ABSTRACT The topic of food safety continues to receive increased attention and has ramifications on various human, environmental, policy, and economic levels worldwide. By garnering feedback from 30 food industry experts, this study was undertaken to identify the most critical issues facing the food industry in relation to food safety. According to expert opinion and after three rounds of Delphi inquiry, food contamination detection, outbreaks, and prevention along with governmental oversight, education for and communication with consumers and employees, and globalization were identified as the main areas at the forefront of food safety. Delphi and constant comparative research methods are explained, and suggestions on how to make meaning from the results to progress in this area are discussed. HIGHLIGHTS Experts identified 42 critical safety issues currently facing the food industry. Contamination, government involvement, education, and globalization are top issues. Issues also have themes related to technology, DNA sequencing, and emerging trends. Government oversight for outbreaks is a priority area of challenge and opportunity.

ABSTRACT Salsa-associated outbreaks, including the large multistate outbreak in the United States in 2008 caused by jalapeño and serrano peppers contaminated with Salmonella Saintpaul, have raised concerns about salsa as a potential vehicle for transmission. Despite these events, there has been relatively limited research on the potential growth of pathogenic bacteria in salsa. The aim of this study was to characterize the survival and growth of Salmonella, including the outbreak strain of Salmonella Saintpaul (E2003001236), in freshly made salsa and its main ingredients. Chopped tomatoes, jalapeño peppers, cilantro, and onions were tested individually or mixed according to different salsa recipes. Samples were inoculated with five Salmonella serotypes at 3 log CFU/g: Saintpaul (various strains), Typhimurium, Montevideo, Newport, or Enteritidis. Samples were then stored at room temperature (23°C) for up to 12 h or 3 days. The Salmonella Saintpaul levels reached approximately 9 log CFU/g after 2 days in tomato, jalapeño pepper, and cilantro. Growth was slower in onions, reaching 6 log CFU/g by day 3. Salsa recipes, with or without lime juice, supported the growth of Salmonella Saintpaul, and final levels were approximately 7 log CFU/g after 3 days at 23°C. In contrast, the counts of Salmonella Typhimurium, Salmonella Montevideo, Salmonella Newport, and Salmonella Enteritidis increased only 2 log CFU/g after 3 days in any of the salsas. Other Salmonella Saintpaul strains were able to grow in salsas containing 10% lime juice, but their final levels were less than 5 log CFU/g. These findings indicate the enhanced ability of the Salmonella Saintpaul outbreak strain to grow in salsa compared with other Salmonella strains. Recipe modifications including but not limited to adding lime juice (at least 10%) and keeping fresh salsa at room temperature for less than 12 h before consumption are strategies that can help mitigate the growth of Salmonella in salsa.

ABSTRACT Community-associated Clostridium difficile infection (CA-CDI) now accounts for approximately 50% of CDI cases in central Minnesota; animals and meat products are potential sources. From November 2011 to July 2013, we cultured retail meat products and fecal samples from food-producing and companion animals in central Minnesota for C. difficile by using standard methods. The resulting 51 C. difficile isolates, plus 30 archived local veterinary C. difficile isolates and 208 human CA-CDI case isolates from central Minnesota (from 2012) from the Minnesota Department of Health, were characterized molecularly, and source groups were compared using discriminant analysis. C. difficile was recovered from 0 (0%) of 342 retail meat samples and 51 (9%) of 559 animal fecal samples. Overall, the 81 animal source isolates and 208 human source isolates were highly diverse genetically. Molecular traits segregated extensively in relation to animal versus human origin. Discriminant analysis classified 95% of isolates correctly by source group; only five (2.5%) human source isolates were classified as animal source. These data do not support meat products or food-producing and companion animals as important sources of CA-CDI in the central Minnesota study region.

