The influence of antibiotics on transitory resistome during gut colonization with CTX-M-15 and OXA-162 producing Klebsiella pneumoniae ST15

Great efforts have been made to limit the transmission of carbapenemase-producing Enterobacteriaceae (CPE), however, the intestinal reservoir of these strains and its modulation by various antibiotics remain largely unexplored. Our aim was to assess the effects of antibiotic administration (ampicillin, ceftazidime, ciprofloxacin) on the establishment and elimination of intestinal colonization with a CTX-M-15 ESBL and OXA-162 carbapenemase producing Klebsiella pneumoniae ST15 (KP5825) in a murine (C57BL/6 male mice) model. Whole genome sequencing of KP5825 strain was performed on an Illumina MiSeq platform. Conjugation assays were carried out by broth mating method. In colonization experiments, 5 × 106 CFU of KP5825 was administered to the animals by orogastric gavage, and antibiotics were administered in their drinking water for two weeks and were changed every day. The gut colonization rates with KP5825 were assessed by cultivation and qPCR. In each of the stool samples, the gene copy number of blaOXA-162 and blaCTX-M-15 were determined by qPCR. Antibiotic concentrations in the stool were determined by high pressure liquid chromatography and a bioanalytical method. The KP5825 contained four different plasmid replicon types, namely IncFII(K), IncL, IncFIB and ColpVC. IncL (containing the blaOXA-162 resistance gene within a Tn1991.2 genetic element) and IncFII(K) (containing the blaCTX-M-15 resistance gene) plasmids were successfully conjugated. During ampicillin and ceftazidime treatments, colonization rate of KP5825 increased, while, ciprofloxacin treatments in both concentrations (0.1 g/L and 0.5 g/L) led to significantly decreased colonization rates. The gene copy number blaOXA-162 correlated with K. pneumoniae in vivo, while a major elevation was observed in the copy number of blaCTX-M-15 from the first day to the fifteenth day in the 0.5 g/L dose ceftazidime treatment group. Our results demonstrate that commonly used antibiotics may have diverse impacts on the colonization rates of intestinally-carried CPE, in addition to affecting the gene copy number of their resistance genes, thus facilitating their stable persistance and dissemination.

www.nature.com/scientificreports/ isolate also carried the plasmid-borne aac(6′)-Ib-cr fluoroquinolone resistance determinant. The isolate carried resistance-determinants for aminoglycoside resistance (aac(3)-IIa, aph(3′)-Ia and aac(6′)Ib-cr as well. In addition, four different plasmid replicon types, namely IncFII(K), IncL, IncFIB and ColpVC were detected in the KP5825. The IncL and IncFII plasmids were successfully conjugated, and the IncL plasmid contained only the bla OXA-162 resistance gene within a Tn1991.2 genetic element, while the IncF(II)K contained the bla CTX-M-15 resistance gene. The KP5825 showed high level resistance against the beta-lactam and fluoroquinolone antibiotics based on MIC values determined in the broth microdilution assay. The broth mating procedure-based in vitro conjugation assay performed was successful, and the conjugated E. coli J53 harbouring the pOXA-162 showed increased resistance in case of ertapenem and meropenem and in case of E. coli J53 harbouring pCTX-M-15, resistance to cephalosporins (ceftazidime, cefotaxime) was detected. The characteristics of KP5825 and the conjugated E. coli J53 strains are shown in Table 1.
In the colonization studies with 6-8 week-old C57BL/6 male mice the antibiotics were administered in their drinking water for two weeks. The concentration of antibiotics in mouse stool was assessed with high pressure liquid chromatography (HPLC) on the 1st and 15th day after KP5825 colonization. The ampicillin concentration in the stool samples in Amp_0.5 group on the first day was 720.2 ± 247.0 µg/g (average ± SD), while on the fifteenth day 739.3 ± 219.4 µg/g. The average ciprofloxacin concentration in the Cip_0.1 group was 17.2 ± 5.96 µg/g on the first day, and 20.7 ± 4.97 µg/g on the fifteenth day; and in Cip_0.5 group, it was 203.8 ± 46.0 µg/g on the first day and 244.8 ± 61.9 µg/g on fifteenth day. Ceftazidime was undetecable from mice stool samples.
