INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Escherichia coli O157:H7 was first identified as a human pathogen in 1982 and has emerged as a significant food-borne pathogen (16). This pathogen causes hemorrhagic colitis and, in some cases, potentially fatal hemolytic-uremic syndrome (15). Outbreaks have been associated with ground beef (16), apple cider (3), and dry, fermented sausage (9). The acid tolerance of O157:H7 strains permits survival in acidic foods and may play a role in the survival of E. coli O157:H7 during passage through the stomach (2, 14).

In E. coli, several stationary-phase survival genes are regulated by the alternate sigma factor, ³⁸, encoded by rpoS (17, 18). ³⁸ has been shown to regulate approximately 30 proteins, some of which enhance survival in the presence of acid, salt, or heat (10, 17, 18). The components and regulation of the rpoS regulon are beginning to be elucidated (5, 6, 21).

In 1963, substrate-accelerated death was proposed by Postgate and Hunter (23). Substrate-accelerated death is a phenomenon whereby death ensues at a much higher rate when a growth-limiting substrate is reintroduced to starved bacteria. Postgate and Hunter (23, 24) noted that substrate-accelerated death caused by glucose was enhanced in E. coli, Serratia marcescens, and Aerobacter aerogenes (Klebsiella aerogenes) when phosphate was present at a pH of >6.5. Glucose has also been implicated as an agent leading to decreased survival of bacteria in growth media when heat sterilized in the presence of phosphate (12). It was shown that Vibrio cholerae, when placed in media that had been autoclaved with both glucose and phosphate present, exhibited a decreased growth rate or inhibition of growth depending on the concentrations of glucose and phosphate. Although these studies were not the first to notice a killing effect of media on certain bacteria (11), they were the first to make the connection between phosphate and glucose. Since the description of substrate-accelerated death, studies have been conducted to determine the products responsible (8, 27), the mechanism of action (8), and the possible mutagenicity of the by-products (22), but none have provided a definitive answer to these questions. In addition, none of these studies examined the levels and conditions necessary for this compound to be toxic to pathogens nor a genetic basis for resistance or susceptibility to the compound.

In this study, the production of toxic by-products from heated glucose and phosphate and the ability to kill E. coli O157:H7 and other enteric pathogens were analyzed. In addition, the role of the rpoS regulon in protection against the glucose-phosphate by-product was examined.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. E. coli O157:H7 strain ATCC 43895; E. coli O157:H7 pO157 (strain ATCC 43895 with the 97-kb plasmid removed); an E. coli O157:H7 rpoS mutant (rpoS::pRR10) (10), strain FRIK 816-3 (from C. W. Kaspar, Food Research Institute); E. coli O127:H6 strain FRIK 345; E. coli MC4100 (25); E. coli RH90 (MC4100 rpoS mutant) (21); Klebsiella sp. strain FRIK 915; Salmonella typhimurium FRIK 49; and Shigella dysenteriae ATCC 29026 were stored in liquid nitrogen in nutrient broth containing 12% glycerol. Prior to each experiment, the frozen stock culture was streaked on tryptic soy agar (TSA) or TSA containing 250 µg of penicillin (Sigma Chemical Co., St. Louis, Mo.) per ml (for FRIK 816-3) and incubated overnight at 37°C. Five milliliters of tryptic soy broth (TSB; plus penicillin for FRIK 816-3) was inoculated with a colony from the overnight plate and incubated with shaking (150 rpm) at 37°C for 8 h. For stationary-phase cultures, 100 µl of an 8-h culture was used to inoculate 100 ml of TSB (with penicillin for FRIK 816-3) and incubated with shaking (150 rpm) at 37°C for 15 h (ca. 10⁹ CFU/ml). For exponential-phase cultures, TSB (100 ml) was inoculated with 100 µl of an 8-h culture and incubated with shaking (150 rpm) at 37°C for 4.5 h (A₆₀₀ = 0.74; ca. 10⁷ CFU/ml). Additional media used in this study included brain heart infusion broth, cooked meat medium, minimal broth plus glucose, and nutrient broth. All growth media were Difco Laboratories (Detroit, Mich.) products.

