Endogenous Metabolites Released by Sanitized Sprouting Alfalfa Seed Inhibit the Growth of Salmonella enterica

Warm, humid, and nutrient-rich conditions that are used to produce sprouts encourage Salmonella enterica to proliferate. However, many disparate sanitation methods exist, and there is currently no single treatment that can guarantee pathogen-free seeds.

methods common to organic farming practices. In an exploration phase, we took a multifaceted approach, examining impacts on Salmonella colonization of alfalfa seed by indigenous bacteria, as well as changes to seed composition and to the seed metabolome. In a validation phase, we investigated whether the distinct metabolome of treated seeds or specific metabolites of interest could be sufficient to inhibit of Salmonella growth if applied to contaminated seeds exogenously. Figure 1 shows our workflow encompassing both stages.

RESULTS AND DISCUSSION
Moderate S. enterica growth occurred on sanitized alfalfa seed. The growth of five S. enterica strains was measured on germinating alfalfa seed subjected to three sanitation treatments, including sodium hypochlorite (CLO) and heat, peroxide, and vinegar (HPA) ( Table 1). Treatment with concentrated chlorine or hydrogen peroxide solutions is currently recommended by regulatory agencies, and the HPA treatment represents an alternative approach compliant with organic production principles that relies on multiple "hurdles" to achieve a synergistic antibacterial effect (26). None of the S. enterica strains were detected on alfalfa seed immediately after treatment by the methods used in this study (limit of detection = 1 log CFU/g). In contrast, populations had increased by 5.1 to 7.6 and 4.8 to 6.6 log CFU/g on germinating CLO-and HPAtreated seeds, respectively, after 1 day (Table 1). After 6 days of germination, S. enterica populations reached 4.5 to 7.6 log CFU/g on untreated seed and 6.0 to 7.1 and 3.9 to 6.0 log CFU/g on CLO-and HPA-treated seed (Table 1). Salmonella enterica serovar Agona PARC 5 populations were highest after 1 day of germination (7.58 6 0.15 log CFU/g), while Salmonella enterica serovar Typhimurium LMFS-S-JF-001 populations were significantly lower (P , 0.05). Overall, S. enterica populations on germinating seeds subjected to the HPA treatment were consistently lower than those measured on seed treated with CLO. These findings are in line with previous studies that have shown the combinatorial treatment of heat, peroxide and acetic acid to be effective in hindering growth of Escherichia coli O157:H7, Listeria monocytogenes, and S. enterica inoculated onto alfalfa and radish seeds (27).
Cocolonization by indigenous bacteria is affected by S. enterica but not by sanitation. The fastest increase in mesophilic aerobic bacteria populations occurred on seed inoculated with S. Typhimurium, reaching more than 9 log CFU/g by day 6 ( Table 1). Overall, the sanitation treatments had less of an effect on indigenous bacterial populations, as detected by aerobic plate counts, than on S. enterica. Antimicrobial treatments did not result in a significant reduction of the population of indigenous aerobic bacteria on alfalfa seed inoculated with S. enterica compared to controls (CTL). These findings are comparable to previous work that showed that populations of indigenous bacteria were not reduced by more than 1 log by treatment with H 2 O 2 , peroxyacetic acid plus H 2 O 2 , acidified NaClO 2 , NaClO 2 , EDTA, Na 3 PO 4 , and NaClO unless they were applied at concentrations that induced phytotoxic effects (28).
Due to the distinct growth characteristics of S. Agona PARC 5 (the most rapidly growing among the strains tested) and S. Typhimurium LMFS-S-JF-001 (the slowest growing), these two strains were selected for comparison across all remaining analyses.
The chemical composition of alfalfa seed changes dynamically during germination. Changes in the chemical composition of alfalfa seed during germination and after sanitation have important implications for seed health (necessary to maintain freshness and germination rate), phenolic antioxidant production (with potentially protective effects for both the seeds and consumers), and nutrient production (supporting sprout growth and impacting developing microbiota).
