LBJMR medium: a new polyvalent culture medium for isolating and selecting vancomycin and colistin-resistant bacteria

Multi-drug resistant bacteria are a phenomenon which is on the increase around the world, particularly with the emergence of colistin-resistant Enterobacteriaceae and vancomycin-resistant enterococci strains. The recent discovery of a plasmid-mediated colistin resistance with the description of the transferable mcr-1 gene raised concerns about the need for an efficient detection method for these pathogens, to isolate infected patients as early as possible. The LBJMR medium was developed to screen for all polymyxin-resistant Gram-negative bacteria, including mcr-1 positive isolates, and vancomycin-resistant Gram-positive bacteria. The LBJMR medium was developed by adding colistin sulfate salt at a low concentration (4 μg/mL) and vancomycin (50 μg/mL), with glucose (7.5 g/L) as a fermentative substrate, to a Purple Agar Base (31 g/L). A total of 143 bacterial strains were used to evaluate this universal culture medium, and the sensitivity and specificity of detection were 100% for the growth of resistant strains. 68 stool samples were cultured on LBJMR, and both colistin-resistant Gram-negative and vancomycin-resistant Gram-positive strains were specifically detected. The LBJMR medium is a multipurpose selective medium which makes it possible to identify bacteria of interest from clinical samples and to isolate contaminated patients in hospital settings. This is a simple medium that could be easily used for screening in clinical microbiology laboratories.


Background
The worldwide emergence of multidrug-resistant (MDR) bacteria represents a major public health issue. Controlling the spread of these bacteria relies upon both reducing the prescription of antibiotics and preventing transmission from carrier patients to others [1]. More specifically, this prevention targets the emerging carbapenemase-producing Enterobacteriaceae and vancomycin-resistant enterococci (VRE) strains, with the development of specific detection methods, mostly based on chromogenic and selective culture media [2].
The increase in infections due to carbapenemaseproducing Enterobacteriaceae has led to the revival of colistin as a last-resort treatment [3]. Its widespread use, particularly in the livestock food in many countries [4], has inevitably led to the emergence of colistin-resistant strains over the past ten years [5,6]. Until recently, all colistin resistance mechanisms which had been described were attributed to chromosomic mutations [7]. In China in 2015, Liu and colleagues were the first to report the plasmid-mediated colistin-resistance gene in animals and humans, which they named mcr-1 [8]. This was followed by descriptions of the variants mcr-1.2 and mcr-1.3 [9,10], mcr-2, mcr-3, mcr-4 and mcr-5 genes [11][12][13][14]. A significant number of human cases (in the majority of cases asymptomatic carriers) was described [15,16]. Particularly, the worldwide analysis of colistin-resistant strains with an unknown mechanism allowed the detection of a consequent number of mcr-1-positive bacteria [17][18][19]. This mobile colistin resistance gene presents low resistance levels, with minimal inhibitory concentrations (MIC) of colistin around 4 μg/ml, which is close to the clinical breakpoint of colistin resistance (> 2 μg/mL), according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) [20]. Those data raised concerns about the detection and isolation of these pathogens, and the screening of the mcr-1 gene into carbapenemase-producing bacteria has been added to recommendations for clinical microbiology laboratories in France [21]. While carbapenemase screening is well-defined with phenotypic and molecular techniques [22], there is a need for colistinresistance screening tools [2]. Given the diversity of mechanisms of colistin resistance [23], phenotypic methods, such as chromogenic culture media [2,24] are preferred for a rapid detection.
In addition to colistin-resistant Enterobacteriaceae, the detection of vancomycin-resistant enterococci isolates is of clinical concern. The prevalence of VRE strains is increasing in Europe, especially Enterococcus faecium [25] and has led to nosocomial infections in the United States [26], and their dissemination is associated with high mortality rates [27]. The development of an effective screening tool for the early detection of those multidrug resistant pathogens has become a priority, in order to adapt treatment and isolate patients who are either infected or carriers.
For such purposes, we developed a polyvalent selective culture medium which can isolate both colistin-resistant Gram-negative bacteria, including Enterobacteriaceae strains harboring the mcr-1 gene, and vancomycin-resistant Gram-positive bacteria. In this study, we evaluated the performances of this medium on a representative set of bacterial strains and fecal samples. This universal culture medium was named LBJMR, standing for Lucie Bardet-Jean-Marc Rolain, and has recently been patented [28].
The colistin-resistance genes which were potentially involved were previously screened by RT-PCR for Enterobacteriaceae [6,44], as were the vancomycin resistance genes vanA and vanB for the VRE strains [45]. MICs of colistin and vancomycin were determined using E-test® strips (BioMérieux, Marcy l'Étoile, France) for Gram-negative and Gram-positive strains, respectively, and MICs of daptomycin were determined for the four reference strains of Enterococcus faecium (DSM 17050, 25,698, 25,697 and 13,590), and were evaluated according to the EUCAST guidelines [20].
In addition, 68 samples were cultivated on LBJMR medium, including 66 stool samples (56 from humans and ten from chickens) that were previously screened for mcr-1 by RT-PCR (56 positive and ten negative) and two clinical rectal swabs obtained from patients in the La Timone hospital, from which a vancomycin-resistant E. faecium VRE strain was isolated. Colonies with different morphologies isolated by culture of samples on LBJMR were isolated by subculture on Trypticase Soy Agar (TSA, Becton-Dickinson, Heidelberg, Germany), and the species were identified using Matrix-Assisted Laser Desorption-Ionization-Time Of Flight Mass Spectrometry (MALDI-TOF MS), as described previously [46].
The Purple Agar Base supplemented with glucose (7.5 g/L), colistin sulfate (0, 4 and 8 μg/mL) and vancomycin (0 and 50 μg/mL) (Sandoz, Levallois-Perret, France) was evaluated with 24 representative strains, including 20 Enterobacteriaceae (five intrinsic genera, eight with acquired resistance and seven colistin-susceptible strains as control) and four Gram-positive strains susceptible to vancomycin, which were inoculated at 10 6 CFU. This experiment was also conducted by replacing glucose with lactose (Laboratoires Humeau, La Chapelle-sur-Erdre, France). The optimum concentration of vancomycin for the medium composition was then determined using 20 Gram-positive strains representative of vancomycin resistance (three intrinsically-resistant genera and 17 enterococci, including seven VRE) and ten chicken stool samples from Algeria, five of which were positive for mcr-1 by RT-PCR, and five of which were negative [19,44].

