Susceptibility to antibiotics in isolates of Lactobacillus plantarum RAPD‐type Lp299v, harvested from antibiotic treated, critically ill patients after administration of probiotics

Abstract Recultured Lactobacillus plantarum 299v‐like strains were tested regarding antibiotic susceptibility, and no decrease was detected. Antibiotics are frequently used to treat patients in intensive care units (ICUs) and are associated with a significant risk of selection of resistant bacterial strains. In particular, it is possible that genetic transfer of antibiotic resistance to the resident gastrointestinal flora, as well as to administered probiotics, may be increased in the ICU setting. The aim of the present investigation was to detect possible changes in antimicrobial susceptibility in reisolates of the probiotic strain Lactobacillus plantarum 299v (Lp299v) given to antibiotic treated, critically ill patients. Lp299v‐like strains were identified in cultures of biopsies and fecal samples from 32 patients given the probiotic strain enterally in two previous ICU studies. The patients received a variety of antibiotics. Isolates with the same genomic RAPD profile (RAPD‐type) as Lp299v were obtained to enable monitoring of antibiotic susceptibility by E‐tests. Forty‐two isolates, collected throughout the course of illness, were tested against 22 different antibiotics. No obvious decrease in susceptibility was found for 21 of the tested antibiotics. There was a tendency toward decreased susceptibility to ampicillin. The stable antibiotic susceptibility profiles of the Lp299v‐like isolates studied here suggests this probiotic is less likely to acquire resistance when administered to critically ill patients treated with broad‐spectrum antibiotics.

controls (Gu, Wei, & Yin, 2012;Wang et al., 2013). However, little is known about a number of critical issues in this context, such as the possible impact of probiotics on the normal microbiological flora (Imperial & Ibana, 2016), and the effects of administered drugs on the antibiotic susceptibility of probiotics. Some experimental animal data support the suggestion that antibiotic resistance genes are transferred to probiotic strains (Mater, Langella, Gérard Corthier, & Flores, 2008), although this aspect has not been studied in the hospital environment.
The highest antibiotic pressure is in the intensive care units (ICUs), where patients with few exceptions are given antibiotics, mainly broad-spectrum agents, as an important part of their treatment. Therefore, in ICUs there is significant ongoing selection of bacteria with resistance to antibiotics (Karam, Chastre, Wilcox, & Jean-Louis Vincent, 2016;Zilahi, Artigas, & Martin-Loeches, 2016). The colon can be regarded as a fermenter that has a high bacterial content, and it is highly likely that there is exchange of different molecules including genetic material. Probiotics are used in many ICUs mainly in order to reduce the prevalence of antibiotic-associated diarrhea. Nonetheless, we hypothesized that transfer of antibiotic resistance to probiotics could occur in this context.
Lactobacillus plantarum is considered to be a genomically stabile Lactobacillus plantarum 299v (Lp299v; DSM9843) is the probiotic component of a number of products that have been commercially available for more than three decades in Sweden and for several years in other countries. Moreover, Lp299v in the form of a fruit drink (ProViva ® ) has been used as a prophylactic remedy in many Swedish hospitals, including ICUs. To date, several ICU studies using the strain Lp299v (Klarin, Johansson, Molin, Larsson, & Jeppsson, 2005;Klarin, Wullt, et al., 2008;McNaught, Woodcock, Anderson, & MacFie, 2005), and the genomically closely related L. plantarum 299 (= DSM 6595)  have not shown any important side effects of these probiotics.
Two earlier ICU-based studies of critically ill patients receiving broad-spectrum antibiotics addressed the following issues: (1) whether administration of Lp299v resulted in adherence of the probiotic to the gut mucosa (Study 1, Klarin et al., 2005) and (2) whether coadministration of Lp299v with antibiotics reduced colonization of Clostridium difficile (Study 2, Klarin, Wullt, et al., 2008). In the present investigation, 42 isolates of RAPD-type Lp299v cultured from rectal mucosal biopsies (Study 1) or fecal specimens (Study 2) were investigated to determine whether antibiotic susceptibility profiles were altered after gastrointestinal transit in patients on broad-spectrum antibiotics. The results showed no significant changes in the antibiotic susceptibility profiles of Lp299v-like isolates screened against a panel of 22 broad-spectrum antibiotics. This finding together with the low incidence of side effects observed in Studies 1 and 2, indicates that the probiotic Lp299v should be a viable candidate for use in critically ill patients receiving antibiotics.