The colonization was performed with the KP5825 strain administered by orogastric gavage and the colonization rate of mice with CTX-M-15 ESBL-and OXA-162 carbapenemase-producing KP5825 was quantified by both a conventional culture analysis method and the qPCR technique, in order to simultaneously determine the absolute and relative colonization rates with the tested isolate. The effect of different treatment regimens on colonization with KP5825 are shown in Fig. 1A,B. The densities of KP5825 detected in feces were assayed on the 3rd, 6th, 9th, 12th and 15th days. If KP5825 organisms were not detected in the stool, the lower limit of detection (~ 2.3 log 10 CFU/g) was assigned. In case of all observation periods of ceftazidime treatments, the rate of KP5825 colonies were the highest, while on the other hand, during ciprofloxacin treatments they were the lowest ( Fig. 2A). During ampicillin (Amp_0.5) and ceftazidime treatments (Caz_0.1 and Caz_0.5), the absolute  www.nature.com/scientificreports/ colonization rate of the carbapenem-resistant KP5825 slightly increased. Upon treatment with ampicillin, a moderate increase of K. pneumoniae cell count was detected. In contrast, during ciprofloxacin treatments in both concentrations (Cip_0.1 and Cip_0.5) and in the control group, the colonization rates have decreased significantly. The most extensive decrease in colonization rate was observed in the group treated with the lower dose of ciprofloxacin (Cip_0.1). These alterations are the most unexpected as the present carbapenemase-producing K. pneumoniae shows high level resistance to fluoroquinolones. These results were consequent with qPCR results by observing the log 10 fold change of rpoB1 housekeeping gene designed for KP5825. The relative colonization rate of KP5825 between the first and fifteenth day of colonization showed also differences between antibiotic treatments. In Caz_0.1 group the colonization rate of carbapenem-resistant KP5825 slightly increased, while on the other hand treatment with ampicillin resulted in a moderate increase of KP5825. An extensive decrease in colonization rate was observed in the groups treated ciprofloxacin (Cip_0.1, Cip_0.5) (Fig. 2B). The effect of antibiotic-treatment regiments on bla CTX-M-15 and bla OXA-162 genes' copy number in the gut was determined by qPCR from each the stool sample and results were calculated as the fold change of gene normalized to the rpoB1 reference gene and relative to the control mice. The relative copy number of the ESBL bla CTX-M - 15 and the carbapenemase bla OXA-162 were determined and these results were correlated to the rpoB1 housekeeping gene of KP5825 on the first and on the fifteenth days from the feces of each mouse used in the experiment (Fig. 3). The relative copy number did not change for bla OXA-162 during the observed period in any treatment group. In contrast, a major elevation was observed from the first day to the fifteenth day in the treatment group with the Caz_0.5 treatment yielding 2 and 400-fold absolute gene copy number increase of the bla CTX-M-15 gene. At the same time, the relative copy number of the bla CTX-M-15 gene (which was controlled with the rate of the rpoB gene) also increased significantly (p < 0.05) from 2-to 5-times relative to the control in the Caz_0.5 treatment group (Fig. 3). Nevertheless, only the original CTX-M-15 and OXA-162-producing K. pneumoniae isolate could be reisolated from various feaces samples from mice during the experiment using the selective CHROMagar plates. We could not isolate other ESBL or carbapenenamse-producing bacteria on the appropriate selective culture media, except the original KP5825.

Discussion
Klebsiella pneumoniae is a prevalent and dangerous cause of hospital-associated infections, especially in ICUs 24,25 . Because of their global spread, high mortality and very limited therapeutic options, carbapenem-resistant K. pneumoniae (CRKP) was declared a major public health threat, rated as priority 1, critical pathogen by the World Health Organization [26][27][28] . Patients with intestinal carriage of CRKP upon admission may act as reservoirs 29 ; moreover, gastrointestinal colonization with MDR K. pneumoniae increases the risk of subsequent infections and mortality 29,30 . Colonization with a carbapenem-resistant Klebsiella has been highlighted as a hallmark of a subsequent extraintestinal infection by these pathogens; therefore, the identification of patients whom are positive for CRKP-colonization may be an important step to introduce infection control interventions and to save patients from developing an infection 31 .
