QUANTITATIVE TRANSFER OF ESCHERICHIA COLI O157:H7 DURING PILOT-PLANT PRODUCTION OF FRESH-CUT LEAFY GREENS By Annemarie Lucia Buchholz A DISSERTATION Submitted to Michigan State University In partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Food Science 2012   ABSTRACT QUANTITATIVE TRANSFER OF ESCHERICHIA COLI O157:H7 DURING PILOT-PLANT PRODUCTION OF FRESH-CUT LEAFY GREENS By Annemarie Lucia Buchholz Cross-contamination/multi-directional transfer of Escherichia coli O157:H7 during commercial shredding, conveying, flume-washing and drying of fresh-cut leafy greens has become a major public health concern. However, the extent of bacterial transfer during the various unit operations involved in post-harvest processing of leafy greens is poorly understood. Consequently, a pilot-scale leafy green processing line consisting of a commercial shredder, conveyor, flume tank, shaker table, and centrifugal dryer was assembled to E. coli O157:H7 transfer during processing of iceberg lettuce, Romaine lettuce, and baby spinach. Using 22.7 kg of leafy greens inoculated with a 4-strain avirulent, GFP-labeled, ampicillin-resistant E. coli O157:H7 cocktail at 6, 4 and 2 log CFU/g, bacterial transfer coefficients were determined between the product, the water, and the equipment surfaces during processing, with the effect of post-inoculation hold time and shred size on transfer also evaluated. During leafy green processing, ~90% of the E. coli O157:H7 inoculum transferred to the wash water. After processing, E. coli O157:H7 populations on equipment were highest on the shredder and conveyor. Shred size did not affect the numbers of E. coli O157:H7 transferred from contaminated leafy greens during processing. However, extending the time between inoculation and processing decreased the removal and transfer of this pathogen during 6 4 processing. When 22.7 kg of product containing 10 and 10 E. coli O157:H7 CFU/g were processed followed by 22.7 kg uninoculated product, E. coli O157:H7 was quantifiable in all iceberg and romaine lettuce samples. At an inoculation level of 2 log CFU/g, E. coli O157:H7   was sporadically detected after processing 81.6 kg of uninoculated iceberg and found throughout the entire batch of uninoculated romaine lettuce. Using Radicchio as a colored surrogate for iceberg lettuce, this study demonstrated that 9.1 kg of E. coli O157:H7-inoculated Radicchio could contaminate 907 kg of subsequently processed uncontaminated iceberg lettuce. After processing, shreds of Radicchio were found in all but one of the ~38 bags of shredded lettuce. Hundreds of contaminated Radicchio shreds remained on the processing line, with the majority found on the conveyor, followed by the shredder, flume tank, and shaker table, showing that contaminated product can be continually spread during leafy green processing long after a contamination event. The effect bacterial attachment time on cross-contamination was determined by processing 0.5 kg of E. coli O157:H7-inoculated Radicchio immediately and after storage for 1 d at 4°C or 5 d at 22°C, followed by 45.5 kg of uninoculated iceberg lettuce. After processing, E. coli O157:H7 was found in all uncontaminated iceberg lettuce samples, with mean counts highest for 1 h- followed by 1 d- and 5 d-held product. Scanning electron microscopy images showed greater bacterial attachment for 5 d-held product, with larger cell aggregates enveloped in a dense extracellular matrix. When transfer between inoculated and uninoculated product was assessed during partial dewatering, faster cross-contamination was seen using an E. coli O157:H7-inoculated Radicchio to uninoculated iceberg lettuce ratio of 1:100 than 1:10,000 (w/w) until a maximum population was reached due to the higher number of E. coli O157:H7 cells available for transfer. These findings, which demonstrate E. coli O157:H7 transfer between product, wash water, and equipment surfaces, and to large quantities of uncontaminated fresh-cut leafy greens, are critical to the development of science-based transfer models for risk analysis. TABLE OF CONTENTS LIST OF TABLES ix LIST OF FIGURES xii KEY TO SYMBOLS AND ABBREVIATIONS xv 1 INTRODUCTION CHAPTER 1: Review of Pertinent Literature 1.1 Microorganisms associated with fresh produce 1.2 Escherichia coli O157:H7 1.3 Leafy green associated recalls and outbreaks 1.4 The leafy green industry 1.5 Post-harvest processing sources and spread of contamination 1.6 E. coli attachment, persistence and biofilm formation on surfaces 1.7 Inoculation of leafy greens and recovery of bacteria 1.8 Bacterial transfer in other food environments 1.9 Predictive modeling of bacterial transfer 1.10. Overall goals and objectives 4 5 6 7 15 16 25 28 30 33 35 CHAPTER 2: Quantitative Transfer of Escherichia coli O157:H7 to Equipment during Small-Scale Production of Fresh-Cut Leafy Greens 2.1 ABSTRACT 2.2 INTRODUCTION 2.3 MATERIALS AND METHODS 2.3.1 Experimental design 2.3.2 Leafy greens 2.3.3 Bacterial strains 2.3.4 E. coli O157:H7 growth rates 2.3.5 Attachment 2.3.6 Inoculation of iceberg lettuce, Romaine lettuce and baby spinach 2.3.7 Impact of post-inoculation hold time on E. coli O157:H7 transfer 2.3.8 Processing equipment 2.3.9 Identification of product contact surfaces 2.3.10 Leafy green processing 2.3.11 Impact of shred size on E. coli O157:H7 transfer 2.3.12 Sample collection 2.3.13 Microbiological analyses 2.3.14 Statistical analysis 2.4 RESULTS 2.4.1 E. coli O157:H7 growth rate and attachment ability 2.4.2 Lettuce and baby spinach 2.4.3 Flume water 2.4.4 Centrifugation water 37 iv 38 39 42 42 42 42 43 43 44 45 45 46 46 47 47 48 49 50 50 50 50 51 2.4.5 Processing equipment surfaces 2.4.6 Impact of shred size and hold time before shredding 2.5 DISCUSSION 2.6 FIGURES 2.7 TABLES CHAPTER 3: Transfer of Escherichia coli O157:H7 from Equipment Surfaces to Iceberg and Romaine Lettuce during Pilot-Plant Production of Fresh-Cut Leafy Greens 3.1 ABSTRACT 3.2 INTRODUCTION 3.3 MATERIALS AND METHODS 3.3.1 Experimental design 3.3.2 Leafy greens 3.3.3 Bacterial strains 3.3.4 Inoculation of iceberg and Romaine lettuce 3.3.5 Processing equipment 3.3.6 Iceberg and Romaine lettuce processing and sample collection 3.3.7 Microbiological analyses 3.3.8 Statistical analyses 3.4 RESULTS 3.4.1 Initial inoculation levels 3.4.2 Flume water 3.4.3 Processing equipment surfaces 3.4.4 Lettuce 3.5 DISCUSSION 3.6 FIGURES 3.7 TABLES CHAPTER 4: Tracking an Escherichia coli O157:H7 Contaminated Batch of Leafy Greens through a Commercial Processing Line 4.1 ABSTRACT 4.2 INTRODUCTION 4.3 MATERIALS AND METHODS 4.3.1 Experimental design 4.3.2 Produce 4.3.3 Bacterial strains 4.3.4 Inoculation of Radicchio 4.3.5 Processing equipment 4.3.6 Radicchio and iceberg lettuce processing and sample collection 4.3.7 Microbiological analyses 4.3.8 Statistical analysis 4.3.9 Modeling of non-linear curves 4.4 RESULTS 4.4.1 Radicchio 4.4.2 Iceberg lettuce v 51 52 54 60 72 89 90 91 94 94 94 94 95 95 96 97 97 99 99 99 99 100 102 107 113 122 123 124 126 126 126 126 127 127 128 129 129 130 131 131 131 4.4.3 Processing equipment surfaces 4.4.4 Flume water 4.5 DISCUSSION 4.6 FIGURES 4.7 TABLES CHAPTER 5: Impact of Post-inoculation Hold Time on Escherichia coli O157:H7 Transfer during Commercial Production of Fresh-cut Leafy Greens 5.1 ABSTRACT 5.2 INTRODUCTION 5.3 MATERIALS AND METHODS 5.3.1 Experimental Design 5.3.2 Produce 5.3.3 Bacterial strains 5.3.4 Inoculation of Radicchio 5.3.5 Scanning electron microscopy (SEM) 5.3.6 Processing equipment 5.3.7 Processing and collection of Radicchio and iceberg lettuce samples 5.3.8 Microbiological analyses 5.3.9 Statistical analysis 5.4 RESULTS 5.4.1 Initial inoculum levels 5.4.2 Attachment and biofilm formation 5.4.3 Radicchio 5.4.4 Iceberg lettuce 5.4.5 Flume water 5.4.6 Processing equipment surfaces 5.5 DISCUSSION 5.6 FIGURES 5.7 TABLES CHAPTER 6: Quantitative Transfer of Escherichia coli O157:H7 from Inoculated to Uninoculated Leafy Greens during Shaker Table Dewatering 6.1 ABSTRACT 6.2 INTRODUCTION 6.3 MATERIALS AND METHODS 6.3.1 Experimental Design 6.3.2 Produce 6.3.3 Bacterial strains 6.3.4 Inoculation of Radicchio 6.3.5 Processing equipment 6.3.6 Processing and collection of Radicchio and iceberg lettuce samples 6.3.7 Microbiological analyses 6.3.8 Statistical analysis 6.4 RESULTS 6.4.1 Radicchio vi 132 132 134 141 147 149 150 151 154 154 154 154 155 155 156 156 157 158 159 159 159 159 160 160 161 162 169 175 179 180 181 183 183 183 183 184 184 184 185 185 187 187 6.4.2 Iceberg lettuce 6.5 DISCUSSION 6.6 FIGURES 6.7 TABLES 187 188 191 192 CONCLUSIONS AND FUTURE RECOMMENDATIONS 193 APPENDICES 197 APPENDIX I: Mean Escherichia coli O157:H7 Populations on Equipment Surfaces after Processing 198 APPENDIX II: Isolation and Identification of Gas-Producing Yeasts from Maraschino Cherries AII.1 ABSTRACT AII.2 INTRODUCTION AII.3 MATERIALS AND METHODS AII.3.1 Sample collection AII.3.2 Enrichment AII.3.4 Isolation AII3.5 Biochemical identification of yeast isolates AII3.6 Identification of yeast isolates through sequencing AII.3.7 Gas confirmation during small-scale maraschino cherry production AII.3.8 Survival curve for yeast isolates AII.3.9 Thermal inactivation AII3.10 Laboratory pasteurization AII.3.11 Statistical analysis AII.4 RESULTS AII.4.1 Isolation of gas-producing yeasts AII.4.2 Identification of yeast isolates AII.4.3 Gas confirmation during small-scale maraschino cherry production AII.4.4 Survival curve for yeast isolates AII.4.5 Thermal inactivation and laboratory pasteurization. AII.5 FIGURES AII.6 TABLES 226 APPENDIX III: Pulsed-Field Gel Electrophoresis as a Predictor of Listeria monocytogenes Biofilm Formation AIII.1 ABSTRACT AIII.2 INTRODUCTION AIII.3 MATERIALS AND METHODS AIII.3.1 Study design AIII.3.2 Serotyping AIII.3.3 DNA isolation and PFGE typing AIII.3.4 Biofilm Formation AIII.4 RESULTS vii 227 228 229 229 229 229 230 230 231 232 232 233 233 234 234 234 235 236 236 238 244 248 249 251 252 252 252 252 252 253 AIII.5 FIGURES AIII.6 TABLES 254 255 259 REFERENCES viii LIST OF TABLES Table 1.1: Leafy green associated outbreak from 1995 through 2011 12 Table 1.2: Sources of Processing Contamination 18 Table 2.1: Bacterial attachment, as measured by the optical density at 630 nm for four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) and three virulent strains of E. coli O157:H7 (K3995, K4495, and K4831) (n=3) 72 Table 2.2: E. coli transfer (%) during processing of leafy green inoculated at 6 log CFU/g 73 Table 2.3: E. coli transfer (%) during processing of leafy green inoculated at 4 log CFU/g 75 Table 2.4: E. coli transfer (%) during processing of leafy green inoculated at 2 log CFU/g 77 Table 2.5: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 6 log CFU/g (n=3) 79 Table 2.6: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 4 log CFU/g (n=3) 82 Table 2.7: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 2 log CFU/g (n=5) 86 Table 3.1: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 6 log CFU/g (average of 3 replicates) 113 Table 3.2: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 4 log CFU/g (average of 3 replicates) 115 Table 3.3: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 2 log CFU/g (average of 5 replicates) 117 Table 3.4: E. coli O157:H7 populations (mean ± SD) recovered from equipment surfaces before and after processing 90.7 kg of uninoculated leafy greens 119 Table 4.1: Weight (g) ± SD and Percent ± SD of Radicchio recovered from 147 ix equipment surfaces after processing Table 4.2: E. coli O157:H7 populations (mean ± SD) recovered from equipment surfaces after leafy green processing 148 Table 5.1: E. coli O157:H7 populations (mean ± SD) recovered from the processing equipment surfaces after leafy green processing 175 Table 5.2: E. coli O157:H7 populations (mean ± SD), Weight ± SD and Percentage ± SD of Radicchio recovered from equipment surfaces after processing 177 Table 6.1: Weibull model parameters fitted to E. coli O157:H7 populations on iceberg lettuce 192 Table AI.1: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 6 log CFU/g (n=3) 199 Table AI.2: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 4 log CFU/g (n=3) 203 Table AI.3: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 2 log CFU/g (n=5) 210 Table AI.4: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 6 log CFU/g followed by 90.8 kg of uninoculated leafy greens (n=3) 214 Table AI.5: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 4 log CFU/g followed by 90.8 kg of uninoculated leafy greens (n=3) 218 Table AI.6: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 2 log CFU/g followed by 90.8 kg of uninoculated leafy greens (n=5) 222 Table AII.1: Sugar concentration (°Brix) and pH-value of cherry syrup 244 Table AII.2: The number gas-producing yeast positive cherry samples 245 Table AII.3: Gas production and minimum generation time in sugar syrup at 22 ± 1 °C within 30 days (n = 2) 246 Table AII.4: The D10-values and Z-values of yeasts isolated from maraschino cherry syrup (n = 3) 247 Table AIII.1: L. monocytogenes serotypes 255 x Table AIII.2: L. monocytogenes biofilm OD’s, serotypes, and PFGE types xi 257 LIST OF FIGURES Figure 1.1: Produce outbreaks due to E. coli O157:H7 between 1998 and 2006 8 Figure 1.2: Produce-linked outbreak vehicles 1998 and 2006 9 Figure 1.3: Leafy green processing equipment (clockwise from top left): lettuce shredder, conveyer belt, flume tank, shaker table, plastic collection basket and dewatering centrifuge 17 Figure 1.4: Scanning electron micrograph showing attachment and biofilm formation by E. coli and/or native microflora on a lettuce leaf 28 Figure 1.5: Simulated number of cross-contaminated bags along the production (22 batches) when one contaminated batch enters the processing line 35 Figure 2.1: Lettuce shredder sampling locations: (A) loading end, (B) interior feed belt and guides, and (C) blade and discharge chute 60 Figure 2.2: Conveyer belt sampling locations 61 Figure 2.3: Flume tank sampling locations 62 Figure 2.4: Shaker table sampling locations 63 Figure 2.5: Dewatering centrifuge sampling locations 64 Figure 2.6: Growth curves at 37°C for four avirulent, GFP-labeled, ampicillinresistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) and three virulent strains of E. coli O157:H7 (K3995, K4495, and K4831) (n=3) 65 Figure 2.7: E. coli O157:H7 populations (mean ± SD) on the product during processing of leafy greens inoculated at: (A) ~6 log CFU/g (n = 3), (B) ~4 log CFU/g (n = 3), and (C) ~2 log CFU/g (n = 5) 66 Figure 2.8: E. coli O157:H7 populations (mean ± SD) in water during processing of leafy green inoculated at: (A) ~6 log CFU/g (n = 3), (B) ~4 log CFU/g (n = 3), and (C) ~2 log CFU/g (n = 5) 69 Figure 3.1: E. coli O157:H7 populations (mean ± SD) on the product during processing of iceberg and Romaine lettuce inoculated at ~6 log CFU/g (n = 3) 107 Figure 3.2: E. coli O157:H7 populations (mean ± SD) on the product during processing of iceberg and Romaine lettuce inoculated at ~ 4 logs CFU/g 108 xii (n = 3) Figure 3.3: E. coli O157:H7 populations (mean ± SD) on the product during processing of iceberg and Romaine lettuce inoculated at ~ 2 logs CFU/g (n = 5) 109 Figure 3.4: E. coli O157:H7 populations (mean ± SD) in water during processing of iceberg and Romaine lettuce inoculated at ~6 logs CFU/g (n = 3) 110 Figure 3.5: E. coli O157:H7 populations (mean ± SD) in water during processing of iceberg and Romaine lettuce inoculated at ~4 logs CFU/g (n = 3) 111 Figure 3.6: E. coli O157:H7 populations (mean ± SD) in water during processing of iceberg and Romaine lettuce inoculated at ~2 logs CFU/g (n = 5) 112 Figure 4.1: Linear trendline fitted to mean (± SD), E. coli O157:H7 populations on the Radicchio after leafy green processing 141 Figure 4.2: Weibull model fitted to percentage (± SD) of remaining Radicchio recovered from the iceberg lettuce after leafy green processing 142 Figure 4.3: Weibull model fitted to mean (± SD), E. coli O157:H7 populations on the iceberg lettuce after leafy green processing 143 Figure 4.4: E. coli O157:H7 populations (mean ± SD) on the iceberg lettuce before and after centrifugal drying 144 Figure 4.5: Radicchio remaining on equipment surfaces after 907 kg of uninoculated product was processed: (A) conveyer belt and flume tank and (B) blade and interior surface of the lettuce shredder 145 Figure 4.6: Weibull model fitted to mean (± SD), E. coli O157:H7 populations in the 900 L of water during processing of leafy green inoculated CFU/g 146 Figure 5.1: Scanning electron micrograph images showing attachment and biofilm formation E. coli and/or native microflora on Radicchio leaves after inoculation and holding for (A) 1 hour at 22 °C, (B) 1 day at 4 °C, and C) 5 days at 22 °C 169 Figure 5.2: E. coli O157:H7 populations (mean ± SD) on Radicchio after processing of iceberg lettuce 171 Figure 5.3: Percentage by weight (± SD) of Radicchio recovered from the iceberg lettuce after processing 172 Figure 5.4: E. coli O157:H7 populations (mean ± SD) on iceberg lettuce after 173 xiii processing Figure 5.5: E. coli O157:H7 populations (mean ± SD) in 890 L of recirculation water during processing of uninoculated iceberg lettuce 174 Figure 6.1: E. coli O157:H7 populations (mean ± SD) on the iceberg lettuce after leafy green processing 191 Figure AII.1: Maraschino cherry production: (A) laboratory-scale set-up and (B) gas visible in the outlet tube 238 Figure AII.2: The correlation between syrup concentration (°Brix) and generation time of yeast isolates 239 Figure AII.3: The correlation between pH of syrup and generation time of yeast isolates 240 Figure AII.4: The survival curve of isolate Y1 in different °Brix syrup at ambient temperature 241 Figure AII.5: The survival curve of isolate Y2 in different °Brix syrup at ambient temperature 242 Figure AII.6: The survival curve of isolate Y3 in different °Brix syrup at ambient temperature 243 Figure AIII.1: L. monocytogenes PFGE patterns obtained using ApaI and AscI 254 xiv KEY TO SYMBOLS AND ABBREVIATIONS ANOVA Analysis of Variance CDC Centers for Disease Control and Prevention CFU Colony Forming Unit(s) CSPI Center for Science in the Public Interest d days(s) EHEC Enterohemorrhagic E. coli FAO The Food and Agriculture Organization of the United Nations FSMA Food Safety Modernization Act g grams(s) GFP Green fluorescent protein h hour(s) HHS United State Department of Health and Human Services HUS Hemolytic uremic syndrome min minute(s) ml milliliter(s) MLVA Multilocus variable number tandem repeat analysis mm millimeter(s) OD Optical Density P3ARRT Pathogen-Produce Pair Attribution Risk Ranking Tool PBS Phosphate Buffered Saline PFGE Pulsed-field Gel Electrophoresis ppm Parts per million xv RTE Ready-to-eat s second(s) SAS Statistical Analysis Systems SEM Scanning electron microscopy Stx Shiga toxin TSA-YE Trypticase Soy Agar with 0.6 % Yeast Extract TSB-YE Trypticase Soy Broth with 0.6 % Yeast Extract US United States of America USDA United States Department of Agriculture USDA United State Department of Agriculture VT Verocytotoxin WHO World Health Organization μl microliter(s) μm micron(s) xvi INTRODUCTION As the United States population attempts to move toward healthier eating habits (more fruits and vegetables) outbreaks associated with produce are becoming more frequent (HHS/USDA 2005). In light of recent contamination problems, consumer confidence in the safety of the produce has been declining. The Center for Science in the Public Interest (CSPI 2009) put leafy greens at the very top of the recently published list of “The 10 Riskiest Foods”, while the United Nations and World Health Organization’s Food and Agriculture Organization report (FAO/WHO 2008), ranked leafy greens a Level 1 Priority for food safety, the highest priority among produce commodities. An increasing number of foodborne outbreaks traced to fresh fruits and vegetables are partially attributed to production, processing, and consumption patterns. Since 1990, 19 outbreaks were specifically associated with the consumption of leafy greens (CSPI 2008a). In September 2006, contaminated pre-bagged baby spinach was linked to an E. coli O157:H7 outbreak in 26 states and Canada that resulted in 205 illnesses and five deaths (Jay and others 2007). A few months later, two additional E. coli O157:H7 outbreaks were traced to California grown iceberg lettuce that was shredded and then purchased by two fast-food Mexican restaurant chains in the Midwest and Northeast (FDA 2006, FDA 2007). These two outbreaks sickened a combined total of over 150 individuals (CSPI 2008b) and raised nationwide concerns regarding the safety of leafy greens along with the various routes for contamination. In the United States, the progression from locally grown produce to centralized production has led to numerous multi-state and nationwide outbreaks of foodborne illness (Calvin 2007). In many cases, fruits and vegetables are grown on centralized large-scale farms in locations that specialize in a specific product. Under these conditions, one contamination 1 incident at a large centralized grower or processor could quickly lead to a multi-state outbreak with near catastrophic consequences for the industry, as was seen in the recent outbreak involving baby spinach (Calvin 2007, Jay and others 2007). By definition, fresh fruits and vegetables generally do not undergo any treatment, other than washing, for the reduction and/or elimination of potentially hazardous microorganisms. Unfortunately, most commercial sanitizers used for washing fresh produce can only reduce the microbial levels by 90 to 99% ,which still makes these products potential vehicles for the transmission of microbial pathogens such as Salmonella and Escherichia coli O157:H7 (Beuchat and others 2004, Burnett and others 2004, Keskinen and others 2009, Sapers 2001, Weissinger and others 2000, Zhang and others 2009). Microbial contamination of fresh produce can occur at any point throughout the farm-tofork continuum (Beuchat 2002). In the field, spoilage and pathogenic microorganisms that adversely impact product safety and shelf-life can come from many sources, including the soil, manure, irrigation water, and both domestic livestock and wild animals (Beuchat 2006, Beuchat and Ryu 1997). During processing, the microbial load often increases as a result of direct contact with contaminated equipment surfaces (e.g., conveyor belts, slicers, dicers) (Abadias and others 2008, Allende and others 2004, Garg and others 1990, Kaneko and others 1999). Human foodborne pathogens can be readily transferred to much larger quantities during subsequent product handing and further processing, raising even greater food safety concerns. It has been demonstrated that foodborne pathogens can also contaminate the product during processing (Stafford and others 2002). Even as the number and magnitude of the outbreaks increase, there is still very little known about the transfer of bacteria during various unit operations involved in post-harvest processing (shredding, conveying, washing and dewatering) of leafy greens. 2 It is hypothesized that E. coli O157:H7 is transferred in quantifiable numbers between leafy greens, wash water, and the food contact surface areas of the processing equipment during simulated commercial processing. The research reported in this dissertation had five primary objectives: 1) quantify E. coli O157:H7 transfer to equipment during pilot-scale production of fresh-cut leafy greens; 2) quantify the transfer of E. coli O157:H7 from equipment surfaces to iceberg and Romaine lettuce during simulated pilot-scale processing; 3) track an E. coli O157:H7 inoculated batch of leafy greens through a pilot-scale processing line; 4) determine the impact of post-inoculation hold time on E. coli transfer during leafy greens processing and 5) quantify the E. coli O157:H7 transfer from inoculated leafy greens to uninoculated leafy greens during partial dewatering. 3 CHAPTER 1 Review of Pertinent Literature 4 1.1 Microorganisms associated with fresh produce Fresh produce is typically host to a highly diverse group of microorganisms, including bacteria, yeasts, molds, parasites and viruses, as well as insects and other plant pests, that may exist on plants as harmless commensals, plant pathogens or potential spoilage organisms. Aerobic bacterial counts on minimally processed vegetables generally range from 5 to 7 log CFU/g with the majority (80 to 90%) of these organisms being Pseudomonas, Enterobacter and Erwinia species (Francis and others 1999; Garg and others 1990; King and others 1991; Johnston and other 2005; Ruiz and others 1987; Thunberg and others 2002). Lactic acid bacteria are also detected on some leafy greens during temperature abuse (Francis and others 1999; Manvell and Ackland 1978). Bacterial counts on fresh produce can vary greatly depending on growth, harvest, processing, packaging, storage and handling conditions. Based on a few field and retail market surveys, coliform and Enterococcus counts ranged from 1 to 4 log CFU/g, with E. coli levels of < 1 to 3 log CFU/g (Johnston and other 2005; Mukherjee and others 2004; Ruiz and others 1987). In another survey, 1.6 and 9.7% of conventionally and organically grown produce yielded E. coli, respectively. For organically grown produce, E. coli populations were 19 times higher using manure or compost aged less than 12 months as compared to older compost (Mukherjee and others 2004). As part of a diverse microbial community, fresh produce may also harbor potential human foodborne pathogens such as Salmonella, E. coli O157:H7, Cryptosporidium and Hepatitis A virus, all of which are indicators of fecal contamination (Doyle and Erickson 2006). Many reviews have addressed the incidence, growth and survival of foodborne pathogens on fresh and processed produce (Beuchat 1996; Beuchat and Bracket 1990; Beuchat and others 2004; Gleeson and O'Beirne 2005; Harris and others 2003). 