Survival studies. Phosphate-buffered saline (PBS; 0.14 M NaCl, 2.7 mM KCl, 10.1 mM Na₂HPO₄, 1.8 mM KH₂PO₄ [pH 7.4]) was sterilized (autoclaved) at 121°C at 15 lb/in² for 15 min. The chemicals used for PBS were from both Sigma and Mallinckrodt, with no differences in results. Starvation medium was sterilized in 100-ml volumes in 250-ml flasks. Glucose (Sigma) was added to PBS and either autoclaved or filter sterilized (0.2-µm-pore-size filter; Gelman Sciences, Ann Arbor, Mich.). The concentrations of the glucose added depended on the experiment but ranged from 0.25 to 2%. In experiments where heat sterilization lasted less than 15 min, the medium was filtered (0.2-µm-pore-size filter) to ensure sterility. Additional sugars (arabinose, fructose, galactose, levoglucosan, ribose, sucrose, trehalose, and xylose) were also added to PBS at a concentration of 2% and sterilized. Exponential- and stationary-phase cells were diluted 10¹ and 10³, respectively, in PBS, and 1.0 ml was added to the appropriate menstruum. The initial determination of CFU per milliliter was ca. 10⁴. Three trials were conducted for each set of conditions. Except where indicated, the flasks were incubated without shaking at 25°C in the dark.

Bacterial enumerations. Samples were removed at appropriate intervals and plated on TSA either by spread plating or by using a model D Spiral Systems (Cincinnati, Ohio) plater. Plates were incubated at 37°C overnight. The percent survivors was determined by using the CFU per milliliter determined immediately after inoculation as 100%. The limit of detection of this plating method was 10 CFU/ml; therefore, a maximum decrease of 3 log₁₀ CFU/ml could be detected. Data from at least three trials was analyzed by the t test using SigmaStat (Jandel Scientific, San Rafael, Calif.) software.

Extraction and high-performance liquid chromatography (HPLC) analysis. Extraction of a toxic by-product was achieved by placing 100 ml of starvation medium in a 500-ml separatory funnel and adding 100 ml of ethyl acetate (Fisher Scientific Co., Pittsburgh, Pa.). The funnel was shaken for 20 s and incubated without shaking for 1 min. This cycle was repeated five times. The ethyl acetate layer was removed, and the ethyl acetate was evaporated in a rotary evaporator. The residue was resuspended in 5 ml of PBS.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of autoclaved glucose-PBS on E. coli O157:H7 strains ATCC 43895 and FRIK 816-3. Initial studies characterizing the survival of the E. coli O157:H7 rpoS mutant (pRR10::rpoS; FRIK 816-3) indicated that the presence of glucose decreased survival (data not shown). Thus, the survival of stationary-phase FRIK 816-3 was examined in 0.25% glucose that was either filtered and added to autoclaved PBS or autoclaved with the PBS. Strain FRIK 816-3 survived in PBS and increased in numbers when filter-sterilized glucose was added to PBS (Fig. 1). In glucose autoclaved in PBS, the numbers of strain 816-3 decreased by >99% by day 1 and >99.9% by day 2 (original numbers were ca. 10⁴ CFU/ml). Approximately 30% of the colonies seen on day 1 were smaller in size (1 to 2 mm in diameter) than the remaining colonies (3 to 4 mm in diameter) (Fig. 2) or those seen on day 0. The limit of detection for all experiments was between 0.01 and 0.1% survivors (10 CFU/ml), depending on the starting number of CFU/ml. Analysis of FRIK 816-3 survival over 24 h demonstrated that no decline in culturable numbers was seen during the first 4 h (Fig. 3); however, after 4 h, culturable numbers had a linear decline for the next 20 h.

View larger version (11K): [in this window] [in a new window]   FIG. 1.   Survival of stationary-phase E. coli O157:H7 rpoS mutant strain FRIK 816-3 in autoclaved PBS (), 0.25% glucose autoclaved in PBS (), and 0.25% glucose (filter sterilized) in PBS (). All points represent the mean from triplicate independent trials. Error bars represent the standard error.

View larger version (100K): [in this window] [in a new window]   FIG. 2.   Morphology of E. coli O157:H7 rpoS mutant strain FRIK 816-3 after incubation for 1 day in 0.25% glucose autoclaved in PBS. This figure is also representative of colonies observed for E. coli O157:H7 strain 43895 in 2% glucose autoclaved in PBS after 1 day.

View larger version (11K): [in this window] [in a new window]   FIG. 3.   Survival of stationary-phase E. coli O157:H7 rpoS mutant strain FRIK 816-3 in autoclaved PBS () and 0.25% glucose autoclaved in PBS (). All points represent the mean from triplicate independent trials. Error bars represent the standard error.