The effect of lactose at the concentration typically found in milk (134 mM) on the ability of ricin to inhibit protein synthesis in HeLa cells was studied. Ricin (0.001 to 300 μg/ml) that was either not treated or treated with 134 mM lactose was added to test tubes containing 1 ml of HeLa cells (approximately 3 × 10 5 cells in a low-leucine medium). After 2 h of incubation at 37°C, 0.5 μCi of l -[U- 14 C]-leucine was added to each tube and incubated for another 60 min. The cells were harvested by centrifugation and lysed, and cellular proteins were separated. The amount of radioactivity incorporated into the proteins was determined by liquid scintillation. The biological activity of ricin, i.e., the amount of radioactivity in a sample relative to that of the control (cells not treated with ricin), was calculated for each treatment. The inhibitory effect of 134 mM lactose on the biological activity of ricin was only significant at concentrations of ricin below 1 μg/ml. At higher ricin concentrations, the effect of 134 mM lactose decreased as the concentration of ricin increased, resulting in an increase in the inhibition of proteins synthesis. Our results also indicated that bovine milk, when used in place of 134 mM lactose, was more effective for reducing the activity of ricin at concentrations below 1 μg/ml but was ineffective against ricin concentrations greater than 1 μg/ml. These results suggest that milk may not protect against ricin intoxication at the concentration (0.89 μg/ml) equivalent to the lowest limit of its 50% lethal dose for a 20-kg child consuming 225 ml (8 oz) of milk.

This study was conducted to compare the identification of Shiga toxin 1 (Stx1) based on its specific biological activity and based on results of a commercial enzyme-linked immunosorbent assay (ELISA) kit. Stx1 was thermally treated for various periods in phosphate-buffered saline, milk, and orange juice. The residual Stx1 concentration was determined with the commercial ELISA kit, and its residual enzymatic activity (amount of adenine released from a 2,551-bp DNA substrate) was determined with a biological activity assay (BAA). Regression analysis indicated that the inactivation of Stx1 as a function of time followed first-order kinetics. The half-lives determined at 60, 65, 70, 75, 80, and 85°C were 9.96, 3.19, 2.67, 0.72, 0.47, and 0.29 min, respectively, using the BAA. The half-lives determined by the ELISA with thermal treatments at 70, 75, 80, and 85°C were 40.47, 11.03, 3.64, and 1.40 min, respectively. The Z, Q 10 , and Arrhenius activation energy values derived by both assays were dissimilar, indicating that the rate of inactivation of the active site of Stx1 was less sensitive to temperature change than was denaturation of the epitope(s) used in the ELISA. These values were 10.28°C and 9.40 and 54.70 kcal/mol, respectively, with the ELISA and 16°C and 4.11 and 34 kcal/mol, respectively, with the BAA. Orange juice enhanced Stx1 inactivation as a function of increasing temperature, whereas inactivation in 2% milk was not very much different from that in phosphate-buffered saline. Our investigation indicates that the ELISA would be a reliable method for detecting the residual toxicity of heat-treated Stx1 because the half-lives determined with the ELISA were greater than those determined with the BAA (faster degradation) at all temperatures and were highly correlated ( R 2 = 0.994) with those determined with the BAA.

In this study, saxitoxin dihydrochloride in skim milk was reacted with sodium hydroxide and hydrogen peroxide to yield nontoxic 8-amino-6-hydroxymethyl-iminopurine-3(2 H )-propionic acid (AHIPA), which was quantified by fluorescence spectroscopy using excitation and emission wavelengths of 330 and 425 nm, respectively. Samples of saxitoxin dihydrochloride (in 20% ethanol, vol/vol) were used as controls. The limits of detection of AHIPA, based on the concentration of saxitoxin prior to inactivation, were 5 and 10 μg/ml for the control and skim milk, respectively. These values are considerably below the concentration of saxitoxin that corresponds to the lethal dosage of 1 mg for an adult of average weight (70 kg). The inactivation of saxitoxin proceeded at a lower rate in skim milk than in the control, as its reaction rate constant was only 0.004 min −1 compared with 0.011 min −1 for the control. We were unable to detect AHIPA in 2% milk contaminated with saxitoxin because of possible interference from what we believed were products of secondary reactions involving milk fat and sodium hydroxide. Our results also indicated that the conversion of saxitoxin to AHIPA increased initially with temperature up to 40°C but decreased thereafter. We observed a decrease in the formation of AHIPA when the concentration of hydrogen peroxide was increased except at 22°C, where there was an initial increase in AHIPA formation between 1.2 and 2.4 mg/ml hydrogen peroxide but its formation decreased thereafter.

