Inhibition of Cereulide Toxin Synthesis by Emetic Bacillus cereus via Long-Chain Polyphosphates

Abstract

Severe intoxications caused by the Bacillus cereus emetic toxin cereulide can hardly be prevented due to the ubiquitous distribution and heat resistance of spores and the extreme thermal and chemical stability of cereulide. It would therefore be desirable to inhibit cereulide synthesis during food manufacturing processes or in prepared foods, which are stored under time-temperature abuse conditions. Toward this end, the impacts of three long-chain polyphosphate (polyP) formulations on growth and cereulide production were examined. The inhibition was dependent on the concentration and the type of the polyP blend, indicating that polyPs and not the orthophosphates were effective. Quantitative PCR (qPCR) monitoring at sublethal concentrations revealed that polyPs reduced the transcription of ces nonribosomal peptide synthetase (NRPS) genes by 3- to 4-fold along with a significantly reduced toxin production level. At lower concentrations, toxin synthesis was decreased, although the growth rate was not affected. These data indicate a differential effect on toxin synthesis independent of growth inhibition. The inhibition of toxin synthesis in food was also observed. Despite the growth of B. cereus, toxin synthesis was reduced by 70 to 100% in two model food systems (reconstituted infant food and oat milk), which were analyzed with HEp-2 cell culture assays and high-performance liquid chromatography (HPLC)/electrospray ionization-time of flight mass spectrometry (ESI-TOF-MS). Accordingly, ces promoter activity was strongly downregulated, as visualized by using a lux-based reporter strain. These data illustrate the potential of polyphosphate formulations to reduce the risk of cereulide synthesis in food and may contribute to targeted hurdle concepts.

A broad variety of bacterial and fungal toxins is produced by multienzyme complexes, such as polyketide synthases (PKS), nonribosomal peptide synthetases (NRPSs), and PKS/NRPS hybrids (7, 11, 38). These include the well-known aflatoxins, ochratoxin A, enniatins, HC-toxin, and cyclopiazonic acid, which account for substantial economic losses in the food and feed industries and represent a central issue regarding public health and food safety worldwide (23, 24, 34, 45). The incidence of food-borne intoxications is steadily increasing in Europe, especially due to toxins produced by pathogenic Bacillus cereus strains (5). Among these, the NRPS toxin cereulide is receiving increasing attention (15, 59a).

Cereulide is an ionophoric cyclododecadepsipeptide of alternating α-amino and α-hydroxy acids (d-O-Leu-d-Ala-l-O-Val-l-Val) (1). Cereulide is produced by a clonal lineage of B. cereus strains that carry a pXO1-like megaplasmid, termed pBCE, which encodes the cereulide synthetase (ces) NRPS (14, 16). Recently, we showed that the ces gene cluster is polycistronically transcribed from a central promoter in a strict temporally regulated way (12, 13).

Usually, cereulide is preformed during the vegetative growth of B. cereus in improperly refrigerated foods and leads to nausea, vomiting, and abdominal cramps when ingested at concentrations around 10 μg per kilogram of body weight (22, 58). Besides these less acute symptoms, a number of studies have indicated a high toxicity potential concerning immunomodulatory and neuro- and hepatotoxic modes of action (4, 39, 44). Due to their lower toxic capacity, children and immunosuppressed individuals are especially vulnerable, and cases of fatal liver failure or encephalopathy have been reported (10, 31, 49, 59). Intoxications can further mimic Reye's syndrome, to which no defined causative agent has been associated so far (21).

Cereulide is extremely stable and withstands a broad pH range and enzymatic cleavage, as well as inactivation by filtration or thermal processing during food manufacturing or reheating of prepared foods (2, 51). Moreover, the occurrence of B. cereus in foods can hardly be prevented due to the ubiquitous distribution and profound resistance of the endospores (42, 59a). Consequently, several previously reported studies focused on antimicrobial compounds to suppress B. cereus growth in food, including bacteriocins, terpenoid substances, organic acids, and others (see, e.g., references 17, 41, 53, 60, and 62). However, targeted strategies to prevent toxin formation in the food production chain and in foods are still missing. Previously reported studies of the influence of long-chain polyphosphates (polyPs) on cell division in nonemetic B. cereus (32) raised the question of whether these substances could also have an impact on cereulide production in emetic strains.

