Development of a novel selective agar for the isolation and detection of Bacillus anthracis

The aim of this study was to develop a novel selective agar for the specific isolation and detection of Bacillus anthracis.


Introduction
Bacillus anthracis is the causative agent of anthrax and occurs naturally in the environment. It is a dangerous pathogen for humans and can affect especially herbivores, due to its high pathogenicity and severe course of disease with high fatality rates depending on the route of infection (World Health Organization 1998;Turnbull 2008). The ability of B. anthracis to form highly stable spores, resistant to heat as well as many chemicals and disinfection agents, has resulted in efforts to abuse this feature for the development of biological weapons in the past (Goel 2015). Unfortunately, culture-based detection of B. anthracis from matrices with high concentrations of accompanying bacterial flora is a serious challenge. Especially soil samples contain only small concentrations of B.
anthracis spores in the presence of large bacterial numbers of closely related species of the Bacillus cereus sensu lato group and other environmental strains which render any attempts for isolation complicated, tedious and often unsuccessful (Silvestri et al. 2015). These difficulties of culture-based detection result to a great extent from the lack of adequate selective media (World Health Organization 1998). PLET agar (polymyxin, lysozyme, ethylenediaminetetraacetic acid and thallium acetate) is a selective agar which is specifically designed to isolate and select B. anthracis (Knisely 1966). However, the incorporation of highly toxic thallium acetate limits its use today due to stricter work safety and environmental protection regulations including a more diligent waste management than in the past (Tomaso et al. 2006;Turnbull 2008). In addition, it was shown that many other Bacillus strains are also able to grow on PLET and that the germination of B. anthracis spores is significantly reduced on this medium (Dragon and Rennie 2001;Klee et al. 2006;Marston et al. 2008). The majority of agars currently used aiming to isolate B. anthracis target the whole B. cereus sensu lato group, which includes species like B. cereus and B. thuringiensis. However, growth on these chromogenic or differential agars such as Cereus Ident TM Agar (CEI), mannitol egg yolk polymyxin B agar or sulfamethoxazole trimethoprim blood agar is not always easy to interpret and, thus, generally prone to misclassification (Juergensmeyer et al. 2006;Tomaso et al. 2006;Marston et al. 2008;Silvestri et al. 2015). Furthermore, many environmental samples, including soil, contain high numbers of members of the B. cereus sensu lato group which can easily overgrow B. anthracis on agar plates (Fasanella et al. 2013).
In this study, we developed and tested a novel selective agar suitable for environmental samples, termed CEFOMA (Bacillus CEreus sensu lato group-specific antibiotics, FOsfomycin, MAcrolides), with superior selectivity for B. anthracis compared to currently used selective or differential agar formulas.

Strains and cultivation
A detailed list of all Bacillus reference strains and environmental isolates (RKI ZBS 2 strain collection) is given in Table 1. All Bacillus strains were cultivated on tryptic soy agar as solid non-selective medium or in lysogeny broth as non-selective liquid medium under aerobic conditions at 37°C. For selective plating, spores or freshly grown bacterial cultures were used and agar plates were incubated for 24-48 h. To avoid CFU losses due to spore clumping, Triton X-100 was added to the working solutions of all spore preparations in a final concentration of 0Á01%. Spores were prepared on manganese sulphate agar following the instructions of DIN EN 14347:2005-08.
Preparation and use of selective agars 1 l CEFOMA agar (for approximately 50 agar plates) was prepared as follows: 38 g brain heart infusion broth modified (BD, Franklin Lakes, NJ) and 13 g agar (no. 1; Oxoid, Hampshire, UK) were dissolved in water and , fosfomycin disodium salt (dissolved in water, final concentration of 20 mg l À1 ; Sigma-Aldrich), erythromycin (dissolved in water, final concentration of 0Á03125 mg l À1 ; Carl Roth), azithromycin dihydrate (dissolved in water, final concentration of 0Á0625 mg l À1 ; Sigma-Aldrich) and defibrinated sheep blood (final concentration of 1%; Oxoid) were added.
In addition to the newly developed CEFOMA agar, two conventional types of selective and differential agar were used for comparison: The commercially available CEI (Merck) is selective for the B. cereus sensu lato group and contains the chromogenic substance 5-bromo-4chloro-3-indoxyl-myoinositol-1-phosphate which is converted by the enzymatic activity of many strains of B. cereus and B. thuringiensis resulting in green-blue colonies, whereas colonies of B. anthracis appear white (Tomaso et al. 2006). Trimethoprim sulfamethoxazole polymyxin blood agar (TSPBA) is also selective for the B. cereus sensu lato group and allows for haemolysis assessment with most strains of B. cereus and B. thuringiensis being typically haemolytic while B. anthracis is not haemolytic. 1 l TSPBA was prepared with Columbia agar and 5% sheep blood and was supplemented with 3Á2 mg trimethoprim, 16 mg sulfamethoxazole and 20 mg polymyxin B (in-house developed formula, adapted from Turnbull 2008).
To test the productivity (the ability to germinate and form colonies) of B. anthracis on CEFOMA, spore preparations were adequately diluted (to reach a theoretical colony number of 10-100 colonies per agar plate) and 100 µl of these dilutions were plated on CEFOMA and sheep blood agar (duplicates for fully virulent strains, at least triplicates for the three attenuated strains Sterne, Pasteur and Wirt). After incubation for 24-48 h, all colonies were enumerated and CFU counts between selective and full medium were compared. Additionally, this experiment was performed in a similar manner for vegetative cells using overnight cultures.
To evaluate the stability of CEFOMA after long-term storage at 4°C for up to 10 weeks, plates were tested every 2 weeks for growth of spores of B. anthracis (Sterne and Wirt) and inhibition of spores of B. cereus, B. thuringiensis and B. subtilis. Additionally, the number of colonies on CEFOMA obtained by plating two heat-treated soil samples (tested negative for B. anthracis) and after incubation for 48 h was enumerated every 2 weeks to screen for any increase in colony numbers due to diminished selectiveness caused by aged agar media.