Foods have been identified as a potential target for bioterrorism due to their essential nature and global distribution. Foods produced in bulk have the potential to have large batches of product intentionally contaminated, which could affect hundreds or thousands of individuals. Bacillus anthracis spores are one potential bioterrorism agent that may survive pasteurization and remain viable throughout the shelf life of fruit juices and cause disease if consumed. This project examined B. anthracis spore survival in orange, apple, and grape juices, as well as wine. Samples of beverages were inoculated with spores of two nonpathogenic B. anthracis strains at approximately 10 6 CFU/ml, and the spore count was determined periodically during storage for 30 days at 4°C. After this time, the counts of survival spores never declined more than 1 log CFU/ml in any of the beverage types. These results indicate that spores can survive, with little to no loss in viability, for at least a month in fruit juices and wine.

The growth of Listeria monocytogenes inoculated on frankfurters at four inoculum levels (0.1, 0.04, 0.01, and 0.007 CFU/g) was examined at 4, 8, and 12°C until the time L. monocytogenes populations reached a detectable limit of at least 2 CFU/g. A scaled-down assumption was made to simulate a 25-g sample from a 100-lb batch size in a factory setting by using a 0.55-g sample from a 1,000-g batch size in a laboratory. Samples of 0.55 g were enriched in PDX-LIB selective medium, and presumptive results were confirmed on modified Oxford agar. Based on the time to detect (TTD) from each inoculum level and at each temperature, a shelf life model was constructed to predict the detection or risk levels reached by L. monocytogenes on frankfurters. The TTD increased with reductions in inoculum size and storage temperature. At 4°C the TTDs (±standard error) observed were 42.0 ± 1.0, 43.5 ± 0.5, 50.7 ± 1.5, and 55.0 ± 3.0 days when the inoculum sizes were 0.1, 0.04, 0.01, and 0.007 CFU/g, respectively. From the same corresponding inoculum sizes, the TTDs at 8°C were 4.5 ± 0.5, 6.5 ± 0.5, 7.0 ± 1.0, and 8.5 ± 0.5 days. Significant differences ( P < 0.05) between TTDs were observed only when the inoculum sizes differed by at least 2 log. On a shelf life plot of ln(TTD) versus temperature, the Q 10 (increase in TTD for a 10°C increase in temperature) values ranged from 24.5 to 44.7 and with no significant influence from the inoculum densities. When the observed TTDs were compared with the expected detection times based on the data obtained from a study with an inoculum size of 10 to 20 CFU/g, significant deviations were noted at lower inoculum levels. These results can be valuable in designing a safety-based shelf life model for frankfurters and in performing quantitative risk assessment of listeriosis at low and practical contamination levels.

This research was conducted to study the growth of Listeria monocytogenes inoculated on frankfurters stored at different conditions as a basis for a safety-based consume by shelf life date label. Three L. monocytogenes strains were separately inoculated at 10 to 20 CFU/cm 2 onto frankfurters that were previously formulated with or without high pressure and with or without added 2% potassium lactate (PL) and 0.2% sodium diacetate (SD). Inoculated frankfurters were air or vacuum packaged; stored at 4, 8, or 12°C; and L. monocytogenes and psychrotrophic plate counts were determined for 90, 60, and 45 days, respectively, or until the stationary phase was reached. The data (log CFU per square centimeter versus time) were fitted using the Baranyi-Roberts model to determine maximum growth rates and lag-phase time. The maximum growth rates and the lag time under each growth condition were used to calculate the time to reach 100-fold the initial Listeria population. In frankfurters lacking PL and SD, the count of all strains increased by 2 log after 18 to 50 days at 4°C and 4 to 13 days at 8°C. The growth was inhibited at 4 and 8°C in frankfurters containing PL and SD, but one ribotype was capable of growing, with the time to reach 100-fold the initial Listeria population ranging from 19 to 35 days at 12°C. In most cases, the time to reach 100-fold the initial Listeria population of L. monocytogenes was significantly longer in vacuum-packaged frankfurters as compared with air-packaged samples. Inclusion of PL and SD also inhibited the growth of psychrotrophs, but at all temperatures the psychrotrophic plate counts were greater than 4 log CFU/cm 2 at the end of the experiments. These results indicated that despite the use of antimicrobials, certain L. monocytogenes strains could be capable of growing under storage-abuse conditions. Growth kinetics data could be useful for establishing a shelf life date label protocol under different handling scenarios.