5 1.2 Escherichia coli O157:H7 One particular organism, E. coli O157:H7, a facultative anaerobic, Gram-negative, rodshaped bacterium, is an unusually virulent enterohemorrhagic strain (EHEC) of E. coli. It is a cause of enteric disease resulting in bloody diarrhea and severe abdominal pain and in some more rare cases, hemolytic uremic syndrome (HUS). Characterized by hemolytic anaemia, thrombocytopenia and hemolytic uremic syndrome (acute kidney failure) (Kaper and others 2004), HUS appears mainly in children, as first described by Karmali and others (1983). E. coli O157:H7 is a common inhabitant of the bovine gastrointestinal tract and has historically been a major source of concern in the ground beef industry (Griffin and others 1991). This organism first became recognized as a foodborne pathogen in 1982 after an outbreak tied it to the consumption of undercooked hamburgers, which ultimately sickened 47 individuals (Riley and others 1983). Since its discovery, E. coli O157:H7 has been linked to outbreaks associated with ground beef (Bell and others 1994, Riley and others 1983), produce (FDA 2006; FDA 2007; Jay and others 2007), fruit juices (CDC 1996; Cody and others 1999), unpasteurized milk (Griffin and Tauxe 1991; Keene and other 1997), and contaminated drinking water (Swerdlow and others 1992). E. coli is characterized by a specific combination of the O (lipopolysaccharide), H (flagellar) and sometimes K (capsular) antigens which define the various serotypes (Kaper and others 2004). Strains of pathogenic E. coli use multistep schemes of pathogenesis starting with colonization of the mucosal site, evading the hosts immune system, multiplication and damage to the host (Kaper and others 2004). E. coli O157:H7 has an extremely low oral infectious dose of less than 100 cells (Meng and others 2007). Enterohemorrhagic strains including E. coli O157:H7, destroy the normal microvillar structure of the colon by expressing the virulence factor 6 protein, intimin, which allows intimate attachment of the bacteria inducing an attaching and effacing lesion. Shiga toxin (Stx) is the major virulence factor of EHEC and also commonly known as verocytotoxin (VT). Stx is a family of structurally similar cytotoxins, with related biological activity, including two main groups, Stx1 and Stx2, which share about 40% amino acid similarity. Stx is composed of five identical B subunits that bind the holotoxin to the glycolipid globotriaosylceramide (Gb3) on the host cell surface and an A subunit, which cleaves ribosomal RNA to disrupt protein synthesis, ultimately killing the epithelial and endothelial cells. Stx causes local damage to the colon resulting in bloody diarrhea, hemorrhagic colitis and intestinal perforation. HUS results when Stx, produced in the colon, travels through the bloodstream to the renal endothelial cells in the kidney, causing inflammation by obstructing the microvasculature through both direct toxicity and inducing local cytokine and chemokine production (Caprioli and others 2005; Kaper and other 2004; Paton and Paton 1998; Tarr and others 2005). 1.3 Leafy green-associated recalls and outbreaks As the United States population attempts to move toward healthier eating habits, fresh produce-associated outbreaks are becoming more frequent. Between 1990 and 2006, consumption of fresh fruits and vegetables led to 768 outbreaks that included 35,060 cases of illness (CSPI 2008b), and these numbers are expected to increase. Furthermore, from 1998 to 2006, the average produce outbreak was responsible for 40.1 cases of illness, which is far greater than that seen for beef (23.4), poultry (24.5), and seafood (8.9) (CSPI 2008a). During the 15-year period from 1990 to 2005, E. coli O157:H7 and Salmonella were respectively responsible for 5 and 23% of these outbreaks (CSPI 2008b). E. coli and Salmonella associated outbreaks, 7 although lower in number, draw attention due to the severity of illness and the higher mortality. Produce-linked outbreaks associated with E. coli O157:H7 have increased significantly in 2005 and 2006 compared to 1998 - 2004, as can be seen in Figure 1.1 (CSPI 2008a). Figure 1.1: Produce outbreaks due to E. coli O157:H7 between 1998 and 2006 (CSPI 2008a) The United Nations and World Health Organization’s Food and Agriculture Organization report (FAO/WHO 2008), ranked produce commodities to determine which have the greatest microbiological hazard based on the following criteria: 1) Frequency and severity of disease, 2) Size and scope of production, 3) Diversity and complexity of the production chain and industry, 4) Potential for amplification of foodborne pathogens through the food chain, 5) Potential for control, and 6) Extent of international trade and economic impact. Leafy greens were given the highest ranking because they are frequently exported in large quantities, have been linked to multiple outbreaks resulting in high numbers of illnesses, and because production is quite complex. This assessment is in agreement with recent information from the Pathogen-Produce Pair Attribution Risk Ranking Tool (P3ARRT), a semi-quantitative risk ranking tool for determining which pathogen-commodity causes the greatest public health impact. P3ARRT 8 showed that enterohemorrhagic E. coli in leafy greens was consistently ranked first (Anderson and others 2011). Based on the data in Figure 1.2, collected by the Center for Science in the Public Interest for specific types of fresh produce, lettuce was most frequently associated with outbreaks of illness, followed by potatoes, tomatoes, melons, sprouts, berries, mushrooms and peppers (CSPI 2008a). Among the broader categories, salads, which would include leafy greens, were responsible for over one-third of all produce-related outbreaks. All of these fresh fruits and vegetables were susceptible to microbial contamination from farm-to-fork. The sources for bacterial pathogens included irrigation water, domestic livestock, wild animals, field workers, processing equipment, consumer cutting boards and home refrigerators (Beuchat 2006). Figure 1.2: Produce-linked outbreak vehicles 1998 and 2006 (CSPI 2008a) Although outbreaks associated with leafy greens are most frequently linked to contaminated lettuce, the term “leafy greens” comprises a wide range of salad and other greens 9 including iceberg lettuce, romaine lettuce, green leaf lettuce, red leaf lettuce, butter lettuce, baby leaf lettuce (i.e. immature lettuce or leafy greens), escarole, endive, spring mix, spinach, cabbage, kale, arugula and chard. Through 2009, 363 outbreaks and 13,568 reported cases of illness in the United States have been associated with the consumption of leafy greens (CSPI 2009b) with many of these outbreaks being traced to E. coli O157:H7-contaminated lettuce and spinach grown in the Salinas Valley of California. The number of outbreaks has increased dramatically since 2004 (CSPI 2008a). Table 1.1, modified from Mandrell (2009) and updated to include recent outbreaks, includes a list of select prominent outbreaks associated with leafy green contamination. In September 2006, contaminated pre-bagged baby spinach triggered an E. coli O157:H7 outbreak in 26 states and Canada resulting in five deaths and 205 illnesses (CDC 2006). This outbreak strain was eventually traced to numerous free-roaming feral swine that were spotted on a cattle ranch roughly a mile away from the implicated spinach field (cattle are the primary reservoir of E. coli O157) (Jay and others 2007). Droppings from these pigs which had access to the spinach fields could thereby contaminate the crop. Two genetically based typing methods multilocus variable number tandem repeat analysis (MLVA) and pulsed-field gel electrophoresis (PFGE), confirmed that the clinical and pig isolates of E. coli O157:H7 were identical. This observation was further supported by the discovery of pig tracks, rooting activity and pig feces in the spinach field and adjacent vineyards (Jay and others 2007). During November and December of 2006, two additional E. coli O157:H7 outbreaks were traced to California-grown iceberg lettuce that was shredded and then purchased by two fast food Mexican restaurant chains in the Midwest and Northeast (FDA 2006 and 2007). These two outbreaks sickened a combined total of over 150 individuals (CSPI 2008b). In one of the 10 outbreaks, state health officials in Minnesota, Iowa, and Wisconsin ultimately matched the strain of E. coli O157:H7 to two environmental samples that were collected from dairy farms near the fields where the lettuce was grown (FDA 2007). The source of the other outbreak strain, however, could not be confirmed (FDA 2006). 11 Table 1.1: Select leafy green associated outbreak from 1995 through 2011 Date Pathogen Location a Illnesses Source Reported Source Region Jul. 95 E. coli O157:H7 MT 74 Lettuce, Romaine MT, WA Sep. 95 E. coli O157:H7 ME 30 Lettuce, Iceberg Unknown Sep. 95 E. coli O157:H7 ID 20 Lettuce, Romaine Unknown Oct. 95 E. coli O157:H7 OH 11 Lettuce Unknown May 96 E. coli O157:H7 IL, CT 61 Lettuce, Mesclun mix CA Jun. 96 E. coli O157:H7 NY 7 Lettuce, Mesclun Unknown May 98 E. coli O157:H7 CA 2 Lettuce, salad Unknown Jul./Aug. 98 Shigella sonnei CA, MA, FL, Canada > 400 Parsley Mexico Sep. 98 E. coli O157:H7 MD 4 Lettuce Unknown Feb. 99 E. coli O157:H7 NE 65 Lettuce, salad Unknown Sep. 99 E. coli O157:H7 CA 8 Lettuce, Romaine CA Sep. 99 E. coli O157:H7 WA 6 Lettuce, Romaine CA Oct. 99 E. coli O157:H7 OH, IN 47 Lettuce, salad Unknown 12 Table 1.1 (cont’d) Oct. 99 E. coli O157:H8 OR 3 Lettuce, Romaine hearts CA Oct. 99 E. coli O157:H7 PA 41 Lettuce, Romaine CA Jul. 02 E. coli O157:H7 WA 29 Lettuce, Romaine CA Nov. 02 E. coli O157:H7 IL, WI, MN, SD, UT 24 Lettuce CA Sep. 03 E. coli O157:H7 CA 57 Lettuce, Iceberg/Romaine CA Sep. 03 E. coli O157:H7 ND 5 Lettuce, mixed with Romaine Unknown Oct. 03 E. coli O157:H7 CA 16 Spinach CA Nov. 04 E. coli O157:H7 NJ 6 Lettuce CA Oct.-Dec. 04 Salmonella Thompson Multi, Europe 21 Rucola (arugula) Italy Aug./Sep. 05 E. coli O157:H7 Sweden 135 Lettuce, Iceberg Sweden Aug./Sep. 06 E. coli O157:H7 Multi, U.S. (26 states) > 200 Baby Spinach, bagged CA Aug./Sep. 06 Yersinia pseudotuberculosis Finland > 400 Lettuce, Iceberg Finland Nov. 06 E. coli O157:H7 NJ, NY, PA, DE 71 Lettuce, Iceberg CA MN, IA, WI 81 Lettuce, Iceberg CA Nov./Dec. 06 E. coli O157:H7 13 Table 1.1 (cont’d) May 08 E. coli O157:H7 WA 10 Lettuce, Romaine CA Jan. 10 E. coli Denmark 260 Lettuce France May 10 E. coli O145 MI, NY, OH, PA, TN 26 Lettuce, Romaine AZ May 10 Salmonella Hvittingfoss IL 90 Lettuce Unknown Jul.-Oct. 10 Salmonella enterica U.K. 136 Salad Unknown May 11 Salmonella Typhimirum IL 15 Salad Unknown Oct./Nov. 11 E. coli O157:H7 Multi, U.S. (10 states) 60 Romaine lettuce Unknown a Information for the January 2010 E. coli outbreak is based on Ethelberg and others (2010), the May 2010 E. coli O145 outbreak is based on CDC (2010), the May 2010 Salmonella Hvittingfoss is based on IDPH (2010), the July – October 2010 Salmonella enterica outbreak is based on Gobin and others (2011), the May 2011 Salmonella Typhimirum outbreak is based on Sun-Times Media LLC (2011), and the October 2011 E. coli O157:H7 outbreak is based on CDC (2011), while all other outbreaks were summarized by Mandrell (2009). 14 1.4 The leafy green industry California is the main producer of spinach in the United States, while smaller-scale production takes place in Texas, Oklahoma, New Jersey, Colorado, Maryland, New York and Ohio (Koike and others 2011). Spinach prices vary throughout the season from between $238/ton to $351/ton (Koike and others 2011). California and Arizona are the main producers of Iceberg lettuce, with California’s peak production season in May and June, while Arizona’s is in December through February (Turini and others 2011). Prior to the series of leafy green outbreaks in the fall of 2006, the pre-washed salad market brought in $2.6 billion annually, with the spinach industry carving out $286 million (Todd and others 2007). Earthbound Farms, under its parent company Natural Selection Farms, had grown from selling spinach at a roadside stand in 1986 into a $360 million industry (Todd and others 2007). The leafy green industry as a whole was hit particularly hard after the outbreaks, with sales plummeting due to the media frenzy that swept the nation during the outbreaks and slow rate of the outbreak investigation. By mid-October of 2006, Natural Selection’s sales of conventional salads were down by 70% and down by 10% for products sold under the Earthbound Farms name (Todd and others 2007). The company was forced to lay off 164 employees and it has been estimated that Earthbound Farms and Dole may have to pay up to $110 million to settle cases with the victims of the outbreaks (Todd and others 2007). Changes to regulations for the produce industry are in the works. With the FDA Food Safety Modernization Act (FSMA) signed into law in 2011, the FDA is required to issue minimum standards for produce safety. It is likely that fresh-cut produce processors will be required to validate that their washing process, i.e. their preventative control, is effective at preventing crosscontamination of pathogenic microorganisms (FDA 2011). 15 1.5 Post-harvest processing sources and spread of contamination The increasing frequency of reported foodborne outbreaks associated with fresh fruits and vegetables is of major concern in the United States. These outbreaks are the driving force for changes needed by the produce industry in the way products are grown, harvested, and processed. Contamination can occur at any point between the farm and the fork. Produce can become contaminated on the farm through fecal contamination from wild and domestic animals, water runoff from livestock operations, green or improperly composted manure, air (dust), insects, contaminated irrigation water, or poor handling practices by the workers during the harvesting process (Beuchat and Ryu 1997). Even when contamination occurs at the growing or harvesting stages, conditions during post-harvest processing can intensify and spread what would normally be just a small, contained contamination event, resulting in a widespread outbreak (Beuchat and Ryu 1997 and Burnett and others 2001). After harvesting, microbial contamination can come from many different sources such as the water used for cooling and washing, equipment surfaces and workers. The equipment used in leafy green processing facilities varies, but most processors use the same general types of equipment. Figure 1.3 depicts some typical equipment pieces used during leafy green postharvest processing including a shredder, conveyer belt, flume tank, shaker table, and dewatering centrifugal dryer. Microbial contaminants can spread to multiple batches of product during processing (e.g., washing, peeling, shredding, slicing, drying and sorting) which can lead to a potential outbreak of illness if the contaminant is a human foodborne pathogen. Table 1.2 shows sources of post-harvest processing contamination for fresh produce. 16 Figure 1.3: Leafy green processing equipment (clockwise from top left): lettuce shredder, conveyer belt, flume tank, shaker table, plastic collection basket and dewatering centrifuge 17 Table 1.2: Sources of processing contamination Stage Contamination Sources Post-harvest Processing Cooling: Water Ice Vacuum Cooling Transport: Human Handling Shipping Vehicle Contamination Storage Containers Fresh cut and Value-added Debris Removal: Human Handling Processing Equipment: Processing - Sifters, Rollers Conveyer: Conveyer belt Washing: Water Value-added Processing: Processing Equipment: - Drying: Shredder, Core Removal, etc. Processing Equipment: - Packaging: Shaker table, Centrifugal dryer Human handling Packaging materials Packing equipment: - 18 Scales, etc. Immediately after harvest, many types of produce, including leafy greens must be cooled to remove field heat and preserve product quality (Kays 1997). Rapidly cooling fresh produce to refrigeration temperatures will minimize the growth of both spoilage and pathogenic bacteria. Commonly used commercial precooling methods for leafy greens include vacuum cooling, and hydro cooling or ice cooling (Boyhan 2004). Both spoilage and pathogenic bacteria can be transmitted via air, condensate and water droplets to the produce during the precooling phase. Doering and others found that the amount of time that passes before cooling iceberg lettuce and baby spinach was significant in controlling the E. coli population. They showed that the lower field temperature (25 °C) and the shortest time held before cooling (0 h) adversely impacted the E. coli counts when compared to the higher field temperature (32 °C) and longer time (10 h) before forced air cooling (2009). Vacuum cooling, which is based on evaporative cooling of a product under low pressure, can be used to rapidly decrease the field temperature of many fruits and vegetables and is now the method of choice for leafy greens. However, this cooling method may lead to the internalization of both spoilage and pathogenic bacteria. A recent study found that vacuum cooling increased the infiltration of E. coli O157:H7 (considering both cell numbers and the depth of penetration) into lettuce tissue by more than 90% when compared to the non-vacuum cooled product with these internalized cells protected during quadruple washing of the product in a produce sanitizer (Li and others 2008). However, Doering and others (2009) failed to see any change in E. coli O157:H7 populations before and after vacuum cooling. Water-based cooling methods carry a much greater risk of contamination, with the microbial populations based on the initial numbers of microorganisms in the water and the numbers of microorganisms transferred between the product and the water. Potable water, 19 defined as water that has a coliform count of less than 1 CFU/ml, an aerobic plate count of less than 100 CFU/ml and is free of Cryptosporidium, Giardia lamblia, Legionella and enteric viruses (EPA 2009), should be used in all cases, including when making ice. In a series of studies looking at the transfer of E. coli O157:H7 first from the contaminated lettuce leaves to uncontaminated lettuce leaves via the melting ice and then from melting ice to Romaine lettuce, it was shown that as the ice melted, water transferred E. coli O157:H7 from inoculated sites on the leaf to the uninoculated sites on all of the three stacked romaine lettuce heads within one container when the top head contained the inoculated site. Most sites on the Romaine lettuce became contaminated with E. coli O157:H7 from the melting ice with populations averaging 3.8 2 and 5.5 CFU/cm for experiments conducted at 4 and 20 °C, respectively. In the same series of experiments, when the melting ice was contaminated with E. coli O157:H7 at 7 log CFU/ml, E. 2 coli O157:H7 populations of 3.5 to 3.8 log CFU/cm were seen on the top leaf of the first of the three heads, with lowest counts on the bottom layer of romaine heads with no difference in numbers of E. coli O157:H7 recovered from each sampling site at 4 and 20 °C (Kim and Harrison 2008). Various chemical sanitizers can be added as appropriate to reduce microbial populations in the water (Beuchat and others 2004, Burnett and others 2004, Keskinen and others 2009, Rangaranjan and others 2000, Sapers 2001, Weissinger and others 2000, Zhang and others 2009). However, the sanitizer concentration must always be maintained at an effective level and be carefully monitored to ensure that the organic load does not decrease the efficacy of the sanitizer. Internalization of both spoilage and pathogenic bacteria during water cooling of fresh produce continues to be a major concern, particularly when the temperature of the produce exceeds that of the water by 10 °F or more (Rangaranjan and others 2000). 20 Human foodborne pathogens, including E. coli O157:H7, Salmonella, Listeria and Cryptosporidium that may inadvertently contaminate fresh fruits and vegetables in the field or at harvest, can be readily transferred to much larger quantities during subsequent product handing and further processing, raising even greater food safety concerns. It has been demonstrated that foodborne pathogens can also contaminate the product during processing. In one outbreak of salmonellosis traced to shredded lettuce, Stafford and others (2002) recovered Salmonella Bovismorbificans phage type 32 from the cutting wheel of a mechanical shredder during an environmental audit with insufficient cleaning and sanitizing of the shredder cited as a key factor in this outbreak. In a study examining the microbial changes of “Lollo Rosso” lettuce during processing and shelf-life, an increase of 1 log CFU/g was observed after shredding (Allende and others 2004), indicating that the shredder may be an important in-plant vehicle for spreading contamination to leafy greens during processing (Garg and others 1990). In a survey of Japanese ready-to-eat fresh vegetable processors, populations of 2 mesophilic aerobic bacteria exceeded 5.0 log CFU/cm or ml in many samples collected from product contact surfaces of the washing, slicing, dehydrating, and blending equipment, surfaces of the slicer/shredder blades, and the processing room floor, all of which were obtained at the end of processing (Kaneko and others 1999). In another survey conducted in Spain, populations of mesophilic aerobic bacteria, psychrotrophic bacteria, yeast/ mold, and lactic acid bacteria on unprocessed (whole) vegetables, mainly leafy greens, were 1.1, 1.2, 0.7, and 2.6 logs lower, respectively, than their fresh cut counterparts (Abadias and others 2008). Leafy green processing leads to plant tissue damage, which Seo and Frank have demonstrated allows E. coli to preferentially attach to cut plant edges, including the trichomes, stomata, and cuticle cracks. They showed E. coli 20 to 100 μm below the surface of the leaf, 21 inside the stomata and cut surfaces (1999). Using confocal scanning laser microscopy, this same group showed that E. coli can penetrate the leaf at the cut tissue surfaces that lack a waxy coating (Takeuchi and others 2000). Mechanically damaged Romaine lettuce leaves that were cut or shredded showed a 4.54 and 11.05-fold increase in E. coli O157:H7 compared to intact leaves, in which populations only increased by 1.95-fold after a 4 h hold time (Brandl 2008). Romaine lettuce lysates obtained after mechanical damage, promote the growth of E. coli O157:H7 with a growth rate equivalent to that of minimal medium with glucose. A microarray-based whole genome transcriptional profiling analysis showed that these E. coli expressed genes indicating they were adjusting to the alternate nutrient source and to a variety of stresses, including the presence of reactive oxygen species, antimicrobial compounds, toxic compounds and osmotic stress. Stress adaptation may also result in the enhanced resistance to chemical sanitizers (Kyle and others 2010). Fruit and vegetable processing typically involves the use of various types of conveyor belt systems manufactured from different belting materials including high density polyethylene, polypropylene and acetyl. Two main conveyor belt designs – interlocking (a series of interlocking pieces that contain many microbial harborage sites) and smooth (a continuous belt with a single seam) can be found in the produce industry with the belting material and design tailored to the specific product and application. Regardless of the belt type or material, all conveyor belts are prone to microbial build-up and the subsequent transfer of microorganisms to incoming product over time. The newer smooth continuous belts, which can be more easily cleaned and sanitized, are now generally preferred over the older interlocking belts that must be dissembled and then manually cleaned and sanitized. 22 Some produce destined for the ready-to-eat market, including baby spinach and fresh -cut lettuce, can be commercially washed up to three times in various types of flume tanks to remove soil and decrease the levels of microorganisms. In general, washing fresh produce in water alone will only decrease microbial populations by 90 to 99% (Sapers 2001). Therefore, flume and wash water are ideal vehicles for the spread of microorganisms throughout an entire batch of product. This is even truer in Europe where processing leafy greens without the addition of a sanitizer is a common practice (Holvoet and others 2012). Unfortunately, when added to flume and wash water, the commonly used chlorine-based sanitizers as well as most others including peroxyacetic acid, chlorine dioxide and ozone have been shown to be only marginally effective (Gil and others 2009; Sapers 2006). Most industrially used sanitizers are best suited for reducing microbial populations in the wash water rather than those on the product being washed. During continued processing, the level of soil and debris, referred to as the organic load, in the recirculating wash water will increase as additional product is processed. Since chlorine will react with any organic material, most chlorinated sanitizers exhibit decreased antimicrobial activity as the organic load in the water increases, with the potential survival of bacterial pathogens and other microorganisms in the wash water becoming a concern. Other alternatives to chlorine include peroxyacetic acid, chlorine dioxide and ozone (Sapers 2001). When lettuce was sprayed with an aqueous solution containing 200 or 2,000 ppm chlorine, populations of Salmonella, E. coli O157:H7, Listeria monocytogenes, yeasts and molds, 2 and mesophilic aerobic bacteria on the products decreased only 0.35 and 2.30 log CFU/cm (Beuchat and others 1998). After a 10 min exposure to an aqueous solution of 200 ppm chlorine at 4 and 22°C, L. monocytogenes populations decreased only 1.3 and 1.7 log CFU/g on lettuce, respectively (Zhang and Farber 1996). Doering and others (2009) were only able to achieve a <1 23 to 3 log CFU/g reduction of E. coli O157:H7 on iceberg lettuce and baby spinach after washing in a 50 ppm free chlorine solution. These findings indicate that the bacteria which are not being killed on the leafy greens could potentially cross-contaminated uncontaminated product that follows it through the processing line. After fluming and washing, excess water must be removed before packing through the use of shaker tables, blowers, or centrifugal dryers or by other means to maintain product quality and an acceptable shelf-life. All drying methods are product dependent with the end goal being the preservation of both product quality and shelf-life. Whereas iceberg and romaine lettuce can withstand the forces associated with centrifugal drying, other products such as baby spinach and parsley would be severely damaged. One study conducted at the U.S. Department of Agriculture (USDA) looked at plant lesions and their ability to promote growth of E. coli O157:H7 on mechanically bruised/damaged post-harvest Romaine lettuce leaves. Bruised plant tissue, leaves cut into larger pieces and leaves shredded into multiple smaller pieces showed a 3.99, 4.54 and 11.05-fold increase in E. coli O157:H7, respectively, compared to the intact leaves which only increased by 1.95-fold after leaves had been held for 4 h at 28°C (Brandl 2008). The increased growth of E. coli on the damaged plant tissues may be related to E. coli’s enhanced ability to attach to cut surfaces (Takeuchi and others 2000) as well as the availability of nutrients leaching out of the broken plant cells. The potential for cross-contamination exists during all stages of transport from the field to the consumer’s plate. Wooden or plastic totes, crates, and other types of storage containers inevitably come in contact with soil or other debris in the field at the time of harvest. They are an important source of microbial contamination when transported to processing facilities. Many produce growers are now using plastic instead of wooden containers for transport and storage 24 due to the increased ease of cleaning and sanitizing. Ailes and others (2008) found that when compared to produce samples taken directly from the field, those items collected from the packing bins had over a six-fold increase in likelihood of E. coli contamination and only a fourfold increase in likelihood for samples that originated from the box or conveyor belt. 1.6 E. coli attachment, persistence and biofilm formation on surfaces The mechanisms that E. coli O157:H7 uses to attach to leafy greens surfaces are an increasing area of interest. One group in the United Kingdom (Shaw and others 2008) has recently demonstrated that EHEC strains of E. coli O157:H7 and non-O157:H7 both attach to the leaf surfaces of arugula, spinach and lettuce through the EspA filamentous type III secretion system (fT3SS) that has been shown to also play an important role in the organisms colonization of bovine and human hosts. This finding indicates that EHEC uses plant leaves solely as a transmission vector rather than acting like a plant pathogen. Research conducted at the University of Arizona (Xicohtencatl-Cortes and others 2009) showed that flagella also play a role in attachment because fliC gene mutants are significantly impaired in their ability to adhere to the surfaces of spinach and lettuce compared to the parental strains. Crevices and surface wounds on the leafy greens protect the organisms from the environment and any decontamination strategies used during processing. Adhesion was shown to be temperature and time dependent. They also explained that while intimin adhesion is vital for attachment to the host organism, it is not required for colonization of leafy greens. Based on this, colonization of leafy greens by EHEC strains of E. coli O157:H7 is likely a means to survive the processing environment and to allow transmission to the human host (Xicohtencatl-Cortes and others 2009). 