LBJMR evaluation
The sensitivity and specificity of the LBJMR medium to detect all colistin-resistant Gram-negative bacteria and all vancomycin-resistant Gram-positive bacteria were evaluated with all the strains listed in Additional file 1: Table S1 and samples listed in Additional file 1: Table S2, which were incubated on LBJMR in an aerobic atmosphere at 37°C for between 24 and 48 h.
Inocula of the strains were prepared at 0.5 McFarland (corresponding to 1.5 × 10 8 CFU/mL according to the EUCAST expert system) and serial 10-fold dilutions were then performed in Phosphate Buffer Saline. 10 μL of these suspensions were deposited on agars and simultaneously on TSA medium to determine the detection limits for each bacterium by assessing the viable concentration by colony count. Growth or inhibition of these microorganisms was observed after overnight incubation at 37°C in an aerobic atmosphere. Those experiments were performed in duplicate.
Finally, as the LBJMR medium was developed so as to avoid Proteus swarming, Proteus mirabilis and Proteus vulgaris strains were inoculated alone or mixed with E. coli P10 (mcr-1) at the same concentration to evaluate the isolation ability of the LBJMR medium. Cultures were incubated at 37°C for 72 h.
The presence of the mcr-1 gene in those stools was first screened using RT-PCR as previously described [44]. Briefly, the DNA of the stools was extracted, after pretreatment by proteinase K and incubation at 95°C, using Biorobot EZ1 Advanced XL (Qiagen, Hilden, Germany). Positive DNA was confirmed using standard PCR. Samples with positive DNA were then cultured on LBJMR, after a prior enrichment step performed by inoculating approximately 1 g of the sample in tryptic soy broth (TSB) medium (BioMérieux, Marcy l'Étoile, France) and incubating for 24 h at 37°C. 100 μL of these liquid media were then plated on agars that were also incubated at 37°C for 24 h.
For each Gram-negative isolate identified by MALDI-TOF MS, colistin MIC was determined with a colistin E-test® strip. For each colistin-resistant isolate, the presence of mcr-1 was investigated by RT-PCR as previously described [44]. Susceptibility testing to 26 antibiotics was performed on mcr-1-positive strains using the disk diffusion method, and interpreted according to the EUCAST guidelines [20].