| Samples
In two separate prospective randomized controlled investigations (Studies 1 and 2) the probiotic strain Lp299v was given enterally to critically ill adult patients anticipated to require intensive care for ≥3 days. Eight patients in Study 1 and 22 patients in Study 2 were given Lp299v. The control groups (seven and 22 patients, respectively) received similar standard treatments but no probiotics isolates were obtained from biopsies of the rectal mucosa (Study 1), or from fecal samples (Study 2). In both studies samples were collected at inclusion and thereafter twice a week until discharge from the ICU. The patients were treated with different types of antibiotics, initially empirically and thereafter in accordance with clinical findings and the results of microbiological cultures. All patients had received one or more doses of antibiotics before any study product was given and before the first biopsy or fecal sample.

| Analyses
Lactobacilli were cultured from rectal mucosal biopsies or fecal specimens at the time of Studies 1 and 2 using Rogosa agar (Oxoid, Basingstoke, Hampshire, England) incubated anaerobically at 37°C for 3 days. Colonies suspected to be Lp299v (large, creamy whiteyellowish, and somewhat irregular in shape) were isolated and further identified by randomly amplified polymorphic DNA typing (RAPD) (Johansson, Quednau, Molin, & Ahrné, 1995). All isolates were stored at −80°C pending analysis. After reconditioning of the frozen strains, Brucella broth suspensions of the respective strain, were inoculated on Brucella agar plates (Oxoid). E-test strips of 22 antibiotics (detailed below in section 2.3) were applied on the inoculated agar plates, and incubated anaerobically at 35°C for 72 hr. All analyses were done in duplicate.
To enable comparisons of MIC values of the harvested isolates and the original strain, the isolates were divided into four groups according to their exposure to antibiotics and administration of Lp299v: (1) isolates cultured from samples collected prior to probiotic intervention in each study; (2) isolates obtained from rectal mucosa biopsies; (3) isolates acquired from fecal samples; (4) isolates obtained from fecal samples from patients in Study 2 given Lp299v (as the fruitdrink ProViva ® ) after conclusion of their participation in the study. The subjects in group 4 were given the control product (without probiotics) during the study period, but then as non-ICU patients at the same Department of Infectious Diseases received the same treatment as other patients in that ward.

| Tested antibiotics
A wide panel of antibiotics (E-tests; AB Biodisk, Solna, Sweden) was tested against the original strains and the Lp299v-like isolates.
The following antibiotics were given to one or more of the study participants were ampicillin, cefotaxime, ceftazidime, cefuroxime, clindamycin, erythromycin, gentamicin, imipenem, levofloxacin, meropenem, metronidazole, netilmicin, piperacillin, tobramicyn, trimethoprim (to patients given in combination with sulfametoxazole), and vancomycin. To fulfill regulatory requirements for the company holding the patent on Lp299v (Probi AB), some antibiotics seldom used in Sweden were also tested namely; cefepime, chloramphenicol, kanamycin, linezolid, quinupristin/dalfopristin, and streptomycin.

| Studied samples
Forty-two cultured Lp299v-like isolates were analyzed together with the original strain (Lp299v), and with the genomically closely related strain Lactobacillus plantarum 299.
The strain L. plantarum 299 (DSM6595) has been used in studies of ICU patients conducted by our group  and by other researcher (Oláh, Belágyi, Issekutz, Gamal, & Bengmark, 2002;, and therefore it was also of interest to evaluate the antibiotic susceptibility of this probiotic. Six isolates were from samples taken at study inclusion (prior to probiotic intervention), and 24 were from samples collected during the study periods. From three patients in the control group (not given Lp299v) in Study 2, 16 samples (12 of them positive for Lp 299v) were obtained after conclusion of the actual study period. These patients required an extended care period with further antibiotic therapy after their stay in ICU, and as ordinary F I G U R E 1 Flow diagram showing the number of patients that participated in Studies 1 and 2 ward patients they received routine care that included the fruit drink ProViva ® .