Because HOCl and H 2 O 2 are known to cause protein modification and damage (29,30), we hypothesized that all three sanitization treatments could affect protein content. However, we did not find significant differences in total protein due to any of the treatments compared to control treatment ( Fig. 2a; also, see Table S1 in the supplemental material). Protein content remained relatively constant during germination, increasing slightly from 7 to 12 mg/g immediately after treatment to 7 to 19 mg/g thereafter, irrespective of S. enterica strain. Soluble carbohydrate content also increased during germination from 1.7 to 4.2 mmol/g at the outset to 5.5 to 10.8 mmol/g after 6 days ( Fig. 2b; Table S1). In line with these findings, Ling and Chang (31) reported that total soluble carbohydrate contents in red beans and guava seeds continuously increase during germination. Accumulation of carbohydrates can be ascribed to continued hydrolysis of 1,6-galactosidic linkages of glycoproteins, glycolipids, and polysaccharides catalyzed by a-galactosidase (32). Interestingly, total phenolic contents increased dramatically during the first 8 h of germination, followed by a gradual decrease between 8 and 24 h and stabilization after 24 to 32 h ( Fig. 2c; Table S1). Untreated seed inoculated with either of the S. enterica strains contained the highest levels of total phenolics after 8 to 24 h. This is consistent with previous studies that showed that sprouting seeds tend to synthesize phenols during imbibition and early embryonic development (1) to support cell wall, hormone, and other plant regulator synthesis (33). After 6 days, we found that levels of phenolics were substantially lower, irrespective of treatment, ranging from 1.9 to 2.9 mg gallic acid equivalents (GAE)/g sprouts. This likely reflects the previously observed trend for phenolics to steadily decrease between 24 h and 8 days after initiation of germination (34). High phenolic concentrations in plants used for food and medicine have been correlated with antimicrobial effects (35); however, we did not clearly observe this phenomenon here. For example, while phenolic counts in HPA-treated seeds were much lower than those in CLO-treated seeds from 24 h to 6 days, we detected similar increases in colonization, as measured by aerobic plate count (APC), in both settings.
Sanitation strongly affects metabolite profiles in germinating alfalfa seed. Metabolomic analysis was performed on exudates from germinating alfalfa seed inoculated with either S. Agona or S. Typhimurium subjected to two sanitation treatments. A heat map representation of metabolite intensities (Fig. 3) visually demonstrates the effect of treatment type on the abundance of 96.1% of the 535 identified metabolites, which spanned almost all major metabolic pathways, as well as the subtler influence of colonizing S. enterica strain. Principal-component analysis (PCA) of the metabolite profiles enabled elimination of uninformative background variables and visualization of small profile changes. The PCA plot was used to reveal information regarding differences among surface metabolite profile of germinating alfalfa seed and to assess which variables of antimicrobial treatments mostly contributed to this difference (Fig. S1). Principal component 1 (PC1) and PC2 of indigenous bacteria, S. Agona, and S. Typhimurium LMFS-S-JF-001, explained 69.8 and 13.6%, 53.9 and 25.5%, and 67.5 and 17.2% of the total variability of the data, accounting for the metabolite composition changes in germinating alfalfa seed. The plot indicated that distinct segregation and clustering were apparent on HPA-treated S. enterica.
Fold changes (FC) of the metabolites were log transformed as 1 log of FC. All values were reported for completeness, and independent t tests were applied to define significant increase or decrease. Compounds that were significantly increased or decreased (P , 0.05) by both the CLO and HPA treatments included a wide range of metabolites from all biochemical classes, especially amino acids and secondary metabolites ( Table S2). Compounds that were significantly increased (P , 0.05) by both treatments relative to the control included some carbohydrates related to oxidation products of larger biochemical compounds (e.g., oxalate) and breakdown products of cell walls or larger oligosaccharides (e.g., fucitol, maltose, and verbascose). Several markers for protein, membrane, and nucleic acid degradation (e.g., dimethylarginine) were significantly increased following HPA treatment but decreased following CLO treatment (P , 0.05). Levels of 25 amino acids were significantly increased upon colonization of both S. Agona PARC 5 and S. Typhimurium LMFS-S-JF-001 after the HPA treatment, while no amino acids were significantly decreased (P , 0.05) (Table S3). In addition, the concentrations of 10 phospholipids were significantly decreased in both HPA-treated S. enterica-colonized samples (P , 0.05) (Table S4). In this study, due to the vast bulk of the sample used for metabolite analysis, which was composed of seed exudate material (estimated 99.5% of sample volume, as opposed the indigenous and pathogenic bacteria contributing roughly 0.5% of sample volume), we attributed the levels of metabolites mainly to seeds. A caveat here is that we cannot completely rule out the possibility that some metabolites produced by bacteria could also contribute to expression changes. However, the fact that we did not detect any amino sugars in our analysis (a metabolite commonly produced by bacteria but not synthesized in any significant amounts by plants) (36) seems to point to the fact that our results are representative of metabolites derived from seed and not the colonizing bacteria.