Comparison of the LBJMR medium with other polymyxincontaining media
The sensitivity and specificity of the LBJMR medium was first compared to the BD™ Cepacia Medium (Cepacia, Becton-Dickinson, Heidelberg, Germany) medium with all the colistin-resistant Enterobacteriaceae strains listed in Additional file 1: Table S1. The LBJMR medium was then compared to the SuperPolymyxin medium with 103 bacterial strains, including 61 of the Enterobacteriaceae, 17 non-fermentative Gram-negative, and 25 Gram-positive strains (Additional file 1: Table S1). Other polymyxin-containing media were also compared concomitantly to the LBJMR medium, using EMB Agar (Sigma-Aldrich, Illkirch, Germany) as a control, with some of these strains as described below. Inocula were prepared following the same protocol as for the LBJMR evaluation.
Columbia Colistin Nalidixic Acid Agar +5% sheep blood (CNA) (BioMérieux, Marcy l'Étoile, France) and three mixed culture media: EMB with colistin sulfate (4 μg/mL) and vancomycin (50 μg/mL), LBJMR with Daptomycin (10 μg/ml, Novartis, Horsham, United Kingdom) in place of vancomycin and LBJMR with Amphotericin B (5 μg/ml, Bristol-Myers Squibb, Rueil-Malmaison, France) were concomitantly tested with 45 Enterobacteriaceae strains representative of colistin resistance (30 mcr-1, ten susceptible and five intrinsically resistant genera). LBJMR with Amphotericin B was also evaluated with the eight VRE strains. Finally, the LBJMR medium was compared to the SuperPolymyxin medium with 12 samples, including the two clinical rectal swabs from the La Timone hospital and ten human stools from Thailand, five of which had been identified as positive by culture for the detection of an isolate harboring the mcr-1 gene, and five controls that were negative for mcr-1 using RT-PCR.

Statistical analysis
The data were analyzed using a Chi-test to compare the Cepacia and LBJMR media and a Student-t test for a pairwise comparison of SuperPolymyxin and LBJMR selective media for the detection of non-fermentative Gram-negative colistin-resistant strains. Significance was assessed at p < 0.05.

LBJMR development
Because colistin is a cationic molecule, we looked for a medium which was deprived of any ions in its original composition and which was also deprived of any electrolytes to avoid Proteus spp. swarming. The optimal conditions for LBJMR medium (growth of colistin-and vancomycin-resistant strains and inhibition of colistinor vancomycin susceptible strains) were obtained on Purple Agar Base with glucose (Additional file 1: Table  S3), 4 μg/mL of colistin sulfate and 50 μg/ml of vancomycin (Additional file 1: Table S4). The culture of ten chicken stools from Algeria enabled the detection of E. coli strains harboring the mcr-1 gene from three of the five samples which were positive for mcr-1 by RT-PCR, including strain 235 which had previously been isolated on Cepacia medium [19] and no Gram-negative strains were detected from the five negative samples (Additional file 1: Table S2). A correlation was observed between the Ct-values that reflect the DNA concentration and the results of culture (Additional file 1: Table S2).
The final preparation of the LBJMR medium was as follows: 31 g/L of Purple Agar Base, 7.5 g/L of glucose, 4 μg/mL of colistin sulfate and 50 μg/mL of vancomycin. 7.5 g/L of lactose was used in place of glucose for the 24 strains initially tested and the same results were obtained. Glucose was selected because, combined with bromocresol purple, a pH indicator, it revealed the polyvalent capacity of the LBJMR medium with an easy and fast visualization of the species of interest in the clinical routine: both Enterobacteriaceae and enterococci gave yellow colonies, contrasting on the purple agar and exhibiting different sizes (2-3 mm and 0.1-1 mm, respectively).

