Impact of the Timing of Antibiotic Administration on Digestive Colonization with Carbapenemase-Producing Enterobacteriaceae in a Murine Model

While antibiotic use is a risk factor of carbapenemase-producing Enterobacteriaceae (CPE) acquisition, the importance of timing of antibiotic administration relative to CPE exposure remains unclear. In a murine model of gut colonization by New Delhi metallo-beta-lactamase-1 (NDM-1)-producing Klebsiella pneumoniae, a single injection of clindamycin within at most 1 week before or after CPE exposure induced colonization persisting up to 100 days.

ABSTRACT

While antibiotic use is a risk factor of carbapenemase-producing Enterobacteriaceae (CPE) acquisition, the importance of timing of antibiotic administration relative to CPE exposure remains unclear. In a murine model of gut colonization by New Delhi metallo-beta-lactamase-1 (NDM-1)-producing Klebsiella pneumoniae, a single injection of clindamycin within at most 1 week before or after CPE exposure induced colonization persisting up to 100 days. The timing of antibiotic administration relative to CPE exposure may be relevant to infection control and antimicrobial stewardship approaches.

INTRODUCTION

Carbapenemase-producing Enterobacteriaceae (CPE) are an emerging public health issue, considered a critical priority by the World Health Organization (1). Among CPE, New Delhi metallo-beta-lactamase-1 (NDM-1)-producing Enterobacteriaceae are particularly preoccupying. Indeed, NDM-1 confers resistance to most β-lactams, including carbapenems, and has spread worldwide (2), raising fears of severe infections without therapeutic options (3).

In the hospital setting, contact with a CPE-colonized patient or prior antibiotic use are major risk factors for CPE acquisition (4, 5). Among antibiotics, antianaerobes (e.g., piperacillin-tazobactam or clindamycin) seem particularly at risk (6, 7).

While antibiotics are a known risk factor, the role of the timing of CPE exposure relative to antibiotic administration is unclear. We describe here a murine model of gut colonization with NDM-1-producing Klebsiella pneumoniae following a single administration of clindamycin and assess the effects of timing of clindamycin administration relative to CPE exposure on effective CPE colonization.

The French Ethical Committee for Animal Experimentation approved this study (APAFIS #7166). Seven-week-old C57BL/6 male mice housed under specific-pathogen-free conditions and a clinical isolate of Klebsiella pneumoniae-producing NDM-1–carbapenemase were used (8).

First, we validated the murine model of CPE gut colonization. Mice were divided into four groups with or without 24-h CPE exposure in drinking water (10⁷ CFU/ml at day 0) and/or intraperitoneal clindamycin (200 μg) (Fig. 1). The CPE load was evaluated by plating stool samples onto selective medium (lysogeny broth agar with 32 mg/liter cefotaxime and 6 mg/liter vancomycin). In mice exposed to both CPE and clindamycin, 8.3- to 8.7-log CPE/g stools were recovered at days 7 through 14 (Fig. 1A). In mice exposed to CPE without clindamycin, CPE load in stools briefly peaked at 5.6 logs of CPE/g of stool at day 2 but fell below the detection threshold from day 7 onward. Without clindamycin administration, there was no effective colonization.

FIG 1.

Gut microbiota alteration using clindamycin is a prerequiste for durable colonization with CPE alongside the intestinal tract. (A) Quantification of CPE load in stools depending on clindamycin administration and CPE exposure (five mice per group). Control group, no CPE exposure and no clindamycin; CPE exposure group, CPE in drinking water for the first 24 h (day 0) and no clindamycin; clindamycin group, no CPE exposure and intraperitoneal injection of clindamycin at day 0; CPE + clindamycin group, CPE in drinking water for the first 24 h (day 0) and intraperitoneal injection of clindamycin at day 0. (B and C) Relative abundance of bacterial phyla in terminal ileum samples using next generation 16S rRNA gene sequencing. The groups were as described for panel A (two mice per group). (D) Quantification of CPE load along the gastrointestinal tract. Five mice were exposed to CPE in drinking water for the first 24 h (day 0) associated with intraperitoneal injection of clindamycin at day 0. Luminal samples were taken at day 14.

Microbiota alterations observed with CPE exposure and/or clindamycin injection were analyzed by next-generation sequencing (n = 2 per group) of intraluminal samples taken at day 7 by 16S rRNA gene amplification by using an Ion 16S Metagenomics kit and sequencing with the Ion PGM System (Life Technologies, Carlsbad, CA). Bioinformatic analyses were performed using QIIME2 (9) and the R phyloseq package (10). CPE exposure without clindamycin did not alter gut microbiota compared to controls: more than 90% of 16S rRNA gene sequences were classified as Firmicutes, either Clostridiales or Lactobacillales (Fig. 1B and C). Clindamycin administration without CPE exposure led to an increase in Bacteroidetes (approximately 45%) and Proteobacteria, mainly Enterobacteriales (up to 26%). Clindamycin administration with CPE exposure resulted in a major increase in Proteobacteria (up to 84%), mostly Enterobacteriales (up to 76%).