Long-chain polyphosphates and derived orthophosphate (P_i) blends are classified as GRAS (generally recognized as safe) and are extensively used in the dairy and meat industries owing to their functional aspects regarding emulsification, stabilization, oxidation prevention, and flavor protection and in a second line as antimicrobial agents (40, 55). PolyPs are straight-chain polymers of condensed orthophosphoric acid residues. Interestingly, these chemically manufactured molecules have natural analogues synthesized in all bacteria and fulfill manifold physiological functions (for a recent review, see reference 52). Due to their polyanionic nature, polyPs sequester divalent cations and render them unavailable to microorganisms. This general deprivation is often assumed to be one cause of their antimicrobial activity, which leads to a loss of membrane integrity, impaired cell division, or lysis (27, 32, 63). However, more-specific effects have also been reported, including interference with ATP generation, protein folding, or enzyme activities (8, 50, 64), indicating that the knowledge concerning the molecular effects of polyPs is still limited. The aim of this work was to investigate the effect of long-chain polyphosphates on cereulide synthesis under conditions of a simulation of temperature abuse. This study presents, to our knowledge, the first data concerning the inhibitory action of polyPs on a nonribosomally synthesized peptide toxin.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

To examine polyP effects on cereulide synthesis, the emetic reference strain B. cereus F4810/72 (AH187), isolated from an emetic food-poisoning case (61), was chosen. A bioluminescent B. cereus F4810/72 reporter strain harboring the pXen1-derived vector pMDX[P₁/luxABCDE] (for details, see reference 12) was used to visualize the activity of the cereulide synthetase promoter in foods. Strains were routinely precultured (16 h at 30°C at 150 rpm) in standard plate count (PC) medium for liquid culture experiments or in standard LB (Luria-Bertani) medium for food spiking assays, respectively. For reporter strain experiments, 5 μg ml⁻¹ chloramphenicol (Cm5) was added to ensure plasmid maintenance.

Polyphosphates.

Three different mixtures of food-grade, long-chain sodium polyphosphate salts (glassy) (SPG) were obtained from BK Giulini GmbH (Ladenburg, Germany). Formulations differed with respect to the total phosphate content (P₂O₅), the average chain length, and the orthophosphate content. The chemical compositions and properties are described in Table 1. Stock solutions of 10% (wt/vol) were prepared with ultrapure water. The pH was adjusted to 6.8 with 3 M NaOH in order to exclude the influence of pH effects on the experiments. According to their solubilities, solutions of polyP 2 and 3 were filter sterilized (0.22-μm pore size), while the polyP 1 solution was autoclaved (121°C for 10 min). Heat treatment has no effect on the functionality of polyPs (28). Solutions were prepared freshly prior to each experiment.

TABLE 1.

Properties of food-grade, long-chain sodium polyphosphates used^b

PolyP blend	Mean P2O5 content (%) ± SD	Mean pH (1% solution) ± SD	Food additive declaration(s)a
PolyP 1	69.0 ± 1.0	6.0 ± 0.5	E 452, E 339
PolyP 2	68.0 ± 1.0	7.0 ± 0.5	E 452, E 339
PolyP 3	68.5 ± 1.0	3.7 ± 0.3	E 452

^aE 452, sodium polyphosphate; E 399, sodium orthophosphate.

^bThe mixtures were obtained from BK Giulini GmbH (Ladenburg, Germany).

Liquid culture experiments.

PolyP stock solutions were added aseptically to PC medium to final concentrations of 0.01% (vol/vol) and 0.03% (vol/vol), respectively. One hundred milliliters was inoculated with precultured B. cereus (10³ CFU ml⁻¹) and incubated under conditions of shaking (150 rpm at 24°C). Growth was recorded by the optical density at 600 nm (OD₆₀₀). Samples of 2 ml were removed at the indicated time points. One milliliter was autoclaved (121°C for 20 min) for the quantification of cereulide in toxicity bioassays. The remaining aliquot was harvested by centrifugation (10,000 × g for 2 min at 4°C), frozen in liquid nitrogen, and used for determinations of ces gene transcripts.

RT-qPCR.

Total RNA isolation, cDNA synthesis, and real-time quantitative PCR (RT-qPCR) were carried out as described previously (12). In brief, total RNA was prepared with TRIzol reagent lysis and chloroform-ethanol extraction followed by DNA digestion. cDNA was synthesized from 100 ng of total RNA using Moloney murine leukemia virus (M-MLV) reverse transcriptase (RNase H minus; Promega) and primers targeting CesA and 16S rRNA genes (14, 37). To determine cesA expression in relation to 16S rRNA transcript levels, SYBR green I-based qPCR was carried out with a SmartCycler instrument (Cepheid) with a total volume of 25 μl using 1 μl of cDNA template (equivalent to 10 ng total RNA). Relative gene expression ratios were calculated with the REST (Relative Expression Software Tool) method (46, 47). Samples of the untreated control harvested at an OD₆₀₀ of 1 were used as external reference conditions.