25 Biofilms are defined as an aggregate of microbes that are embedded in an extracellular matrix, which encapsulates the cells and provides a physical barrier against environmental stresses such as chemicals (sanitizers), temperature changes, desiccation or ultraviolet rays (Annous and others 2009, Ryu and others 2005, Tarver 2009). A major concern in the food industry is the formation of these biofilms on processing equipment. E. coli are capable of producing biofilms on inert surfaces such as: glass, stainless steel, high density polyethylene, polyamide-6, polyvinyl chloride, teflon coupons, glass wool, polystyrene microtiter plates, and glass coverslips (Annous and others 2009). A large portion of a food processing plant uses equipment composed of stainless steel, because of the durability of the material. Specific strains may be more difficult to remove/kill on stainless steel food contact surfaces and more aggressive sanitizing and cleaning procedures are necessary (Ryu and Beuchat 2005, Ryu and others 2004). In a recent study it was shown that when E. coli O157:H7 is allowed to become incorporated into a biofilm of other good biofilm-forming companion strains, they show greater resistance to peroxide based sanitizers compared to planktonic cells (Uhlich and others 2010). If biofilms are not removed or prevented portions of these bacterial aggregates persist as transient residents on equipment surfaces throughout the processing line and slough off over time to contaminate large quantities of previously uncontaminated product during processing Once present on the food contact surfaces of processing equipment, bacteria can transfer to the produce during processing. In a field-coring study conducted at the University of Georgia, when field coring devices were inoculated with soil containing E. coli O157:H7 populations of 2.72 and 1.67 log CFU/g the pathogen transferred to 10 and 5 consecutively processed heads of lettuce, respectively (Taormina and others 2009). Moore and others (2003) demonstrated the ability of Campylobacter jejuni and Salmonella enterica serovar Typhimurium to transfer from 26 stainless steel surfaces to both wet and dry lettuce via direct contact with 66% percent of the Salmonella population transferring from a dry stainless steel surface to a dry product after 60 minutes of direct contact. In contrast, 23 to 31% of the Salmonella population transferred from dry stainless steel to wet lettuce after 120 minutes of direct contact. Transfer of C. jejuni ranged from 16 to 38% for dry lettuce and 15 to 27% for wet lettuce after 80 minutes of contact. These results show the ability for C. jejuni and Salmonella Typhimurium to persist on surfaces and maintain the ability to transfer to food surfaces hours after they become contaminated. Bacterial attachment and biofilm formation on the surfaces of leafy greens greatly hinder the removal of microbial contaminants during conventional washing (Carmichael and others 1999, Sapers 2006). When Lang and others (2004) assessed the impact of post-inoculation drying times on E. coli O157:H7 survival on iceberg lettuce and parsley, higher (or equal to) populations were recovered from the leafy greens after 2 h of drying at 22°C rather than 2 h drying at 22°C followed by 22 h at 4°C. Using scanning electron microscopy (SEM), another study showed biofilm formation on the stem ends of freshly harvested lettuce leaves after submersion in an E. coli cocktail for 1 week at 4°C (Figure 1.4) (Annous and others 2009). SEM is frequently used to observe the survival, colonization and biofilm formation of microbes on the surface of produce (Annous and others 2009, Carmichael and others 1999, Keskinen and others 2009 and Morris and others 1997). These results suggest that transfer of E. coli O157:H7 to processing equipment should be examined by SEM and microbiological analysis both after a short and long post-inoculation holding time. This would determine if transfer also follows the same trend as in a study by Lang and others (2004) and decreases after longer holding times. 27 Figure 1.4: An SEM (Scanning electron microscopy) showing attachment and biofilm formation of E. coli cells on a lettuce leaf (Annous and others 2009) 1.7 Inoculation of leafy greens and recovery of bacteria To study the degree to which transfer occurs between leafy greens and the processing environment, product needs to be coated with the inoculum consistently to ensure accurate results that could be replicated and used for predictive modeling. Lang and others (2004) studied the effects of different inoculation methods on survival and recovery of foodborne pathogens on iceberg lettuce and parsley. Their results showed that when using either an E. coli O157:H7 or Salmonella cocktail, more bacteria were recovered from the dip-inoculated produce than from spot- or spray-inoculated produce, due to a greater numbers of bacterial cells attaching initially. When inoculating large quantities of product for a pilot scale study, safety is essential to prevent laboratory-acquired infections and avoid escape to the laboratory or research environment. One common work around to avoid using large quantities of a dangerous pathogen 28 is to use an avirulent strain as a surrogate for the virulent outbreak strain. This technique has become especially common in studies taking place in a processing environment or field. Erickson and others (2010) have demonstrated the application of four avirulent E. coli O157:H7 strains (ATCC 43888, CV2B7, 6980-2, and 6982-2) while studying surface and internalization on field grown spinach and lettuce treated with spray-contaminated irrigation water. Shepherd and others (2011) have also applied avirulent strains in a study assessing the use of physical coverings to control E. coli at the surface of a compost heap. In another study, a comparison was done between avirulent E. coli O157:H7 and generic E. coli isolated from irrigation water, soil and lettuce (Tomás-Callejas and others 2011). They showed that populations of both the generic and avirulent strains rapidly declined on the spray inoculated fresh-cut baby leafy greens (Mizuna, Red Chard and Tatsoi), but that minimal processing was not able to disinfect the products (Tomás-Callejas and others 2011). In addition to the avirulent feature, these strains have been transformed with a pGFPuv plasmid containing a green fluorescent protein (GFP) gene and an ampicillin-resistance gene to allow for differentiation from native E. coli strains that would be found in the filed environment. To ensure that these labeled strains maintain the plasmid during the study, the stability is tested through a series of propagations without selective pressure and the rate of plasmid-loss is determined. Most GFP labeled E. coli O157:H7 have proved to be very stable according to Ma and others (2011). After two consecutive subcultures (~2 generations) in a nonselective media, the rate of plasmid loss for labeled E. coli O157:H7, Salmonella and Listeria strains tested in this study were 0 – 30%, 15.8 – 99.9% and 8.1 – 93.4%, respectively (Ma and others 2011). Once the inoculated leafy greens are processed, a viable recovery method for pathogens must be implemented. Sampling/swabbing the surfaces of processing equipment pieces creates 29 unique hurdles. Vorst and others (2004) determined which of four sampling devices, a sterile environmental sponge, a sterile cotton-tipped swab, a sterile calcium alginate fiber-tipped swab, and a one-ply composite tissue, would recover the most Listeria monocytogenes from food-grade stainless steel surfaces. They showed that using one-ply composite tissue, rehydrated in 0.1% peptone before swabbing and vortexed in 0.1% peptone for 1 min. after swabbing, showed statistically more recovery that the other methods. As the authors point out, this method is both inexpensive and easy to use and will be very useful for sampling the bacteria that have transferred to the equipment surfaces from the inoculated leafy greens during processing. 1.8 Bacterial transfer in other food environments Bacterial transfer has also been demonstrated in various other food preparation scenarios including the grinding of meat (Farrell and others 1998; Flores and Stewart 2004, Flores and Tamplin 2002), slicing deli meats (Keskinen and others 2008a; Keskinen and others 2008b; Lin and others 2006; Sheen and Hwang 2008; Vorst and others 2006), between cutting boards (Chai and others 2008; Chen and others 2001; Fravalo and others 2009; Jiménez and others 2009; Ravishankar and others 2010), and through the use of cloths, hands, gloves, utensils and various food products (Chen and others 2001; Gill and Jones 2002; Montville and others 2001; Scott and Bloomfield 1990). Bacteria transfer in a small ground beef table-top grinder demonstrated a linear relationship between the amount of inoculum and the beef trim added to the grinder and the amount of contaminated ground fractions collected. Contaminated ground beef was added to the grinder after an estimated 20% of the beef had entered the grinder, but was collected from the grinder after just 8% of the beef had been collected, so before the point at which the inoculated beef was expected to exit the grinder. This indicates that the bacterial transfer takes 30 place in part from the conveying, mixing, extrusion and cutting stages of processing and that product does not move though the grinder linearly (Flores and Tamplin 2002). Building on this work, Flores and Stewart conducted an experiment to track E. coli O157:H7 during mid-sized commercial grinding of inoculated followed by uninoculated beef. They were able to detect E. coli in 12.7 to 86.2 % of the ~6600 g beef trim batch using initial inoculums of 2 to 6 log CFU/g. As the inoculum level increased, the level of contamination in the ground beef increased. Using a Chi-squared algorithm based distribution model, these researchers also showed that the contamination level was a function of the ground beef batch fractions collected (Flores and Stewart 2004). One notable study measured the distribution of L. monocytogenes during slicing of roast turkey breast, Genoa hard salami, and bologna with kitchen knives. These researchers demonstrated that the majority for bacterial transfer takes place within the first 5 to 15 slices depending on the stainless steel type and product being cut (Vorst and others 2006). In a similar fashion, this research group also demonstrated that L. monocytogenes counts dropped between 3 and 5 log CFU/g after slicing 16 slices roast turkey breast, even at very low inoculation levels (Keskinen and other 2008b). In agreement with these studies, L. monocytogenes was shown to transfer to a larger number of meat samples when the inoculation level was higher (Lin and others 2006). Biofilm-forming ability and inoculation hold time have also been shown to impact transfer of L. monocytogenes to deli meats, with significantly more transfer taking place when strong biofilm-formers were used as the inoculum verses the weaker strains and greater transfer taking place 6 h post-inoculation rather than after 12 h (Keskinen and other 2008a). A non-linear model for these L. monocytogenes cross contamination scenarios has been developed by Sheen 31 and Hwang to predict the number of ham slices that may become contaminated by a L. monocytogenes-contaminated slicer blade (2008). Bacterial transfer readily occurs in commercial food processing environments, foodservice establishments and household kitchens. Campylobacter jejuni was shown to transfer from vegetables to wash water (30.1 – 38.2%), wash water to cucumbers (26.3 – 47.2%), vegetables to the cutting board (1.6 – 10.3%) and from the cutting board to cucumbers (22.6 – 73.3%), while simulating the preparation of a salad in a household kitchen (Chai and others 2008). The log percent transfer of Enterobacter aerogenes from chicken to hands (0.94%), a cutting board to lettuce (0.90%), the faucet to hands (0.36%), hands to lettuce (-0.12%), unwashed hands to washed hands (-0.20%), and from hands to the faucet (-0.80%) has also been recorded (Chen and others 2001). Temperature did not affect the transfer of Salmonella to a cutting board, but the amount of time the inoculated product remained in contact with the cutting board did impact transfer (Jimenez and others 2009). One study showed that 60.5% of the Campylobacter on naturally contaminated chicken legs transferred to a cutting board with this number increasing to 80.6% if the contaminated chicken legs remained on the cutting board for over 10 min (Fravalo and others 2009). Proper sanitation of kitchen utensils is needed to prevent cross contamination, Ravishankar and others (2010) studied different scenarios where Salmonella enterica could move between cutting boards, knives, chicken and lettuce. They showed that when knives and cutting boards are not washed after cutting chicken inoculated with 6 log CFU/g and used for slicing lettuce, ~ 2.97 log CFU/g of the Salmonella enterica is able to transfer. If the utensils are quickly rinsed prior to slicing, there is a ~2.36 log CFU/g transfer, while thoroughly washing the utensils with soapy, hot water results in less than 1 log CFU/g detectable on the lettuce. Gloves may be an effective means to reduce the bacterial transfer 32 between food and the hands of foodservice employees and in transfer from contaminated hands back to food as was demonstrated in a study looking at transfer between chicken, lettuce, bare and gloved hands. Montville and others (2001) showed that 0.01% Enterobacter aerogenes (a nonpathogenic surrogate with similar attachment characteristics to Salmonella) transfer still takes place from food to hands and from hands to food when the volunteers wore gloves. Significantly more transfer (10%) took place without the glove barrier. 1.9 Predictive modeling of bacterial transfer A mathematical risk model for use in a leafy green processing line was created by Pérez Rodríguez and others (2011) which looked at E. coli O157:H7 cross-contamination in three different contamination scenarios, to estimate the prevalence and percent of E. coli O157:H7 transferred during a cross contamination event in fresh-cut bagged leafy greens. In this simulation, the model’s process parameters were locked to ensure comparable data among the different simulations. These parameters consisted of 22 batches in a day’s product, with a batch size of 1000 kg and a bag size of 100 g. Initial concentrations of the contamination were 0.01, 1 and 100 CFU/g. Based on their simulation, when a low concentration of the contamination enters the processing line, product that is processed in the first hour will become contaminated with subsequent bags of product being sporadically contaminated. The prevalence rate will increase with the inoculum level. This simulation predicts that, when processing a batch of leafy greens inoculated at 1 and 100 CFU/g in sanitizer-free wash water, the prevalence would be ~ 3.05 and 13.39 %, respectively. Based on the simulation, chlorination of the water would reduce this percentage to 0.14 – 0.20 % and 3.28 – 4.00 %, for the 1 and 100 CFU/g inoculation levels, respectively. As shown in Figure 1.5, the predictive model indicates that after a contaminated 33 batch enters the processing line, the lower the inoculum, the more gradual the decline in the number of cross-contaminated bags. In the simulation using the lowest inoculum, crosscontamination only occurs to the first three batches before dropping off steeply. At the medium inoculation level, cross-contamination to the first two batches was very high (10,000 crosscontaminated bags/batch), with a slight decline to only 100 cross-contaminated bags/batch, followed by a gradual drop-off to only 10 cross-contaminated bags/batch in the batch. The highest inoculation level yielded cross-contamination of between 10,000 and 1000 crosscontaminated bags/batch for the entire production run. More research will need to be conducted to validate the predictive model by quantifying the extent of cross-contamination to subsequent product by tracking contaminated product though a fresh-cut leafy green processing line. In the event of an outbreak, these finding could be used as a guide to help estimate the amount of product which may have became cross-contaminated during processing and would need to be recalled. 34 Figure 1.5: Simulated number of cross-contaminated bags along the production (22 batches) when one contaminated batch enters the processing line (Pérez Rodríguez and others 2011), for these different initial contamination levels (S1, S2, and S3) 1.10 Overall goals and objectives The number of documented cases of foodborne illness associated with consumption of produce, such as leafy greens, continues to rise. Consumers expect safe products, and the food industry must make changes to their current processing procedures to be able to provide these products to their customers. It is the researcher’s duty to establish where adjustments can be made to help prevent future outbreaks. Without modifications to the present processing methods, leafy green outbreaks will continue to increase in both number and magnitude. The overall objective of this research is to quantify the extent of E. coli O157:H7 transfer between leafy greens, wash water and food contact surface areas during simulated commercial processing. 35 The following series of experiments were conducted to quantify the transfer of Escherichia coli O157:H7 between the leafy greens, the wash water and the food contact surfaces in a pilot-scale processing line that included a commercial shredder, conveyor, flume tank, shaker table and centrifugal dryer. Using three different types of leafy greens (iceberg, Romaine lettuce and baby spinach) dip-inoculated with a 4-strain avirulent, GFP-labeled cocktail 2 4 6 of E. coli O157:H7 at three different levels (10 , 10 and 10 CFU/g), transfer coefficients were determined for the numbers of E. coli O157:H7 transferred between the product, the water and the equipment surfaces during processing. The effects of post-inoculation holding time and shred size on the transfer coefficients were subsequently evaluated. Inoculated Radicchio - used as a surrogate for contaminated leafy greens, was also tracked through the pilot-scale processing line to visually assess the spread of contaminated product during processing. Finally, the transfer rate between inoculated and uninoculated leafy greens was assessed during shaker table dewatering. 36 CHAPTER 2: Quantitative Transfer of Escherichia coli O157:H7 to Equipment during Small-Scale Production of Fresh-Cut Leafy Greens 37 2.1 ABSTRACT Post-harvest contamination and subsequent spread of Escherichia coli O157:H7 can occur during shredding, conveying, fluming and dewatering of fresh-cut leafy greens. This study quantified E. coli O157:H7 transfer from leafy greens to equipment surfaces during simulated small-scale commercial processing. Three to five batches (22.7 kg) of baby spinach, iceberg lettuce and romaine lettuce were dip-inoculated with a 4-strain cocktail of avirulent, green fluorescent protein (GFP)-labeled, ampicillin-resistant E. coli O157:H7 to contain ~6, 4, and 2 log CFU/g, and then processed after 1 h of draining at ~23oC or 24 h of storage at 4°C. Lettuce was shredded using an Urschel TransSlicer at two different blade and belt speeds to obtain normal (5 x 5 cm) and more finely shredded (0.5 x 5 cm) lettuce. Thereafter, the lettuce was step-conveyed to a flume tank, washed and then dried using a shaker table and centrifugal dryer. Product (25 g) and water (40 ml) samples were collected at various points during processing. 2 After processing, product contact surfaces (100 cm ) on the shredder (n = 14), conveyer (n = 8), flume tank (n = 11), shaker table (n = 9) and centrifugal dryer (n = 8) were sampled using 1-ply composite tissues. Sample homogenates diluted in phosphate or neutralizing buffer were plated, with or without prior 0.45 μm membrane filtration, on trypticase soy agar containing 0.6% yeast extract supplemented with 100 ppm ampicillin to quantify GFP-labeled E. coli O157:H7 under UV light. During leafy green processing, ~90% of the E. coli O157:H7 inoculum transferred to the wash water. After processing, E. coli O157:H7 populations were highest on the conveyor and shredder (P < 0.05), followed by the centrifugal dryer, flume tank and shaker table, with ~29% of the remaining product inoculum lost during centrifugal drying. Overall, a significantly (P < 0.05) higher percentage of the inoculum was shed from 1 h (92%) compared to 24 h-held lettuce (58%) after centrifugal drying ,with shred size not impacting the rate of transfer. 38 2.2 INTRODUCTION Given the current emphasis on healthier dietary habits in the US, including increased consumption of salads, foodborne outbreaks associated with leafy greens unfortunately are becoming more frequent (HHS/USDA 2005). Consumption of leafy greens increased only 9% from 1996 to 2005 compared to the previous decade; however, the number of leafy greenassociated outbreaks increased 38.6% during this same period (Herman and others 2008). Consequently, consumer confidence in the safety of leafy greens has decreased. In 2009, the Center for Science in the Public Interest (CSPI 2009b) ranked leafy greens, including spinach, at the very top of “The 10 Riskiest Foods.” Changes in agricultural practices including large centralized production and harvesting practices are partly responsible for the increasing number of leafy greens outbreaks. Since 1990, 19 outbreaks were specifically associated with the consumption of leafy greens (CSPI 2008a), many of which were traced to produce grown on large-scale farms in the Salinas Valley of California. In September 2006, contaminated pre-bagged baby spinach triggered an E. coli O157:H7 outbreak in 26 states and Canada, resulting in five deaths and 205 illnesses (Jay and others 2007). A- few months later, two additional E. coli O157:H7 outbreaks were traced to California-grown iceberg lettuce that was shredded and then purchased by two fast-food Mexican restaurant chains in the Midwest and Northeast (FDA 2006, FDA 2007). These two outbreaks sickened a combined total of over 150 individuals (CSPI 2008b) and raised nationwide concerns regarding the safety of leafy greens along with the various routes for contamination. Microbial pathogens can contaminate fresh produce at any point from farm-to-fork, with sources of contamination in the field including feces from wild and domestic animals, water runoff from livestock operations, green or improperly composted manure, air (dust), insects, and 39 irrigation water. During and after harvest, pathogens can come in contact with fresh produce through field workers, harvesting equipment, hand coring/trimming, and water used for cooling or washing, as well as improperly cleaned crates, totes and storage bins and the vehicles used for transport (Beuchat 1996, Beuchat and Ryu 1997, Gorny 2006). While most pathogen contamination occurs in the field, conditions during post-harvest processing can intensify and spread what would normally be a small isolated contamination event to far larger quantities of product that can lead to widespread outbreaks (Beuchat and Ryu 1997, Burnett and Beuchat 2001). This was the case in a 2001 Australian outbreak of Salmonella Bovismorbificans that was later traced to residual product behind the cutting wheel rim of a lettuce shredder (Stafford and other 2002). Similarly, a Dutch processing plant was implicated in an E. coli O157 outbreak associated with shredded and pre-packaged lettuce, which sickened individuals in both the Netherlands and Finland (Friesema and others 2008), and a shigellosis outbreak in Texas caused restaurant patrons to become ill after consuming shredded lettuce, which may have become contaminated during processing and further handling (Davis and others 1988). Few strategies other than irradiation can ensure the safety of fresh-cut salad greens, with current washing and sanitizing practices reducing E. coli O157:H7, Salmonella and other pathogens only 1 to 2 log CFU/g, (Sapers 2006). Many factors associated with microbial attachment contribute to the ineffectiveness of these intervention strategies. Fett (2000) suggested that naturally occurring biofilms on sprouts may protect Salmonella and E. coli O157:H7 from inactivation by chemical sanitizers. During fluming and washing, E. coli also can penetrate into the cracks, crevices, cut/injured surfaces and intercellular spaces of lettuce (Seo and Frank 1999, Takeuchi and Frank 2000), making disinfection strategies or physical removal impossible. 40 Despite the continually increasing market for pre-bagged fresh-cut, ready-to-eat salad greens, the potential for spread of both pathogenic and spoilage microorganisms during commercial shredding, conveying, washing and dewatering of leafy greens is still poorly understood. However, previous studies have quantified the numbers of bacteria transferred during field coring of leafy greens (Taormina and others 2009), grinding of meat (Farrell and others 1998, Flores and Tamplin 2002, Flores and Stewart 2004) and slicing of deli meats (Keskinen and others 2008a, Keskinen and others 2008b, Lin and others 2006, Sheen and Hwang 2008, Vorst and others 2006), as well as between cutting boards (Chai and others 2008, Chen and others 2001, Fravaol and others 2009, Jimenez and others 2009, Ravishankar and others 2010), cloths, hands, gloves, utensils and various food products (Chen and others 2001, Gill and Jones 2002, Montville and others 2001, Scott and Bloomfield 1990). The objectives of this study were: 1) to quantify the transfer of an avirulent GFP-labeled 4-strain cocktail of E. coli O157:H7 from inoculated iceberg lettuce, Romaine lettuce and baby spinach to the water and different pieces of equipment used for commercial processing of freshcut leafy greens, and 2) to assess the impact of post-inoculation hold time and lettuce shred size on E. coli O157:H7 transfer. However, before beginning this work, the growth rates and attachment forming abilities of the four GFP-labeled, avirulent E. coli O157:H7 strains were compared to three virulent E. coli O157:H7 leafy green outbreak strains to ensure that the GFP genetically engineered strains behaved identically to the parent strains. 