosure statement
was determined using the standard curve of glucuro-
No potential conflict of interest was reported by the
nate lactone. author(s).

Mouse infection model Funding

The potential toxicity of several Klebsiella pneumo- This research was funded by grant from the National Natu-
niae strains was tested using the mouse bacteremia ral Science Foundation of China [grant number 82072334].
model [29,30]. Female C57BL/6 mice (7 weeks old)
were acquired from Guangzhou Yancheng Biotech-
Ethical approval
nology Service Company and acclimatized for 7
days. The mice were allowed to eat and drink All the research involving bacteria were approved by the
water during the investigation. Simultaneously, the Medical Ethics Committee of Southern Medical Univer-
mice were divided into five groups at random (n = sity (number:AF/SQ-04/01.0). All the research involving
5 in each group). Mice in each group were inocu- animal were approved by the Medical Ethics Committee
lated with 5.0 × 105 colony-forming units (CFU) of Southern Medical University (number:K2021016).
/mouse of each tested bacterial strain intravenously.
Phosphate-buffered saline (PBS) was used as the
Statistical analyses
negative control. Symptoms and mortality rates of
the test mice were examined and reported for Prism 8.0 (GraphPad Software) was used for statistical
120 h after infection. To ensure data consistency, analysis. A P value ≤ 0.05 was used to determine sig-
animal experiments were repeated at least twice. nificance. Each data point was repeated at least three
Log-rank tests (Mantel–Cox) were used to evaluate times independently.
survival curves.

ORCID
Antimicrobial susceptibility test Qian Wang http://orcid.org/0000-0002-0299-4462
The antimicrobial susceptibility of the donor, trans-
formants and rerecipient was conducted using the References
broth microdilution method according to the Clinical
and Laboratory Standards Institute (CLSI) M07 [31]. [1] Russo TA, Marr CM. Hypervirulent Klebsiella pneu-
moniae. Clin Microbiol Rev. 2019;32(3):e00001-19.
[2] Shon AS, Russo TA. Hypervirulent Klebsiella pneu-
moniae: the next superbug? Future Microbiol. 2012;7
REP-PCR (6):669–671.
[3] Ye M, Tu J, Jiang J, et al. Clinical and genomic analysis
REP-PCR was used to validate that the recipient of liver abscess-causing Klebsiella pneumoniae ident-
strains had received the plasmid. Table 1 lists the pri- ifies new liver abscess-associated virulence genes.
mers used to perform REP-PCR (repetitive extragenic Front Cell Infect Microbiol. 2016;6(165). Epub 2016/
palindromic sequence-based polymerase chain reac- 12/15.
tion). The process was performed as described pre- [4] Wyres KL, Lam MMC, Holt KE. Population genomics
of Klebsiella pneumoniae. Nat Rev Microbiol. 2020;18
viously [22]. When two strains had the same pattern
(6):344–359. Epub 2020/02/15.
of bands, they were believed to belong to the same [5] Paczosa MK, Mecsas J. Klebsiella pneumoniae: going
genotype, and a maximum difference between the on the offense with a strong defense. Microbiol Mol
two bands was acceptable. Biol Rev. 2016;80(3):629–661. Epub 2016/06/17.
1292 Y. Hua et al.