Model food experiments.

PolyP-free organic oat milk and a polyP-free dairy-based infant food formula obtained from German retail sources served as model food matrices. Selected foods either were reported previously to promote the synthesis of large amounts of cereulide (54) or had been analyzed as a B. cereus food spoilage case in our laboratory (oat milk) (our unpublished data). The infant food was reconstituted according to the manufacturer's instructions and cooled to room temperature. PolyP stock solutions (10%, wt/vol) were added aseptically to the desired concentrations. Batches were homogenized with a stomacher, and 30-g portions were filled into petri dishes. Depending on the experiment (see below), foods were spiked with B. cereus F4810/72 or B. cereus F4810/72(pMDX[P₁/luxABCDE]), resulting in 10³ CFU per gram. Immediately after inoculation as well as after incubation (24 h at 24°C), enumeration of viable cell counts was carried out for both strains. Whole portions of food (30 g) were comminuted with 270 ml of a 0.025% Tween 80 solution (pH 7.0) using a stomacher. The homogenates were serially 10-fold diluted in LB broth, and 100 μl was spread in double onto LB (wild type) or LB-Cm5 (reporter strain) plates for conventional plate counting. The increase in viable cell counts per gram of food within 24 h (N_i, expressed as log₁₀ CFU g⁻¹) was determined as follows: N_i = N₂₄ − N₀ (where N₂₄ is the log₁₀ CFU g⁻¹ after 24 h and N₀ is the log₁₀ CFU g⁻¹ directly after inoculation). Food samples prepared in parallel were used to determine cereulide production after 24 h (wild-type B. cereus). Samples spiked with the luminescent reporter strain were used to monitor cereulide synthetase promoter activity (see below).

Reporter gene analysis of ces gene transcription in model food.

For the visualization of ces P1 promoter-driven ces gene expression, B. cereus F4810/72(pMDX[P₁/luxABCDE]) was grown in foods in the presence or absence of polyPs as described above. Images were captured after 24 h with a photon-counting intensified charge-coupled device (ICCD) camera (model 2400-32; Hamamatsu Photonics), and bioluminescence intensity was superimposed onto gray-scale images as false color renderings. Quantification of the total photon counts via region-of-interest (ROI) analysis was performed according to the instructions provided with Living Image 2.01 software (Caliper Life Sciences) and the implemented IGOR Pro 4.01 software tool (WaveMetrics) as described elsewhere previously (12).

Extraction of foods for cereulide quantification.

For cereulide quantification by high-performance liquid chromatography (HPLC)-mass spectrometry (MS) and HEp-2 assays, food was spiked with the B. cereus wild type as described above. Each food sample (30 g) was extracted with 20 ml of 96% ethanol by shaking on a rocking table (24 h at 23°C). The extracts were centrifuged twice (20 min at 8,500 × g at 24°C) to remove extraction residuals and food debris. Prior to measurements, the supernatants were membrane filtrated (0.2-μm polytetrafluorothylene [PTFE] membrane; Phenomenex, Germany). Control samples additionally spiked with 25 μg g⁻¹ valinomycin (Fluka, Germany) served as an internal standard for toxin recovery rate calculations. To exclude an interfering toxic background effect of the food extracts on the HEp-2 cells in the cytotoxicity bioassay and to exclude the possibility that foods inherently contained cereulide, extracts of unspiked control food samples (with or without polyP) were additionally included.

Cereulide quantification (cytotoxicity bioassay).

Amounts of toxin in liquid culture samples and food extracts were assessed with a HEp-2 cell-based cell culture assay as previously described (30). Valinomycin served as an internal standard, and cereulide amounts were accordingly determined as valinomycin equivalents (VE).

Cereulide quantification (HPLC/ESI-TOF-MS analysis).