The combined effect of heat and hydrogen peroxide (HP) on the inactivation of avirulent Bacillus anthracis spores (Sterne strain 7702; strain ANR-1, an avirulent Ames derivative lacking the pXO2 plasmid; and strain 9131, a plasmid-less Sterne strain) was evaluated in milk. The study temperature ranged from 90 to 95°C, and the concentration of added HP varied from 0.05 to 0.5%. Decimal reduction times ( D -values) were determined using a sealed capillary tube technique. The mean D - and z -values of hydrated freeze-dried spores of all three strains in milk ranged from 550 s at 90C to180s at 94°C and from 8.6 to 9.0°C, respectively. When 0.05% HP was added to the milk, the D -values were decreased at least threefold, and at 0.5% HP the D -values ranged from 1 to 10 s. At 90°C, all three strains had similar D -values when 0.05% HP was added. Increasing the concentration of HP to 0.5% had a greater reducing effect on the D -value for strain 7702 than on the values for strains ANR-1 and 9131. The rate of inactivation of each strain followed first-order reaction kinetics at each temperature-peroxide combination. Equations in the form of D = Constant × (HP concentration) n had R 2 values greater than 0.97 for strains ANR-1 and 7702 and of at least 0.7 for strain 9131. This study suggests that a combination of high temperature (from 90 to 95°C) and HP could be used for inactivation of B. anthracis spores in the event of deliberate contamination of milk such that the contaminated milk could be disposed of safely.

Microbiological analyses of fruits and vegetables produced by farms in Minnesota and Wisconsin were conducted to determine coliform and Escherichia coli counts and the prevalence of E. coli , Salmonella , and E. coli O157:H7. During the 2003 and 2004 harvest seasons, 14 organic farms (certified by accredited organic agencies), 30 semiorganic farms (used organic practices but not certified), and 19 conventional farms were sampled to analyze 2,029 preharvest produce samples (473 organic, 911 semiorganic, and 645 conventional). Produce varieties included mainly lettuces, leafy greens, cabbages, broccoli, peppers, tomatoes, zucchini, summer squash, cucumber, and berries. Semiorganic and organic farms provided the majority of leafy greens and lettuces. Produce samples from the three farm types had average coliform counts of 1.5 to 2.4 log most probable number per g. Conventional produce had either significantly lower or similar coliform populations compared with the semiorganic and organic produce. None of the produce samples collected during the 2 years of this study were contaminated with Salmonella or E. coli O157:H7. E. coli contamination was detected in 8% of the samples, and leafy greens, lettuces, and cabbages had significantly higher E. coli prevalence than did all the other produce types in both years for the three farm types. The prevalence of E. coli contamination by produce type was not significantly different between the three farm types during these 2 years, with the exception of organic leafy greens, in which E. coli prevalence was one-third that of semiorganic leafy greens in 2003. These results indicate that the preharvest microbiological quality of produce from the three types of farms was very similar during these two seasons and that produce type appears to be more likely than farm type to influence E. coli contamination.

Samples were collected from 26 organic and conventional farms and 12 county fairs in Minnesota during 2001 and 2002 to identify the presence of Escherichia coli O157. Immunomagnetic separation was used for isolation of E. coli O157. Isolates were further characterized by the presence of virulence marker genes ( stx 1 , stx 2 , eaeA, E-hly, katP, etpD, and espP ), antimicrobial susceptibility profiles, and genotypes. During 2001, E. coli O157 was isolated from 16 (5.2%) of 305 fecal samples and from 7 (36.8%) of 19 farms. During 2002, E. coli O157 was isolated from 6 (4.5%) of 132 fecal samples from weaned calves at 4 (23.5%) of 17 farms. During 2001 and 2002, cattle manure samples were collected from 12 county fairs, and E. coli O157 was isolated from 19 (11%) of 178 samples and 9 (75%) of 12 county fairs. Among 40 E. coli O157 isolates, 17 isolates (43%) had both the stx 1 and stx 2 genes, and 21 strains (53%) had the stx 2 gene only. Thirteen percent of O157 isolates were resistant to tetracycline, and 25% were resistant to sulfadimethoxine. Heterogeneity of E. coli O157 strains was demonstrated by the presence of 22 different pulsed-field gel electrophoresis (PFGE) patterns. Four PFGE patterns matched those of isolates previously found in humans. The presence of E. coli O157 at county fairs suggests the potential for transmission to the public, who may have contact with cattle or their environment.