41 2.3 MATERIALS AND METHODS 2.3.1 Experimental design. Three to five batches (22.7 kg) of baby spinach, iceberg and 6 4 2 Romaine lettuce were dip-inoculated with E. coli O157:H7 to contain ~10 , 10 and 10 CFU/g, and processed by shredding, conveying, fluming, shaker table dewatering and/or centrifugal drying, during and/or after which various product, water and equipment surface samples were collected for E. coli O157:H7 quantification. The impact of post-inoculation hold time and shred size on E. coli O157:H7 transfer during processing was also determined using triplicate batches 4 of iceberg lettuce inoculated at 10 CFU/g. 2.3.2 Leafy greens. Pre-washed, pre-bagged baby spinach (Spinacia oleracea L.) from 5 or 10 lb (2.3 or 4.5 kg) bags, as well as individually wrapped heads of iceberg (Lactuca sativa L.) (24 heads/case) and Romaine (Lactuca sativa L. var. longifolia) (12 heads/case) lettuce, were obtained from a local wholesaler (Stan Setas Produce Co., Lansing, MI), stored at 4 °C and used within 5 days of delivery. These three products originated from California or Arizona, depending on the growing season. All heads of lettuce were hand-cored immediately before use. 2.3.3 Bacterial strains. Four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) and three virulent E. coli O157:H7 strains from three separate leafy green outbreaks in 2006 (K3995 - spinach, K4492 - lettuce, and K4831 - lettuce) were obtained from Dr. Michael Doyle at the Center for Food Safety, University of Georgia, Griffin, GA. Upon arrival, stock cultures of each strain were prepared in trypticase soy broth (Difco, Becton Dickinson, Sparks, MD) containing 0.6% yeast extract (Difco, Becton Dickinson) (TSBYE) and 10% (v/v) glycerol (Sigma Chemical Co., St. Louis, MO) and stored at -80°C until needed. Working cultures were prepared by streaking each stock culture on trypticase soy agar plates (Difco, Becton Dickinson) containing 0.6% yeast extract 42 (TSAYE). After 18 - 24 h of incubation at 37°C, a single colony was transferred to 9 ml of TSBYE and similarly incubated. TSAYE and TSBYE used to grow the four avirulent, GFPlabeled ampicillin-resistant E. coli O157:H7 strains were supplemented with 100 ppm ampicillin (ampicillin sodium salt, Sigma Life Science, St. Louis, MO) (TSAYE + amp and TSBYE + amp). 2.3.4 E. coli O157:H7 growth rates. For the growth rate study, 20 µl aliquots of the 4 avirulent and 3 virulent E. coli O157:H7 strains were separately inoculated in triplicate into flasks containing 200 ml of TSBYE + amp and TSBYE, respectively. Immediately after inoculation and again after 2, 4, 6, 15, 18, 21 and 24 h of incubation at 37°C without shaking, 100 µl aliquots were removed, appropriately diluted in sterile 0.1% phosphate buffer solution (PBS) and plated on TSAYE + amp (avirulent strains) or TSAYE (virulent strains) with all plates counted after 18 - 20 h of incubation at 37°C. The equation to calculate Generation time (Td) was as follows: Td = (t2 – t1) * log(2)/log(q2/q1), where q1 is the number of E. coli at the t1 and q2 is the E. coli population at t2. 2.3.5 Attachment. A modification of the microtiter plate assay described by Jackson and others (Jackson and others 2002) was conducted in triplicate to assess attachment ability. The seven E. coli O157:H7 strains were separately grown in 9 ml of colony forming antigen (CFA) medium containing (per liter) 10 g of casamino acids (BD Bacto™, Becton Dickinson), 1.5 g of yeast extract (BD Bacto™, Becton Dickinson), 50 mg of MgSO4 (J.T. Baker, Mallinckrodt Baker Inc.), and 5 mg of MnCl2 (J.T. Baker, Mallinckrodt Baker Inc.) for 18 h at 37°C, with this same medium supplemented with 100 ppm ampicillin for the GFP-labeled strains. Thereafter, each 43 culture was diluted 1:100 in fresh CFA and pipetted (200 μl) into triplicate wells of a 96-well untreated polystyrene microtiter tissue culture plate (BD Falcon MicrotestTM Flat Bottom, Becton Dickinson, Franklin Lakes, NJ) with three wells per plate containing 200 μl of CFA with/without 100 ppm ampicillin serving as negative controls. Following 48 h of incubation at 22°C, the microtiter wells were emptied, stained with 200 μl of 2.0% crystal violet (Remel, Lenexa, KS) for 5 min, rinsed five times with deionized water and air-dried with the remaining dye resolubilized in 160 μl of 33% (v/v) glacial acetic acid (Sigma Chemical Co.). Optical densities were then read at 630 nm using a Synergy HT Microplate Reader (Bio-Tek Instruments, Inc, Winooski, VT). 2.3.6 Inoculation of iceberg lettuce, Romaine lettuce and baby spinach. A 0.2 ml aliquot of each avirulent E. coli O157:H7 strain was transferred to 200 ml of TSBYE + amp, incubated for 18 – 20 h at 37°C and then combined to obtain 800 ml of the 4-strain cocktail. Based on similar growth rates, the four strains were combined in equal volumes, after which 800, o 8 and 0.08 ml of the cocktail was added to 75 L of tap water (~15 C) in a 121 L plastic container 7 5 3 (Rubbermaid, Wooster, OH) to obtain populations of ~10 , 10 and 10 CFU/ml, respectively. Cored heads of unwashed iceberg (~24 heads) and Romaine lettuce (~48 heads) or commercially washed and bagged baby spinach leaves, each weighing 22.7 kg, were immersed in the E. coli suspension for 15 (iceberg and Romaine) or 1 min (baby spinach) to ensure that each leaf was uniformly inoculated. The heads of iceberg and Romaine lettuce were spun in a dewatering centrifuge (described below) to remove residual inoculum, and then drained/air-dried for 1 h at 22°C before processing. Baby spinach was drained and air-dried for 1 h before processing. Duplicate 25-g samples of each product were then aseptically collected to determine the initial inoculation level at the time of processing. 44 2.3.7 Impact of post-inoculation hold time on E. coli O157:H7 transfer. In order to assess the impact of post-inoculation hold time on E. coli O157:H7 transfer, three 22.7 kg 4 batches of iceberg lettuce were inoculated at ~10 log CFU/g as just described and then o drained/air-dried for 24 h in a 4 C walk-in cooler before processing. 2.3.8 Processing equipment. A small-scale commercial leafy green processing line capable of processing ~3,500 kg/h was assembled that consisted of a lettuce shredder, step conveyer, flume tank, shaker table and dewatering centrifuge (Figures 2.1 – 2.5). The commercial lettuce shredder (model TRS 2500 Urschel TranSlicer, Valparaiso, IN) was operated at a feed belt/slicing wheel speed of 198 m/min and 905 RPM, respectively, to obtain a shred size of approximately 5 x 5 cm. The polyurethane step conveyer belt (ThermoDrive, Mol Industries, Grand Rapids, MI) on the conveyor (Dorner model 736018 mc series, Dorner Manufacturing, Hartland, WI) was operated at 0.11 m/sec. A stainless steel water recirculation o tank (~1000 L capacity) was filled with 890 L of sanitizer-free tap water (~15 C), because the objective was to quantify E. coli O157:H7 transfer rather than to assess inactivation by sanitizers during processing. This water tank was connected to a 3.6-m long by stainless steel flume tank (Heinzen Manufacturing, Inc., Gilroy, CA) - equipped with two overhead spray jets (1 m from the start) by a 4.14 m long, 10 cm-diameter hard plastic discharge hose and a centrifugal pump (model XB754FHA, Sterling Electric, Inc., Irvine, CA) that circulated the water at ~15 L/sec. The stainless steel shaker table for partial dewatering was operated by a 1 HP Baldor washdown duty motor (Baldor Electric Co., Ft. Smith, AR) at 1760 RPM. Water removed from the leafy greens during mechanical shaking passed through a fine mesh screen and was fed into the water holding tank by a water recirculation spout underneath the shaker table. A 50-lb (22.7-kg) 45 capacity centrifugal Spin Dryer (model SD50-LT, Heinzen Manufacturing, Inc.) with three internally timed spin cycles totaling 60 s was used for centrifugal drying. 2.3.9 Identification of product contact surfaces. A 0.35% (w/v) suspension of Glo Germ™ (Glo Germ Co., Moab, Utah) was prepared by dissolving 264 g in 1 L of 95% EtOH, which was then added to 75.7 L of distilled water. Uninoculated retail baby spinach leaves and heads of cored iceberg lettuce (22.7 kg each) were divided into five small batches of equal size and immersed in the Glo Germ™ suspension for 1 min. After draining and drying for 18-24 h at 4°C, the lettuce was processed through the entire processing line, beginning with shredding, whereas baby spinach processing began with fluming. Immediately after processing, the entire line was viewed under ultraviolet light (365 nm, Model SB-100XR , Spectronics Corporation, Westbury, NY) to identify the major product contact points. Based on these observations, 14, 8, 11, 9 and 8 product contact surfaces on the shredder, conveyor, flume tank, shaker table and dewatering centrifuge, respectively, were targeted for subsequent quantitative recovery of E. coli O157:H7. Glo Germ™ was removed from the equipment surfaces with 95% EtOH before any E. coli O157:H7-inoculated product was processed, so as not to affect bacterial transfer. 2.3.10 Leafy green processing. Inoculated heads of cored iceberg and Romaine lettuce (22.7 kg) were hand-fed into the shredder, with the shredded product then step-conveyed to the flume tank for washing in 890 L of recirculating sanitizer-free water, partially dewatered on the shaker table, collected in a single centrifugation basket and centrifugally dried. Inoculated baby spinach (22.7 kg) was manually dumped into the flume tank, washed in the flume water, partially dried on the shaker table, collected in two centrifuge baskets and then centrifugally dried. Whole heads of iceberg and Romaine lettuce and baby spinach leaves (22.7 kg each) were processed at 46 a rate of about 0.35 kg per second, with the entire 22.7 kg of product ready for centrifugal drying after 60-70 sec. 2.3.11 Impact of shred size on E. coli O157:H7 transfer. Three 22.7 kg batches of 4 cored iceberg lettuce were inoculated with the E. coli O157:H7 cocktail at ~10 log CFU/g. In order to assess the impact of a smaller shred size of approximately 0.5 x 5 cm on subsequent E. coli O157:H7 transfer, the feed belt and slicing wheel speeds were changed from 198 to 76 m/min and from 905 to 2012 RPM, respectively, with the lettuce then processed as previously described. 2.3.12 Sample collection. During processing, five iceberg lettuce, eight Romaine lettuce and ten spinach samples (25 g each) were collected from the conveyer belt (iceberg and Romaine lettuce only) and the end of the flume tank (3 m from the start) at regular intervals, with the number of samples corresponding to the capacity of the centrifugation basket for the three different products. Five 40-ml water samples were also obtained from the flume tank and water recirculation spout. Immediately after shaker table dewatering, five iceberg lettuce, eight Romaine lettuce and ten baby spinach samples (25 g each) were collected through the 2.75 cmdiameter holes that were spaced 10 cm apart along the side of the centrifugation basket. Product in the basket was then dried in the pre-set Spin Dryer. During centrifugal drying, four 40 ml water samples were collected from the centrifuge drain at 15 sec intervals. After centrifugation, four iceberg lettuce, six Romaine lettuce and eight baby spinach samples (25 g each) also were collected from the centrifugation basket as previously described. Twenty-eight baby spinach and 50 iceberg and Romaine lettuce contact areas on the equipment (14 shredder, 8 conveyer, 11 flume tank, 9 shaker table and 8 dewatering centrifuge samples (Figures 2.1 – 2.5)) measuring 2 100 cm as previously identified using Glo Germ™ were sampled as described by Vorst et al. 47 (2004) using 1-ply composite tissues moistened with 1 ml of sterile 1% (w/v) phosphate buffer (8.5 g/L NaCl, 1.44 g/L Na2HPO4, and 0.24 g/L KH2PO4, J.T. Baker, Mallinckrodt Baker Inc., Phillipsburg, NJ). 2.3.13 Microbiological analyses. All lettuce and baby spinach samples (25 g) were added to 100 ml of sterile 1% phosphate buffer in a Whirl-Pak™ filter bag (Nasco, Fort Atkinson, WI) and homogenized in a stomacher (Stomacher 400 Circulator, Seward, 2 Worthington, UK) for 1 min at 260 rpm. The only exception was lettuce inoculated at ~10 CFU/g, which was processed in a pulsifier (Pulsifier, Filtaflex Ltd., Almonte, Ontario, Canada) rather than the stomacher, to reduce the amount of lettuce exudate/particulates in the samples to be filtered. Preliminary data indicated similar release of bacteria from leafy greens during stomaching and purifying. These sample homogenates were then either appropriately diluted in sterile 1% phosphate buffer and plated on TSAYE + amp or processed using 0.45 μm membrane filters (Millipore, Millipore Corporation, Billerica, MA) (lower detection limit of 0.04 CFU/g), which were placed on 60-mm dia. Petri plates containing TSAYE + amp to quantify E. coli O157:H7. The 1-ply composite tissue samples were added to 15 ml of sterile 1% phosphate buffer in a Whirl-Pak™ bag, stomached for 1 min at 260 rpm and then plated identically to the 2 produce samples, giving a lower detection limit of 1 CFU/100 cm when the entire 15 ml sample was passed through the membrane filter. The 40 ml water samples were either appropriately diluted in sterile 1% phosphate buffer and plated on TSAYE + amp or processed by membrane filtration, which gave a minimum detection limit of 0.03 CFU/ml. Following 18 - 20 h of incubation at 37°C, all green fluorescing colonies as seen under ultraviolet light (365 nm, BlakRay, Ultra-violet Product Inc. San Gabriel, CA) were counted as E. coli O157:H7. 48 2.3.14 Statistical analysis. E. coli O157:H7 counts were converted to log CFU per g, ml 2 or 100 cm and subjected to ANOVA using JMP 8.0 (SAS Institute Inc., Cary, NC). For all tests, α = 0.05. E. coli generation times were calculated by comparing the time in hours to the CFU/ml during the exponential phase growth. The Tukey-Kramer HSD test was used to compare E. coli growth rates and attachment abilities of the strains, identify significant differences in transfer to the individual product contact surfaces and water samples collected at various locations throughout the process. The same tests were applied to identify significant differences in the numbers of E. coli O157:H7 shed from iceberg lettuce, Romaine lettuce and baby spinach at the various locations sampled during processing. Impact of shred size and postinoculation hold time on the numbers of E. coli O157:H7 transferred to the water and various equipment surfaces during processing were also assessed using this same procedure. 49 2.4 RESULTS 2.4.1 E. coli O157:H7 growth rate and attachment ability. The three virulent E. coli O157:H7 strains (K3995, K4495, and K4831) and three of the four avirulent strains (CV2b7, 6980-2, and 6982-2) exhibited similar growth rates (P < 0.05) with generation times at 37°C ranging from 21.5 to 25.1 min. (Figure 2.6). E. coli O157:H7 ATCC43888 - one of the four avirulent GFP-labeled strains used in the cocktail for leafy green inoculation, grew significantly more slowly (generation time of 36.8 min) and yielded a higher OD value of 1.321 than the other six strains tested (OD 0.062 to 0.146) (P < 0.05) in the microtiter plate attachment assay (Table 2.1). 2.4.2 Lettuce and baby spinach. After inoculating each product at the three different targeted levels, iceberg lettuce, Romaine lettuce and baby spinach contained mean E. coli O157:H7 populations of 6.0 – 6.2, 4.0 – 5.0 and 1.8 – 2.5 log CFU/g at the time of processing, with Romaine lettuce containing fewer E. coli O157:H7 (1.8 log CFU/g) than baby spinach (2.5 log CFU/g) at the lowest target inoculation level of 2 log CFU/g (P < 0.05). At the medium inoculation level, Romaine lettuce (5.0 log CFU/g) retained more of the inoculum than did iceberg lettuce and baby spinach, 4.0 and 4.2 log CFU/g, respectively (P < 0.05) (Figure 2.7). 2.4.3 Flume water. Overall, 82.3 to 118.2, 48.5 to 64.6 and 48.9 to 147.2% of the original E. coli O157:H7 inoculum was respectively shed from iceberg lettuce, Romaine lettuce and baby spinach into the 890 L of processing water used for fluming (Tables 2.2 – 2.4). These percentages are based on an estimation of the total number of E. coli O157:H7 cells having been transferred to the 890 L of flume water from the 50 lbs of inoculated leafy greens. During fluming, E. coli O157:H7 populations on inoculated iceberg and Romaine lettuce decreased (P < 0.05) by 0.7 - 0.9 and 1.0 - 1.2 log, respectively, with a decrease of 0.9 to 1.3 log seen for baby 50 spinach directly entering the flume tank. More of the inoculum was removed from baby spinach than from iceberg lettuce (P < 0.05). Thereafter, relatively few significant differences (P < 0.05) were seen between the different sampling locations, with the numbers of E. coli O157:H7 on the three products generally decreasing ~0.1 log CFU/g during shaker table dewatering and centrifugal drying (Figure 2.7), with no significant differences between the three different products. After centrifugal drying, iceberg lettuce, Romaine lettuce, and baby spinach, respectively, retained 7.9 - 12.6, 8.3 - 14.1, and 3.2 - 17.4% of the original E. coli O157:H7 inoculum (Tables 2.2 – 2.4). 2.4.4 Centrifugation water. Water exiting the centrifuge drain after spin-drying yielded E. coli O157:H7 populations that were 1.1 - 2.1 log CFU/ml higher (P < 0.05) compared to the flume water (Figure 2.8). Overall, 30.5 - 38.6, 17.8 - 26.8 and 6.3 - 15.6% of the original inoculum on iceberg lettuce, Romaine lettuce, and baby spinach, respectively, was recovered from the ~3.1 L of water exiting the centrifuge drain (Tables 2.2 – 2.4). At the 6 log CFU/g inoculation level, E. coli O157:H7 was more easily shed from iceberg lettuce than from baby spinach (P < 0.05) during centrifugation. Similarly, both iceberg and Romaine lettuce shed more E. coli O157:H7 during drying than baby spinach, when initially inoculated to contain 2 log CFU/g (P < 0.05). At 4 log CFU/g, all three products differed significantly from each other, with iceberg lettuce losing the most and baby spinach the least E. coli O157:H7 (P < 0.05). As expected, the decrease in numbers of E. coli O157:H7 transferred to the water paralleled the three different lettuce inoculation levels (Figures 2.7 and 2.8). 2.4.5 Processing equipment surfaces. After processing iceberg and Romaine lettuce inoculated at the highest and lowest levels, significantly higher percentages of the original E. coli O157:H7 inoculum were frequently transferred to surfaces on the shredder (0.20 to 0.53%), 51 conveyer belt (0.21 to 0.93%) and centrifugal dryer (0.15 to 0.48%), as compared to the flume tank (0.01 to 0.09%) and shaker table (0.01 to 0.15%), with the only exception being iceberg lettuce inoculated at the lowest level, where the surfaces of the shredder yielded higher populations (P < 0.05). Baby spinach followed a similar trend, with the most heavily inoculated products retaining the most bacteria after centrifugal drying (0.05%), followed by the flume tank (0.01%) and shaker table (0.00%), while the least inoculated products did not show any significant differences between location (P < 0.05) (Tables 2.2 – 2.4). When inoculated at 4 log CFU/g, Romaine and iceberg lettuce transferred higher numbers of E. coli O157:H7 to the shredder (0.31 and 0.53%), followed by the conveyer belt (0.35 and 1.39%) and lastly the dewatering centrifuge (0.03 and 0.22%), flume tank (0.02 and 0.04%), and shaker table (0.01 and 0.04%) (Tables 2.2 – 2.4). During spinach processing, highest numbers of E. coli O157:H7 transferred to sampling locations on the dewatering centrifuge (0.03%) (P < 0.05) followed by the flume tank (0.01%) and shaker table (0.00%) (Tables 2.2 – 2.4). After conveying, transfer was generally greatest when the product initially contacted the slanted front end of the flume tank (Figure 2.1, flume tank sample 1) and when the contaminated water contacted the drain of the centrifugal dryer (Figure 2.5, dewatering centrifuge sample 4) (Tables 2.5 – 2.7). Overall, no significant differences (P > 0.05) were observed in the percentage of E. coli O157:H7 cells transferred to surfaces on the five different pieces of processing equipment in terms of leafy green variety, except for samples collected from the conveyer belt after processing Romaine and iceberg lettuce at the medium inoculation level and on the centrifugal dryer after processing baby 2 spinach and Romaine lettuce inoculated at 10 CFU/g (Tables 2.2 – 2.4). 2.4.6 Impact of shred size and hold time before shredding. Before processing, inoculated iceberg lettuce to be coarse-shredded, fine-shredded, or held for 24 h and then coarse- 52 shredded contained E. coli O157:H7 populations of 4.0, 3.8 and 3.7 log CFU/g, respectively (Figure 2.7B). Overall, no significant differences (P > 0.05) in percentages of inocula shed in the flume water were seen between coarse- (86.5%) and fine-shredded lettuce (83.1%) or lettuce held for 24 h at 4°C before processing (49.9%). After centrifugal drying, 42.2% of the original E. coli O157:H7 inoculum remained on product that was stored for 24 h before processing, which was greater (P < 0.05) than for iceberg lettuce that was processed 1 h after inoculation (8.1%). For coarse and fine-shredded, and 24 h-held lettuce, only 0.0 to 1.4%, 0.0 to 1.5%, and 0.0 to 0.5% of the original inoculum, respectively, transferred to the equipment surfaces, with no significant differences between these values (P > 0.05) (Table 2.3). 53 2.5 DISCUSSION Four avirulent strains of E. coli O157:H7 were used in this study, rather than the virulent outbreak strains, due to the production of potentially infectious aerosols that could have been generated during processing. As has been reported previously (Taormina and others 2009, Wachtel and others 2002), the GFP label that was inserted via a plasmid into these four strains also allowed easy differentiation of our inoculum from the background microflora, which often exceeded 7 log CFU/g. Except for the avirulent GFP-labeled strain ATCC43888, which exhibited slower growth and greater adherence in the microtiter plate assay, the remaining three GFP-labeled strains behaved similarly to the three outbreak strains. While Strain ATCC43888 grew more slowly, it was included in the study to represent the worst case scenario in terms of adherence. Uniform contamination of the inoculated heads via dipping was essential for obtaining repeatable results. However, the method of inoculation clearly has an impact on bacterial transfer, with Lang et al. (2004) having previously shown that greater numbers of E. coli O157:H7 could be recovered from dip- as opposed to spot- or spray-inoculated lettuce when the same numbers of bacterial cells were initially applied. Three inoculation levels - 6, 4, and 2 log CFU/g were chosen to represent high, medium, and low levels of contamination. While the 6 and 4 log CFU/g levels used in our work may be unrealistically high, previous studies have shown that cattle (the main carriers of E. coli O157:H7) normally shed less than 100 CFU/g of feces, with certain high-level shedders, known as “super-shedders,” shedding up to 1,000,000 CFU/g in their feces (Chase-Topping and others 2007, Low and others 2005). Hence, the inoculation levels used represent direct contamination of product in the field via fecal matter from domestic and wild animals. In addition, these high 54 inoculation levels were also needed to obtain quantitative transfer data that could be modeled and incorporated into a subsequent risk assessment as reported elsewhere (Pérez Rodríguez and others 2011). Direct contact of produce with manure represents only one route of pre-harvest contamination. This same contaminated manure also can enter irrigation water used for crops (Beuchat 2002, Ibekwe and others 2004), resulting in lower numbers of pathogens distributed over a far wider area as opposed to a localized fecal event from an animal. Although closer to realistic estimates for microbial contamination, our results using an inoculation level of 2 log CFU/g yielded considerably greater variability than the medium and high levels. Commercial washing of leafy greens in sanitizer solutions containing up to 100 ppm chlorine or 80 ppm peroxyacetic acid is now a universally accepted practice (FDA 2009). While most effective in reducing microbial populations in the flume water, these sanitizers typically reduce bacterial populations on fresh produce only 90 to 99%, which represents only a marginal improvement over the use of water alone (Beuchat and others 2004, Burnett and others 2004, Keskinen and others 2009, Sapers 2001, Weissinger and others 2000, Zhang and others 2009). Because the goal of this study was to quantify the numbers of E. coli O157:H7 cells transferred during leafy green processing, all lettuce and baby spinach was processed using sanitizer-free water. If chemical sanitizers had been used, the numbers of E. coli O157:H7 would have likely decreased to non-detectable levels in the flume water, especially in experiments using low inoculation level, thereby negating our effort to obtain quantitative transfer data for subsequent mathematical modeling. Our data align well with the typical bacterial reductions of 90 – 99% seen during washing of fresh produce in water alone. During processing of iceberg lettuce inoculated at 2.9 – 3.5 log 55 CFU/g, approximately 90% of the E. coli O157:H7 inoculum was shed during fluming, with populations of 1.1 to 1.6 log CFU/g remaining on the product after centrifugal drying. Hence, given an oral infectious dose for E. coli O157:H7 of less than 100 cells (Meng and others 2007), and the fact that our counts, even at the lowest inoculation level, fell within this range, the ineffectiveness of current processing/washing methods is clearly a major concern, as evidenced by the three nationwide spinach and lettuce outbreaks in 2006 (FDA 2006, FDA 2007, Jay and others 2007) and a series of more recent product recalls involving E. coli O157:H7, Salmonella and Listeria monocytogenes (FDA 2010a, FDA 2010b, FDA 2010c). During spin drying, about 22% of the remaining inoculum was shed in ~3 L of water exiting the drain of the centrifugal dryer. Even at the lowest inoculation level, E. coli O157:H7 was quantifiable at levels of 0.9 to 1.6 log CFU/ml in the centrifugation water. Because microbial contaminants are presumed to be non-uniformly distributed in commercially processed leafy greens, routine monitoring of the spent centrifugation water for E. coli or other indicator organisms appears to be a more promising means than product sampling, to better ensure end product safety. Methods are currently being developed to concentrate and rapidly detect pathogens in larger volumes of processing water with ultrafilters yielding E. coli recovery rates of 74 to 96% (Hill and others 2005, Morales-Morales and others 2003). The levels and point at which leafy greens become contaminated are both key factors that partially dictate the extent of E. coli O157:H7 transfer during processing. One study conducted at Rutgers University used chicken, lettuce, cutting boards, hands, and a spigot to show that as the inoculum size increased, the number of bacteria transferred remained constant, with the transfer rate decreasing (Montville and Schaffner 2003). In our study, increased bacterial transfer to the water and equipment surfaces paralleled the increase in the produce inoculation level. 