Analysis of food extracts was performed by using an Agilent rapid-resolution HPLC system (series 1200) connected to a time-of-flight (TOF) mass spectrometer (G1969A; 6210 TOF LC/MS) running in the positive electrospray ionization (ESI) mode. Chromatographic separation was performed with a Prontosil 120-3 C₁₈ column kept at 50°C (100-mm by 2-mm internal diameter [i.d.], 3 μm; Bischoff Chromatography, Germany) by 20-μl sample injection and MeOH-H₂O as the mobile phase (flow rate of 200 μl min⁻¹). Prior to starting a chromatographic run, the composition of the mobile phase was kept at 100% solution A (MeOH-H₂O [10:90, vol/vol], 10 mM ammonium acetate [pH 7.4]). Separation was developed by increasing solvent B (MeOH-H₂O [90:10, vol/vol], 10 mM ammonium acetate [pH 7.4]) to 30% within 1 min, followed by a linear gradient of solvent B from 30% to 100% within 8 min. The effective mobile phase was held for 16 min, with an increased flow rate (250 μl min⁻¹) in the last 4 min. After each run, the composition was reset to 100% solution A. HP-0921 and purine (ES-TOF Reference Mass solutions; Agilent, Germany) in solution B were used as mass-correcting standards, being continuously mixed in the postcolumn flow (100 μl min⁻¹). ESI interface parameters were as follows: 350°C drying gas temperature, 7-liter min⁻¹ drying gas flow, and 45-psig nebulizer gas pressure. For mass spectrometric detection (positive-ion-polarity mode), the capillary and “fragmentor” voltages were set to 5,000 V and 250 V, respectively.

Each sample was measured at least in triplicate. Data were acquired and processed with Mass Hunter Workstation software (B.01.02; Agilent, Germany). Valinomycin (Sigma, Germany) served as a surrogate calibration standard for cereulide quantification as described previously (19). Standard dilutions were analyzed at least once for each sample sequence. To quantify cereulide and valinomycin abundances, integrated extracted ion current (EIC) chromatograms with ion ranges (m/z) of 1,170.5 to 1,193.5 for cereulide and of 1,128.5 to 1,151.0 for valinomycin targeting the NH₄⁺ and K⁺ adducts, respectively, were used (48). Calibration curves extrapolated from integrated peak areas were plotted to calculate cereulide amounts via linear regression.

RESULTS

Impact of polyPs on B. cereus growth, ces gene transcription, and toxin synthesis in liquid culture.

To initially evaluate the effects induced by polyP exposure, B. cereus F4810/72 was grown in PC medium in the presence of three long-chain polyP mixtures (Table 1). MICs of all polyP blends were determined by an automated turbidimetry-based system (Bioscreen C; Labsystems Oy, Finland). MICs ranged from 0.07% to 0.1%, and microscopic analysis revealed that concentrations above 0.05% caused lytic effects (data not shown). MICs were in good agreement with data from previously reported studies of Gram-positive bacteria (28, 32, 40). To avoid lysis of the cells, a sublethal concentration of 0.03% was used. Under this condition, cell growth was impaired and the commencement of logarithmic growth was delayed (Fig. 1). Under the influence of polyPs 2 and 3, cell division was arrested, and cell shape deformation occurred (Fig. 1). The growth inhibition efficacy was strongly dependent on the type of polyP blend, with the mixture containing the lowest orthophosphate content being the most effective (Table 1 and Fig. 1). In the presence of polyP 2 and polyP 3, cells were 3- to 4-fold longer than cells in the untreated control culture. Concordantly, we observed that exponential growth with normal division was resumed after 17 h in the presence of polyP 1, whereas the onset of exponential growth was substantially delayed under the influence of polyP 2 and polyP 3 (Fig. 1).

FIG. 1.

Effects of sublethal polyP concentrations on growth and cell shape of emetic B. cereus in liquid culture. B. cereus F4810/72 cells were grown in PC medium in the absence (control [○]) and in the presence of 0.03% food-grade, long-chain polyphosphates (polyP 1 [▾], polyP 2 [▪], and polyP 3 [⧫]). Growth was monitored at an optical density at 600 nm. Standard deviations are derived from three independent measurements. Phase-contrast microscope images were taken 21 h after inoculation (C, control; polyP 1 to polyP 3, elongated cells under the influence of 0.03% polyphosphates; the white scale bar represents 5 μm).

To determine whether polyP exposure would also affect the transcription of cereulide synthetase (ces) genes, RT-qPCR was applied. In control cultures, maximal transcription was observed in the late exponential growth phase (17 h after inoculation) (Fig. 1); thereafter, transcription was strongly downregulated (Fig. 2). This strictly temporal expression of the NRPS genes is a characteristic feature of cereulide synthesis in emetic B. cereus strains (13). In accordance with this finding, maximum transcription was delayed by polyP exposure (Fig. 2) but was detected when cells had resumed exponential-growth-like stages with shorter cell lengths (Fig. 1). Importantly, exposure to all three polyP formulations significantly reduced the level of ces gene transcription, which was 3- to 4-fold lower than that of the control. When cultures entered stationary phase, ces gene transcripts fell below the detection limit (32 h after inoculation) (Fig. 2).

FIG. 2.