[6] Xu Y, Zhang J, Wang M, et al. Mobilization of the Annu Rev Microbiol. 2010;64:163–184. Epub 2010/
nonconjugative virulence plasmid from hypervirulent 09/10.
Klebsiella pneumoniae. Genome Med. 2021;13(1):119, [20] Tsatsaronis JA, Franch-Arroyo S, Resch U, et al.
Epub 2021/07/24. Extracellular vesicle RNA: A universal mediator of
[7] Wyres KL, Wick RR, Judd LM, et al. Distinct evol- microbial communication? Trends Microbiol.
utionary dynamics of horizontal gene transfer in 2018;26(5):401–410. Epub 2018/03/20.
drug resistant and virulent clones of Klebsiella pneu- [21] Chatterjee S, Mondal A, Mitra S, et al. Acinetobacter
moniae. PLoS Genet. 2019;15(4):e1008114, Epub baumannii transfers the blaNDM-1 gene via outer
2019/04/16. membrane vesicles. J Antimicrob Chemother.
[8] Lam MMC, Wyres KL, Duchene S, et al. Population 2017;72(8):2201–2207. Epub 2017/05/16.
genomics of hypervirulent Klebsiella pneumoniae clo- [22] Bou G, Cervero G, Dominguez MA, et al. PCR-based
nal-group 23 reveals early emergence and rapid global DNA fingerprinting (REP-PCR, AP-PCR) and pulsed-
dissemination. Nat Commun. 2018;9(1):2703, Epub field gel electrophoresis characterization of a nosocomial
2018/07/15. outbreak caused by imipenem- and meropenem-resist-
[9] Li W, Sun G, Yu Y, et al. Increasing occurrence of anti- ant Acinetobacter baumannii. Clin Microbiol Infect.
microbial-resistant hypervirulent (hypermucoviscous) 2000;6(12):635–643. Epub 2001/04/04.
Klebsiella pneumoniae isolates in China. Clin Infect [23] Yaron S, Kolling GL, Simon L, et al. Vesicle-mediated
Dis. 2014;58(2):225–232. Epub 2013/10/09. transfer of virulence genes from Escherichia coli O157:
[10] Toyofuku M, Nomura N, Eberl L. Types and origins of H7 to other enteric bacteria. Appl Environ Microbiol.
bacterial membrane vesicles. Nat Rev Microbiol. 2000;66(10):4414–4420.
2019;17(1):13–24. Epub 2018/11/07. [24] Bielska E, Sisquella MA, Aldeieg M, et al. Pathogen-
[11] Kim JH, Lee J, Park J, et al. Gram-negative and gram- derived extracellular vesicles mediate virulence in the
positive bacterial extracellular vesicles. Semin Cell Dev fatal human pathogen Cryptococcus gattii. Nat
Biol. 2015;40:97–104. Epub 2015/02/24. Commun. 2018;9(1):1556, Epub 2018/04/21.
[12] Kalra H, Drummen GP, Mathivanan S. Focus on [25] Russo TA, Olson R, Fang C-T, et al. Identification of
extracellular vesicles: introducing the next small big biomarkers for differentiation of hypervirulent
thing. Int J Mol Sci. 2016;17(2):170, Epub 2016/02/11. Klebsiella pneumoniae from classical K. pneumoniae.
[13] Schwechheimer C, Kuehn MJ. Outer-membrane ves- J Clin Microbiol. 2018;56(9):e00776–18.
icles from gram-negative bacteria: biogenesis and [26] Dell’Annunziata F, Ilisso CP, Dell’Aversana C, et al.
functions. Nat Rev Microbiol. 2015;13(10):605–619. Outer membrane vesicles derived from Klebsiella
Epub 2015/09/17. pneumoniae influence the miRNA expression profile
[14] Dell’Annunziata F, Dell’Aversana C, Doti N, et al. in human bronchial epithelial BEAS-2B cells.
Outer membrane vesicles derived from Klebsiella Microorganisms. 2020;8(12):1985. Epub 2020/12/17.
pneumoniae are a driving force for horizontal gene [27] Pan YJ, Lin TL, Hsu CR, et al. Use of a dictyoste-
transfer. Int J Mol Sci. 2021;22(16):8732. lium model for isolation of genetic loci associated
[15] Zhang J, Zhao J, Li J, et al. Outer membrane vesicles with phagocytosis and virulence in Klebsiella pneu-
derived from hypervirulent Klebsiella pneumoniae moniae. Infect Immun. 2011;79(3):997–1006. Epub
stimulate the inflammatory response. Microb Pathog. 2010/12/22.
2021;154:104841, Epub 2021/03/11. [28] Cheng HY, Chen YS, Wu CY, et al. Rmpa regulation of
[16] Zhang Y, Jin L, Ouyang P, et al. Evolution of capsular polysaccharide biosynthesis in Klebsiella
hypervirulence in carbapenem-resistant Klebsiella pneumoniae CG43. J Bacteriol. 2010;192(12):3144–
pneumoniae in China: a multicentre, molecular epide- 3158. Epub 2010/04/13.
miological analysis. J Antimicrob Chemother. 2020;75 [29] Russo TA, MacDonald U, Hassan S, et al. An assess-
(2):327–336. Epub 2019/11/13. ment of siderophore production, mucoviscosity, and
[17] Lee CR, Lee JH, Park KS, et al. Antimicrobial resist- mouse infection models for defining the virulence
ance of hypervirulent Klebsiella pneumoniae: epide- spectrum of hypervirulent klebsiella pneumoniae.
miology, hypervirulence-associated determinants, mSphere. 2021;6(2):e00045-21. Epub 2021/03/26.
and resistance mechanisms. Front Cell Infect [30] Pletzer D, Mansour SC, Wuerth K, et al. New mouse
Microbiol. 2017;7:483, Epub 2017/12/07. model for chronic infections by gram-negative bac-
[18] Yang X, Ye L, Chan EW-C, et al. Tracking recombina- teria enabling the study of anti-infective efficacy and
tion events that occur in conjugative virulence plasmid host-Microbe interactions. mBio. 2017;8(1):e00140-7.
p15WZ-82_Vir during the transmission process. Epub 2017/03/02.
Msystems. 2020;5(4):e00140–20. [31] Methods for Dilution Antimicrobial Susceptibility
[19] Kulp A, Kuehn MJ. Biological functions and biogen- Tests for Bacteria That Grow Aerobically—Tenth
esis of secreted bacterial outer membrane vesicles. Edition: M07. (2015).