Germinating seed exudates have moderate antimicrobial effects against S. enterica. Seed exudates harvested from CLO-or HPA-treated seeds after 24 h of germination were applied to CLO-or HPA-treated seeds inoculated with S. enterica to determine whether metabolites released by the treatments affected colonization. The variable effects of the treatments are shown in Fig. 4a and b (see also Tables S5 and S6). HPA-treated sprouts injected daily with exudates from HPA-treated seeds exhibited the most significant antimicrobial effects (P , 0.001) up to 2.4 log CFU/g for S. Agona PARC 5 and 2.7 log CFU/g for S. Typhimurium LMFS-S-JF-001 compared to HPA-treated sprouts with no additional injection of exudates. One caveat to these findings is that although seed exudates were filtered before application, residual indigenous bacteria could still have affected microbiomes and secondary metabolites of inoculated seeds (37). Such interactions between indigenous bacterial inoculants and plant microbial communities, in addition to changes in plant production of bioactive metabolites, have been reported (38).
Polyamines significantly reduce S. enterica growth on germinating alfalfa seed. The polyamines N-acetylputrescine and N 1 -and N 8 -acetylspermidine were significantly increased (P , 0.05) on control and CLO-treated germinating seed inoculated with S. enterica compared to control and CLO-treated germinating seed with indigenous bacteria (data not shown). We therefore examined the putative antimicrobial effects of the polyamines spermidine, agmatine, and N 1 -and N 8 -acetylspermidine on the growth of both S. Agona and S. Typhimurium (Fig. 4c and d; Tables S7 and S8). Strikingly, populations of both S. Agona and S. Typhimurium decreased significantly after 24 and 48 h, respectively, particularly on seed injected with N 1 -and N 8 -acetylspermidine (P , 0.05) despite substantial initial population increases of 8.0 to 8.91 log CFU/g (S. Agona PARC 5) and 7.47 to 8.27 log CFU/g (S. Typhimurium LMFS-S-JF-001) during the first day of germination (Tables S7 and S8). Effects may have been more apparent at later time points (after 24 h) due to repeated daily polyamine injections. In a previous study, the application of 2 to 4 mM exogenous spermine was found be sufficient to inhibit bacterial growth on isolated bacterial cultures of S. Typhimurium LT2 and E. coli K-10 (39). Our study builds upon this prior work by demonstrating similar inhibitory effects at similar concentrations in vivo on germinating alfalfa seed. Specifically, our exogenous applications of 500 ppm acetylspermidine were equivalent to 2.7 mM, and while the concentration of endogenous spermidine in the alfalfa exudates was somewhat lower at 1.2 mM, antimicrobial effects were still observed.
In addition, data on the growth of S. Agona PARC 5 and S. Typhimurium LMFS-S-JF-001 on germinating alfalfa seed supplemented with polyamines were suggestive of longer lag phases leading to delays in the initiation of exponential growth (Table 2). Taken together, these observations indicate that N 1 -and N 8 -acetylspermidine and other polyamines contribute antimicrobial effects during the germination of seed treated with hypochlorite. However, spermidine has been reported to cause acute toxicity at 600 mg/kg body weight and the no-observed-adverse-effect level of 1,000 ppm (83 mg/kg body weight/day) (40). For this reason, the extent to which polyamines can be applied to germinating seeds without toxicity, as well as negative effects impacting sprout palatability, nutritional content, storage, germination, and productivity for growers, remains to be seen.
Until effective and reliable treatments can be developed, multihurdle approaches and microbial testing of spent irrigation water remain of utmost importance to ensure that contaminated sprouts do not enter the marketplace. Although we find that the regrowth of S. enterica in sprouts following sanitization treatments is strain dependent and treatment dependent, the mechanisms responsible for this effect require further investigation. Moreover, deeper studies of the microbiomes in alfalfa seed could be necessary to enhance our understanding of the role of microbiota in germinating alfalfa sprouts. The results of this study provide new insights on mechanisms that influence interactions between enterobacterial pathogens and germinating sprouting seed. With better understanding of the interactions between S. enterica and indigenous bacteria and sprouting of seeds, more targeted interventions can be developed for the production of pathogen-free sprouted vegetables.
Alfalfa seeds and seed inoculation. Certified organic alfalfa sprouting seed was obtained from a Canadian seed producer (Mumm's Sprouting Seeds, Saskatchewan, Canada). A single colony of each S. enterica strain was transferred to 10 ml tryptic soy broth (TSB) (BD Canada, Ontario, Canada) and incubated at 37°C for 18 6 0.25 h with agitation to achieve a cell density of approximately 8 log CFU/ml. The culture was then spun in a centrifuge at 3,000 rpm for 10 min at 22°C, the supernatant was discarded, and the pellet was resuspended in 10 ml of a 0.85% saline solution. Subsequently, 500-ml aliquots from each culture were added to 9.5 ml 0.85% saline, and 10 ml of the diluted culture was added to 100.0 6 0.1 g alfalfa seeds in 500-ml sterile Schott bottles. The bottles were rolled for 1 min to distribute the inoculum through the seed mass, and the contents were transferred to sterile paper sheets in a biosafety cabinet, where they were held for 2 h.