Influence of sublethal polyP concentrations on ces gene transcription and cereulide synthesis in liquid culture. B. cereus F4810/72 cells were grown in PC medium in the absence (black bar) and in the presence of 0.03% of three food-grade, long-chain polyphosphate mixtures (polyP 1 to polyP 3, gray and white bars) (see also Fig. 1). Transcript levels of cesA were determined at the indicated time points in relation to 16S rRNA transcript levels of the same RNA preparations. Error bars indicate standard deviations from three independent cultures and nine independent RT-qPCR measurements. Cereulide quantity, given as valinomycin equivalents (VE), was assessed with the HEp-2 cell cytotoxicity assay. All values are means, and error bars show the standard deviations from three independent cytotoxicity assays.

ces gene transcript levels were further compared with toxin synthesis. Without polyPs, cereulide became detectable 17 h after inoculation and accumulated progressively during late logarithmic and stationary phases (Fig. 2). In contrast, cultures exposed to food additives remained nontoxic for at least 24 h. In the stationary phase cereulide was detectable in all polyP-treated cultures; however, toxin amounts were 3-fold reduced. This is in agreement with a lower level of transcription of the ces genes.

To finally distinguish the simple inhibition of growth from the inhibition of toxin synthesis, the impact of polyPs at subinhibitory concentrations with respect to vegetative cell growth was investigated. Figure 3 illustrates that the addition of 0.01% polyPs had no inhibitory effect on the growth rate of B. cereus F4810/72, and anomalies in the cell form were not observed (Fig. 3). However, cereulide synthesis was significantly reduced by 30 to 60% after 24 and 32 h of cultivation, respectively.

FIG. 3.

Growth and toxin production of emetic B. cereus in the presence of 0.01% polyP. B. cereus F4810/72 growth in PC medium in the absence (○, control culture) and in the presence of three food-grade polyphosphate blends (polyP 1 [▾], polyP 2 [▪], and polyP 3 [⧫]) was monitored by measurements of the optical density at 600 nm. Phase-contrast microscope images were captured 21 h after inoculation (C, control; polyP 1 to polyP 3, cells treated with 0.01% polyphosphates). The black scale bar represents 5 μm. Cereulide was quantified with the HEp-2 cell-based toxicity assay after 19, 24, and 32 h, respectively. The toxin quantity was set to 100% for the untreated culture (white bars) to display the reduction of toxin synthesis under the influence of polyP 1 (pale gray bars), polyP 2 (dark gray bars), and polyP 3 (black bars). Error bars correspond to standard deviations derived from three independent experiments.

Impact of polyPs on B. cereus growth, ces synthetase promoter activity, and toxin synthesis in model foods.

To examine whether polyP blends would also inhibit cereulide production in complex food matrices, two different model systems were chosen. Dairy-based infant food supports cereulide formation to extremely high levels and was regarded to be a “high-challenge” trial product. Industrial oat milk was the subject of a B. cereus spoilage case recently analyzed in our laboratory. All foods included in the spiking studies tested negative for natural contamination with emetic B. cereus by using selective plating media and PCR targeting the ces genes (data not shown). Furthermore, control measurements and measurements of samples directly extracted after inoculation with B. cereus showed that cereulide was absent in the retail food and that cereulide was not introduced from the precultures in the food (data not shown). A toxic effect on HEp-2 cells was excluded by measuring extracts from unspiked food or unspiked foods treated with polyP (data not shown).

In line with data from previous studies, cereulide was detected by liquid chromatography (LC)/MS analysis predominantly in its NH₄⁺ adduct form, followed by the toxicologically important, highly stable K⁺ complex (33, 48) (data not shown). The occurrence of the NH₄⁺ adduct is a typical bias introduced by analytical procedures. LC/MS data were in good agreement with the HEp-2 assay data, indicating that the higher structure of the ionophor necessary to induce toxicity was not altered by the addition of polyP (Tables 2 and 3).

TABLE 2.

Effects of polyPs on cell count, cereulide synthetase promoter activity, and cereulide production by emetic B. cereus in infant food after 24 h at 24°C^a

PolyP formulation	Mean Ni (log10 CFU g−1) ± SD of B. cereus F4810/72b	Mean Ni (log10 CFU g−1) ± SD of B. cereus F4810/72(pMDX[P1/luxABCDE])b	ces synthetase promoter activity (total ROI count)c	Mean cereulide concn (μg g−1 food) by LC/MS analysis ± SD	Cereulide concn (%) by cytotoxicity assayd
Control	4.0 ± 0.3	3.8 ± 0.1	6.3E + 07	6.4 ± 1.5	100
5% polyP 1	4.1 ± 0.2	3.9 ± 0.1	5.8E + 06	2.0 ± 0.8	31
5% polyP 2	−0.1 ± 0.4	−0.2 ± 0.2	7.1E + 03	0.4 ± 0.8	6.8
5% polyP 3	3.2 ± 0.2	3.0 ± 0.2	3.1E + 03	0.1 ± 0.1	1.5

^aBefore incubation, food samples were inoculated with the B. cereus wild type or the bioluminescent reporter strain F4810/72(pMDX[P₁/luxABCDE]), resulting in 10³ CFU g⁻¹. All experiments were performed at least in triplicate. Values are given as means with standard deviations.