Microbiological analyses of fresh fruits and vegetables produced by organic and conventional farmers in Minnesota were conducted to determine the coliform count and the prevalence of Escherichia coli , Salmonella , and E. coli O157:H7. A total of 476 and 129 produce samples were collected from 32 organic and 8 conventional farms, respectively. The samples included tomatoes, leafy greens, lettuce, green peppers, cabbage, cucumbers, broccoli, strawberries, apples, and seven other types of produce. The numbers of fruits and vegetables was influenced by their availability at participating farms and varied from 11 strawberry samples to 108 tomato samples. Among the organic farms, eight were certified by accredited agencies and the rest reported the use of organic practices. All organic farms used aged or composted animal manure as fertilizer. The average coliform counts in both organic and conventional produce were 2.9 log most probable number per g. The percentages of E. coli –positive samples in conventional and organic produce were 1.6 and 9.7%, respectively. However, the E. coli prevalence in certified organic produce was 4.3%, a level not statistically different from that in conventional samples. Organic lettuce had the largest prevalence of E. coli (22.4%) compared with other produce types. Organic samples from farms that used manure or compost aged less than 12 months had a prevalence of E. coli 19 times greater than that of farms that used older materials. Serotype O157:H7 was not detected in any produce samples, but Salmonella was isolated from one organic lettuce and one organic green pepper. These results provide the first microbiological assessment of organic fruits and vegetables at the farm level.

A previously identified set of anti– Escherichia coli O157:H7 colicinogenic E. coli were characterized to assess the suitability of these isolates as a preharvest food safety intervention in cattle. This collection of 23 E. coli strains were screened for virulence factors, antibiotic resistance, type of colicin(s) present, and their ability to inhibit other pathogenic E. coli . With the use of PCR, pathogen genes were detected in six of the 23 colicinogenic E. coli . When the nonpathogenic strains were assessed for antibiotic resistance, four strains showed resistance to at least one antibiotic. The remaining set of 14 strains were evaluated for the presence of previously identified colicins. Seven colicins (B, E1, E2/E7, E7, Ia/Ib, K, and M) were detected. One half of the strains possessed multiple types of colicins. The most commonly detected colicins were B, E2/E7, and M, which were found in six strains each. DNA sequencing was also performed in order to classify the E2/E7 colicins separately from E7 colicins. The 14 colicinogenic E. coli also were evaluated for their ability to inhibit 10 different non-O157 pathogenic E. coli . Six of the colicinogenic E. coli were capable of inhibiting all 10 pathogens, and the remaining eight strains could each inhibit between six to eight of the pathogenic E. coli . This strain collection has great potential for inhibiting E. coli O157:H7 in cattle.

Escherichia coli strains were screened for their ability to inhibit E. coli O157:H7. An initial evaluation of 18 strains carrying previously characterized colicins determined that only colicin E7 inhibited all of the E. coli O157:H7 strains tested. A total of 540 strains that had recently been isolated from humans and nine different animal species (cats, cattle, chickens, deer, dogs, ducks, horses, pigs, and sheep) were tested by a flip-plating technique. Approximately 38% of these strains were found to inhibit noncolicinogenic E. coli K12 strains. The percentage of potentially colicinogenic E. coli per animal species ranged from 14% for horse isolates to 64% for sheep strains. Those isolates that inhibited E. coli K12 were screened against E. coli O157:H7, and 42 strains were found to be capable of inhibiting all 22 pathogenic strains tested. None of these 42 strains produced bacteriophages, and only 24 isolates inhibited serotype O157:H7 in liquid culture. The inhibitory activity of these strains was completely eliminated by treatment with proteinase K. When mixtures of these 24 colicinogenic strains were grown in anaerobic continuous culture, the four-strain E. coli O157:H7 population was reduced at a rate of 0.25 log 10 cells per ml per h, which was fivefold faster than the washout rate. Two strains originally isolated from cat feces (F16) and human feces (H30) were identified by repetitive sequences polymerase chain reaction as the predominant isolates in continuous cultures. The results of this work indicate that animal species other than cattle can be sources of anti-O157 colicinogenic strains, and these results also lead to the identification of at least two isolates that could potentially be used in preharvest control strategies.