View larger version (14K): [in this window] [in a new window]   FIG. 4.   Survival of stationary-phase E. coli O157:H7 in autoclaved PBS () alone or with 0.25% (), 0.5% (), 1% (), 1.5% (), or 2% () glucose autoclaved in PBS. All points represent the mean from triplicate independent trials. Error bars represent the standard error.

rpoS protection from glucose-phosphate by-product. To determine if the decline in rpoS mutant strain FRIK 816-3 CFU per milliliter in 0.25% glucose autoclaved in PBS was due to a nonfunctional rpoS system or serotype, a non-O157 rpoS mutant strain (RH90) of E. coli was examined. A significantly greater decline (P < 0.001) in CFU per milliliter was seen in E. coli RH90 than in the parent strain (MC4100) (Fig. 5), although there was an 80% decline in number of E. coli MC4100 CFU per milliliter on day 4. Therefore, there is serotype-to-serotype variation in the sensitivity to the glucose-phosphate by-product, but the rpoS regulon played a role in protection against the toxic glucose by-product for both serotypes.

View larger version (12K): [in this window] [in a new window]   FIG. 5.   Survival of stationary-phase E. coli O157:H7 rpoS mutant strain FRIK 816-3 in PBS autoclaved with () or without () 0.25% glucose and stationary-phase E. coli MC4100 () or E. coli RH90 (rpoS mutant of MC4100) () in 0.25% glucose autoclaved in PBS. All points represent the mean from triplicate independent trials. Error bars represent the standard error.

Requirement of phosphate and heat in the generation of the glucose-phosphate by-product. To determine the essential components for the generation of the toxic by-product(s), glucose was autoclaved in either PBS or ddH₂O (with filter-sterilized PBS added after autoclaving) and assayed for activity against strain FRIK 816-3. When glucose was autoclaved in ddH₂O, growth occurred during incubation over 3 days at the ambient temperature (data not shown). FRIK 816-3 added to glucose autoclaved in PBS declined >3 log₁₀ CFU/ml after 2 days. Thus, PBS or a component was essential to the generation of the toxic glucose by-product. Each component of PBS was then added separately to ddH₂O and 0.25% glucose, and the mixture was autoclaved. The addition of Na₂HPO₄ or KH₂PO₄ resulted in a FRIK 816-3 survival curve similar to that observed when 0.25% glucose was autoclaved in PBS (Fig. 1). NaCl or KCl autoclaved with 0.25% glucose did not have a detrimental effect on FRIK 816-3 survival and produced a survival curve similar to that seen with filter-sterilized glucose added to PBS (Fig. 1). The addition of glucose to other buffers (TBS, MOPS, or HEPES) and sterilization did not reduce the numbers of FRIK 816-3 (data not shown).

View larger version (14K): [in this window] [in a new window]   FIG. 6.   Survival of stationary-phase E. coli O157:H7 rpoS mutant FRIK 816-3 in autoclaved PBS () and 0.25% glucose autoclaved in PBS for 0 (), 5 (), 10 (), and 15 () min. All points represent the mean from triplicate independent trials. Error bars represent the standard error.

Effect of incubation temperature on the bactericidal activity of the glucose-phosphate by-product. The rpoS mutant (FRIK 816-3) and the parental strain of E. coli O157:H7 (ATCC 43895) were incubated in 0.25 and 2% glucose autoclaved in PBS at 4°C, respectively. The number of CFU of strain 816-3 per milliliter remained constant for 4 days of incubation but declined linearly to <0.1% survivors by day 8 (Fig. 7). Strain 43895 did not decline in number of CFU per milliliter until day 8 and was undetectable at day 10. Therefore, the bactericidal activity of the glucose by-product was decreased at low temperature (i.e., 4°C) but was not eliminated.

View larger version (13K): [in this window] [in a new window]   FIG. 7.   Survival of stationary-phase E. coli O157:H7 (circles) and E. coli O157:H7 rpoS mutant strain FRIK 816-3 (squares) in autoclaved PBS (filled symbols) or glucose (2% for wild-type strain and 0.25% for the rpoS mutant) autoclaved in PBS (open symbols) at 4°C. All points represent the mean from triplicate independent trials. Error bars represent the standard error.

Relationship of the glucose-phosphate by-product to substrate-accelerated death. Bacterial growth media with and without glucose were utilized to determine if substrate-accelerated death might explain the decline in culturable numbers of the bacteria (23, 24). Media containing glucose as the limiting substrate (i.e., TSB, BHI broth, cooked meat medium, or minimal medium with glucose) and without glucose (nutrient broth) were tested. E. coli O157:H7 ATCC 43895 added to 2% glucose autoclaved in PBS after growth in media either with or without glucose exhibited a decrease in numbers of CFU per milliliter (data not shown). Growth in glucose-containing media resulted in a >99.9% decline in CFU per milliliter by day 2, while cells grown in nutrient broth decreased >99% in CFU per milliliter by day 2.