Our experiments aimed to investigate the effects of various antibiotic treatments on the gastrointestinal colonization, gene dynamics and role in the resistome of the high-risk clone K. pneumoniae ST15, producing the www.nature.com/scientificreports/ IncFII of the high risk clone K. pneumoniae play an important role in its international dissemination. In our experiment, both plasmids were shown to be conjugable. OXA-48-like enzymes itselves hydrolyze carbapenems to a lesser extent, as we also observed, however their co-occurrence with other β-lactam resistance mechanisms, such as membrane impermeability, may result in high-level carbapenem-resistance [9][10][11] . Three antibiotics were included in our study, namely ampicillin, ceftazidime (representatives of β-lactams) and ciprofloxacin. Ampicillin and its derivates (i.e. the aminopenicillins) and ciprofloxacin (a member of the fluoroquinolones) are still one the most widely used drugs in the community, therefore, the assessment of their effect on the gut resistome is of utmost importance 32 . Ceftazidime has been recently sidelined in therapy, due to its availability and the emergence of ESBLs worldwide. Nevertheless, the introduction of ceftazidime and avibactam, a novel cephalosporin/β-lactamase inhibitor combination into the clinical practice-especially for the treatment of OXA-48-type (Class D) carbapenemase producing MDR Gram-negative organisms has provided renewed relevance to this drug 33,34 . Results of our experiments have shown that the studied antibiotic treatment regiments affected the resistome of mice in different ways.
Previous antibiotic therapy is an independent risk factor for colonization with ESBLproducing Enterobacteriaceae as demonstrated several studies 35,36 . During our studies, ampicillin pre-treatment was used (for a duration of 14 days) to maintain and promote the colonization of KP5825 in all treatment groups, which was done to model the natural colonization of the host with the microorganism. The rationale behind this was that-based on literature findings-gastro-intestinal colonization with K. pneumoniae is difficult to establish in mice via gavage treatment and that antibiotic (ampicillin) pre-treatment has been noted to play a role in disrupting the microbiota of the desired host to allow for the colonization of K. pneumoniae 37,38 .
There are controversial data regarding the effects of beta-lactam treatment on the gastrointestinal colonization with multi-drug reistant organisms. Several authors have noted that exposure to various β-lactam antibiotics allow for the colonization by ESBLs (an ST131 E. coli strain was used in the experiments), regardless of negatively affecting (clindamycin) the members of the Bacteroidales order or not (cefuroxime and dicloxacillin) 39 . Conversely, others reported that the treatment with cephalosporins at the ICU did not increase the acquisition rate of carbapenem-resistant Enterobacteriaceae 40 . In our study, as a consequence of the treatment with β-lactam antibiotics, both the colonization rate and-independently from this-the gene copy number of bla CTX-M-15 both increased. Nevertheless, the copy number of bla OXA-162 correlated with the colonization rate of KP5825. In the case of bla CTX-M-15 located on IncF(II)K plasmid, a higher gene copy number was detected in mice stool samples after cephalosporin treatment, thus indicating a shift in resistome. The measurement of the replicon' copy number could have additionally provided valuable information on the underlying reasion for the observed increase, however, this experiment was unfortunately not performed. Given that the CTX-15-producing transconjugant could not be isolated from stool samples, highlights that either recipient Enterobacteriales was not detectable www.nature.com/scientificreports/ (or culturable) in feces of mice or the copy number of bla CTX-M-15 resistance genes was increased only in the host cells. Thus, it may also be assumed that the bla CTX-M-15 gene may have been transferred to non-culturable bacteria. However, this does not change the fact that bla CTX-M-15 gene was present in higher levels and the plasmid is capable of conjugation in present of susceptible recipient bacterium. The fact that ciprofloxacin reduced the colonization rate in our experiments is particularly interesting, especially in light of the fact that the colonizing K. pneumoniae strain itself had high-level fluoroquinolone resistance as it had both chromosomal and plasmid-mediated quinolone resistance determinants. Regardless, our carbapenem-resistant K. pneumoniae isolate disappeared or its load has significantly decreased in the feces of ciprofloxacin treated mice. These findings support earlier studies where ciprofloxacin did not increase the abundance of antibiotic resistance genes-carrying plasmids and failed to promote colonization with MDR Gramnegative bacteria 37,41 . A potential explanation involves the limited antimicrobial effects of ciprofloxacin on the anaerobic intestinal microbiota 42 .
Based on the results of our experiments, it may be assumed that the differences in the colonization effects of the tested antibiotics are mainly rooted in their structre-activity relationships and biological targets, rather then the doses in which they were applied (there were no difference between different doses of the same antibiotic). These results highlights the fact that the that timing of the antimicrobial adiministration relative to CPE exposure is also an important parameter to consider in providing ecological space for the implantation and expansion of the MDR strain.