Soil samples and deliberate spiking
Two different soil samples were used: Soil sample I (pH of 7Á7 after 1 : 10 dilution in water), characterized by a rather sandy composition with a low content of organic matter, was directly taken from the environment (a lawn in Berlin), soil sample II (pH of 7Á0 after 1 : 10 dilution in water) was taken from commercial potting soil (Edeka, Gut & G€ unstig) and was rich in humus. Analysis of soil composition and nutrient content of the two samples was carried out by Reblu GmbH (Bodenanalyse Zentrum, Filderstadt, Germany). Both samples were confirmed to be free of B. anthracis by PCR (after 45 min of incubation in LB medium, heat treatment at 95°C for 60 min and processing by DNeasy â Blood & Tissue Kit (Qiagen, Hilden, Germany) according to the pretreatment instructions for Gram-positive bacteria) and extensive plating on CEFOMA and CEI. 1 g of each soil sample was diluted 1 : 10 in water and heat treated for 30 min at 65°C to kill off the majority of vegetative cells. For deliberate spiking, 10 3 spores of B. anthracis Sterne or Wirt were added to the diluted soil samples (resulting in a final concentration of 100 spores per ml) prior to heat treatment. All samples were rigorously shaken after spiking and again after heat treatment before taking a sample for plating. 100 µl of this mixture was then plated directly on agar and incubated for 24-48 h at 37°C (theoretically resulting in, on average, 10 colonies of B. anthracis per plate). All colonies resembling typical morphologies of B. anthracis were picked and transferred to fresh plates. Heat lysates (95°C for at least 60 min) were prepared and directly used for colony PCR analysis. All experiments were performed as triplicates. For each experiment, soil samples were plated on two CEFOMA plates (100 µl each).

PCR
For the confirmation of picked isolates of B. anthracis, Sterne from spiked soil samples, real-time PCR was used to screen for the presence of the protective antigen (PA) using CGGATCAAGTATATGGGAATATAGCAA as forward primer, CCGGTTTAGTCGTTTCTAATGGAT as reverse primer and CTCGAACTGGAGTGAAGTGT-TACCGCAAA as the probe (modified with FAM-TAMRA) (Ellerbrok et al. 2002). Likewise, for the confirmation of picked isolates of B. anthracis Wirt, real-time PCR was used to screen for the dhp-locus using CGTAAGGACAATAAAAGCCGTTGT as forward primer, CGATACAGACATTTATTGGGAACTACAC as reverse TGCAATCGATGAGCTAATGAACAAT-GACCCT as probe (modified with FAM-TAMRA) (Antwerpen et al. 2008). All sequences are given in 5'-3' orientation. Amplification and detection were done with ABI PRISM 7500 (40 cycles: denaturing for 15 s at 95°C, annealing and elongation for 60 s at 60°C).