Production of the bactericidal compound from other sugars. Other sugars (2%) were tested to determine if they produced a bactericidal compound when autoclaved in PBS. All reducing sugars tested (arabinose, fructose, galactose, ribose, and xylose) decreased the survival of E. coli O157:H7 ATCC 43895 by 99.9% within 2 days (data not shown). The nonreducing sugars tested (levoglucosan, sucrose, and trehalose) did not generate an inhibitory compound or decrease survival.

Sensitivity of enteric bacteria other than E. coli O157:H7 to the glucose-phosphate by-product. Stationary-phase E. coli O127:H6 was tested for its ability to survive in 0.25% glucose autoclaved in PBS (Table 1). A 90% decline in CFU per milliliter was observed by day 6. Three other enteric bacteria (Klebsiella, Salmonella typhimurium, and Shigella dysenteriae) were also examined for their ability to survive in 2% glucose autoclaved in PBS after growth to stationary phase in TSB. A decline in CFU per milliliter of 99.9% was seen by day 1 for S. dysenteriae and by day 4 for Klebsiella sp. and S. typhimurium. Survival of these bacteria in PBS remained essentially unchanged over the sampling period.

HPLC analysis of the glucose-phosphate by-product. HPLC analysis of 4% glucose autoclaved in PBS yielded two fractions (fractions 12 and 19) that showed bactericidal activity (against E. coli O157:H7 FRIK 816-3) relative to the results with eluent recovered from the column prior to injection of the glucose-phosphate by-product (data not shown). Although bactericidal activity was low (a reduction of 1.0 log₁₀ CFU/ml by day 1 for FRIK 816-3) for these two fractions, no decrease in numbers was seen with the other 20 fractions or with the control fractions.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Over 50 years ago, Corper and Clark (11) demonstrated that media containing caramelized glucose retarded the growth of tubercle bacilli. Our study demonstrates that heating reducing sugars in the presence of phosphates produces bactericidal by-products. The by-product, exact mechanism of its formation, and bactericidal activity have not been identified or elucidated. Additionally, it remains unclear whether this glucose-phosphate by-product is associated with substrate-accelerated death (23).

In this study, we found that E. coli strains without a functioning rpoS system were highly sensitive to the glucose by-product. Although E. coli strains with a functioning rpoS system were sensitive to high concentrations of the by-product, the rpoS mutant strains (FRIK 816-3 and RH90) had significantly reduced survival (P < 0.001) in comparison to the parental strains at low concentrations of the by-product. This may explain why this phenomenon was originally noticed in bacteria other than E. coli (11, 12). Previous studies examining the effect of this by-product with E. coli used 5% glucose (27), which is not comparable to the standard glucose concentrations (0.25 to 0.5%) found in growth media (i.e., tryptic soy broth). Likewise, 0.5 to 1.0% glucose autoclaved in PBS was detrimental to the parental strain of E. coli O157:H7 (ATCC 43895), whereas a concentration of glucose as low as 0.05% autoclaved in PBS was inhibitory to E. coli FRIK 816-3 (data not shown). Therefore, the level of the glucose-phosphate by-product is not high enough to inhibit growth or affect survival of most bacteria in standard media if they have a functional rpoS or analogous system.

A major concern generated by these findings is that the formation of the glucose-phosphate by-product in bacteriological growth media during sterilization may hinder recovery of bacteria from foods, particularly injured bacteria. Variation in sterilization times and conditions between labs could enhance the disparity in detection of these sublethally injured bacteria. Since the present research cannot solve this problem, it is an area that needs further exploration.

Along similar lines, Postgate and Hunter (23, 24) proposed that substrate-accelerated death, caused by glucose, was enhanced in E. coli when phosphate was present at a pH of >6.5. We subjected E. coli to the glucose-phosphate by-product after growth in media with and without glucose. The presence or absence of glucose did not affect the sensitivity of the E. coli O157:H7 to the glucose-phosphate by-product. Further studies examining recovery media should take into account the possible presence of this glucose-phosphate by-product. Our studies explain neither the reports of substrate-accelerated death occurring with noncarbohydrate substrates, like phosphate and nitrogen, nor those instances in which phosphates were not present and the carbohydrates were heated (23, 24).