| MIC determinations
The MIC values determined for Lp299v and L. plantarum 299 were within a twofold (1 + 1) dilution (Table 1)  The tested isolates were exposed to different antibiotics and combinations of antibiotics and to the environment in the GI tract for varying periods of time. In almost all cases the drugs were given intravenously, and cephalosporins and carbapenems were used most frequently. In patients with several subsequent isolates, we found no gradual change in susceptibility. Four of the six patients with isolates from the sample taken at study inclusion had been treated with antibiotics for 1-20 days before start of the investigation. Our results demonstrate that the antibiotic susceptibility of the probiotic bacterium Lactobacillus plantarum 299v remains stable also after passage through the GI tract in patients treated with broad-spectrum antibiotics.

| D ISCUSS I ON
The present evaluation was performed on frozen samples that had been collected in two previous studies and stored at −80°C (Studies 1 and 2) and our aim was to detect changes in antibiotic susceptibility of a probiotic Lactobacillus strain given to intensive care patients.
These patients were treated with broad-spectrum antibiotics administered parenterally, and hence major changes in the microbiota could be anticipated, in particular selection of antibiotic-resistant bacterial species or clones. Conceivably, this situation would favor dissemination of antibiotic resistance that would include the probiotic strain in focus in our study. However, in two cohorts of patients, we were unable to detect any changes in susceptibility of the probiotic strain to a number of antibiotics, most of which had been used clinically at our ICU. Furthermore, we found no changes from susceptibility to resistance for any of the tested antibiotics. It was impossible for us to conclude that transitions cannot occur, because MIC breakpoints are lacking for many of the antibiotics widely used to treat critically ill patients. Notwithstanding, the MIC values for Lp299v for the antibiotics considered, (e.g., cephalosporins and carbapenems) were all low, indicating that development of resistance is less likely.   Lp299v S1/6 S1/8 S1/13 S2/III-20 S2/IV-1 S2/IV-4 Lp299v S1 2 -4 S1 2 -11 S1 2 -14 S1 8 -22 S1 14 -8 S1 15 -4 S1 15 -7 S1 X -Y Other species and strains used as probiotics show intrinsic resistance to several antibiotics (Wong, Ngu, Dan, Ooi, & Lim, 2015), although with different patterns compared with L. plantarum 299 and

SZ/X-Y
Lp299v. However, the changes in antibiotic susceptibility of bacterial strains used as probiotics have been studied and have in some cases been found to be due to increased activity in cellular efflux pump mechanisms (Thumu & Halami, 2012). This may explain the observed change in the MIC of ampicillin.
Lp299v and L. plantarum 299 have been used in several clinical studies without any reports of infections with pathogens showing extended antimicrobial resistance that might have originated from these two strains. Therefore, we focused the present postexposure survey focused on these probiotics.
The GI tract is estimated to harbor 600 or more bacterial species, some of which exhibit intrinsic or acquired resistance to various antibiotics. Inevitably, such species or strains may be positively selected to varying degrees during antibiotic treatment, as exemplified by the occurrence of antibiotic-associated diarrhea, which is often caused by the opportunistic pathogen Clostridium difficile (Kevin, Brown, Khanafer, Nick Daneman, & Fismana, 2013;Napolitano & Edmiston, 2017;Vardakas, Trigkidis, Boukouvala, & Falagas, 2016).