Conclusions
In summary, our results have shown that in the presence of β-lactam antibiotics, the amount of the high-risk clone of K. pneumoniae showed an increase in the absolute and relative colonization rate, as well as gene copy number of bla CTX-M-15 on the IncF(II) conjugative plasmid. In contrast, gene copy of bla OXA-162 -which was also conjugative in vitro on IncL plasmid-correlated with K. pneumoniae cell count in vivo. Increases in the degree of colonization in the presence of antibiotics has been described by previous studies, however, a clone-independent change in the copy number of bla CTX-M-15 resistance genes in vivo has not been previously described. In contrast, a parallel decrease in both the clone and the resistance genes was observed after the treatment of fluoroquinolones. This has already been observed by others, but contrasting observations have also been published. Gastrointestinal colonization of MDR bacteria poses a serious clinical problem, both in community-based and nosocomial settings, and in our study we demonstrated a diverse influence of commonly administered antibiotics (ampicillin, ceftazidime, ciprofloxacin) on intestinally carried multidrug-resistant K. pneumoniae.

Methods
Bacterial strains. K. pneumoniae ST15 (KP5825) was obtained from National Public Health Centre (Budapest, Hungary) 23 . Azide-resistant E. coli J53 was used in the conjugation assays.
Antibacterial susceptibility testing. Antibacterial susceptibility testing was performed by the broth microdilution method according to the EUCAST guidelines v.9.0 (www. eucast. org) 43 . Incubation was performed at 35 °C for 16-20 h and minimum inhibitory concentrations (MICs) were determined visually. E. coli ATCC 25922 was used as control strain.
Conjugation assay. Conjugation assays were carried out by broth mating procedure in Lurian-Bertani (LB) broth (Sigma-Aldrich, USA) with the KP5825 isolate as donor and the E. coli J53 azide resistant strain as recipient 43 . Overnight cultures of donor and recipient strains grown in LB broth were added to 8 mL fresh LB broth at a donor-recipient ratio of 1:1 (300 μL of cultures each), and incubated for 4 h at 37 °C. The mixed cultures were centrifuged and the supernatant was removed in order to get rid of the antibiotics, to avoid the inhibitory effect against E. coli J53. The pellet was re-suspended in fresh culture and plated onto a LB-agar containing 100 μg/mL azide (Sigma-Aldrich) and 0.1 μg/mL of cefotaxime (Sigma-Aldrich) and/or 0.1 μg/mL of ertapenem (Sigma-Aldrich) 44 . Colonies growing on the selective agar plates and again on subculture agar were subjected to confirmatory tests of ESBLs and carbapenemase by CTX-M Multi and Carba 5 immunochromatographic assays (NG Biotech, Guipry, France).

Mouse model of in vivo colonization with KP5825.
All experiments were carried out using 6-8 weekold C57BL/6 male mice weighted 24-26 g (Jackson Laboratory, Bar Harbor, Maine, USA) and housed in sterile cages with irradiated food and acidified water. Each group contained seven mice. For experiments involving antibiotic treatment, 0.5 g/L of ampicillin (Sandoz GmbH) was administered to animals in the drinking water for fourteen days and changed every day. For colonization experiments, 5 × 10 6 CFU of K. pneumoniae KP5825 was administered by orogastric gavage in a 200 μl volume on the fourteenth and fifteenth day of ampicillin pretreatment. After the oral colonization with KP5825 the following antibiotics-0.5 g/L ampicillin (Amp_0.5), 0.1 g/L ceftazidime (GlaxoSmithKline) (Caz_0.1), 0.5 g/L ceftazidime (Caz_0.5), 0.1 g/L ciprofloxacin (Bayer AG) (Cip_0.1) and 0.5 g/L ciprofloxacin (Cip_0.5)-were further administered to the animals in the drinking water for two weeks and changed every day (Fig. 1).
Mice were single-housed at the time of colonization experiment. Animals were maintained in a specific pathogen-free facility at Institute of Medical Microbiology, Semmelweis University. All mouse handling, cage changes and feacal pellet collection were performed in a biosafety level 2 (BSL-2) facility, with personnel wearing sterile gowns, masks and gloves.   Table 2). The qPCR was carried out in a Step One Real-Time PCR System (Applied BioSystems, Thermo Fisher Scientific) in default setting. The copy number of resistance gene results were evaluated using the 2 −ΔΔCt method 46 . Utilizing the 2 −ΔΔCt method, results are presented as the fold change of gene normalized to the rpoB1 reference gene and relative to the control mice. The number of rpoB1 housekeeping gene for the determinaton of the K. pneumoniae relative amount in the feces, and the bla CTX-M-15 and bla OXA-162 genes for determining the relative amount of resistance genes compared to KP5825 were determined on the first and on the fifteenth days.  www.nature.com/scientificreports/