Antimicrobial seed treatments and germination of alfalfa seed. Alfalfa seed inoculated with each S. enterica strain was subjected to three antimicrobial treatments: immersion in a 5,000-ppm sodium hypochlorite solution (NaClO) (Ricca Chemical Company, Texas, USA) for 20 min (CLO) and a treatment compliant with organic production practices (HPA) consisting of immersion in water heated to 50°C for 10 min followed by immersion in 2% hydrogen peroxide (BDH, Alberta, Canada) with 0.1% acetic acid (white vinegar; Heinz Company, Illinois, USA) for 10 min. The concentration of 5,000 ppm NaClO was chosen because 20,000 ppm calcium hypochlorite as recommended by FDA (9) is caustic and can be a potential health hazard to workers who are handling the agent. On the other hand, the Canadian recommendation of 2,000 ppm calcium hypochlorite was not sufficient to kill S. enterica on alfalfa seed in previous studies (27). Twenty-five grams of alfalfa seed in sterile 500-ml glass bottles was treated with 125 ml of each solution or sterile water (control [CTL]). The bottles were shaken vigorously every 2 min for 10 s. After treatment, the seeds were filtered through Whatman no. 4 filter paper (20 mm) to remove the antimicrobial solution and were rinsed with 125 ml sterile distilled water (DW). Immediately after seed treatment, 20.00 6 0.05 g treated or untreated seeds was placed in a sterile pipette tip box with a sterile gauze pad moistened with 20 ml sterile DW. The boxes were closed and placed in a darkened incubator set at 22.0 6 0.5°C for 6 days. Sterile DW (5 ml) was added to each box 24 h after initiation of germination and every day thereafter.
Enumeration of S. enterica and aerobic mesophilic bacteria on germinating alfalfa seed. S. enterica populations were estimated from colony counts obtained on the selective medium xylose lysine deoxycholate agar (XLD; BD Canada, ON, Canada). Total aerobic mesophilic populations were estimated on tryptic soy agar (TSA). Seed from each box was sampled at 0 h (immediately after treatment) and after 1 and 6 days of germination. At each sampling interval, 3 g of seed was placed in a 50-ml sterile centrifuge tube with 27 ml phosphate-buffered saline (PBS) (Amresco Inc., Ohio, USA). After mixing on a vortex mixer for 15 s, serial dilutions of the supernatant were prepared in PBS, and 50 ml or 1 ml of the diluted samples was applied to XLD, in duplicate, using a spiral plater (easySpiral Pro; Interscience Laboratories, Inc., Massachusetts, USA). The plates were incubated at 37°C for 24 6 2 h. Additionally, 50ml aliquots were applied to TSA using the spiral plater for estimation of aerobic plate count (APC). The TSA agar plates were incubated at 37°C for 24 6 2 h. Indigenous mesophilic aerobic populations were calculated by subtraction of S. enterica colony counts on XLD agar from colony counts on TSA agar.
Lyophilization of sprout exudates. Germinated seeds were collected after 0, 8, 16, 24, and 32 h and 6 days for proximate analyses and 24 h for metabolomics analysis. Samples (10 g) were placed in sterile filtered stomacher bags (Filtra-Bag; VWR, Alberta, Canada) with 10 ml sterile DW and were gently massaged by hand for 2 min. The filtrates were transferred to 100-ml flat-bottom specimen containers (VWR, Alberta, Canada) with 8 ml 100% ethanol and were stored at 220°C for 24 h. The containers were then placed in a high-vacuum chamber for 6 h to remove ethanol, and the residual materials were frozen at 280°C overnight in preparation for lyophilization in a bulk tray freeze-dryer (Labconco, Missouri, USA).
Proximate analyses. Germinating seed lyophilizates were dissolved in 1 ml sterile DW for proximate analysis. Protein content was determined by Bradford assay (41); total phenolic content was determined using a Folin-Ciocalteu assay with minor modifications as described by Chen et al. (42); total soluble carbohydrate content was determined by phenol-sulfuric acid methods (43).