56 When leafy greens were dip-inoculated and processed 1 h later, approximately 83% of the inoculum transferred to the flume water (Tables 2.2 – 2.4). In a bench-top study, Zhang et al. (Zhang and others 2009) showed that a single iceberg lettuce leaf inoculated with E. coli O157:H7 at 5.6 log CFU/leaf transferred nearly 99% of the inoculum to 500 ml of sanitizer-free water. These differences in reduction are likely due in part to differences in water turbulence during washing. Such a high reduction is especially surprising, because their leaves were spot inoculated, which has been shown to result in lower E. coli O157:H7 recovery as compared to the dip inoculation method that was used in our work (Lang and others 2004). They also held their inoculated lettuce for an additional hour at 4°C before washing, which would suggest more bacterial attachment. In the present study, a small, but consistent fraction of the original E. coli O157:H7 inoculum transferred to the flume tank, shaker table, and centrifugal dryer during processing. Therefore, the populations of E. coli O157:H7 recovered from these wet surfaces most likely reflect the number of cells present in the thin film of water that coated the 100 cm2 surface before sampling. In terms of the wet equipment surfaces, greatest transfer was seen to the interior walls, basket carrier, and drain of the dewatering centrifuge, with the centrifugation water also yielding E. coli O157:H7 populations that were 1 to 2 logs higher than the processing water. In addition to the wet surfaces, direct transfer of E. coli O157:H7 was also seen between the product and product contact surfaces of the shredder and conveyor belt. After processing iceberg and Romaine lettuce, the shredder and conveyer belt generally yielded higher E. coli O157:H7 counts than the flume tank, shaker table, or centrifugal dryer. These microbial ‘hot spots’ are not surprising, since exudate from shredded lettuce was visible on the cutting wheel and discharge chute of the shredder, as well as on all product contact areas of the conveyor. This 57 liquid likely enhanced E. coli O157:H7 transfer from lettuce to the stainless steel and polyurethane surfaces of the shredder and conveyor belt, as was also shown by Moore et al. (Moore and others 2003). In a survey of fresh vegetable processing facilities in Japan, 2 populations of mesophilic aerobic bacteria exceeded 5.0 log CFU/cm or ml in many samples collected from washing, slicing, dehydrating, and blending equipment, surfaces of slicer/shredder blades, and the processing room floor after processing (Kaneko and others 1999). Although the percentage of the E. coli O157:H7 inoculum transferred to equipment surfaces was minimal in our study, only a few cells are needed to cross-contaminate subsequent product during continuous processing. In support of these observations, contaminated lettuce particulates lingering behind the cutting wheel rim of a commercial lettuce shredder were the probable cause of an Australian Salmonella Bovismorbificans outbreak that resulted in 41 cases of illness (Stafford and other 2002). Lettuce shred size did not significantly impact E. coli O157:H7 transfer during subsequent processing. These results were somewhat surprising, given that an increase in the cut surface area would be expected to release more lettuce exudates, which would in turn increase E. coli O157:H7 transfer during shredding and conveying. One reason for this unexpected outcome may be the modest difference in shred size – approximately 5 x 5 cm for large-shredded compared to 0.5 x 5 cm for small-shredded lettuce. Bacterial attachment and biofilm formation on the surfaces of leafy greens are two key factors that negatively impact the removal of microorganisms during conventional washing (Carmichael amd others 1999, Sapers 2006). When Lang et al. (2004) assessed the impact of different post-inoculation drying times on recovery of E. coli O157:H7 from iceberg lettuce and parsley, higher (or similar) populations were obtained from leafy greens that were dried for 2 h at 58 22°C as opposed to 2 h at 22°C followed by 22 h at 4°C. Consistent with those findings, our results also showed that increasing the time from inoculation to processing from 1 h at 23oC to 1 h at 23oC followed by 24 h at 4°C decreased the number of E. coli O157:H7 cells transferred from iceberg lettuce to the water and equipment surfaces during processing. While this implies that refrigerating the product for 1 day post-harvest could decrease the potential spread of E. coli O157:H7 during processing and lead to potentially smaller-scale outbreaks, current commercial produce washing/sanitizing practices are still not sufficiently efficacious against foodborne pathogens. Based on several surveys indicating Salmonella and Listeria monocytogenes incidence rates of 0 to 35.7% and 4.8%, respectively, on farm and retail produce (Doyle and Erickson 2008), ample opportunity exists for foodborne pathogens to survive, grow, and enter produce packing houses on product coming from the field. This is the first report to quantify the number of E. coli O157:H7 cells transferred from product to equipment surfaces during pilot-scale production of fresh-cut leafy greens. Based on our findings, approximately 90% of the E. coli O157:H7 inoculum was shed in the sanitizer-free water, with this pathogen also contaminating the product contact surfaces of the shredder, conveyor, flume tank, shaker table, and dewatering centrifuge to various degrees during processing. Routine microbial monitoring of the centrifugation water may prove to be a useful strategy for minimizing and preventing the continued spread of bacterial pathogens to large quantities of product, with these findings providing important insight into potential lapses in current cleaning/sanitizing practices and improved equipment designs. 59 2.6 FIGURES Figure 2.1: Lettuce shredder sampling locations: (A) loading end, (B) interior feed belt and guides, and (C) blade and discharge chute A B 1 3 3 2 3 1 2 3 C 5 6 6 6 60 6 4 Figure 2.2: Conveyer belt sampling locations 1 2 2 1 2 1 2 1 61 Figure 2.3: Flume tank sampling locations 1 2 2 2 2 2 2 2 2 2 2 62 Figure 2.4: Shaker table sampling locations 2 2 3 4 1 3 6 5 63 6 Figure 2.5: Dewatering centrifuge sampling locations 3 1 1 2 1 1 2 4 64 Figure 2.6: Growth curves at 37°C for four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) in TSBYE + amp and three virulent strains of E. coli O157:H7 (K3995, K4495, and K4831) TSBYE (n=3) 65 a Figure 2.7: E. coli O157:H7 populations (mean ± SD) on the product during processing of leafy greens inoculated at: (A) ~6 log CFU/g (n = 3), (B) ~4 log CFU/g (n = 3), and (C) ~2 log CFU/g (n = 5). A 66 Figure 2.7 (cont’d) B 67 Figure 2.7 (cont’d) C a One over the detection limit was used to calculate the mean log value for samples not yielding colonies by direct plate count. Means of the same product type with different letters are significantly different (P ≤ 0.05) 68 a Figure 2.8: E. coli O157:H7 populations (mean ± SD) in water during processing of leafy green inoculated at: (A) ~6 log CFU/g (n = 3), (B) ~4 log CFU/g (n = 3), and (C) ~2 log CFU/g (n = 5). A 69 Figure 2.8 (cont’d) B 70 Figure 2.8 (cont’d) C a One over the detection limit was used to calculate the mean log value for samples not yielding colonies by direct plate count. Means of the same product type with different letters are significantly different (P ≤ 0.05). 71 2.7 TABLES Table 2.1. Bacterial attachment, as measured by the optical density at 630 nm for four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) and three virulent strains of E. coli O157:H7 (K3995, K4495, and K4831) (n=3) Strains a Mean Optical Density ± SD Virulent K3995 0.062 ± 0.007 K4831 0.081 ± 0.012 K4492 0.061 ± 0.007 Control 0.065 ± 0.005 CV2b7 0.059 ± 0.003 ATCC43888 1.321 ± 0.232 6982-2 0.095 ± 0.009 6980-2 0.146 ± 0.012 Control with 100 ppm ampicillin 0.065 ± 0.014 B B B B Avirulent, GFP-labeled, ampicillin-resistant a B A B B B Means of the same product type with different letters are significantly different (P ≤ 0.05). 72 Table 2.2: E. coli transfer (%) during processing of leafy green inoculated at 6 log CFU/g b,c Mean ± SD E. coli O157:H7 recovered (%) Processing Step a Conveyor Belt Flume Tank a 52.6 ± 43.0 82.3 ± 24.4 0.2 ± 0.4 NA a Shredder Iceberg Lettuce 147.2 ± 50.4 a Romaine Lettuce NA Flume Water Baby Spinach 0.2 ± 0.2 0.0 ± 0.1 0.0 ± 0.0 0.2 ± 0.2 0.0 0.0 a Shaker Table Dewatering Centrifuge a ± 0.0 ± 0.0 0.1 ± 0.1 % Original Inoculum on Product after Processing 17.4 a Ba Ba Aa ± 10.8 11.1 ± 5.5 Aa Aa Ba Ba ABa 0.2 ± 0.3 0.7 ± 0.8 0.0 ± 0.0 0.0 ± 0.0 0.4 ± 0.8 7.9 ± 2.6 Aa ABa Ba Ba ABa The total equipment surface areas for the shredder, conveyer belt, flume tank, shaker table and dewatering centrifuge were 1.74 m2, 2.88 m2, 2.60 m2, 2.47 m2, and 4.55 m2, respectively. b Means with different capital letter represent E. coli O157:H7 counts that differ significantly in terms of processing equipment location while means with different lowercase letter differ significantly in terms of leafy green variety (P ≤ 0.05). 73 Table 2.2 (cont’d) c These percentages are based on an estimation of the total number of E. coli O157:H7 cells having been transferred to the 890 L of flume water or equipment surfaces from the 50 lbs of inoculated leafy greens. 74 Table 2.3: E. coli transfer (%) during processing of leafy green inoculated at 4 log CFU/g b,c Mean ± SD E. coli O157:H7 recovered (%) Baby Romaine Iceberg Iceberg Iceberg Lettuce Spinach Lettuce Lettuce Lettuce (held 24 h before (finely Processing Step processing) shredded) a 96.9 ± 2.4 a Conveyor Belt Flume Tank a 0.1 ± 0.1 NA a Shredder 48.5 ± 47.6 NA Flume Water 0.4 ± 0.3 0.0 ± 0.1 0.0 ± 0.0 0.0 ± 0.0 8.3 ± 6.1 0.0 ± 0.0 a 0.0 ± 0.0 Shaker Table Dewatering Centrifuge a 0.0 ± 0.1 % Original Inoculum on Product 3.2 ± 3.1 Ba Ca Aa 83.1 ± 78.6 86.5 ± 38.4 Aa Bb Ca Ca Ca 0.4 ± 0.5 1.4 ± 1.0 0.0 ± 0.1 0.0 ± 0.0 0.2 ± 0.4 8.1 ± 2.8 after Processing 75 Aa Ba Ca Ca BCa 0.0 ± 0.1 0.2 ± 0.2 0.1 ± 0.1 0.4 ± 0.6 1.6 ± 2.0 12.5 ± 5.7 49.9 B B B B A ± 29.8 0.1 ± 0.4 0.4 ± 0.8 0.2 ± 0.5 0.0 ± 0.1 0.5 ± 0.5 42.2 ± 19.8 B AB AB B A Table 2.3 (cont’d) a The total equipment surface areas for the shredder, conveyer belt, flume tank, shaker table and dewatering centrifuge are 1.74 m2, 2.88 m2, 2.60 m2, 2.47 m2, and 4.55 m2, respectively. b Means with different capital letter represent E. coli O157:H7 counts that differ significantly in terms of processing equipment location while means with different lowercase letter differ significantly in terms of leafy green variety (P ≤ 0.05). c Percentages are based on an estimation of the total number of E. coli O157:H7 cells having been transferred to the 890 L of flume water or equipment surfaces from the 50 lbs of inoculated leafy greens. 76 Table 2.4: E. coli transfer (%) during processing of leafy green inoculated at 2 log CFU/g b,c Mean ± SD E. coli O157:H7 recovered (%) Processing Step a Conveyor Belt a 64.6 ± 64.2 118.2 ± 4.1 0.5 ± 1.4 NA a Shredder Iceberg Lettuce 48.9 ± 21.8 a Romaine Lettuce NA Flume Water Baby Spinach 0.8 ± 1.5 0.0 ± 0.1 0.0 ± 0.0 0.5 ± 0.9 0.1 ± 0.2 0.0 ± 0.0 0.0 ± 0.0 % Original Inoculum on Product after Processing 7.3 ± 9.6 Flume Tank a Shaker Table Dewatering Centrifuge a a Aa Aa Ab 14.1 ± 24.3 Aa ABa Ba Ba ABa 0.3 ± 0.3 0.9 ± 0.8 0.1 ± 0.2 0.2 ± 0.5 0.4 ± 0.7 12.6 ± 4.5 Aa Ba Ba Ba Bab The total equipment surface areas for the shredder, conveyer belt, flume tank, shaker table and dewatering centrifuge are 1.74 m2, 2.88 m2, 2.60 m2, 2.47 m2, and 4.55 m2, respectively. b Means with different capital letter represent E. coli O157:H7 counts that differ significantly in terms of processing equipment location while means with different lowercase letter differ significantly in terms of leafy green variety (P ≤ 0.05). 77 Table 2.4 (cont’d) c Percentages are based on an estimation of the total number of E. coli O157:H7 cells having been transferred to the 890 L of flume water or equipment surfaces from the 50 lbs of inoculated leafy greens. 78 Table 2.5: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 6 log CFU/g (n=3) Mean ± SD E. coli O157:H7 recovered (log CFU/100 cm2) Location on Processing Equipment Baby Spinach Romaine Lettuce Iceberg Lettuce 1. front, stainless steel (n=2) ND 5.0 ± 0.5 5.7 ± 0.3 2. ND 4.6 ± 0.9 3. feed belt guide, stainless steel (n=4) ND 3.5 ± 2.1 4. below the cutting wheel, stainless steel (n=1) ND 4.2 ± 0.9 5. cutting wheel, stainless steel (n=1) ND 4.6 ± 0.3 6. discharge chute, stainless steel (n=4) ND 5.6 ± 0.5 ND 5.1 ± 0.6 a Shredder feed belt, plastic (n=2) AB ABC BC ABC ABC A 5.3 ± 0.3 4.8 ± 0.8 5.0 ± 0.7 4.7 ± 0.3 5.8 ± 0.5 AB ABC BCDE ABCDE ABCDE A Conveyer Belt 1. conveyer belt, polyurethane (n=4) 79 A 5.7 ± 0.4 A Table 2.5 (cont’d) 2. belt guide, plastic (n=4) ND 4.8 ± 0.6 AB 5.7 ± 0.5 A Flume Tank 1. slanted front end, stainless steel (n=1) 5.0 ± 0.5 2. interior wall, stainless steel (n=10) 2.6 ± 0.8 A D 5.3 ± 0.4 1.9 ± 0.9 AB DE 5.1 ± 0.2 2.5 ± 0.8 ABCDE F Shaker table 1. front, stainless steel (n=1) 3.1 ± 0.6 2. perforated dewatering screen, stainless steel (n=2) 3.6 ± 0.7 3. center, stainless steel (n=2) 3.8 ± 0.3 4. end (vertical portion), stainless steel (n=1) 3.0 ± 0.5 5. end (slanted portion), stainless steel (n=1) 2.5 ± 1.3 6. recirculation spout, stainless steel (n=2) 1.5 ± 0.5 Dewatering Centrifuge 80 BCD ABCD ABC BCDE CDE E 3.9 ± 0.3 4.4 ± 0.2 4.5 ± 0.2 3.6 ± 0.2 2.4 ± 0.2 1.3 ± 0.5 ABC ABC ABC ABCD CDE E 3.7 ± 0.3 4.7 ± 0.0 4.6 ± 0.1 3.8 ± 0.5 3.5 ± 0.4 2.6 ± 0.3 CDEF ABCDE BCDE DEF EF F Table 2.5 (cont’d) 1. interior basket carrier, stainless steel (n=4) 2. interior wall, stainless steel (n=2) 4.0 ± 0.2 3. viewing window in the lid, plastic (n=1) 2.6 ± 1.3 4. drain, stainless steel (n=1) a 4.1 ± 0.3 5.2 ± 0.2 AB ABC CDE A 4.8 ± 0.3 4.1 ± 0.2 1.0 ± 0.0 5.5 ± 0.1 AB ABC E AB 4.9 ± 0.6 4.2 ± 0.5 2.1 ± 0.3 6.0 ± 0.4 ABCD CDE F AB One over the detection limit was used to calculate the mean log value when a sample did not yield any colonies by direct plate count. Means of the same product type with different letters are significantly different (P ≤ 0.05). The numbers preceding the equipment location descriptions correspond with locations identified in Figures 2.1 through 2.5. 81 Table 2.6: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 4 log CFU/g (n=3) Mean ± SD E. coli O157:H7 recovered (log CFU/100 cm2) Location on Processing Equipment Baby Spinach Romaine Iceberg Lettuce Lettuce a Iceberg Lettuce Iceberg Lettuce (finely (held 24 h before shredded) processing) Shredder 1. front, stainless steel (n=2) ND 4.2 ± 0.2 2. ND 4.2 ± 0.1 ND 3.8 ± 0.7 4. below the cutting wheel, stainless ND 3.0 ± 0.9 feed belt, plastic (n=2) 3. feed belt guide, stainless steel AB AB AB 3.8 ± 0.2 3.7 ± 0.3 3.4 ± 0.9 AB ABC ABCDE 1.8 ± 1.1 1.1 ± 0.7 1.2 ± 0.5 ABC ABC BC 2.0 ± 0.8 1.7 ± 0.7 1.8 ± 0.9 2.2 ± 1.7 ABC ABC ABC (n=4) steel (n=1) 82 ABCD 3.3 ± 0.4 ABCDEF 0.9 ± 0.1 ABC ABC Table 2.6 (cont’d) 5. cutting wheel, stainless steel ND 3.9 ± 0.1 ND 4.5 ± 0.3 1. conveyer belt, polyurethane (n=4) ND 4.4 ± 0.4 2. belt guide, plastic (n=4) 4.3 ± 0.3 ABC 2.3 ± 0.2 DEFG 1.3 ± 1.1 ABC 2.3 ± 0.3 2.1 ± 1.1 2.6 ± 0.9 2.5 ± 0.6 2.6 ± 0.7 0.6 ± 1.6 0.7 ± 1.5 ABC (n=1) 6. discharge chute, stainless steel A 3.7 ± 0.7 AB 2.4 ± 1.0 ABC AC (n=4) Conveyer Belt ND A AB 4.0 ± 0.5 3.9 ± 0.4 A A 2.8 ± 0.6 1.9 ± 1.2 AC ABC A A Flume Tank 1. slanted front end, stainless steel 2.9 ± 0.9 AB 4.5 ± 0.3 AB 3.3 ± 0.1 ABCDEF 2.5 ± 0.1 ABC ABC (n=1) 2. interior wall, stainless steel 0.7 ± 0.9 D 1.7 ± 1.1 DE 1.1 ± 0.6 GH 1.3 ± 1.7 B B (n=10) Shaker table 1. front, stainless steel (n=1) 1.2 ± 1.0 ABCD 1.2 ± 1.0 83 DE 2.2 ± 0.3 EFG 2.2 ± 0.5 ABC ABC Table 2.6 (cont’d) 2. perforated dewatering screen, 1.8 ± 0.8 ABCD 2.9 ± 0.1 ABCDE 2.8 ± 0.1 ABCDEF 0.8 ± 1.6 ABC 1.0 ± 1.7 1.5 ± 1.9 1.4 ± 0.8 1.3 ± 0.4 0.3 ± 1.6 3.0 ± 0.6 2.4 ± 0.5 1.8 ± 0.8 ABC stainless steel (n=2) 3. center, stainless steel (n=2) 1.7 ± 0.4 4. end (vertical portion), stainless 1.4 ± 0.3 ABC ABCD 3.1 ± 3 3.1 ± 0.1 BC ABCD 2.9 ± 0.2 2.3 ± 0.2 BCDEF CDEFG 2.7 ± 0.8 1.7 ± 0.1 ABC ABC ABC ABC steel (n=1) 5. end (slanted portion), stainless 0.9 ± 0.9 BCD 3.2 ± 0.2 ABCD 2.0 ± 0.5 FG 3.0 ± 1.6 ABC ABC steel (n=1) 6. recirculation spout, stainless steel 1.0 ± 1.3 CD 0.8 ± 1.2 E 0.4 ± 0.9 H 2.3 ± 2.5 ABC BC (n=2) Dewatering Centrifuge 1. interior basket carrier, stainless 2.3 ± 0.3 ABC 2.4 ± 1.5 CD 2.7 ± 0.2 CDEF 3.2 ± 1.0 A A steel (n=4) 2. interior wall, stainless steel (n=2) 1.7 ± 0.2 3. viewing window in the lid, plastic 0.9 ± 0.6 ABCD BCD 3.1 ± 0.2 3.1 ± 0.3 (n=1) 84 ABCD ABCD 2.1 ± 0.3 3 ± 1.6 F H 3.1 ± 0.8 3.1 ± 1.9 ABC ABC ABC ABC Table 2.6 (cont’d) 4. drain, stainless steel (n=1) a 3.1 0.3 A 0.9 ± 0.8 DE 3.9 ± 0.1 ABCD 3.5 ± 0.7 ABC 3.5 ± 0.3 A One over the detection limit was used to calculate the mean log value when a sample did not yield any colonies by direct plate count. Means of the same product with different letters are significantly different (P ≤ 0.05). The numbers preceding the equipment location descriptions correspond with locations identified in Figures 2.1 through 2.5. 85 Table 2.7: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 2 log CFU/g (n=5) Mean ± SD E. coli O157:H7 recovered (log CFU/100 cm2) Location on Processing Equipment a Baby Spinach Romaine Lettuce Iceberg Lettuce 1. front, stainless steel (n=2) ND 1.3 ± 0.6 1.6 ± 0.6 2. ND 0.9 ± 1.1 3. feed belt guide, stainless steel (n=4) ND 0.0 ± 1.2 4. below the cutting wheel, stainless steel (n=1) ND 0.2 ± 0.8 5. cutting wheel, stainless steel (n=1) ND 0.9 ± 0.4 6. discharge chute, stainless steel (n=4) ND 1.8 ± 0.6 ND 1.1 ± 1.1 Shredder feed belt, plastic (n=2) ABC ABCD DEF BCDEFG ABCDE A 1.3 ± 0.6 1.0 ± 1.2 1.3 ± 0.6 1.0 ± 0.6 1.5 ± 0.6 ABCD ABCDEF ABCDEF ABCDEFG ABCDEFGH ABC Conveyer Belt 1. conveyer belt, polyurethane (n=4) 86 ABC 1.8 ± 0.5 A Table 2.7 (cont’d) 2. belt guide, plastic (n=4) ND 1.0 ± 0.9 ABC 1.7 ± 0.5 AC Flume Tank 1. slanted front end, stainless steel (n=1) 1.9 ± 1.0 2. interior wall, stainless steel (n=10) -0.7 ± 0.8 A C 0.3 ± 1.2 -1.1 ± 0.3 BCDEFG G 1.5 ± 0.2 -0.4 ± 1.0 ABCDEF H Shaker table 1. front, stainless steel (n=1) -0.3 ± 1.3 2. perforated dewatering screen, stainless steel (n=2) -0.4 ± 1.3 3. center, stainless steel (n=2) 0.1 ± 1.2 4. end (vertical portion), stainless steel (n=1) -0.5 ± 1.0 5. end (slanted portion), stainless steel (n=1) 0.1 ± 1.3 6. recirculation spout, stainless steel (n=2) -1.0 ± 0.7 Dewatering Centrifuge 87 BC BC BC BC ABC C -0.8 ± 0.8 0.0 ± 0.8 0.1 ± 0.9 -0.8 ± 0.7 -1.1 ± 0.1 -1.0 ± 0.5 EFG BCDEFG CDEF EFG FG FG -0.1 ± 0.7 0.5 ± 1.0 0.6 ± 1.0 -0.3 ± 0.8 0.2 ± 1.6 -0.2 ± 1.3 DEFGH ABCDEFGH BDEFG FGH BCDEFGH GH Table 2.7 (cont’d) 1. interior basket carrier, stainless steel (n=4) 2. interior wall, stainless steel (n=2) 0.0 ± 0.9 3. viewing window in the lid, plastic (n=1) -0.9 ± 0.5 4. drain, stainless steel (n=1) a 0.5 ± 0.7 0.8 ± 1.2 AB BC BC AB 0.6 ± 1.1 -0.1 ± 0.8 -1.0 ± 0.0 1.8 ± 0.6 BCDE DEFG FG AB 1.0 ± 0.7 0.0 ± 0.9 -0.1 ± 1.1 1.7 ± 0.5 ABCDEFG EFGH EFGH ABCDE One over the detection limit was used to calculate the mean log value when a sample did not yield any colonies by direct plate count. Means of the same product with different letters are significantly different (P ≤ 0.05). The numbers preceding the equipment location descriptions correspond with locations identified in Figures 2.1 through 2.5. 88 CHAPTER 3: Transfer of Escherichia coli O157:H7 from Equipment Surfaces to Iceberg and Romaine Lettuce during Pilot-Plant Production of Fresh-Cut Leafy Greens 89 3.1 ABSTRACT Transfer of Escherichia coli O157:H7 during commercial production of ready-to-eat salad greens has become a public health concern due to several large foodborne outbreaks. The goal of this study was to quantify E. coli O157:H7 transfer from product-inoculated equipment surfaces to uninoculated lettuce during post-harvest processing. Uninoculated cored heads of iceberg and Romaine lettuce (22.7 kg) were processed using a commercial shredder, step conveyor, 3.3-m flume tank with sanitizer-free tap water, shaker table and centrifugal dryer, followed by 22.7 kg of product that was dip-inoculated to contain ~6, 4, or 2 log CFU/g of a 4strain avirulent, GFP-labeled, ampicillin-resistant E. coli O157:H7 cocktail. After draining the flume tank and refilling the holding tank with tap water, 90.8 kg of uninoculated product was similarly processed and collected in ~5 kg aliquots. After processing, 42 equipment surface samples along with 46 iceberg or 36 romaine lettuce samples (25 g each) from the collection baskets were quantitatively examined for GFP-labeled E. coli O157:H7 by direct plating or membrane filtration using trypticase soy agar containing 0.6% yeast extract and 100 ppm ampicillin. Initially, greatest E. coli O157:H7 transfer was seen from inoculated lettuce to the shredder and conveyer belt, with all equipment surface populations decreasing 90 - 99% after 6 4 processing 90.8 kg of uncontaminated product. After processing lettuce containing 10 or 10 E. coli O157:H7 CFU/g followed by uninoculated lettuce, E. coli O157:H7 was quantifiable throughout the entire 90.8 kg of product. At an inoculation level of 2 log CFU/g, E. coli O157:H7 was consistently detected in the first 21.2 and 68.0 kg of previously uninoculated lettuce at 2 - 3 log CFU/100 g and transferred out to 75.6 and 86.9 kg product, respectively. These findings demonstrate the persistence of E. coli O157:H7 on equipment and subsequent contamination of potentially large quantities of uncontaminated product during processing. 90 3.2 INTRODUCTION Changes in food production and consumption patterns are repeatedly blamed for the increasing frequency of produce-associated outbreaks. In the United States between 1998 and 2007, fresh produce was implicated in 684 outbreaks that included 26,735 cases of illness (CSPI 2009). Norovirus was linked to 51% of these outbreaks, whereas Salmonella and E. coli O157:H7 were respectively responsible for 17 and 7% of the outbreaks (CSPI 2009). The food industry has evolved from many local producers to a far smaller number of large-scale operations with national distribution patterns. This change means that one contamination event at a large centralized grower or processor can have far-reaching consequences. This was the case in September of 2006, when contaminated pre-bagged baby spinach was linked to an E. coli O157:H7 outbreak in 26 states and Canada that resulted in 205 illnesses and five deaths (Jay and others 2007). Only three months later, two additional nationwide E. coli O157:H7 outbreaks were traced to California-grown iceberg lettuce that was shredded and then purchased by two fast-food Mexican restaurant chains in the Midwest and Northeast (FDA 2006, FDA 2007), with these two outbreaks involving over 150 individuals (CSPI 2009). Leafy greens are prone to contamination throughout the farm-to-fork continuum (Beuchat 2002), including during growth in the field (Erickson and others 2010, Solomon and others 2002), harvesting (Taormina and others 2009), post-harvest handling and processing (Kaneko and others 1999, Stafford and others 2002), distribution (Beuchat and Ryu 1997), and preparation in the home or food service settings (Chai and others 2008, Chen and others 2001, Montville and others 2001, Ravishankar and others 2010). Even though E. coli O157:H7 contamination from fecal material is presumed to be highly localized in the field, this organism can be spread easily to previously pathogen-free produce during and after harvest, thus 91 intensifying what was once a small contamination event (Beuchat and Ryu 1997, Burnett and Beuchat 2001). The introduction and spread of bacterial pathogens during commercial leafy green processing remains a major industry concern. In May and June of 2001, 10 cases of Salmonella Bovismorbificans infection phage type 32 were traced to commercially shredded lettuce in Australia (Stafford and others 2002). Investigators eventually recovered the outbreak strain of S. Bovismorbificans from the back rim of the lettuce shredder cutting wheel, with improper maintenance of the shredder being a key factor in this outbreak. More recently, shredded and pre-packaged lettuce from a Dutch processing plant was traced to an E. coli O157 outbreak that sickened a total of 50 individuals in the Netherlands and Iceland (Friesema and others 2008). This outbreak, which again was likely due to the spread of the contaminant during processing, exemplifies the problems that can occur when a contaminated product is simultaneously distributed across borders. Despite these recent large-scale outbreaks, the extent of bacterial transfer and crosscontamination that can occur during commercial processing of leafy greens remains poorly understood, with most bacterial transfer work previously focused on slicing of deli meats (Keskinen and others 2008a, Keskinen and others 2008b, Lin and others 2006, Sheen and Hwang 2008, Vorst and others 2006), grinding of beef (Farrell and others 1998, Floras and Tamplin 2002, Floras and Stewart 2004), and various food service and consumer food preparation/handling practices (Chai and others 2008, Chen and others 2001, Fravalo and others 2009, Gill and Jones 2002, Montville and others 2001, Ravishankar and others 2010, Scott and Bloomfield 1990). In limited work with E. coli O157:H7-inoculated leafy greens, our laboratory demonstrated that approximately 90% of the inoculum was shed in sanitizer-free 92 water, with this pathogen also contaminating the product contact surfaces of a shredder, conveyor, flume tank, shaker table and dewatering centrifuge to various degrees during processing (Buchholz and others 2012b). When researchers in Japan examined factories that processed ready-to-eat fresh vegetables, samples from equipment surfaces used for trimming, washing, slicing, soaking, dehydrating, blending, packaging, the surfaces of workers’ plastic gloves and from the floor and air of processing rooms proved to be most heavily contaminated (Kaneko and others 1999). Elsewhere in the food chain, E. coli O157:H7 was shown to transfer from inoculated coring blades to multiple heads of iceberg lettuce during field coring (Taormina and others 2009), with this pathogen also shown to remain viable on dry stainless steel surfaces for up to 4 days (Kusumaningnum and others 2003). The overall goal of this study was to determine the numbers of E. coli O157:H7 transferred from inoculated to uninoculated iceberg and Romaine lettuce during small-scale processing, which included shredding, conveying, fluming/washing, shaker table dewatering and centrifugal drying, with these findings needed in the development of more accurate risk assessments for leafy greens. 