Because the identity of the glucose-phosphate by-product is unknown, it is not surprising that the mechanism of bactericidal activity has not been determined. Carlsson et al. (8) found that glucose-phosphate solutions autooxidized when heated for more than 5 min, producing hydrogen peroxide. Hydrogen peroxide may not be a significant antimicrobial agent against E. coli O157:H7 at the levels produced since this strain produces high concentrations of catalase and the genes for stationary-phase catalase are regulated by rpoS (7). E. coli O157:H7 without the 97-kb plasmid (harboring katP coding for catalase) survived as well as the parental strain in low concentrations of glucose (0.25%) autoclaved in PBS (data not shown), indicating that reduced catalase levels do not impart significant sensitivity to the glucose-phosphate by-product. Moreover, addition of catalase to autoclaved glucose-PBS did not significantly reduce bactericidal activity (data not shown).

Maillard reactions between hexoses and amino acids have been shown to produce compounds, such as 2,5-dimethyl-4-hydroxy-3(2H)-furanone, that have been proposed to break DNA strands by the production of hydroxyl radicals (19). Maillard reaction by-products do influence the survival of E. coli (20), but whether this type of compound is produced in heated glucose-phosphate solutions, without the presence of amino acids, has not been determined. If intracellular release of hydroxyl radicals from the glucose-phosphate by-product is the mechanism of action, it is conceivable that rpoS-regulated DNA binding proteins, such as Dps (1), could provide protection. This would explain the difference in sensitivity between the wild-type and rpoS mutant strains. Glucose by-products have also been shown to promote conversion of a nonmutagenic compound to a mutagenic compound. Majeska and McGregor (22) added a heated glucose-phosphate solution to the nonmutagenic chemical 2,6-dimethyl[2-(2-thienyl)ethenyl]phenol and recovered more bacterial revertants than if the two solutions were tested separately. These results support the hypothesis that the glucose-phosphate by-product promotes some interaction with the DNA via hydroxyl radicals or a mutagen.

The wide array of products formed when glucose is heated in neutral to alkaline solutions in the absence of phosphates has been reviewed by Forsskahl (13), but most of the compounds have not been identified. One product, 5-(hydroxymethyl) furfural (5-HMF), was identified (26, 29) and found to be present in high concentrations, especially when glucose was heated for a long time (>50 min at 120°C) in water. Neither HPLC analysis nor thin-layer chromatography of the heated glucose-phosphate solutions detected 5-HMF (data not shown). Although 5-HMF has been shown to break down during extended heating to formic, levulinic, acetic, and lactic acids (28), these acids do not seem to be the toxic glucose-phosphate by-product since the pathway of glucose breakdown does not appear to go through 5-HMF. In addition, the toxicity of the by-products was not evident until 4 h after their introduction (Fig. 3), which is not consistent with acid toxicity. To verify that 5-HMF was not toxic to E. coli O157:H7 FRIK 816-3, a filter-sterilized solution was tested and found to have no effect on survival (data not shown). This was not totally surprising since it has been shown that E. coli has the ability to biotransform 5-HMF to 5-hydroxymethyl furfuryl alcohol, presumably to detoxify the 5-HMF (4).

Several methods have been used in an attempt to isolate and analyze the by-products of heated sugar-phosphate solutions. Suortti and Mälkki (27) found seven different HPLC fractions that were bactericidal to E. coli but were unable to isolate or identify the compounds. By utilizing different HPLC conditions, we were able to separate the compounds and narrow down the toxic bactericidal activity to two fractions. However, attempts to concentrate the by-product for further analysis were unsuccessful. Additional studies found that while the 2% glucose autoclaved in PBS and stored at 25°C for 6 days prior to inoculation was bactericidal to E. coli O157:H7 ATCC 43895, the bactericidal activity was lower than that found in a freshly made solution. In addition, UV light absorption (200 to 330 nm) studies of the by-product showed that the product is unstable at room temperature (data not shown). This instability most likely contributed to our inability to concentrate the by-product. Clearly, much work is needed before these compounds can be identified.

The production of a bactericidal by-product from readily available substrates, such as glucose (fructose) and phosphate, makes it an appealing candidate for application as an antimicrobial in processed foods. However, this goal will require isolation, identification, and stabilization of the glucose-phosphate by-product. In addition, work is needed to determine which genes of the rpoS regulon protect against the glucose-phosphate by-product and if the by-product plays a role in substrate-accelerated death. These findings will yield new insight into the survival and recovery of stationary-phase bacteria in foods and the environment.