An investigation of pharyngeal streptococci in healthy volunteers
demonstrated that even short periods of treatment with macrolides Increasing occurrence of multidrug-resistant cocci is becoming a serious problem in many clinics and above all in critically ill patients (Lee, Lee, Park, Jeong, & Lee, 2015;Munita, Bayer, & Arias, 2015).
Selective decontamination has been applied for many years and has been considered not to carry any risk of development of resistant bacteria, although modern techniques have disproved that assumption (Buelow, Bello González, & Verslius, 2014;Buelow et al., 2017).
In contrast the use of probiotics has appeared as a complement in the prevention of emergence of antibiotic resistance (Ouwehand, Forssten, Hibberd, Lyra, & Stahl, 2016). use of probiotics as a complement in restoring a microbial balance has provided promising results. It is necessary to find suitable commensals of human origin for use in approaches developed to reestablish balance and colonization resistance (Blaser, 2016;Pamer, 2016) is of vital importance. The results of our investigation identify Lp299v, alone or more likely in combination with other strains as a possible candidate for use in this context. Lp299v was originally detected in human feces.
Clearly, it is necessary to consider the question of whether the analyzed strains actually constituted bacteria that had been exposed long enough to the conditions that prevail in the endogenous fermenter-the colon. We have previously demonstrated that Lp299v becomes established on the rectal mucosa to the same extent in antibiotic treated, critically ill patients as in healthy volunteers (Klarin et al., 2005). Introduced exogenous bacteria and fungi (probiotics or other microorganisms) are chiefly visitors to the GI tract, whereas the commensal microbiome is by definition prone to remain. Administered Lp299v bacteria adhere to the mucosa and stay there for shorter or longer periods of time. The mechanism for this adhesion occurs via a mannose entity on the epithelium cell site (Adlerberth et al., 1996) and hence it is plausible that Lp299v first adheres to the mucosa, after which the bonding at that location is disrupted and the bacteria are transported with the luminal content and then adhere at another binding site a bit further down the gut.
This process is repeated and might be considered analogous to a chromatography column. In addition, since the Lp299v thrive in the gut, new generations of this strain-with the same genetic profileshould be released. Inasmuch as Lp299v has been found to remain adhered to the rectal mucosa long after cessation of oral administration (Johansson et al., 1993), it is reasonable to assume that the mannose-to-bacterium adherence bond is fairly strong. This implies that the exposure time necessary for exchange of genetic material in the GI tract should be sufficient, and also that the feces samples taken can actually can represent exposed bacteria and not merely bacteria that have rapidly passed through the gut.
In our study, the MIC values were not identical for the exposed bacteria and the native strain, which strengthens our argument that the cultured strains represent bacteria that had been properly exposed to the fermentative environment of the GI tract. It is distinctly possible that the retrieved Lp299v bacteria had adhered to the mucosal epithelium several times in different locations during their journey to the rectum or that they were indeed new generations of this strain.
We found no changes in antibiotic susceptibility, and thus we conclude that the Lp299v genome is stable. Therefore it can also be assumed that there is little risk that Lp299v genetic material will spread to other species.
Conditions in the human (and animal) GI tract, and especially in the colon, are suitable for genetic exchange between species (Salyers, Gupta, & Wang, 2004), with some species being more prone than others to act as donors and/or recipients. Furthermore, animal studies have provided overwhelming evidence for in vivo   (Moubareck, Bourgeois, Courvalin, & Doucet-Populaire, 2003;Tannock, Bateup, & Jenkinson, 1993). It also seems likely that bacteria transiently colonizing the intestine (e.g., probiotics) can take part in the exchange of resistance genes. Accordingly, as part of the safety profile of probiotics in preparations marketed as health products and for prophylactic use in hospitals, it should be confirmed that the bacteria strains in these products are not prone to development of antibiotic resistance, and this should be demonstrated by available in vitro and in vivo methodology.
In conclusion, we found no evidence that Lactobacillus plantarum 299v is prone to acquiring genetic material coding for antimicrobial resistance in antibiotic treated, critically ill patients. In the context of susceptibility, L. plantarum 299v is also stable against antimicrobial agents in clinical settings with high antimicrobial pressure.

ACK N OWLED G M ENTS
This study was supported by grants from Region Skåne, Sweden, the Scandinavian Society for Antimicrobial Chemotherapy Foundation, and an unconditional grant from Probi AB, Lund, Sweden.

CO N FLI C T O F I NTE R E S T
GM and BJ are stock holders in Probi AB.