93 3.3 MATERIALS AND METHODS 3.3.1 Experimental design. Transfer of E coli O157:H7 from various surfaces of a small-scale commercial leafy green processing line to iceberg and Romaine lettuce was assessed by first processing 22.7 kg of uninoculated lettuce to prime the processing line, followed by 22.7 kg of lettuce inoculated with E. coli O157:H7 at 106 and 104 (3 replicates each) or 102 CFU/g (5 replicates). After draining and refilling the water recirculation tank with tap water, 90.8 kg of uninoculated lettuce was processed, collected in a series of baskets and quantitatively examined for E. coli O157:H7 as described below. 3.3.2 Leafy greens. Individually wrapped heads of iceberg (Lactuca sativa L.) (24 heads per case) and Romaine (Lactuca sativa L. var. longifolia) (12 heads per case) lettuce were obtained from a local wholesaler (Stan Setas Produce Co., Lansing, MI), stored in a 4°C walk-incooler and used within 5 days of delivery. Products originated from California or Arizona, depending on the growing season. All heads were hand-cored immediately before use. 3.3.3 Bacterial strains. Four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) were obtained from Dr. Michael Doyle at the Center for Food Safety, University of Georgia, Griffin, GA. Upon arrival, stock cultures of each strain were grown in trypticase soy broth (Difco, Becton Dickinson, Sparks, MD) containing 0.6% yeast extract (Difco, Becton Dickinson) supplemented with 100 ppm ampicillin (ampicillin sodium salt, Sigma Life Science, St. Louis, MO) (TSBYE + amp). After 24 h of incubation at 37oC, 10% (v/v) glycerol (Sigma Chemical Company, St. Louis, MO) was added to these cultures which were then stored at -80°C until needed. Working cultures were prepared by streaking each stock culture on trypticase soy agar (Difco, Becton Dickinson) containing 0.6% yeast extract and 100 ppm ampicillin (TSAYE + amp). After 18 - 24 h of 94 incubation at 37°C, cells from a single colony were transferred to 9 ml of TSBYE + amp and similarly incubated. Thereafter, a 0.2 ml aliquot of each culture was transferred to 200 ml of TSBYE + amp, incubated for 18 - 20 h at 37°C and then combined in equal volumes to obtain the 4-strain cocktail. 3.3.4 Inoculation of iceberg and Romaine lettuce. E. coli O157:H7 suspensions containing 107, 105 and 103 CFU/ml were respectively obtained by adding 800, 8 and 0.08 ml of the 4-strain cocktail to 75 L of tap water (~15oC) in a 121 L plastic container (Rubbermaid, Wooster, OH). Thereafter, 22.7 kg of iceberg (~24 heads) or Romaine lettuce (~48 heads) was immersed in the E. coli suspension for 15 min, removed, spun in a 50-lb capacity centrifugal Spin Dryer (model SD50-LT, Heinzen Manufacturing, Inc., Gilroy, CA), with one internally timed spin cycle totaling ~20 sec to remove the residual inoculum, and then drained/air-dried for 1 h at 22°C before processing. Using sterilized tongs, two representative 25-g samples were taken from two inoculated lettuce heads to determine the average inoculation level at the time of processing. 3.3.5 Processing equipment. The pilot-scale size leafy green processing line housed in the Department of Food Science and Human Nutrition consisted of a lettuce shredder, conveyer, flume tank, shaker table, and water recirculation tank. The commercial lettuce shredder (model TRS 2500 Urschel TranSlicer, Valparaiso, IN) operated at a feed belt speed of 198 m/min and slicing wheel speed of 905 RPM, which yielded shreds of lettuce measuring approximately 5 x 5 cm. The motorized conveyor (Dorner model 736018 mc series, Hartland, WI), which was equipped with a smooth polyurethane step conveyer belt (ThermoDrive, Mol Industries), operated at 0.11 m/sec. The stainless steel water recirculation tank (~1000 L capacity) was filled with 890 L of sanitizer-free tap water (~15oC). This water was centrifugally pumped (model 95 XB754FHA, Sterling Electric, Inc., Irvine, CA) at a rate of ~15 L/sec through a 4.14 m long, 10cm diameter hard plastic connector hose to a 3.6-m long stainless steel flume tank (Heinzen Manufacturing, Inc.) that was equipped with two overhead spray jets located 1 m from the start of the flume tank. The stainless steel shaker table for partial dewatering was equipped with a wash down duty motor (1 HP, Baldor Electric Co., Ft. Smith, AR) that operated at 1760 rpm. Water exiting the flume tank passed through a screen on the shaker table into a water recirculation spout, which returned the water to the holding tank for recirculation. 3.3.6 Iceberg and Romaine lettuce processing and sample collection. Cored heads of uninoculated iceberg and Romaine lettuce (22.7 kg) were fed into the shredder for continuous processing followed by 22.7 kg of inoculated lettuce. The lettuce was then step-conveyed to the flume tank and washed, with ~5 kg aliquots of the partially dried product exiting the shaker table then collected in each of ~12 plastic mesh laundry baskets (United Plastics, Leomister, MA). After draining the flume and recirculation tanks, 11 previously identified (Buchholz and others 2012b) product contact surface samples (100 cm2) on the equipment (3 shredder, 2 conveyer belt, 3 flume tank and 3 shaker table samples) were collected using one-ply composite tissues (Kimwipes®, Kimberly-Clark Corp., Irving, TX) moistened in 1 ml of sterile 1% (w/v) phosphate buffer (8.5 g/L NaCl, 1.44 g/L Na2HPO4, and 0.24 g/L KH2PO4, J.T. Baker, Mallinckrodt Baker Inc., Phillipsburg, NJ) as described by Vorst et al. (2006). The water recirculation tank was then refilled with ~890 L of sanitizer-free tap water (~15oC), after which 90.8 kg of uninoculated lettuce was similarly processed. During processing, 200-ml aliquots of water (~13 from iceberg and ~18 Romaine from lettuce processing) were collected from the water return spout on the shaker table. After processing, the 90.8 kg of previously uninoculated lettuce was collected in ~24 baskets for later enumeration of E. coli O157:H7. Finally, all 50 product contact surface 96 areas (14 shredder, 8 conveyer belt, 11 flume tank and 9 shaker table samples) previously identified by Buchholz et al. (2012b) using Glo Germ™ (Glo Germ Company, Moab, Utah) were sampled using one-ply composite tissues moistened with 1 ml of sterile phosphate buffer. 3.3.7 Microbiological analyses. Romaine and iceberg lettuce samples (25 g) composited from 5 separate locations within each of the ~24 baskets were added to 100 ml of sterile phosphate buffer in a Whirl-Pak™ filter bag (Nasco, Fort Atkinson, WI) and pulsified (Pulsifier, Filtaflex Ltd., Almonte, Ontario, Canada) for 1 min. Lettuce samples were either serially diluted in sterile phosphate buffer and plated on TSAYE + amp or processed by filtration using 0.45 μm membrane filters (Millipore, Millipore Corporation, Billerica, MA) (lower detectable limit of 0.04 CFU/g), which were placed on 60-mm dia. Petri plates containing TSAYE + amp to quantify E. coli O157:H7. The one-ply composite tissue samples were added to 15 ml of phosphate buffer in a Whirl-Pak™ bag, homogenized by stomaching (Stomacher 400 Circulator, Seward, Worthington, UK) for 1 min at 260 rpm and then plated on TSAYE + amp as well, giving a lower detectable limit of 1 CFU/100 cm2. The 200 ml water samples were either appropriately diluted and plated on TSAYE + amp or processed by membrane filtration, which gave a minimum detectable limit of 0.01 CFU/ml. After 18 – 20 h of incubation at 37°C, all green fluorescing colonies as seen under ultraviolet light (Blak-Ray, Ultra-violet Product Inc., San Gabriel, CA) were counted as E. coli O157:H7. 3.3.8 Statistical analyses. Bacterial counts were converted to log CFU per gram, ml or 100 cm2 to determine the numbers of E. coli O157:H7 transferred to uncontaminated product, equipment surfaces, and water, respectively, during processing of iceberg and Romaine lettuce. An ANOVA with the Tukey-Kramer HSD test were then performed using JMP 8.0 (SAS 97 Institute Inc., Cary, NC) to determine statistically significant differences (P < 0.05) in the numbers of E. coli O157:H7 transferred between the equipment surfaces, lettuce and water. 98 3.4 RESULTS 3.4.1 Initial inoculation levels. The three four-strain E. coli O157:H7 cocktails used to inoculate iceberg lettuce contained 6.98 ± 0.07, 4.98 ± 0.11 and 2.96 ± 0.24 log CFU/ml, and the cocktails for Romaine lettuce contained 7.07 ± 0.47, 5.96 ± 0.15 and 3.37 ± 1.16 log CFU/ml. After immersing for 15 min followed by 1 h of air drying and 20 sec of centrifugal drying to remove excess inoculum, iceberg lettuce contained E. coli O157:H7 populations of 6.08 ± 0.63, 4.14 ± 0.45 and 2.21 ± 0.71 log CFU/g, and Romaine lettuce contained 6.44 ± 0.59, 4.47 ± 0.65 and 1.98 ± 0.18 log CFU/g (Figure 3.1, 3.2 and 3.3). No significant differences in inoculation levels (P > 0.05) were seen between product types. 3.4.2 Flume water. When 22.7 kg of lettuce containing three different levels of E. coli O157:H7 was processed, approximately 90% of the inoculum transferred to the sanitizer-free water. This water, which was subsequently drained from the flume and water recirculation tanks to avoid re-contaminating the uninoculated product to follow, contained E. coli O157:H7 populations of 2.23, 2.21 and -0.27 log, and 2.95, 1.94 and -0.25 log CFU/ml after processing iceberg and Romaine lettuce inoculated at the three levels, respectively. After refilling the recirculation tank with ~890 L of tap water (~15oC) this same water yielded E. coli O157:H7 populations of -1.40, 0.01 and -2.18 log CFU/ml, and 1.98, 0.12 and -1.61 log CFU/ml after processing the remaining 90.8 kg of uninoculated iceberg and Romaine lettuce, respectively. 3.4.3 Processing equipment surfaces. After processing, populations of E. coli O157:H7 were higher (P < 0.05)on the conveyer belt for both products at all three inoculation levels as well as on the shredder after processing Romaine and iceberg lettuce at the medium and low inoculation levels, respectively. Statistically higher (P < 0.05) E. coli populations were seen on the polyurethane conveyer belt compared to the other equipment surfaces (Tables 3.1, 3.2, and 99 3.3). Overall, E. coli O157:H7 populations decreased 0.24 to 2.51 and 0.11 to 1.98 log CFU/100 cm2 on equipment surfaces after processing the uninoculated iceberg and Romaine lettuce, respectively, with greatest losses seen for the shredder and shaker table (Table 3.4). Based on these observations, processing additional uninoculated product would likely have resulted in further cross-contamination from the equipment. 3.4.4 Lettuce. GFP-labeled E. coli O157:H7 was sporadically recovered from uninoculated lettuce used to prime the processing line immediately, before introducing inoculated product. This phenomenon can be explained by pieces of inoculated lettuce sporadically exiting the processing line with uninoculated lettuce since not all shreds of lettuce moved through the line at the same rate. In addition, some GFP-labeled cells of E. coli O157:H7 likely transferred from the inoculated lettuce to the recirculating wash water and then to the last portion of uninoculated lettuce exiting the processing line. When 22.7 kg of lettuce inoculated at 2 log CFU/g was processed followed by 90.8 kg of uninoculated lettuce, E. coli O157:H7 th transferred to the 78 - 82 nd th th kg of uninoculated iceberg lettuce (the 127 - 131 kg of total processed iceberg lettuce) at -1.29 log CFU/g, whereas at inoculation levels of 4 and 6 CFU/g, the last kg of previously uninoculated product processed (the 91st kg) yielded E. coli O157:H7 populations of 0.34 and 1.51 log CFU/g, respectively. Regardless of inoculation level, contamination was seen throughout the uninoculated Romaine lettuce after processing. E. coli O157:H7 counts after Romaine lettuce processing were 2.90, 0.72 and -1.10 log CFU/g for the high, medium, and low inoculation levels, respectively. Differences in E. coli O157:H7 contamination levels were observed between replicates for each inoculation level; however, the distribution followed the same pattern (Figure 3.4, 3.5, and 3.6). 100 A 2 log reduction was observed after 6.98 and 16.21 kg of uninoculated Romaine lettuce was processed in the high and medium inoculation level experiments, respectively. Similarly, in the high and medium inoculation level experiments, 5.92 and 8.88 kg of uninoculated iceberg lettuce needed to be processed to achieve a 2 log reduction, respectively. However, E. coli O157:H7 reductions of only a 1.74 and 1.98 log were observed in the low inoculation level experiments using Romaine and iceberg lettuce, respectively. After processing uninoculated followed by inoculated product, mean recovery was significantly greater (P > 0.05) for Romaine as compared to iceberg lettuce regardless of the inoculation level. 101 3.5 DISCUSSION Quantitative bacterial transfer has been previously assessed during field coring of iceberg lettuce (Taormina and others 2009), production of ground beef (Farrell and others 1998, Floras and Tamplin 2002, Floras and Stewart 2004), and slicing of delicatessen meats (Keskinen and others 2008a, Keskinen and others 2008b, Lin and others 2006, Sheen and Hwang 2008, Vorst and others 2006). In one study modeling the distribution of E. coli O157:H7 during commercial grinding of inoculated followed by uninoculated beef, results were obtained using initial inoculums of 2, 4 and 6 log CFU/g with E. coli detected in 12.7, 32.1, and 86.2% of ~6600 g of beef trim, respectively (Floras and Stewart 2004). In terms of leafy greens, Taormina et al. (2009) at the University of Georgia demonstrated that E. coli O157:H7 can transfer to successive uncontaminated heads of iceberg lettuce during harvesting. When field coring implements were inoculated through contact with contaminated soil containing E. coli O157:H7 at 2.72 and 1.67 log CFU/g and then used to cut and core 10 lettuce heads, the pathogen transferred to 10 and 5 consecutively processed heads, respectively, again demonstrating a direct correlation between increasing bacterial load and subsequent transfer. This, however, is the first study to systematically quantify the extent of product crosscontamination for key unit operations, including shredding, conveying, fluming, shaker table dewatering, and centrifugal drying, that are found in most commercial processing lines for production of fresh-cut salad greens. In this work, three inoculation levels were evaluated - 6, 4 and 2 log CFU/g, which represent high, medium, and low bacterial loads on leafy greens, respectively. One study conducted at Rutgers University showed that as the inoculum size increases, the number of bacteria transferred remains constant and the apparent percent transfer rate will decrease (Montville and Schaffner 2003). In their study focusing on cross- 102 contamination between chicken, lettuce, cutting boards, hands, and a spigot, a significant correlation was seen between inoculum size and percent transfer during a cross-contamination event. While our starting inocula of 6 and 4 log CFU/g might be considered unusually high, these levels were needed to generate the necessary quantitative data for subsequent predictive models and eventual incorporation into risk assessments for leafy greens. The two higher inoculation levels may represent direct contamination in the field from fecal matter, because cattle are a major source of this pathogen in the environment (Williams and others 2008). Although direct contact with manure is one source of pre-harvest contamination, the same contaminated manure can also be introduced into the farm’s irrigation water, which may be lead to far lower numbers of bacteria being distributed to a much wider area of the field (Beuchat 2002, Ibekwe and others 2004). As expected, the E. coli O157:H7 transfer rates were more variable using lettuce inoculated at 2 log CFU/g, as compared to the two higher levels. A uniform inoculum, as could be achieved only by dipping, was essential for generating repeatable transfer data. While other inoculation methods also exist, Lang et al. (2004) reported that higher E. coli O157:H7 populations were recovered from dip- as opposed to spot- or sprayinoculated lettuce when the same populations were initially applied, because a larger surface area of the leaf could be covered. In our study, we chose to evaluate products held for one hour after dip inoculation to simulate contamination at the processing facility. However, extending the post-inoculation hold time before processing enhances the attachment of E. coli O157:H7 to leafy greens and lowers the transfer rate during processing, as previously reported by Buchholz et al. (2012b). Lang et al. (2004) reported similar findings when comparing E. coli O157:H7 recovery from leafy greens that were dried for either 2 h at 22°C or 2 h at 22°C followed by 22 h 103 at 4°C. Additional work is needed to determine the extent of cross-contamination during processing of field-contaminated leafy greens. In the United States, commercially prepared leafy greens intended for the fresh-cut readyto-eat market undergo multiple washings in sanitizer-containing flume systems to remove soil and decrease the microbial load. However, microbial contaminants can be easily spread to previously uncontaminated product via water and equipment surfaces during processing. Because the goal of this study was to quantify E. coli O157:H7 transfer during leafy green processing, both iceberg and Romaine lettuce were processed without the use of a chemical sanitizer, which is also common practice in Europe (Holvoet and others 2012). If sanitizers had been used, E. coli O157:H7 populations in the flume water would have most likely been reduced to near non-detectable levels, decreasing our ability to quantify the transfer of viable cells to previously uncontaminated product. When iceberg and Romaine lettuce were inoculated to contain ~2 log CFU/g, E. coli was consistently detected in the flume water at levels between -1.5 and -2.0 log CFU/g. However, given an estimated oral infectious dose of less than 100 cells for E. coli O157:H7 (Meng and others 2007), presence of any viable E. coli O157:H7 cells in flume water, as seen in our study, is reason for concern if appropriate intervention strategies are not taken to prevent cross-contamination. The potential for cross-contamination during flume washing was also previously demonstrated by immersing apples in a dump tank (Buchanan and others 1999); however, a few years later a group of USDA researchers showed negligible crosscontamination from inoculated to uninoculated apples via the wash water (Annous and others 2001). Not only was E. coli O157:H7 quantifiable during and after processing of the 22.7 kg of inoculated lettuce, but small numbers of the pathogen were also present in water collected from 104 the shaker table before the inoculated product reached that same point in the process. This phenomenon may be attributed to the tumbling movement of the product through the flume tank, and the velocity differential between the water and the product, with the overhead jets also forcing the leafy greens below the waterline. Since the uninoculated and inoculated products were of the same type and therefore visually indistinguishable, the exact point at which the inoculated leaves exited the shaker table could only be estimated based on knowing that the initial uninoculated and inoculated batches were of equal size at the start. As the inoculation level increased, both the amount of contaminated product and the E. coli O157:H7 populations on the product increased. The design of this study provides an overall estimate of the numbers of E. coli O157:H7 accumulating over time after a contaminated batch of product is processed, rather than the numbers of E. coli O157:H7 exclusively transferred to previously uncontaminated lettuce. Cross contamination of uninoculated leafy greens during processing may have occured through contact with 1) the water (~2 L) remaining in the flume tank, shaker table, centrifugal pump, hose and water recirculation tank after draining, 2) inoculated lettuce clinging to the shredder and conveyer belt surfaces, and 3) any contaminated leafy greens lingering in the processing line. In terms of equipment surfaces, the conveyer belt, shredder discharge chute and front panel of the flume tank were frequent collection points for E. coli O157:H7 after uninoculated product was processed. These locations, which were not exposed to water, were frequently coated with lettuce juice and many small shreds of residual product after processing, providing ideal bacterial harborage sites. This finding was not surprising, since two studies showed that relatively high numbers of Salmonella and Campylobacter were transferred to Romaine lettuce even 1 to 2 h after the food contact surface became contaminated (Moore and others 2003), with 105 Salmonella remaining viable on surfaces for up to four days (Kusumaningnum and others 2003). In our study, some of the small shreds of lettuce lingering on the processing equipment could still be the initial inoculated product, which could break free from the equipment surface as additional product is processed. In a study examining the microbial changes of “Lollo Rosso” lettuce during processing and subsequent storage, an increase of 1 log CFU/g was observed after shredding (Allende and others 2004), reaffirming the role of the shredder in spreading contamination to additional product during processing (Garg and others 1990, Stafford and others 2002). Locations that were constantly exposed to the wash water, such as the interior walls of the flume tank and the water recirculation spout on the shaker table, yielded the lowest E. coli O157:H7 populations after processing. Comparing these findings with data collected after processing 22.7 kg of contaminated iceberg or Romaine lettuce though the same processing line (Buchholz and others 2012b) illustrates how easily this pathogen can persist on food contact surfaces and cross-contaminate product still to be processed, especially if sanitizers are not used or monitored properly. In summary, this study shows that one contaminated batch of leafy greens can easily contaminate subsequent batches of previously uncontaminated product in a processing facility if an effective microbial intervention strategy is not implemented. Such cross-contamination can occur even when E. coli O157:H7 is present in very low levels, with 0.06 and 0.34% of the original 2 log CFU/g inoculum spread to previously uninoculated product after processing. This study supports the importance of sanitizers in wash waters, or other interventions, to adequately reduce the microbial load on the product and in flume water during commercial processing to prevent future product recalls and outbreaks. 106 3.6 FIGURES Figure 3.1: E. coli O157:H7 populations (mean ± SD) on the product during processing of iceberg and Romaine lettuce inoculated at ~6 log CFU/g (n = 3). Iceberg 8 Romaine E. coli O157:H7 (log CFU/g) 7 6 5 4 3 2 1 0 -1 -2 a 0 20 40 60 80 100 120 140 Cumulative Mass (kg) Mean log values for samples without counts by direct plating were calculated as 1/detection limit. 107 a Figure 3.2: E. coli O157:H7 populations (mean ± SD) on the product during processing of iceberg and Romaine lettuce inoculated at ~ 4 logs CFU/g (n = 3). Iceberg 6 Romaine E. coli O157:H7 (log CFU/g) 5 4 3 2 1 0 -1 0 20 40 60 80 100 120 140 -2 -3 -4 a Cumulative Mass (kg) Mean log values for samples without counts by direct plating were calculated as 1/detection limit. 108 a Figure 3.3: E. coli O157:H7 populations (mean ± SD) on the product during processing of iceberg and Romaine lettuce inoculated at ~ 2 logs CFU/g (n = 5). 3 Iceberg E. coli O157:H7 (log CFU/g) Romaine 2 1 0 0 40 60 80 100 120 140 -1 -2 a 20 Cumulative Mass (kg) Mean log values for samples without counts by direct plating were calculated as 1/detection limit. 109 a Figure 3.4: E. coli O157:H7 populations (mean ± SD) in water during processing of iceberg and Romaine lettuce inoculated at ~6 logs CFU/g (n = 3). Iceberg 6 E. coli O157:H7 (log CFU/g) 7 Romaine 5 4 3 2 1 0 -1 0 20 40 60 80 100 120 140 -2 -3 -4 a Cumulative Mass (kg) Mean log values for samples without counts by direct plating were calculated as 1/detection limit. 110 a Figure 3.5: E. coli O157:H7 populations (mean ± SD) in water during processing of iceberg and Romaine lettuce inoculated at ~4 logs CFU/g (n = 3). Iceberg 4 E. coli O157:H7 (log CFU/g) Romaine 3 2 1 0 0 20 40 60 80 100 120 140 -1 -2 -3 a Cumulative Mass (kg) Mean log values for samples without counts by direct plating were calculated as 1/detection limit. 111 a Figure 3.6: E. coli O157:H7 populations (mean ± SD) in water during processing of iceberg and Romaine lettuce inoculated at ~2 logs CFU/g (n = 5). E. coli O157:H7 (log CFU/g) Iceberg 0 0 20 40 60 80 100 120 140 -1 -2 -3 a Romaine 1 Cumulative Mass (kg) Mean log values for samples without counts by direct plating were calculated as 1/detection limit. 112 3.7 TABLES Table 3.1: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 6 log CFU/g (average of 3 replicates) Mean (± SD) E. coli O157:H7 (log CFU/g) Location on Processing Equipment Romaine Lettuce Iceberg Lettuce 1. front, stainless steel (n=2) 3.7 ± 1.8 2.1 ± 0.5 2. 1.2 ± 0.4 2.9 ± 0.4 2.0 ± 1.0 2.8 ± 0.5 1.6 ± 0.7 3.2 ± 1.0 3.8 ± 0.8 3.1 ± 0.7 3.6 ± 0.0 0.6 ± 1.0 Shredder feed belt, plastic (n=2) 3. feed belt guide, stainless steel (n=4) 1.6 ± 0.7 4. below the cutting wheel, stainless steel (n=1) 3.5 ± 1.0 5. cutting wheel, stainless steel (n=1) 3.9 ± 0.6 6. discharge chute, stainless steel (n=4) 4.0 ± 1.3 ABC DE DE ABC ABC AB BCD ABCD D ABCD BCDE ABC Conveyer Belt 1. conveyer belt, polyurethane (n=4) 4.4 ± 0.8 2. belt guide, plastic (n=4) 3.7 ± 0.8 A AB A AB Flume Tank 3. slanted front end, stainless steel (n=1) 4.2 ± 0.8 4. interior wall, stainless steel (n=10) 0.8 ± 0.9 113 ABC E ABCD EF Table 3.1 (cont’d) Shaker table 1. front, stainless steel (n=1) 2.2 ± 0.2 2. perforated dewatering screen, stainless steel 2.7 ± 0.2 BCDE BCD 1.4 ± 0.3 2.4 ± 0.2 2.2 ± 0.6 2.0 ± 0.3 1.9 ± 0.1 CDEF BCD (n=2) 3. center, stainless steel (n=2) 4. end (vertical portion), stainless steel (n=1) 2.6 ± 0.56 5. end (slanted portion), stainless steel (n=1) 1.9 ± 0.5 6. recirculation spout, stainless steel (n=2) a 2.7 ± 0.4 0.5 ± 0.8 BCD BCD CDE E -0.5 ± 0.8 BCD BCDE BCDE F Mean log values for samples without counts by direct plating were calculated as 1/detection limit. Means of the same product not followed by the same letter are significantly different (P ≤ 0.05). b The numbers preceding the equipment location descriptions correspond with locations previously identified by Buchholz et al. (2012b). 