LBJMR evaluation
All the polymyxin-resistant Enterobacteriaceae strains (n = 81) tested were able to grow on the LBJMR medium, with the lowest detection limit (10 1 CFU), including those with the mcr-1 gene, as well as all the colistin-resistant non-fermentative Gram-negative (n = 10) and vancomycin-resistant Gram-positive strains (n = 11), with detection limits dependent on their MIC, as shown in Table 1. Meanwhile, all the bacteria susceptible to colistin or vancomycin were inhibited (not detected at 10 5 CFU), giving rise to 100% sensitivity and specificity for the growth of Gram-negative colistin-resistant and Gram-positive vancomycin-resistant strains ( Table 1). The swarming of Proteus sp. strains was fully inhibited on LBJMR, even after 48 h of incubation.
Of the 1052 stools screened for mcr-1 by RT-PCR, 66 were cultured on LBJMR, including 56 that were positive after 32 cycles of RT-PCR, as well as ten negative stool samples as controls. 16 cultures were positive on LBJMR medium, enabling the detection of 17 colistin-resistant Enterobacteriaceae strains harboring the mcr-1 gene (colistin MICs ranging from 2 to 8 μg/mL): 15 were identified as E. coli and two as K. pneumoniae by MALDI-TOF MS (Additional file 1: Table S2). Indeed, two bacterial species, E. coli and K. pneumoniae, were detected from the culture of the same sample (FHM128) (Additional file 1: Table S2). Five isolates were already known from a previous study on Cepacia medium [6,19]. All the E. coli isolates from France were Extended-Spectrum Beta-Lactamase-producing strains (data not shown). In conclusion, the LBJMR medium enabled the isolation of new colistin-resistant strains harboring the mcr-1 gene. The weak culture results compared to PCR could be explained by the length of storage of the tested samples (up to four years) at −80°C, possibly inducing the death of bacteria while their DNA was still detectable [47]. However, it should be noted that, for some samples, we isolated another kind of strain, mostly Gram-positive bacteria which have an intrinsic resistance to vancomycin, largely Pediococcus pentosaceus and Weissela cibaria, as shown in Additional file 1: Table S2.
Finally, the culture of two clinical samples from the hospital enabled the isolation of two vancomycin-resistant E. faecium isolates (VRE1 and VRE2). In addition, two colistin-resistant Enterobacteriaceae isolates were also detected in the VRE2 sample, identified as K. pneumoniae LB3 (MIC = 64 μg/mL) and E. coli LB4 (MIC = 16 μg/mL), both negative for the mcr-1 gene (Fig. 1).

Comparison between LBJMR and other selective polymyxin-containing media
The LBJMR medium was more sensitive than the Cepacia and CNA media for detecting the strains exhibiting the mcr-1 gene (Additional file 1: Table S5). Indeed, only four of those strains could grow on the CNA medium, because it also contains nalidixic acid and fewer than half of them grew on the Cepacia medium. Comparison of the sensitivities of the LBJMR (100%) and Cepacia (47%) media using a Chi-square statistical test gave a significant p-value (<10 −4 ).
LBJMR and SuperPolymyxin media presented the same sensitivity for Enterobacteriaceae, with the growth of all colistin-resistant strains (detection limit of 10 1 CFU), including the 30 mcr-1 strains tested, and the inhibition of all susceptible strains (Table 1). In addition, the culture of ten human stools gave the same results: five were positive, with the detection of an E. coli positive for mcr-1, and the five controls remained negative (Additional file 1: Table S2). In contrast, the LBJMR medium showed a significantly higher sensitivity than the SuperPolymyxin medium for the detection of colistin-resistant non-fermentative Gram-negative strains, with a significant p-value of 0.019 (<0.01) with a pairwise comparison using a Student-t test (Table 1). More  specifically, the LBJMR medium was able to detect those screened in the cystic fibrosis samples, such as Burkholderia cepacia, even at low concentrations, and could replace specific media such as the Cepacia medium for their isolation. The addition of amphotericin B or the replacement of vancomycin by daptomycin in the LBJMR medium did not affect the growth of Enterobacteriaceae strains (Additional file 1: Table S5). Finally, all the VRE strains were inhibited on the SuperPolymyxin medium because of the presence of daptomycin, while they were detected on the LBJMR medium (Table 1).
These results were confirmed by the culture of the two clinical samples from the hospital: vancomycin-resistant E. faecium could only grow on the LBJMR medium and not on the SuperPolymyxin medium for both samples, and the two Enterobacteriaceae from the VRE2 sample found on the LBJMR medium were also detected on the Super-Polymyxin medium.
Finally, the concomitant addition of colistin and vancomycin to EMB agar showed the systematic inhibition of E. coli strains harboring the mcr-1 gene, corresponding to the tested strains with the lowest MIC.