^bN_i, increase in viable cell counts per gram of food after 24 h (see Materials and Methods).

^cBioluminescence emitted by the emetic reporter strain quantified by a region-of-interest (ROI) analysis deduced from the whole petri dish area (Fig. 4). Data are from an average of four independent experiments.

^dData are derived from three independent HEp-2 cytotoxicity assays and represent mean values. The toxin amount produced in the absence of polyP was set to 100%.

TABLE 3.

Effects of polyPs on cell count and cereulide production by emetic B. cereus in organic oat milk after 24 h at 24°C^a

PolyP formulation and concn (vol%)	Mean Ni (log10 CFU g−1) ± SD of B. cereus F4810/72b	Mean cereulide concn (μg g−1 food) ± SD by LC/MS analysisd	Cereulide concn (%) by cytotoxicity assayc
Control	5.3 ± 0.1	1.3 ± 0.3	100
polyP 1
0.1	5.3 ± 0.1	0.7 ± 0.2	51
0.2	5.0 ± 0.0	0.4 ± 0.1	14
0.3	4.9 ± 0.0	0.3 ± 0.1	11
0.5	3.8 ± 0.3	ND	0
polyP 2
0.1	5.2 ± 0.1	0.5 ± 0.1	31
0.2	4.8 ± 0.1	0.4 ± 0.0	11
0.3	4.3 ± 0.1	0.2 ± 0.2	8.6
0.5	0.0 ± 0.2	ND	0
polyP 3
0.1	5.1 ± 0.1	0.4 ± 0.1	11
0.2	4.8 ± 0.1	0.3 ± 0.1	8.6
0.3	4.0 ± 0.0	0.2 ± 0.2	5.7
0.5	−0.2 ± 0.1	ND	0

^aBefore incubation, food samples were inoculated with B. cereus F4810/72, resulting in 10³ CFU g⁻¹. Data are given as means ± standard deviations from three independent experiments.

^bN_i, increase in viable cell counts per gram of food within 24 h at 24°C (see Materials and Methods).

^cData are derived from three independent HEp-2 cytotoxicity assays and represent mean values. The toxin amount produced in the absence of polyP was set to 100%.

^dND, not detectable (detection limit of ≤50 ng ml⁻¹ cereulide).

High cereulide concentrations of 6 μg/g were detected in artificially inoculated dairy-based infant food after simulation of food spoilage at room temperature (Table 2). Correspondingly, the ces synthetase promoter was highly active, which was visualized and quantified by using the B. cereus lux reporter strain (Fig. 4). Treatment with 5% polyP 2 and polyP 3 almost completely repressed bioluminescence, whereas moderate signal intensities were observed with polyP 1 (Fig. 4 and Table 2). To decipher whether lux activity was reduced due to cell lysis or due to transcription inhibition and to further correlate growth and toxin production, viable cell counts for both B. cereus wild-type and reporter strains were determined. As in the liquid culture experiments, the three polyP mixtures caused different effects on growth and toxin formation (Table 2). Growth was not affected by 5% polyP blend 1. In comparison, polyP 3 inhibited cell growth by 1 log unit, and treatment with 5% polyP 2 led to bacteriostatic and slight bactericidal effects since only the initial cell inoculum was recovered. In correlation with ces promoter activities, cereulide concentrations were significantly reduced, to 31% (polyP 1), 6.0% (polyP 2), and 1.6% (polyP 3). Thus, consistent with the data obtained with broth experiments, polyP exposure clearly reduced the level of ces promoter-driven NRPS expression (polyPs 1 and 3) or abated cell growth (polyP 2) and generally inhibited cereulide production.

FIG. 4.

Influence of polyPs on cereulide synthetase promoter activity in infant food. Reconstituted infant food was inoculated with the bioluminescent B. cereus reporter strain F4810/72(pMDX[P₁/luxABCDE]) (12), and food spoilage was simulated by incubation at 24°C for 24 h. The influence of 5% polyPs on ces promoter activity was visualized and quantified after 24 h with a photon-counting ICCD camera (model 2400-32; Hamamatsu Photonics) (Table 2). C, untreated control; polyP, food-grade, long-chain polyphosphate mixtures. Representative images are shown.