Agar development and specificity testing
A comprehensive literature research was performed to gather all available knowledge dealing with the selective isolation of B. anthracis and with available data on antimicrobial resistance. With the exception of PLET, all other agars used components which are thought to provide specificity for the whole B. cereus sensu lato group (Turnbull 2008). Therefore, we combined all these ingredients, that is, polymyxin and lysozyme (PLET) and the combination of trimethoprim and sulfamethoxazole (resulting in the Bacillus cereus sensu lato group-specific component, termed 'CE' component). Thallium acetate was omitted due to its high toxicity. EDTA was omitted because its effect is believed to be linked to the use of thallium acetate in PLET and its chelating activity (Bowen 2000). As expected, all members of the B. cereus sensu lato group were able to grow on this agar while growth of other Bacillus species like B. subtilis was reduced (data not shown). Next, the literature was systematically screened for antibiotic susceptibility testing of members of the B. cereus sensu lato group, including B. anthracis, and five antibiotics were identified by this literature search, which seemed to be suitable for a novel selective agar (Table 2). Not surprisingly, the susceptibility to most antibiotics among the B. cereus sensu lato group was, in general, rather similar. However, in case of the two macrolide antibiotics erythromycin and azithromycin, it was noted that strains of B. anthracis exhibited higher tolerance towards these substances than many other strains of the B. cereus sensu lato group (Yamamoto et al. 2001;Bryskier 2002;Cavallo et al. 2002;Mohammed et al. 2002;Frean et al. 2003;Jones et al. 2003;Schlegelova et al. 2003;Turnbull et al. 2004;Citron and Appleman 2006;Luna et al. 2007;Ikeda et al. 2015;Kreizinger et al. 2016). It is important to note that many strains of B. anthracis are defined as being sensitive to erythromycin and azithromycin (Turnbull 2008). However, they can tolerate the presence of these antibiotics in small amounts, while many other strains of the B. cereus sensu lato group cannot grow in the presence of the same antibiotic concentration (Table 2). When these antibiotics (macrolide component; 'MA') were combined with the CE component of the selective agar, strong synergistic effects (i.e. minimal inhibitory concentrations were smaller in the presence of other antibiotics than in non-selective growth medium) were observed and, thus, concentrations of azithromycin and erythromycin were reduced by 75% of the initially theoretically assumed tolerance limits based on the literature search (Table 2). Additionally, the natural resistance of B. anthracis against fosfomycin (while most other members of the B. cereus sensu lato group are susceptible) was exploited and consequently incorporated as the 'FO' component (Schuch and Fischetti 2006). Initially, the same concentration of fosfomycin was used as described in the Ground Anthrax Bacillus Refined Isolation (GABRI) method by Fasanella et al. (2013); but again synergistic effects (resulting from the presence of other antibiotics in the growth medium) were encountered and the fosfomycin concentration was lowered from initially 50 to 20 mg l À1 . Finally, sheep blood was added to CEFOMA as a non-selective component which allows for haemolysis testing and identification of non-target colonies. 5% sheep blood tended to reduce the selective properties of CEFOMA as observed in subsequent experiments (data not shown); thus, only 1% sheep blood was used, which was sufficient to distinguish between haemolytic and non-haemolytic strains.
As a first evaluation of the sensitivity and specificity of CEFOMA, 12 B. anthracis strains (including six fully virulent reference strains, three attenuated strains and three environmental isolates) and 19 other reference strains of the B. cereus sensu lato group (predominantly belonging to the species B. cereus and B. thuringiensis) were tested. All B. anthracis strains were able to grow on CEFOMA after 24-48 h of incubation (Fig. 1), albeit at a reduced speed, roughly estimated 2-3 times slower, compared to media without antibiotics like sheep blood agar (but apparently still faster than on PLET agar; Knisely 1966). Interestingly, some attenuated strains showed slower growth and more irregular, often translucent colony shapes compared with fully virulent strains (Fig. 1), which complicated enumeration after only 24 h of incubation resulting in slightly lower colony number counts compared to 48 h of incubation. In contrast, colony number counts for fully virulent stains did not differ between 24 and 48 h of incubation.
Next, colony numbers were enumerated to determine whether all plated bacteria (spores or vegetative cells) could form colonies. No colony loss was recorded for B. anthracis on CEFOMA agar compared to sheep blood agar without any additives as, on average, colony numbers were equal on the two growth media after plating equal amounts of bacteria (spores or vegetative cells). In contrast, all tested non-target Bacillus strains were either completely inhibited on CEFOMA or at least the CFU number was remarkably reduced by several orders of  (Table 1). Importantly, even unusual strains of B. cereus (e.g. B. cereus BW-A and BW-B) which behave very similar to B. anthracis as they do not show haemolysis or phospholipase C activity (and are thus hardly distinguishable from B. anthracis on differential media like CEI or TSPBA agar) were not able to grow on CEFOMA (Klee et al. 2006). Furthermore, long-time storage of CEFOMA plates at 4°C up to 1 month had no obvious effect on the inhibition of growth of non-target strains, nor did it effect growth of B. anthracis strains (data not shown). Only publications containing complete resistance data on B. anthracis are displayed (thus, exact upper and lower limit of MICs were stated in the publications). Data on antimicrobial resistance of other members of the B. cereus sensu lato group without data on B. anthracis (or with insufficient information) were also collected and analysed. These data are not shown here since those (inter-laboratory) results can vary remarkably due to different experimental procedures for MIC testing making direct comparisons between the studies difficult and unreliable. The MIC 50 value is the concentration required to inhibit 50% of the bacterial strains tested. Commercially available reference strains only reflect a small proportion of the true diversity found in nature.
To test the selectiveness of CEFOMA under more realistic conditions, two soil samples (tested negative for B. anthracis) were used to compare the amount of environmental background flora obtained on CEFOMA with conventional selective and differential media like TSPBA and CEI. Despite the heat treatment at 65°C to kill off most vegetative cells, plates of TSPBA as well as of CEI were overgrown within a day (Fig. 2). Notably, neither haemolysis activity on TSPBA nor the presence of 5bromo-4-chloro-3-indoxyl myo-inositol-1-phosphate in CEI were adequate to rule out the presence of B. anthracis. In contrast, overall colony numbers on CEFOMA were reduced by at least 1-2 log 10 levels compared with conventional selective agar (Fig. 2). Moreover, the remaining colonies were significantly smaller than B. anthracis and, additionally, haemolysis could be used to distinguish them from B. anthracis. Next, soil samples were spiked beforehand with B. anthracis Wirt or B. anthracis Sterne (resulting in a theoretical mean of 10 spores of B. anthracis per plate) and treated as described above (Fig. 3). B. anthracis could easily be identified on CEFOMA after 24 or 48 h and all potential isolates were, subsequently, confirmed by PCR.
Notably, 57 of 60 picked colonies on six CEFOMA plates were confirmed to be B. anthracis resulting in a recovery rate (or sensitivity) as well as a specificity of 95%.