Discussion
In accordance with recommendations aiming to isolate patients carrying multi-drug resistant bacteria, it is becoming essential to detect colistin-resistant Gramnegative pathogens, particularly with the recent description of plasmid-mediated colistin-resistant genes among Enterobacteriaceae strains [8][9][10][11][12]. We do not currently know the potential consequences of patient colonization by those resistant bacteria, as it has already been demonstrated that acquisition of resistance is associated with a decrease in virulence [48][49][50], but it is necessary to implement efficient tools to prevent and monitor as closely as possible any epidemic outbreak and develop a rapid therapeutic strategy. Currently, the only available method for detecting colistin resistance, according to joint EUCAST and Clinical Laboratory Standard Institute (CLSI) expert systems, is antibiotic susceptibility testing using broth microdilution, which is not suitable for daily screening in clinical microbiology laboratories. Current polymyxin-containing culture media are also not appropriate because of their high concentration of polymyxin. Indeed, they were not developed to isolate Enterobacteriaceae but to specifically detect bacteria that are intrinsically resistant to polymyxins, avoiding contaminants such as Enterobacteriaceae species. Thus, the establishment of an effective protocol to detect colistin resistance with the development of a reliable culture medium was necessary in clinical microbiology laboratory.
The significant number of strains tested on our selective medium showed that the LBJMR medium allows all colistin-resistant Enterobacteriaceae strains to grow, even those with a very low MIC for colistin and harboring the mcr-1 gene, when all the susceptible strains were inhibited. Furthermore, the screening conducted on stool samples that were positive for mcr-1 by PCR allowed the isolation of new mcr-1 strains (Additional file 1: Table S2). The LBJMR medium could also detect colistin-resistant non-fermentative Gram-negative bacteria, including pathogens which are often found in patients suffering from cystic fibrosis (Table 1). We thereby created a universal culture medium, able to replace specific polymyxincontaining culture media usually used in routine diagnosis. All the vancomycin-resistant bacteria tested were detected at low concentrations on the LBJMR medium, including the VRE strains, when susceptible were inhibited. To attest this capacity, the LBJMR medium should be tested on more VRE strains and comparatively to VRE selective culture media.
The performances of the LBJMR medium are summarized in Table 2: samples can be cultured directly on the LBJMR selective medium, without a previous decontamination step, and colonies can be directly analyzed from the primary culture, without subcultures. Bacteria of interest are easily recognizable, as shown in Fig. 1: both colistin-resistant Enterobacteriaceae and vancomycinresistant enterococci exhibit yellow colonies (2-3 mm and 0.1-1 mm respectively), on a purple agar base. Because LBJMR also permits the growth of intrinsic resistant bacteria, screening should be completed with a rapid identification using MALDI-TOF MS [51], directly from the LBJMR medium. Antibiotic susceptibility testing and PCR screening can be performed in the same time.
Nordmann et al. developed a different selective medium called SuperPolymyxin, based on EMB agar and containing 3.5 μg/mL of colistin sulfate, 10 μg/mL of daptomycin and 5 μg/mL of amphotericin B [52], associated with a complementary phenotypic test, the rapid polymyxin NP [53], to screen for colistin-resistant Gram-negative bacteria [54]. These two media are differential and are able to recognize bacterial species based on color. Sensitivities for colistin-resistant Enterobacteriaceae were the same for the SuperPolymyxin and LBJMR media, but only the LBJMR medium could detect VRE strains (Table 1). Indeed, the SuperPolymyxin medium is composed of daptomycin and amphotericin B, which prevents the growth of contaminants including bacteria of interest such as VRE strains, because they are mostly susceptible to daptomycin. In our study, we tested stool samples and obtained some contaminants, including Gram-positive bacteria which were intrinsically resistant to vancomycin (Additional file 1: Table S2) or some yeasts.
More recently, the culture medium ChromAgar COL-APSE was developed and compared to the SuperPolymyxin medium [55]. The comparison of the LBJMR and the CHROMAgar COL-APSE performances have to be assessed. All those media should be also evaluated on heteroresistant strains, as they are difficult to detect and their frequence is widely underestimated [56]. The advantage of the LBJMR medium in terms of isolating vancomycin resistant enterococci can be also a disadvantage in a few cases, because Gram-positive bacteria which are intrinsically resistant to vancomycin bacteria, including Pediococcus, can also be isolated. In the LBJMR medium, Gram-positive bacteria appear with small colonies and a simple identification by MALDI-TOF can identify them. For the elimination of yeast in samples, amphotericin B can be added in our medium.

Conclusion
The LBJMR medium is an adequate screening tool for all colistin-resistant isolates from clinical samples, independently of their resistance level or mechanism. LBJMR could be used in routine laboratory work to detect colistinresistant and vancomycin-resistant bacteria, allowing for the direct analysis of colonies and, thus, the early isolation of contaminated patients in hospital settings. This medium is currently being investigated as a routine medium in our institute to determine its usefulness in detecting such bacteria.

Additional file
Additional file 1: Table S1. List of the studied strains. The MIC of colistin (CT) or vancomycin (VCN) are indicated in μg/mL. Table S2