114 Table 3.2: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 4 log CFU/g (average of 3 replicates) Mean (± SD) E. coli O157:H7 (log CFU/g) Location on Processing Equipment Romaine Lettuce Iceberg Lettuce 1. front, stainless steel (n=2) 1.9 ± 0.5 0.4 ± 1.0 2. 1.1 ± 1.1 Shredder feed belt, plastic (n=2) 3. feed belt guide, stainless steel (n=4) 1.2 ± 0.6 4. below the cutting wheel, stainless steel (n=1) 1.9 ± 1.1 5. cutting wheel, stainless steel (n=1) 1.3 ± 0.5 6. discharge chute, stainless steel (n=4) 2.5 ± 0.7 ABCD BCD CD ABCD ABCDE AB 1.0 ± 1.1 -0.4 ± 1.0 1.0 ± 0.6 1.0 ± 0.5 0.7 ± 1.3 BCDE ABCD CE ABCD ABCD ABCD Conveyer Belt 1. conveyer belt, polyurethane (n=4) 2.6 ± 0.7 2. belt guide, plastic (n=4) 1.6 ± 1.2 A ABCD 1.7 ± 0.7 0.9 ± 1.1 A ABD Flume Tank 1. slanted front end, stainless steel (n=1) 3.0 ± 0.5 2. interior wall, stainless steel (n=10) -0.3 ± 1.0 Shaker table 115 ABC EF 2.2 ± 0.3 -0.8 ± 0.6 AB E Table 3.2 (cont’d) 1. front, stainless steel (n=1) 1.0 ± 0.2 2. perforated dewatering screen, stainless steel 1.3 ± 0.2 BCDE BCD 0.1 ± 0.2 0.3 ± 0.7 ABCDE BCDE (n=2) 3. center, stainless steel (n=2) 4. end (vertical portion), stainless steel (n=1) 1.6 ± 0.2 5. end (slanted portion), stainless steel (n=1) 0.5 ± 0.6 6. recirculation spout, stainless steel (n=2) a 1.3 ± 0.2 -1.2 ± 0.0 BCD ABCD DEF F 0.3 ± 0.87 0.3 ± 0.2 -0.7 ± 0.9 -1.2 ± 0.0 BCDE ABCDE CDE E Mean log values for samples without counts by direct plating were calculated as 1/detection limit. Means of the same product not followed by the same letter are significantly different (P ≤ 0.05). b The numbers preceding the equipment location descriptions correspond with locations previously identified by Buchholz et al. (2012b). 116 Table 3.3: E. coli O157:H7 populations (mean ± SD) on equipment surfaces during processing of leafy greens inoculated at 2 log CFU/g (average of 5 replicates) Mean (± SD) E. coli O157:H7 recovered (log CFU/g) Location on Processing Equipment Romaine Lettuce Iceberg Lettuce 1. front, stainless steel (n=2) -0.4 ± 0.9 -1.2 ± 0.0 2. -0.7 ± 0.8 -0.9 ± 0.5 3. feed belt guide, stainless steel (n=4) -0.6 ± 0.9 -1.2 ± 0.0 4. below the cutting wheel, stainless steel -0.3 ± 1.3 -1.2 ± 0.0 5. cutting wheel, stainless steel (n=1) -0.9 ± 0.7 -1.2 ± 0.0 6. discharge chute, stainless steel (n=4) 0.5 ± 1.3 -0.9 ± 0.7 1. conveyer belt, polyurethane (n=4) 0.4 ± 1.1 -0.6 ± 0.8 2. belt guide, plastic (n=4) 0.2 ± 1.1 -1.1 ± 0.5 1. slanted front end, stainless steel (n=1) 0.4 ± 1.0 -0.7 ± 0.6 2. interior wall, stainless steel (n=10) -1.1 ± 0.4 -1.2 ± 0.0 Shredder feed belt, plastic (n=2) ABC ABC BC ABC AB AB B AB (n=1) ABC A AB AB Conveyer Belt A AB A B Flume Tank 117 ABC C AB B Table 3.3 (cont’d) Shaker table 1. front, stainless steel (n=1) -0.5 ± 0.9 2. perforated dewatering screen, stainless -0.5 ± 0.9 3. center, stainless steel (n=2) -0.7 ± 1.1 4. end (vertical portion), stainless steel -0.4 ± 1.1 -1.2 ± 0.0 6. recirculation spout, stainless steel (n=2) -1.2 ± 0.0 ABC ABC -1.2 ± 0.0 -1.2 ± 0.0 -1.1 ± 0.4 -0.9 ± 0.5 -1.2 ± 0.0 -1.2 ± 0.0 AB B steel (n=2) ABC ABC AB AB (n=1) 5. end (slanted portion), stainless steel BC AB (n=1) a C B Mean log values for samples without counts by direct plating were calculated as 1/detection limit. Means of the same product not followed by the same letter are significantly different (P ≤ 0.05). b The numbers preceding the equipment location descriptions correspond with locations previously identified by Buchholz et al. (2012b). 118 a Table 3.4: E. coli O157:H7 populations (mean ± SD) recovered from equipment surfaces before and after processing 90.7 kg of uninoculated leafy greens Mean (± SD) E. coli O157:H7 (log CFU) Product Inoculation Level Romaine Lettuce 6 log CFU/g 7.8 6.6 ± 7.0 1.7 ± 0.8 8.6 ± 8.4 7.3 ± 7.3 1.3 ± 1.1 6.6 ± 7.0 6.0 ± 6.8 0.6 ± 0.2 6.9 ± 6.8 4.9 ± 5.0 2.0 ± 1.8 Shredder 6.7 ± 6.8 4.7 ± 5.0 2.0 ± 1.8 6.9 ± 6.9 5.4 ± 5.6 1.5 ± 1.3 Flume Tank 4.7 ± 4.8 4.4 ± 5.1 0.3 ± -0.3 Shaker Table c ± Conveyor Belt 2 log CFU/g 8.3 Shaker Table b Shredder Flume Tank 4 log CFU/g Before Conveyor Belt b Processing Step 5.3 ± 5.1 3.6 ± 3.6 1.7 ± 1.5 Shredder 4.2 ± 4.1 3.4 ± 3.8 0.9 ± 0.3 Conveyor Belt 4.4 ± 4.2 3.5 ± 3.7 1.0 ± 0.5 Flume Tank 2.5 ± 2.8 2.4 ± 3.0 0.1 ± -0.2 119 After Reduction Table 3.4 (cont’d) Shaker Table 3.5 0.2 ± -0.6 Shredder 7.5 ± 7.8 5.7 ± 6.2 1.8 ± 1.6 7.8 ± 8.0 6.8 ± 7.0 1.0 ± 1.0 5.8 ± 5.8 4.6 ± 5.3 1.3 ± 0.6 7.0 ± 7.1 4.6 ± 4.7 2.4 ± 2.4 Shredder 7.4 ± 7.7 3.7 ± 4.1 3.7 ± 3.6 5.6 ± 5.5 4.6 ± 5.0 1.0 ± 0.5 Flume Tank 3.9 ± 3.9 3.7 ± 4.2 0.2 ± -0.3 Shaker Table 5.2 ± 5.1 2.7 ± 2.7 2.5 ± 2.4 Shredder 3.2 ± 3.3 1.8 ± 2.4 1.4 ± 0.9 Conveyor Belt 3.7 ± 3.9 2.4 ± 2.7 1.3 ± 1.2 Flume Tank 3.0 ± 3.3 1.4 ± 1.7 1.6 ± 1.6 Shaker Table a ± Conveyor Belt 2 log CFU/g 2.8 Shaker Table 4 log CFU/g 2.9 Flume Tank 6 log CFU/g ± Conveyor Belt Iceberg Lettuce 3.1 3.3 ± 3.6 1.4 ± 1.7 1.9 ± 1.9 2 Total product contact surface areas for the shredder, conveyer belt, flume tank and shaker table are 1.74, 2.88, 2.60, and 2.47 m , respectively. 120 Table 3.4 (cont’d) b n = 3 for 6 and 4 log CFU/g c n = 5 for 2 log CFU/g 121 CHAPTER 4: Tracking an Escherichia coli O157:H7 Contaminated Batch of Leafy Greens through a Commercial Processing Line 122 4.1 ABSTRACT Cross-contamination of fresh-cut leafy greens with small quantities of Escherichia coli O157:H7-contaminated product during commercial processing is likely to be at least partially responsible for several recent multistate outbreaks. In this study, Radicchio (9.1 kg) was dip6 inoculated to contain 10 CFU/g of a 4-strain avirulent, GFP-labeled, ampicillin-resistant E. coli O157:H7 cocktail and then used as a visual marker to track and quantify the spread of E. coli O157:H7 to iceberg lettuce in a pilot-scale processing line that included a commercial shredder, step conveyor, flume tank, shaker table and centrifugal dryer. After priming this line with 45 kg of uninoculated iceberg lettuce, the inoculated Radicchio was processed followed by 907 kg of 2 iceberg lettuce. Forty water samples (500 ml), 50 equipment swabs (100 cm ) and ~38 bags of lettuce/Radicchio (~22.7 kg/bag) were collected. All visible shreds of Radicchio were retrieved from the bags of shredded product, the equipment, and the floor and were weighed and counted. Twenty-five grams of iceberg lettuce was collected from each bag of product and from 10 of the bags after centrifugal drying. All samples were examined for E. coli O157:H7 by direct plating or membrane filtration with trypticase soy agar containing 0.6% yeast extract and 100 ppm ampicillin. After processing, the weight of inoculated Radicchio in shredded lettuce averaged 614.9 g (93.6 %), 6.9 g (1.3 %), 5.0 g (0.8 %) and 2.8 g (0.5 %) for bags 1 - 10, 11 - 20, 21 - 30, and 31 – 40, respectively. These same bag groupings contained average E. coli O157:H7 populations of 1.69, 1.22, 1.10 and 1.11 log CFU/g. After processing, shreds of Radicchio were most prevalent on the conveyor (9.8 g) followed by the shredder (8.3 g), flume tank (3.5 g) and shaker table (0.1 g), showing that contaminated product can continually spread during leafy green processing long after the contamination event. 123 4.2 INTRODUCTION Leafy greens from the Salinas Valley of California have been implicated in nine Escherichia coli O157:H7 outbreaks from 1995 to 2006, with the last three receiving considerable national attention. In September 2006, contaminated pre-bagged baby spinach triggered an Escherichia coli O157:H7 outbreak in 26 states and Canada, resulting in five deaths and 205 illnesses (Jay and others 2007). Only a few months later, two additional E. coli O157:H7 outbreaks were traced to California-grown iceberg lettuce that was shredded and then purchased by two fast-food Mexican restaurant chains in the Midwest and Northeast (FDA 2006; FDA 2007). These two outbreaks sickened a combined total of over 150 individuals (CSPI 2009a). Changes in food production and consumption patterns are repeatedly blamed for the increasing frequency of produce-associated outbreaks. Evolution of this industry from many small local to far fewer large-scale operations now means that one contamination event at a large centralized grower or processor can have a lasting negative impact on the industry as seen by the spinach outbreak that decimated baby spinach sales for months (Calvin 2007). Leafy greens are prone to contamination throughout the farm-to-fork continuum. Preharvest sources include feces from wild and domestic animals, water runoff from livestock operations, green or improperly composted manure, air, dust, insects, and irrigation water. Pathogens can also come in contact with fresh produce through field workers, harvesting equipment, hand coring/trimming, water used for cooling/washing, improperly cleaned crates, totes and storage bins and the vehicles used for transport (Beuchat 1996; Beuchat and Ryu 1997; Gorny 2006). A significant portion of these leafy-greens become fresh-cut ready-to-eat products that place consumers at increased risk due to the potential spread of microbial pathogens during commercial processing (Beuchat and Ryu 1997; Burnett and Beuchat 2001). This situation is 124 exacerbated by the fact that most commonly used commercial sanitizers are only capable of decreasing the microbial load 90 – 99% on the product during flume washing (Beuchat and others 2004; Burnett and others 2004; Keskinen and others 2009; Sapers 2001; Weissinger and others 2000; Zhang and others 2009). Shredded lettuce can easily remain in a leafy green processing line and contaminate subsequent product as is supported by a 2001 outbreak of Salmonella Bovismorbificans in Australia that sickened 10 people. This outbreak was later attributed to residual shredded lettuce that was found behind the cutting wheel of the shredder during an environmental audit (Stafford and others 2002). Previous work in our laboratory has confirmed that E. coli O157:H7 can readily transfer to the product contact surfaces of a leafy green processing line (Buchholz and others 2012b) and persist on these equipment surfaces, subsequently contaminating large quantities of previously uncontaminated product during processing (Buchholz and others 2012d). The goals of this study were to track an E. coli O157:H7-inoculated batch of Radicchio through a pilot-scale processing line to determine: 1) the numbers of E. coli O157:H7 transferred to a subsequently processed much larger batch of iceberg lettuce as well as to the equipment and water during processing, and 2) the amount of inoculated Radicchio transferred to the shredded lettuce and the various pieces of equipment after processing. 125 4.3 MATERIALS AND METHODS 4.3.1 Experimental design. Initially 45.4 kg of uninoculated iceberg lettuce was processed in a small-scale commercial processing line, which included shredding, conveying, fluming/washing, shaker table dewatering and centrifugal drying. Thereafter, Radicchio heads 6 inoculated at 10 CFU/g E. coli O157:H7 and held for 1 h at 22°C were similarly processed followed by 907.2 kg of uninoculated lettuce, with the red color of Radicchio allowing easy differentiation between inoculated and uninoculated product. All experiments were conducted in triplicate. 4.3.2 Produce. Pallets of pre-cored iceberg lettuce (Lactuca sativa L.) heads (~250 kg/pallet) were received from Aunt Mid’s Produce Company (Detroit, MI) and stored at 4°C at Michigan State University Food Stores and used with 2 days of delivery. Radicchio heads (Cichorium intybus) (12 heads per case) were obtained from a local wholesaler (Stan Setas Produce Co., Lansing, MI), stored at 4°C in a walk-in-cooler and used within 5 days of delivery. Both products originated from California or Arizona, depending on the growing season. 4.3.3 Bacterial strains. Four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) were obtained from Dr. Michael Doyle at the Center for Food Safety, University of Georgia, Griffin, GA. Upon arrival, stock cultures of each strain were prepared in trypticase soy broth (Difco, Becton Dickinson, Sparks, MD) containing 0.6% (w/v) yeast extract (Difco, Becton Dickinson) supplemented with 100 ppm ampicillin (ampicillin sodium salt, Sigma Life Science, St. Louis, MO) (TSBYE + amp) and 10% (v/v) glycerol (Sigma Chemical Company, St. Louis, MO) and stored at -80°C until needed. Working cultures were prepared by streaking each stock culture on plates of trypticase soy agar (Difco, Becton Dickinson) containing 0.6% yeast extract and 100 ppm ampicillin (TSAYE + 126 amp). After 18 - 24 h of incubation at 37°C, a single colony was transferred to 9 ml of TSBYE + amp and similarly incubated. Thereafter, a 0.2 ml aliquot of each culture was transferred to 200 ml of TSBYE + amp, incubated for 18 – 20 h at 37°C and then combined in equal volumes to obtain the 4-strain cocktail. 4.3.4 Inoculation of Radicchio. An E. coli O157:H7 suspension containing 107 CFU/mL was obtained by adding 800 mL (quantities were determined based on the growth rates from Buchholz and others 2012b) of the 4-strain cocktail to 75 L of tap water (~15oC) in a 121 L plastic container (Rubbermaid, Wooster, Ohio). Hand-cored heads of Radicchio (9.1 kg) were then immersed in the E. coli suspension for 15 min, spun in the dewatering centrifuge to remove residual inoculum and drained/air-dried for 1 h at 22°C before processing. Using sterilized tongs, two 25-g samples were taken from two inoculated Radicchio heads to determine the average inoculation level at the time of processing. 4.3.5 Processing equipment. The pilot-scale size leafy green processing line consisted of a lettuce shredder, conveyer belt, flume tank, shaker table and centrifugal dryer. The commercial lettuce shredder (model TRS 2500 Urschel TranSlicer, Valparaiso, IN) was set to operate at a feed belt speed of 198 m/min and a slicing wheel speed of 905 RPM. The polyurethane step conveyer belt (ThermoDrive, Mol Industries, Grand Rapids, MI) on the conveyor (Dorner model 736018 mc series, Hartland, WI) operated at 0.11 m/sec. A stainless steel, ~1000 L capacity, water holding tank was filled with 890 L of tap water (~15oC) for fluming. This water holding tank was connected to a 3.6 m stainless steel flume tank (Heinzen Manufacturing, Inc., Gilroy, CA) equipped with two overhead spray jets (1 m from the start) by a 4.14 m long, 10 cm diameter hard plastic circulation hose with a centrifugal pump (model XB754FHA, Sterling Electric, Inc., Irvine, CA) that operated at a rate of ~15 L/sec. The stainless 127 steel shaker table, which partially dewatered the leafy greens, was equipped with a 1 HP Baldor washdown duty motor (Baldor Electric Co., Ft. Smith, AR) that operated at 1760 RPM. Water drained from the leafy greens through a screen on the shaker table and was transferred to the water holding tank via a water recirculation spout on the shaker table. A 50-lb capacity centrifugal Spin Dryer (model SD50-LT, Heinzen Manufacturing, Inc.) with three internally timed spin cycles totaling 60 sec. was used for centrifugal drying. 4.3.6 Radicchio and iceberg lettuce processing and sample collection. Intact heads of uninoculated cored iceberg lettuce (45.4 kg dry weight) were processed using the previously described pilot scale processing line, beginning with the shredder, followed by 9.1 kg (dry weight) of inoculated Radicchio and collected in three 121 L containers (~23 kg wet weight/ container). After draining and refilling the water recirculation tank with ~900 L of tap water (~15oC), 907.2 kg of uninoculated iceberg lettuce were similarly processed in about 22 min with an additional 121-L of tap water (~15oC) added to the water recirculation tank midway through processing to maintain proper water flow. During processing ~38 containers (~23 kg wet weight/container) of lettuce/Radicchio were sequentially collected for analysis along with ~38 500-mL water samples from the water return spout on the shaker table. One 25-g sample of exclusively iceberg lettuce was removed from each container for microbiological analysis after which all visible shreds of Radicchio in the ~38 containers and on the shredder, conveyor, flume tank, shaker table and the floor were retrieved and weighed. Iceberg lettuce container numbers 1, 3, 5, 10, 15, 20, 25, 30, 35 and 40 were centrifugally dried, after which one 25-g 2 sample/container was collected for analysis. Finally, 42 equipment surface samples (100 cm ) previously identified using Glo Germ™ (14 shredder, 8 conveyer belt, 11 flume tank and 9 shaker table samples) were collected using one-ply composite tissues moistened in 1 mL of 128 sterile phosphate buffer and composited based on the piece of equipment to obtain a total of four samples. 4.3.7 Microbiological analyses. Iceberg lettuce samples (25 g) were added to 100 ml of phosphate buffer in a Whirl-Pak™ filter bag and homogenized in a pulsifier (Pulsifier, Filtaflex Ltd., Almonte, Ontario, Canada) for 1 min. Radicchio pieces from containers 1 - 2, 3 - 4, 5 - 9, 10 - 19, 20 - 24, 25 - 29, 30 - 34, 35 - ~38 were composited to obtain a maximum sample size of 25 g, diluted 1:5 in phosphate buffer in Whirl-Pak™ bag (Nasco, Fort Atkinson, WI) and then either homogenized in a stomacher (Stomacher 400 Circulator, Seward, Worthington, UK) for 1 min at 260 rpm or, for very small samples, hand-massaged in the bag for 1 min. Following the same procedure described in Buchholz and others (2012b), the lettuce, Radicchio and water samples were either appropriately diluted or 10 to 100 ml were filtered through a 0.45 μm membrane filter (Millipore, Millipore Corporation, Billerica, MA) and then plated on TSAYE + amp to quantify E. coli O157:H7. The one-ply composite tissue samples were added to 100 ml of phosphate buffer in a Whirl-Pak™ bag, homogenized in a Stomacher for 1 min at 260 rpm and then plated on TSAYE + amp. The 100 ml water samples were processed by membrane filtration and placed on TSAYE + amp. The plates were incubated for 18 – 20 h at 37°C, after which all green fluorescing colonies seen under ultraviolet light were counted as E. coli O157:H7. 4.3.8 Statistical analysis. E. coli O157:H7 populations transferring from the inoculated Radicchio to the equipment surfaces, uninoculated iceberg lettuce and water during processing 2 were converted to log CFU/g, ml or 100 cm . The wet-weight percentage of inoculated Radicchio in each container of shredded iceberg lettuce was also determined. An ANOVA followed by the Tukey-Kramer HSD test (α = 0.05) was conducted using JMP 8.0 software (SAS 129 Institute Inc., Cary, NC) to determine significant differences in E. coli O157:H7 numbers transferred between the equipment surfaces, water and iceberg lettuce. 4.3.9 Modeling of non-linear curves. JMP 8.0 software was also used to conduct a nonlinear regression analysis and calculate 95% confidence intervals for the curves fitted to the E. coli O157:H7 counts on the Radicchio, lettuce and water, as well as the weights of Radicchio transferred over time. The cumulative Weibull model was used as follows: n log N/N0 = -bt , where b and n are the scale and shape parameters, respectively (Chen 2007, Gonzalez and others 2009, Marfart and others 2002, Peleg and Cole 1998). 130 4.4 RESULTS 4.4.1 Radicchio. A four-strain E. coli O157:H7 suspension containing 7.2 ± 0.5 log CFU/mL was used to inoculate the heads of Radicchio. After immersion, the Radicchio contained 5.9 ± 0.5 log CFU/g E. coli O157:H7. Numbers of E. coli O157:H7 decreased to 5.0 ± 0.4 log CFU/g on the Radicchio after shredding, conveying, washing and shaker table partial dewatering with similar populations seen on Radicchio shreds recovered from the ~38 containers of iceberg lettuce after processing (Figure 4.1). After single 45.4 kg batches of uninoculated iceberg lettuce and Radicchio were processed following by 907.2 kg of uninoculated lettuce, shreds of Radicchio were found in all but one bag of the ~38 bags for each of the three independent experiments. The weight of Radicchio ranged from 858.3 to 0.001 g (Figure 4.2) with most of the inoculated Radicchio appearing in the first of the ~38 containers of the uncontaminated iceberg collected for each of three replicates. When the shredded lettuce bags were divided into four groups including bags 1 - 10, 11 - 20, 21 - 30, and 31 – 40, then after processing, the weight of inoculated Radicchio averaged 614.9 g (93.6 %), 6.9 g (1.3 %), 5.0 g (0.8 %) and 2.8 g (0.5 %) in the shredded lettuce bag groupings, respectively. 4.4.2 Iceberg lettuce. E. coli O157:H7 populations on the 907.2 kg of previously uninoculated lettuce ranged from -0.1 to 3.0 log CFU/g, with an overall average of 1.2 log CFU/g (Figure 4.3). Slight differences in contamination levels were observed between replicates; however, the distribution followed the same pattern. Using the same bag grouping system as described above, bags 1 - 10, 11 - 20, 21 - 30, and 31 – 40, contained average E. coli O157:H7 populations of 1.69, 1.22, 1.10 and 1.11 log CFU/g, respectively. 131 The weight of the of iceberg lettuce increased from 907.2 kg to ~1194 kg due to the water that remained on the product after partial dewatering on the shaker table. No significant decreases (α = 0.05) in E. coli O157:H7 populations were observed on the iceberg lettuce following centrifugal drying for each sample basket pair (Figure 4.4). Unlike the findings in Buchholz and others (2012b), which showed a 30% E. coli population decrease on the iceberg lettuce after centrifugal drying, this study indicates between a 3.8 and 82.0 % decrease. This difference in findings could be due to the large standard deviation and the very low levels of E. coli on this Iceberg lettuce in this particular study. 4.4.3 Processing equipment surfaces. While a few pieces of shredded Radicchio remained on each piece of processing equipment (Figure 4.5), a larger percentage in weight of these pieces (P < 0.05) remained on the conveyer belt as compared to the shredder, flume tank and shaker table (Table 4.1). Radicchio build-up was most evident on the discharge chute and area surrounding the cutting wheel of the shredder, the front of the flume tank and the polyurethane conveyer belt, especially the junction between the guard rails and the belt. No significant differences in average numbers of E. coli O157:H7 (1.88 – 3.15 log CFU/100 cm2) remaining at the end of processing (P > 0.05) were seen between the four equipment surface areas (Table 4.2). These findings clearly indicate that E. coli O157:H7 persisted in this processing line after 907.2 kg of clean uncontaminated iceberg lettuce was processed. 4.4.4 Flume water. After processing 907.2 kg of uninoculated iceberg lettuce, E. coli O157:H7 populations in the sanitizer-free flume water ranged from the limit of detection (LOD) to 3.6 log CFU/100 ml. Differences between the three replicates highlight the variability in 132 transfer during processing (Figure 4.6). This finding is most likely related to the amount of residual water remaining in the system prior to refilling the water holding tank. 133 4.5 DISCUSSION Previous studies have demonstrated the potential for bacterial cross-contamination during field coring (Taormina and others 2009) and post-harvest processing of leafy greens (Buchholz and others 2012b and 2012d) as well as washing of produce (Buchanan and others 1999; LopezGalvez and others 2009; Lopez-Galvez and others 2010; Nou and Lou 2010). Bacterial transfer has also been demonstrated in various other food preparation scenarios, including grinding of meat (Farrell and others 1998; Flores and Stewart 2004, Flores and Tamplin 2002) and slicing deli meats (Keskinen and others 2008a; Keskinen and others 2008b; Lin and others 2006; Sheen and Hwang 2008; Vorst and others 2006), as well as between lettuce, chicken, and cutting boards (Chai and others 2008; Chen and others 2001; Fravalo and others 2009; Jiménez and others 2009; Ravishankar and others 2010), and cloths, hands, gloves, utensils and various food products (Chen and others 2001; Gill and Jones 2002; Montville and others 2001; Scott and Bloomfield 1990). One notable study measured the distribution of Listeria monocytogenes during slicing of roast turkey breast, Genoa hard salami, and bologna with kitchen knives. These researchers demonstrated that most bacteria transfer to the first 5 to 15 slices depended on the stainless steel type and product being cut (Vorst and others 2006). Work conducted at University of Georgia also demonstrated that E. coli O157:H7 transferred to successive uncontaminated heads of iceberg lettuce during harvesting. When devices used in field coring were inoculated with soil containing 2.72 and 1.67 log E. coli O157:H7 CFU/g, the pathogen transferred to 10 and 5 consecutively processed heads of lettuce, respectively (Taormina et al. 2009). Flores and Stewart (2004) tracked the spread of E. coli O157:H7 during commercial grinding of inoculated 134 followed by uninoculated beef with E. coli present in 12.7 to 86.2 % of the ~6600 g beef trim batch using initial inoculums of 2 to 6 log CFU/g. 6 Work in our own laboratory has also shown that after processing lettuce containing 10 4 or 10 E. coli O157:H7 CFU/g followed by uninoculated lettuce, E. coli O157:H7 was 2 quantifiable throughout the entire 90.8 kg of product. At a lower inoculation level of 10 CFU/g, E. coli O157:H7 was consistently detected in the first 21.2 and 68.0 kg of previously uninoculated iceberg and romaine lettuce at 3 and 2 CFU/100 g and transferred out to 75.6 and 86.9 kg product, respectively (Buchholz and others 2012d). Using this data, a simulation model was created to estimate the prevalence and percent of E. coli O157:H7 transferred during a cross contamination event in fresh-cut bagged leafy greens (Pérez Rodríguez and others 2011). This simulation predicts that when processing without sanitized wash water a batch of leafy greens inoculated at 1 and 100 CFU/g the prevalence would be ~ 3.05 and 13.39 %, respectively. Based on the simulation, chlorination of the water would reduce this percentage to 0.14 – 0.20 % and 3.28 – 4.00 %, for the 1 and 100 CFU/g inoculation levels, respectively (Pérez Rodríguez and others 2011). Keeping in mind that grinding of meat and post-harvest processing of leafy greens are fundamentally different processes, it is interesting to note that washing produce with nonchlorinated wash water results in similar E. coli cross-contamination as when a contaminated batch of ground beef enters a commercial grinding operation. As seen in the previously mentioned studies involving ground beef and post-harvest leafy green processing, inoculated and uninoculated food products will inevitably become commingled during processing, which complicates any assessment of transfer from equipment and contaminated product to previously uncontaminated product. In this study E. coli