B. cereus F4810/72 produced 1 μg toxin per gram of food in untreated oat milk samples. This corresponds to one-sixth of the toxin amount in infant food, although final cell counts were higher in oat milk (Tables 2 and 3). This finding illustrates that toxin production is not dependent strictly on the growth rate. Also, these data support previously reported findings that cereulide synthesis is substantially influenced by external stimuli, including the type of food in which cells multiply (12, 13). Again, the efficacies of the polyP blends differed with respect to growth and the inhibition of toxin synthesis (Table 3). While cell multiplication was only marginally affected in the presence of 0.1% polyPs, cereulide concentrations were reduced by 50 to 89% (Table 3). Similar to the dose dependency observed with PC broth, enhanced concentrations progressively reduced cereulide synthesis. Accordingly, amounts of cereulide dropped by 89% to 94% under the influence of 0.3% polyPs. Treatment with 0.5% caused a bacteriostatic (polyP 2) or slightly bacteriolytic (polyP 3) effect, whereas B. cereus was reduced by only 1 log unit in the presence of the mixture with the highest orthophosphate content, polyP 1. Importantly, the remaining viable cells did not produce cereulide, and toxin synthesis was completely prevented under these conditions.

DISCUSSION

The antimicrobial activity of polyP blends was dose dependent, and sublethal, nonlytic concentrations caused bacteriostasis and cell elongation (Fig. 1). The effect was most pronounced with polyP 3, strongly indicating that the effects were caused mainly by long-chain polyphosphates and not by orthophosphate (Table 1). Defective cell division upon polyP treatment was also reported previously for other Gram-positive bacteria (32, 64, 65). It was assumed that the sequestration of divalent cations caused the defect in septum formation due to an indirect effect on the GTPase activity of FtsZ proteins (32). Alternatively, an enhanced water-binding capacity due to the polyP interaction with intracellular protein components was discussed (64). Since the vegetative-cell shape normalized after the commencement of a delayed exponential growth phase, we hypothesize that polyP is slowly degraded to orthophosphate (P_i) residues lacking antimicrobial activity (55). Consistent with this, our BLAST analysis revealed that all key enzymes for polyP hydrolysis and synthesis (polyP kinase [PPK], polyP/AMP phosphotransferase [PAP], and exopolyphosphatase [PPX]) (57) are present in B. cereus F4810/72 (data not shown). Multiphase growth in the presence of polyPs was also observed previously for Bacillus subtilis (43). While the role of intracellular polyPs in B. cereus biofilm formation, sporulation, and motility was demonstrated previously (57), the fate of exogenously added polyP has not been dissected so far. Endogenous polyP concentrations are strictly regulated in prokaryotes. The concentration is low during rapid growth but increases under conditions of nutritional imbalance unfavorable for growth (63). As B. cereus usually does not accumulate high polyP levels (57), the presence of exogenous polyPs might increase the intracellular concentration, which may result in imbalances affecting numerous cellular processes. Our data indicate that polyPs interfere with early stages of the toxin formation process by delaying and reducing ces gene transcription. Consistently, delayed and reduced cereulide formation was observed (Fig. 2). Interestingly, the polyP-induced delay of toxicity despite vegetative growth was reported previously for botulinum toxin production by Clostridium botulinum (64) and aflatoxin B1 and G1 production by Aspergillus spp. (36). It was assumed that polyPs inhibited the activity of an external protease necessary for the activation of the botulinum toxin. In Aspergillus, a possible influence on the PKS enzyme was not determined. However, the underlying effect was remotely attributed to disturbances of primary metabolic pathways.