Discussion
Despite the fact that the formula of PLET agar was first published more than half a century ago and contains highly toxic components, it remained the only truly selective agar for B. anthracis since other selective agars do not suppress the growth of the various B. cereus sensu lato group members. The difficulties to develop a selective agar can be attributed to the fact that no single antibiotic resistance or physiological property of B. anthracis (apart from the slightly higher tolerance towards thallium acetate) could be identified and exploited efficiently to suppress closely related species like B. thuringiensis and B. cereus (Bowen 2000;Turnbull 2008). Here, we present CEFOMA as a promising alternative to the highly toxic PLET agar as this novel agar was shown to be highly selective for B. anthracis enabling the efficient isolation and reliable detection of this pathogen without CFU loss. This agar is not relying on a single substance but instead combining fosfomycin, azithromycin and erythromycin, each suppressing a fraction of the closely related non-target strains. Due to the numerous antibiotics employed, it is conceivable that not all spores are able to germinate or, additionally, that the presence of fosfomycin might induce prophages in the genome causing reduced colony numbers (Schuch and Fischetti 2006). However, no such effects were observed in our experiments. During our literature search, three other antibiotics, rifampicin, daptomycin and linezolid, were also identified as potentially interesting substances (Table 2) which, apparently, can be tolerated by B. anthracis in higher concentrations than its related species (Dabbs et al. 1995;Bryskier 2002;Cavallo et al. 2002;Mohammed et al. 2002;Citron and Appleman 2006;Luna et al. 2007;Horii et al. 2011;Ikeda et al. 2015;Kreizinger et al. 2016). During our experiments, these substances were not able to improve CEFOMA selectivity with the tested panel of non-target reference strains. However, it cannot be ruled out that the inclusion of these antibiotics might further optimize the reduction of accompanying flora in complex soil samples. Furthermore, the incorporation of cycloheximide in CEFOMA as a fungicide might be useful, especially if the crucial heat pretreatment at 65°C to reduce the vegetative accompanying bacterial flora is not desired (e.g. in case vegetative forms of B. anthracis should be detected). 1% blood was added to CEFOMA to assess haemolytic activity. Alternatively, blood (which might influence selectivity) may be replaced by 5bromo-4-chloro-3-indoxyl myo-inositol-1-phosphate, the chromogenic component from CEI, to obtain similar information. However, this proprietary substance is considerably more expensive. Interestingly, the translucent appearance of strains of B. anthracis with only one or no virulence plasmid (e.g. Pasteur or Sterne) was already described for PLET agar as well (Marston et al. 2008;Luna et al. 2009).
Efficient plating media are an integral part to obtain bacterial isolates. In addition, liquid enrichment media are often used beforehand to lower the limit of detection. This would be highly useful for B. anthracis as well because the amount of spores in environmental samples is typically low. More research is thus needed to check whether the composition of CEFOMA can be a suitable starting point for an improved enrichment media.