To determine CPE colonization localization along the intestinal tract, intraluminal contents from the terminal ileum, cecum, and colon, as well as stool samples, were taken at day 7 in five CPE-exposed and clindamycin-treated mice. A higher load of CPE/g of digestive contents was found in the cecum (8.2 log ± 0.5), colon (8.0 log ± 0.4), and feces (8.1 log ± 0.2) than in the ileum (5.0 log ± 0.8, P < 0.0001) (Fig. 1D).

Finally, to determine the effects of timing of clindamycin administration relative to CPE exposure on effective CPE colonization, CPE-exposed mice received clindamycin at different days before or after CPE exposure (Fig. 2). Stool samples were collected twice weekly for a month and then once weekly for 2 months to assess the CPE load. Clindamycin injection at most 1 week before or after CPE exposure was necessary to achieve gut colonization (Fig. 2). Indeed, when clindamycin was injected at either days –21 and –14 or at days +14 and +21, the mice were not durably colonized, despite CPE exposure at day 0.

FIG 2.

Effective CPE colonization depends on the timing of clindamycin administration in relation to CPE exposure. A total of 21 mice were exposed to CPE for 24 h in drinking water (day 0). Clindamycin was injected once at one of the following time points before or after CPE exposure: day –21, day –14, day –7, day 0, day +7, day +14, or day +21.

The timing of antibiotics relative to CPE exposure is a key factor of effective CPE colonization. In our murine model, a single clindamycin injection within a week before or after CPE exposure induced gut colonization for at least 100 days. Remarkably, in a study predating the “omics” era, van der Waaij et al. also demonstrated 2-week selective window for persistent digestive colonization with streptomycin/neomycin-resistant Gram-negative bacteria following oral administration of these antibiotics (11). Other murine models of digestive CPE colonization have been described. In one model, when mice were exposed to KPC-producing Klebsiella pneumoniae 3 days after the first administration of clindamycin, the CPE load initially attained 10 log CFU/g of stool and then decreased to 5 to 6 log CFU/g of stool 5 days after the last clindamycin injection (6). In another model, in which mice were exposed to NDM-1-producing Escherichia coli 4 days after the first administration of vancomycin, metronidazole, and ceftriaxone (12), CPE loads in stools were 2 to 3 log CFU/g (close to the detection limit) 20 days after ending antimicrobial administration. In our model, with a single injection of clindamycin, a CPE load of >7 log CFU/g of stool persisted for 100 days. Furthermore, our study is original in assessing not only the window of opportunity for colonization after the administration of antibiotics but also before and showing that this opportunity exists both 1 week before and after the administration of antibiotics.

Gut microbiota analysis confirms that CPE exposure alone does not lead to colonization since Enterobacteriales, which CPE belong to, remained undetected 7 days after exposure, similar to unexposed controls. At 7 days after clindamycin administration, there was a marked decrease in Clostridiales from more than 75 to 30% at most, an increase in Bacteroidetes (45%), and an appearance of Enterobacteriales (23%), even without CPE exposure. Enterobacteriales are resistant to clindamycin; therefore, their expansion can be enhanced by clindamycin, as described previously in a murine model of Clostridium difficile infection (13). Interestingly, in their pre-“omics” demonstration of a selective window for digestive colonization induced by antibiotics, van der Waaij et al., using germfree mice recolonized with the flora from mice in which colonization was no longer possible late after antibiotic exposure, found (in preliminary experiments) that this flora conferring what was termed “colonization resistance” was characterized by a major proportion of Clostridiales (11). A more recent study using conventional culture methods showed that antibiotics allowed colonization by the extended-spectrum β-lactamases (ESBL) E. coli strain ST 131 regardless of effect on Bacteroidales (clindamycin) or not (cefuroxime and dicloxacillin) (14). These results suggest that the timing of antimicrobials relative to CPE exposure in providing ecologic space for implantation and expansion is an important parameter to consider.

Clinical studies focusing on healthy travelers at high risk of being exposed to CPE or other multidrug-resistant Enterobacteriaceae found a high rate of acquisition of multidrug-resistant Enterobacteriaceae (15). Although initially at 51%, colonization was short-lived since only 5% remained colonized 3 months after their return. In this population of healthy travelers, only 10% reported antibiotic use during their trip. While studies of acquisition and persistence of colonization by NDM-1-producing Enterobacteriaceae after travel to countries of endemicity concern few subjects, larger studies concentrating on ESBL-producing Enterobacteriaceae show that a significant risk factor is the use of antimicrobials during travel (16). These studies suggest that a healthy gut microbiota protects travelers exposed to CPE or other multidrug-resistant Enterobacteriaceae from long-term colonization (17).

Antibiotic use is a major risk factor of CPE colonization (5). In a context of up to 50% of antibiotic misuse (18, 19), avoiding an epidemic spread of CPE requires antimicrobial stewardship approaches. Our study suggests that more than the general notion of prior antibiotic use, it is the timing of antibiotics relative to CPE exposure that may be the main factor explaining colonization with CPE. This could be a crucial parameter to take into account in infection control and antimicrobial stewardship strategies. Indeed, our study shows there may be a specific window of opportunity for CPE colonization relative to antibiotics administration.