Lower polyP concentrations did not affect cell integrity or multiplication in PC broth but diminished toxin synthesis (Fig. 3). These data indicate that the cereulide synthesis mechanism responds more sensitively to the presence of exogenous polyP than does the main cellular metabolism. Interestingly, the synthesis of aflatoxins also has a much narrower tolerance range for phosphate concentrations than the vegetative growth of the fungus (29). Thus, polyP-mediated chelation of metal ions might not be the ultimate cause of reduced toxin synthesis, as main cellular functions like binary fission, which is highly dependent on cations (35), apparently were not influenced. However, the cation dependency of the catalytic ces NRPS modules has not been analyzed so far. It should be noted that polyPs are potent phosphorylating agents and likely participate in protein phosphorylation and thus are influencing metabolic regulation in bacteria (for reviews, see, e.g., references 25 and 52). Recently, it was shown that the phosphorylation of the master regulator of sporulation (Spo0A) is required to activate cereulide production (30). Therefore, imbalances in the intracellular polyP/P_i ratio might interfere with phosphorelay signal transduction pathways. Indeed, previous studies indicated that Spo0A might be a direct target of inherent polyP regulation in B. cereus (57). Similarly, feedback inhibition by phosphorylated intermediates to biosynthetic pathways was taken into account to explain effects leading to aflatoxin synthesis inhibition (29). Additionally, it was demonstrated that an NRPS of Bacillus brevis synthesizes not only the antibiotic tyrocidine but also (di)adenosine polyphosphates (9). It is therefore tempting to speculate that the exogenous polyP overflow might affect catalytic processes of NRPS modules and lead to dysfunction.

The inhibitory potential of polyPs in reconstituted infant food was assessed, since this substrate is known to allow the production of exceptionally large quantities of cereulide when spoilage under temperature abuse conditions is simulated (54). In comparison to broth culture experiments, higher polyP concentrations (5 vol%) were necessary to achieve growth inhibition (Table 2). This is in agreement with data from previous studies, which demonstrated that the efficiency of polyPs in culture media with comparably simple compositions is not necessarily matched by the activity in complex food matrices (50, 55). Thus, the outcome of growth inhibition strongly depends on the complexity of the environment (26, 66), and the analyzed infant formula was enriched with proteins and vitamins. PolyP blend 2 displayed enhanced inhibitory effects and prevented the outgrowth of the initial cell inoculum in the infant formula but not in PC broth and oat milk. This might be explained by the finding that the antimicrobial activity of food-grade phosphates is modulated in response to different food matrices or ingredients (18, 56). PolyP 1, the blend with the highest orthophosphate content, did not influence cell growth (Table 2). However, cereulide titers were reduced by 70%, which strengthens the assumption that the cereulide synthesis process sensitively reacts to the presence of polyphosphates. PolyP 3 most efficiently inhibited toxin production, by 98%, albeit at a concentration (5 vol%, corresponding to approximately a 3.5% P₂O₅ content) (Table 1) which is somewhat above the legal limit for this food additive. Considering the high initial inoculation level, it should be noted, however, that this might not necessarily reflect naturally occurring food spoilage cases. Different studies of the prevalence of B. cereus in powdered infant foods showed that the pathogen is usually found in quantities below 10³ CFU/g (for a recent report, see reference 20). Therefore, lower polyP concentrations may be sufficient to prevent toxin synthesis under common household conditions even in sensitive foods such as infant formulas.

Data comparing the effects of 0.1 to 0.5% of the three polyP blends in the model food oat milk are summarized in Table 3. Most importantly, the use of 0.5 vol% polyPs (approximately 0.35% P₂O₅) completely prevented toxin formation within concentrations commonly used in the food industry worldwide (0.2 to 0.8% P₂O₅ content) (3). Furthermore, these concentrations are within the maximum use level of 0.5 to 2% P₂O₅ for dairy products prescribed by the European Scientific Committee on Food (SCF) (http://ec.europa.eu/food/fs/sc/scf/reports_en.html) and the joint FAO/WHO Codex Alimentarius Commission (CAC) (codex general standard for food additives 192-1995, version 2009 [8a]).

In conclusion, this study indicates that the use of commercial polyphosphate blends could serve as one module of a hurdle concept, which is targeted at the inhibition of cereulide synthesis in food ingredients or foods, especially where B. cereus spores cannot be excluded or inactivated during processing or preparation steps. Considering their broad applicability in diverse types of food (50), polyPs may also contribute to the control of cereulide synthesis in other food systems. However, before a firm conclusion of the general applicability of the polyPs can be drawn, the efficiency of the food additives described in this work must be further tested on additional model foods. One should also bear in mind that the effective control of cereulide contamination in complex food matrices requires a well-balanced interplay of several factors, such as polyP addition, water and nutrient availability, and temperature, which jointly affect toxin synthesis.

The underlying mechanism(s) leading to the inhibitory effect remains unknown. Nevertheless, it is reasonable to speculate that nonribosomal toxin formation is impaired by a multifactorial mechanism due to a narrow tolerance range regarding external phosphate concentrations, and polyPs might also be useful to combat the production of other economically and medically relevant toxins produced by NRPS or PKS multienzymes. However, an elucidation of the molecular basis of the action of exogenously added polyPs on NRPS or PKS enzymes is beyond the scope of this study.