Metabolic profiling. Sample extraction and preparation were conducted at Metabolon, Inc. (Research Triangle Park, NC, USA), with their mass spectrometer platforms and instrument settings and conditions, as described previously in detail (44,45) with a few modifications. The major components of the process are summarized as follows. The sample preparation process was carried out using an automated MicroLab STAR system (Hamilton, Nevada, USA). Recovery standards were added prior to the first step in the extraction process for quality control purposes. Sample preparation was conducted using a proprietary series of organic and aqueous extractions to remove the protein fraction while allowing maximum recovery of small molecules. The resulting extract was divided into equal fractions for analysis on three independent platforms: ultrahigh-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS2) optimized for basic molecules (LC/MS-NEG), UHPLC/MS/MS2 optimized for acidic molecules (LC/MS-POS), and UHPLC/MS/MS2 optimized for polar molecules (LC/MS-Polar).
The mass spectrometry data files were extracted, and ion peaks were identified and integrated using Metabolon's proprietary software. Metabolites were then identified by automated comparison of the ion features in the experimental samples to an in-house reference library of chemical standard entries that included retention time, molecular weight (m/z), preferred adducts, and in-source fragments as well as associated MS spectra and curated by visual inspection for quality control using software developed at Metabolon. The reference library, including known and novel metabolites, was developed by analyzing more than 3,000 pure reference standards of known metabolite structures with the identical LC/MS methods and cataloging all the ions that were produced.
Validation of the effects of germinating seed metabolites on S. enterica growth. The effects of germinating seed metabolites on S. enterica growth were evaluated by two methods: first, using complete leachates from CLO-or HPA-treated alfalfa seeds, encompassing all seed metabolites but filtered to remove bacterial contamination; second, using specific metabolites of interest, specifically, two polyamines previously implicated in pathogenic virulence (46) and found in this study to be significantly increased in CTL and CLO-treated alfalfa seeds inoculated with S. enterica (P , 0.05).
To evaluate the effect of metabolites using complete leachates, 25 g of uninoculated seed was rinsed with CLO or HPA over Whatman no. 4 filter paper and germinated for 24 h. The germinated seed was transferred to sterile filter stomacher bags with 10 ml sterile DW, which were gently massaged by hand for 2 min, and passed through a 0.2-mm syringe filter (VWR, Canada) to prepare leachate. Alfalfa seed inoculated with S. Agona PARC 5 or S. Typhimurium LMFS-S-JF-001 was concomitantly treated with CLO or HPA and were transferred to pipette tip boxes along with 2 ml of leachate. During seed incubation, fresh leachate was injected daily. As a CTL, inoculated alfalfa seed in a pipette tip box was treated with 2 ml of sterile water. Populations of S. Agona PARC 5 and S, Typhimurium LMFS-S-JF-001 on germinating seeds were enumerated immediately after addition of the leachate and after 4, 8, 12, and 24 h and 6 days.
To evaluate the effect of specific metabolites of interest, 20 ml of polyamine solutions containing either spermidine (Alfa Aesar, Massachusetts, USA), agmatine (TCI America, Oregon, USA), or N 1 -and N 8acetylspermidine (Musechem, New Jersey, USA) at concentrations of 10, 100, and 500 ppm was added daily to 20 g alfalfa seed inoculated with S. Agona PARC 5 or S. Typhimurium LMFS-S-JF-001 (a range of 10 to 500 ppm was found to have the largest effect on S. enterica growth in preliminary analyses; data not shown). Twenty grams of alfalfa seed in sterile 500-ml glass bottles was treated with 125 ml of each solution or sterile water (CTL). Populations of each S. enterica strain were determined after 0, 4, 8, 12, 24, and 48 h and 6 days of germination by the procedure described above. Growth curves drawn from the data were fitted to the Baranyi and Roberts predictive model using DMfit (v3.5) available on the ComBase browser (http://browser.combase.cc/DMFit.aspx), and the growth kinetic parameters were recorded (47).
Statistical analysis. All experiments were repeated at least three times using independently prepared samples. Bacterial populations during germination, protein content, total phenolic content, and soluble carbohydrate content were examined by one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) test (a = 0.05) using Minitab 17 (Minitab, Inc., Pennsylvania, USA). P values of ,0.05 were considered statistically significant.
Statistical comparisons of metabolites derived from metabolomics analyses were performed by independent t tests using Minitab 17. An estimate of the false discovery rate (q value) was calculated to take into account the multiple comparisons that normally occur in metabolomics-based studies. Principalcomponent analysis (PCA) and all other statistical analyses were performed using ArrayStudio (Omicsoft, North Carolina, USA) on natural log-transformed scaled imputed data.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, PDF file, 0.2 MB.