O157:H7- 135 inoculated Radicchio was used as an iceberg lettuce surrogate to track the spread of this pathogen and contaminated product during small-scale commercial processing of leafy greens. Various methods for differentiating product are available based on colored dyes, physical markings (Zhang and others 2009), and similar products of different colors with Annous and others (2001) having used both inoculated Golden Delicious and uninoculated Fuji apples to assess the efficacy of a commercial flatbed brush washer (Annous and others 2001). Similarly, Nou and Luo (2010) used inoculated “Lollo Rossa” and uninoculated green-leaf lettuce leaves to evaluate the efficacy of a chlorine wash in the production of mixed salad greens. In our study, Radicchio - a red colored leaf chicory, was used as a surrogate for iceberg lettuce in order to visually track the inoculated product through the processing line. Although not a member of the same genus as lettuce, Radicchio is in the same family and both are considered part of the “leafy vegetable” category. Radicchio is frequently included in fresh-cut leafy green mixtures such as spring mix, Italian salad and fancy romaine salad packages. The vivid red color of the Radicchio allowed us to visually track the product through the processing line and quickly retrieve the small pieces from the processed lettuce, while the head shape mimicked that of a small head of iceberg, which was used as the uninoculated product. Preliminary unpublished experiments showed that Radicchio acted as a better iceberg lettuce surrogate than red leaf lettuce, which clung to the equipment surfaces (unlike iceberg). Dyeing the iceberg heads was also investigated, but it was quickly determined that the color uptake was not sufficient to allow for quick visual differentiation between inoculated and uninoculated pieces. Radicchio was inoculated at the ~6 log CFU/g, to represent a high pathogen load and a worst-case scenario for incoming product from an animal contamination event in the field. This targeted inoculation level at the time of processing was based on previous work in Buchholz and 136 others (2012b and 2012d), which showed the needed low variability for future mathematical modeling and leafy green risk assessments. While less common in a real world scenario than lower pathogen loads, our high inoculation levels represent direct contamination from fecal matter in a field, because cattle (the main carriers of E. coli O157:H7) can present a major contamination source of the environment (Williams and others 2008). While direct contact with produce is one source of pre-harvest contamination, contaminated manure can also enter irrigation water to become a source for the spread of pathogens (Beuchat 2002; Ibekwe and others 2004). Compared to the work in Buchholz and others (2012d), this study used a much smaller inoculated to uninoculated batch size ratio. This was done to simulate only a few heads of lettuce (or a small area of baby spinach) out of an entire field experiencing a contamination event, as would be seen with direct fecal contamination. While direct contact with the produce is one source of pre-harvest contamination, manure can also contaminate a produce field’s water supply, causing the irrigation or rain water to become a source for the spread of pathogens (Beuchat 2002; Ibekwe and others 2004). Dip inoculation provided the needed uniformity for obtaining repeatable cross-contamination results. Ensuring that the entire Radicchio head was coated in the inoculum was essential because once these heads passed through the shredder each shred acted as its own contaminated piece. This study determined the E. coli O157:H7 levels transferred to one batch of previously uncontaminated leafy greens over time after processing a contaminated batch of product. These findings are similar to those discussed in Buchholz and others (2012d) where E. coli O157:H7 was quantifiable throughout 90.8 kg of iceberg and Romaine lettuce with average counts of 1.51 and 2.90 log CFU/g, respectively. The main differences between these studies were batch size and the ability to remove the inoculated pieces before microbial analysis. Spread of E. coli 137 O157:H7 from inoculated to uninoculated product during processing occurred through contact with residual water in the leafy green processing line, inoculated Radicchio retained on the shredder and conveyer belt, and contaminated product lingering throughout the processing line. Since the number of collected lettuce containers varied in the three replicates, 95% confidence intervals were calculated for the E. coli O157:H7 counts based on non-linear regression analysis. When fitted to the data, the nonlinear Weibull model produced an initial decline in counts followed by diminishing numbers of E. coli O157:H7 as processing continued. As expected, the curve for the percentage of Radicchio recovered from each container of iceberg lettuce followed a similar pattern, indicating that transfer decreases as smaller amounts of inoculated product remain in the system. Like other types of fresh-cut produce, leafy greens must be washed during post-harvest processing to remove any dirt and debris as well as to reduce the microbial load. The Weibull model fit to the E. coli counts in the process water shows a very different trend compared to those on the Iceberg lettuce. In this case, the line curves upward showing an increased transfer of the pathogen from contaminated equipment surfaces, from residual pieces of Radicchio, to the wash water. If low bacterial populations remain in flume water as a result of decreased sanitizer efficacy, any product will immediately be contaminated by direct contact with this water. Greater variability was seen in the E. coli O157H:7 counts from water as opposed to lettuce. Cross-contamination of apples during flume washing was also demonstrated by immersing apples in a dump tank (Buchanan and others 1999). Many research studies have shown that tradition of leafy green washing practices are only capable of reducing microbial populations by at best 90 to 99% (Burnett and others 2006; Keskinen and others 2009; Sapers 2001; Weissinger and others 2000; Zhang and others 2009). 138 Most industrially used sanitizers are best suited for reducing microbial populations in the wash water rather than on the product being washed. The organic load in recirculating flume water will increase as additional product is processed. Unless properly monitored, the antimicrobial activity of chlorinated sanitizers will similarly decrease as the organic load increases, with the increasing bacterial load in the wash water negatively impacting end product shelf-life and safety. However, if a sanitizer had been used, fewer E. coli O157:H7 cells would have been available for cross-contamination during processing, and thereby reducing our ability to obtain quantitative data for subsequent mathematical modeling. Hundreds of contaminated Radicchio shreds were recovered from the leafy green processing line after 907.2 kg of uninoculated iceberg lettuce had been processed. Over time, these lingering pieces of inoculated Radicchio will eventually break free from the equipment surface and contaminate additional product as evidenced by the presence of Radicchio in the last container of uninoculated lettuce. Although the E. coli O157:H7 counts were relatively high on these Radicchio pieces, only low numbers of E. coli O157:H7 transferred to the product contact surfaces. Unlike in Buchholz and others (2012b and 2012d) where the shredder and conveyer belt yielded higher populations of E. coli O157:H7, in this study no significant differences (P > 0.05) were seen between the shredder, conveyor, flume tank or shaker table. This difference may be linked to the fact that in this experiment processed a much larger quantity of uninoculated product after the inoculated batch, which could have ‘cleaned’ the equipment surfaces. In summary, this study quantified the extent of cross-contamination to subsequent product by tracking contaminated product though a fresh-cut leafy green processing line. Visually demonstrating the spread of contaminated product during processing illustrates the ease 139 by which E. coli O157:H7 can contaminate multiple batches of product. In the event of an outbreak these finding could be used as a guide to help estimate the amount of product which may have become cross-contaminated during processing and would need to be recalled. 140 4.6 FIGURES Figure 4.1: Linear trend line fitted to mean (± SD) E. coli O157:H7 populations on Radicchio a after leafy green processing . a The wet weight of the iceberg lettuce after processing is the cumulative mass. 141 a Figure 4.2: Weibull model fitted to percentage (± SD) of Radicchio recovered from iceberg b lettuce after leafy green processing . 100 80 % Radicchio 60 40 20 5 4 3 2 1 0 0 200 400 600 800 1000 1200 1400 Cumulative Mass (kg) a Weibull parameters are b = 81,353.30 ± 50,962.54 with the low 95% confidence interval at 30,499.68 and the high 95% confidence interval at 27,5530.86, and n = -2.10 ± 0.19 with the low 95% confidence interval at -2.47 and the high 95% confidence interval at -1.81 b The wet weight of the iceberg lettuce after processing is the cumulative mass. 142 a Figure 4.3: Weibull model fitted to mean (± SD) E. coli O157:H7 populations on iceberg b lettuce after leafy green processing . a Weibull parameters are b = 3.89 ± 0.73 with the low 95% confidence interval at 2.61 and the high 95% confidence interval at 5.58, and n = -0.19 ± 0.03 with the low 95% confidence interval at -0.26 and the high 95% confidence interval at -0.13. b The wet weight of the iceberg lettuce after processing is the cumulative mass. 143 a Figure 4.4: E. coli O157:H7 populations (mean ± SD) , on iceberg lettuce before and after centrifugal drying. a Means of the same sample baskets did not differ significantly (P > 0.05) before and after centrifugal drying. 144 Figure 4.5: Radicchio remaining on equipment surfaces after 907 kg of uninoculated product was processed: (A) conveyer belt and flume tank and (B) blade and interior surface of the lettuce shredder. A B 145 a Figure 4.6: Weibull model fitted to mean (± SD) , E. coli O157:H7 populations in 900 L of water during processing of leafy green inoculated CFU/g a Weibull parameters are b = 1.20 ± 0.28 with the low 95% confidence interval at 0.75 and the high 95% confidence interval at 1.83, and n = 0.21 ± 0.04 with the low 95% confidence interval at 0.05 and the high 95% confidence interval at 0.20. 146 4.7 TABLES Table 4.1: Weight (g) ± SD and Percent ± SD of Radicchio recovered from equipment surfaces after processing a Processing Equipment Weight Shredder 8.3 ± 5.1 Conveyer Belt 9.8 ± 2.8 Flume Tank 3.5 ± 3.4 Shaker Table 0.1 ± 0.1 a Percent AB A AB B 1.5 ± 1.0 1.7 ± 1.0 0.7 ± 0.6 0.0 ± 0.0 Means of not labeled with by the same letter are significantly different (P ≤ 0.05). 147 Table 4.2: E. coli O157:H7 populations (mean ± SD) recovered from equipment surfaces after leafy green processing a Processing Equipment E. coli O157:H7 2 (log CFU/100 cm ) Shredder 2.4 ± 0.4 Conveyer Belt 3.2 ± 0.7 Flume Tank 2.3 ± 0.5 Shaker Table 1.9 ± 0.4 a A A A A Means of not labeled with by the same letter are significantly different (P ≤ 0.05). 148 CHAPTER 5: Impact of Post-inoculation Hold Time on Escherichia coli O157:H7 Transfer during Commercial Production of Fresh-cut Leafy Greens 149 5.1 ABSTRACT Leafy greens are prone to pathogen contamination in the field as well as during subsequent handling and processing, with the strength of bacterial attachment to the leaf surface being highly variable. This study assessed the impact of post-inoculation hold-times and temperatures on Escherichia coli O157:H7 transfer during pilot-scale lettuce processing that included a shredder, step conveyor, flume tank and shaker table. Triplicate batches (0.5 kg) of Radicchio were dip-inoculated with a 4-strain cocktail of avirulent, GFP-labeled, ampicillin5 resistant E. coli O157:H7 (~10 CFU/g) and then processed either immediately or after storage for 1 d at 4°C or 5 d at 22°C. After processing 2.3 kg of uninoculated iceberg lettuce, inoculated Radicchio was processed, followed by 45.5 kg of uninoculated iceberg lettuce. Eight 2 lettuce/Radicchio (25 g), 8 water (500 ml) and 50 equipment swab (100 cm ) samples were collected during processing. After processing, all Radicchio pieces from eight 5-kg baskets of shredded lettuce, the equipment and the floor were weighed after which three iceberg lettuce (25 g) samples from each basket were examined for E. coli O157:H7 by direct plating or membrane filtration using trypticase soy agar containing 0.6% yeast extract and 100 ppm ampicillin. After processing, E. coli O157:H7 was found in all iceberg lettuce samples, with mean counts of 0.3, 0.0 and -0.4 log CFU/g for the 1 h-, 1 d- and 5 d-held product, respectively (P > 0.05). Scanning electron micrographs showed greater bacterial attachment for 5 d-held product. E. coli O157:H7 populations in the flume water did not differ (P > 0.05) for 1 h compared to 1 d and 5 d-held product with a similar trend seen for the equipment surface samples. Based on these findings, the point at which contamination occurs will impact the extent to which E. coli O157:H7 will detach from leafy greens during processing. 150 5.2 INTRODUCTION The high number of foodborne outbreaks traced to fresh fruits and vegetables is partially attributed to current large-scale, centralized production and processing practices. Since 1990, 19 outbreaks have been specifically associated with the consumption of leafy greens (CSPI 2008). The problems associated with contamination of leafy greens became national news in the fall of 2006, when consumption of iceberg lettuce and pre-bagged baby spinach caused 3 large-scale E. coli O157:H7 outbreaks (CSPI 2009b, FDA 2006, FDA 2007, Jay and others 2007). While pathogen contamination likely originates on the farm, current harvesting and post-harvest processing practices can spread pathogens to previously uncontaminated product during field coring (Taormina and others 2009), cooling, shredding, washing, conveying drying and subsequent handling (Beuchat 1996; Beuchat and Ryu 1997; Gorny 2006), and increase the impact of an outbreak as evidenced by the recent multi-state outbreaks (Beuchat and Ryu 1997; Burnett and Beuchat 2001). When present on the surface of leafy greens, E. coli O157:H7 can readily transfer to food contact surfaces on the processing equipment and to the wash water as demonstrated in Buchholz and others (2012b). In addition, both E. coli O157:H7 and the contaminated leafy greens will persist on equipment surfaces throughout the processing line and slough off over time to contaminate large quantities of uncontaminated product as shown in Buchholz and others (2012c and 2012d). Once contaminated, little can be done to reduce the microbial load on leafy greens with currently accepted commercial washing and sanitizing practices only able to generally decrease bacterial populations 1 to 2 log CFU/g (Sapers 2006). Bacterial attachment and biofilm formation on the surfaces of leafy greens make it nearly impossible to completely eliminate foodborne pathogens during conventional washing (Carmichael and others 1999, Sapers 2006). 151 Biofilms, defined as an aggregate of adherent bacterial cells within an exopolysaccharide matrix, provide a physical barrier against environmental stresses including chemicals sanitizers, temperature changes, desiccation and ultraviolet radiation (Annous and others 2009, Niemira and Cooke 2010, Ryu and Beuchat 2005, and Tarver 2009). Scanning electron microscopy (SEM) and confocal laser scanning microscopy have been frequently used to observe bacterial colonization and biofilm formation on fresh produce (Annous and others 2009, Carmichael and others 1999, Fett 2000, Keskinen and others 2009 and Morris and others 1997). One major concern to the produce industry is that pathogens, including Salmonella or E. coli O157:H7, can produce biofilms or become incorporated into biofilms of produced by background microflora. Using SEM, biofilms containing E. coli O157:H7 have been observed on Romaine and baby spinach leaves after 24 to 72 h of storage at 4°C for (Niemira and Cooke 2010) and on the stem ends lettuce leaves after submersion in an E. coli cocktail for 1 wk at 4°C (Annous and others 2009). o Increasing the time between dip inoculation and processing of lettuce from 1 h at 22 C to 1 d at 4°C decreased the number of E. coli O157:H7 cells transferred to the water and equipment surfaces during processing (Buchholz and others 2012b). Lang and others (2004) have also shown that reduced (or equal to) numbers of E. coli O157:H7 and Salmonella were recovered from leafy greens held for a longer period of time. These findings suggest that longer refrigerated hold times before processing should decrease the potential spread of E. coli O157:H7 to uncontaminated product and lead to both fewer and smaller outbreaks. This current study expands on previous work to determine how extending the time between the contamination event and post-harvest processing, and the production of a biofilm on the produce, effects the extent of bacterial transfer to previously uncontaminated produce during 152 processing. The objectives of this study were to: 1) assess the extent of E. coli O157:H7 attachment and biofilm formation on dip inoculated Radicchio leaves following three different time/temperature hold times and then 2) determine the impact of these three different postinoculation time/temperature hold times on the quantitative transfer of E. coli O157:H7 to iceberg lettuce and the equipment used in commercial processing of leafy greens. 153 5.3 MATERIALS AND METHODS 5.3.1 Experimental Design. Radicchio was inoculated with E. coli O157:H7 at 105 CFU/g, held for 1 h at 22°C, 1 d at 4°C or 5 d at 22°C and then used to quantify transfer from equipment surfaces to water and uninoculated iceberg lettuce during simulated commercial processing. E. coli O157:H7 transfer was examined by processing 2.27 kg of uninoculated iceberg lettuce followed by 0.45 kg of inoculated Radicchio and then 45.5 kg of uninoculated iceberg lettuce with the red color of Radicchio allowing the separation of inoculated and uninoculated product before analysis. All experiments were conducted in triplicate. 5.3.2 Produce. Individually wrapped heads of iceberg lettuce (Lactuca sativa L.) (24 heads per case) and Radicchio (Cichorium intybus), (12 heads per case) were obtained from a local wholesaler (Stan Setas Produce Co., Lansing, MI). These products, which originated from California or Arizona depending on the growing season, were stored in a 4°C walk-in-cooler for a maximum of 5 days and then hand-cored immediately before use. 5.3.3 Bacterial strains. Four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) were obtained from Dr. Michael Doyle at the Center for Food Safety, University of Georgia, Griffin, GA. Upon arrival, stock cultures of each strain were prepared in trypticase soy broth (Difco, Becton Dickinson, Sparks, MD) containing 0.6% yeast extract (Difco, Becton Dickinson) supplemented with 100 ppm ampicillin (ampicillin sodium salt, Sigma Life Science, St. Louis, MO) (TSBYE + amp) and 10% (v/v) glycerol (Sigma Chemical Company, St. Louis, MO) and stored at -80°C until needed. Working cultures were prepared by streaking each stock culture on trypticase soy agar (Difco, Becton Dickinson) containing 0.6% yeast extract and 100 ppm ampicillin (TSAYE + amp). After 18 - 24 h of incubation at 37°C, a single colony was transferred to 9 ml of TSBYE + amp and 154 similarly incubated. Thereafter, a 0.2 ml aliquot of each culture was transferred to 200 ml of TSBYE + amp, incubated for 18 – 20 h at 37°C and then combined in equal volumes to obtain the 4-strain cocktail. 5.3.4 Inoculation of Radicchio. An E. coli O157:H7 suspension containing 106 CFU/ml was obtained by adding 3 ml of the 4-strain cocktail to 3 L of tap water (~15oC), which was used as the inoculum for Radicchio heads that were held 1 h or 1 d before processing. Similarly, an E. coli O157:H7 suspension containing 105 CFU/ml, which was obtained by adding 0.3 ml of the 4strain cocktail to 3 L of tap water (~15oC), served as the inoculum, for Radicchio heads that were held 5 d before processing. For inoculation, 0.6 kg of cored Radicchio heads (~2 heads) were immersed in the E. coli suspension for 15 min, drained, air-dried for 1 h at 22°C and then spun in a salad spinner (Oxo 32480 Good Grips Salad Spinner, Oxo, New York, NY) to remove residual inoculum. The inoculated heads were either processed immediately or after storage for 1 d at 4°C or 5 d at 22°C. Duplicate 25-g samples were aseptically collected to determine the initial inoculation level at the time of processing. 5.3.5 Scanning electron microscopy (SEM). Pieces of Radicchio were cut (1 x 1 cm), using a ethanol wiped razor blade, from upper half of the Radicchio leaf which had been held for 1 h, 1 d or 5 d before processing. Pieces were fixed at 4°C in 4% glutaraldehyde buffer with 0.1 M sodium phosphate at pH 7.4 for 1 to 2 h. After a quick rinse in the buffer solution, samples were dehydrated in an ethanol series (25%, 50%, 75% and 95%) for 10 to 15 min at each gradation followed by three final 10 min changes in 100% ethanol. They were then critical point dried in a Balzers Model 010 critical point dryer (Balzers Union Ltd., Balzers, Liechtenstein) using liquid carbon dioxide as the transitional fluid, mounted on aluminum stubs using Carbon suspension cement (SPI Supplies, West Chester, PA) and Carbon tape (Ted Pella, Inc., Redding, 155 CA), and sputter-coated with a thin layer of osmium (~10 nm thickness) in a NEOC-AT osmium coater (Meiwafosis Co. Ltd., Osaka, Japan). Samples were imaged using a JEOL JSM-7500F (cold field emission electron emitter) scanning electron microscope (JEOL Ltd., Tokyo, Japan), located at the Center for Advance Microscopy at Michigan State University. 5.3.6 Processing equipment. The pilot-scale size leafy green processing line that was used included a lettuce shredder, step conveyer, flume tank and shaker table. The commercial lettuce shredder (model TRS 2500 Urschel TranSlicer, Valparaiso, IN) was operated at a feed belt/slicing wheel speed of 198 m/min and 905 RPM, respectively, to obtain a shred size of approximately 5 x 5 cm. The step conveyor (Dorner model 736018 mc series, Hartland, WI) was fitted with a polyurethane step conveyer belt (ThermoDrive, Mol Industries, Grand Rapids, MI) that operated at 0.11 m/sec. A stainless steel, ~1000 L capacity, water recirculation tank was filled with 890 L of sanitizer-free Michigan State University tap water (~15oC) since inclusion of chemical sanitizers leading to inactivation of E. coli O157:H7 would have confounded our ability to assess transfer during processing. This water tank was connected to a 3.6-m long stainless steel flume tank (Heinzen Manufacturing, Inc., Gilroy, CA) - equipped with two overhead spray jets (1 m from the start), by a 4.14 m long, 10 cm diameter hard plastic discharge hose and a centrifugal pump (model XB754FHA, Sterling Electric, Inc., Irvine, CA) that circulated the water at a rate of ~15 L/sec. The stainless steel shaker table for partial dewatering was operated by a 1 HP Baldor washdown motor (Baldor Electric Co., Ft. Smith, AR) at 1760 RPM. Water removed from the leafy greens during mechanical shaking passed through a fine mesh screen in the shaker table and was fed into the water holding tank by a water recirculation spout 5.3.7 Processing and collection of Radicchio and iceberg lettuce samples. Cored heads of uninoculated iceberg lettuce (2.3 kg), followed by 0.5 kg of Radicchio, were shredded, step- 156 conveyed to the flume tank, washed, partially dried on the shaker table and collected in one plastic mesh basket (United Plastics, Leomister, MA). After draining the water holding tank, flume tank and centrifugal pump hose, the water holding tank was refilled with ~890 L of tap water (~15oC). The remaining 45.5 kg of uninoculated iceberg lettuce were then identically processed and sequentially collected in ~ 8 baskets. During processing, about eight 500-ml aliquots of water were collected from the water return spout on the shaker table. After processing, each of the ~8 baskets containing previously uninoculated lettuce was subdivided into three (~2 kg) layers (top, middle and bottom) with one 25-g iceberg lettuce sample aseptically removed from each layer (~24 samples) for E. coli O157:H7 quantification. In addition, all of the red colored Radicchio pieces in each basket and on the shredder, conveyor, flume tank, shaker table and floor were collected using sterilized tongs, weighed and then similarly assessed for numbers of E. coli O157:H7. Finally, 50 product contact areas (100 cm2) on the equipment (14 shredder, 8 conveyer belt, 11 flume tank and 9 shaker table samples) previously identified using Glo Germ™ (Glo Germ Company, Moab, Utah) as described in Buchholz and others (2012b), were sampled using one-ply composite tissues (KimWipes, Kimberly-Clark Corp., Irving, TX) moistened in 1 ml of sterile 1% (w/v) phosphate buffer (8.5 g/L NaCl, 1.44 g/L Na2HPO4, and 0.24 g/L KH2PO4, J.T. Baker, Mallinckrodt Baker Inc., Phillipsburg, NJ) as described by Vorst and others (2004). 5.3.8 Microbiological analyses. All iceberg lettuce samples (25 g) were added to 100 ml of sterile 1% phosphate buffer in a Whirl-Pak™ filter bag (Nasco, Fort Atkinson, WI) and processed in a pulsifier (Pulsifier, Filtaflex Ltd., Almonte, Ontario, Canada) for 1 min. Radicchio samples were weighed in Whirl-Pak™ filter bags, diluted 1:10 in sterile 1% phosphate buffer, and then homogenized by stomaching (Stomacher 400 Circulator, Seward, Worthington, 157 UK) for 1 min at 260 rpm. Iceberg lettuce and Radicchio samples were either appropriately diluted and plated on TSBYE + amp or processed using a 0.45 μm membrane filter which was placed on TSAYE + amp to quantify E. coli O157:H7. The surface samples from each piece of equipment were composited, added to 100 ml of neutralizing buffer in a Whirl-Pak™ bag, homogenized by stomaching for 1 min. at 260 rpm and then plated identically to the product samples. 100 ml of each water sample were either appropriately diluted and plated on TSAYE + amp or processed by membrane filtration. Following 18 – 20 h of incubation at 37°C, all green fluorescing colonies as seen under ultraviolet light were counted as E. coli O157:H7. 5.3.9 Statistical analysis. E. coli O157:H7 counts were converted to log CFU per g, ml or 100 cm2 and subjected to an ANOVA using JMP 8.0 (SAS Institute Inc., Cary, NC). For all tests, a P value of < 0.05 was considered significant. The Tukey-Kramer HSD test was used to determine if the numbers of E. coli O157:H7 transferred to the equipment surfaces, processing water and uninoculated product were influenced by the Radicchio’s post-inoculation holding time. 158 5.4 RESULTS 5.4.1 Initial inoculum levels. Four-strain E. coli O157:H7 suspensions of 6.0 ± 0.2, 5.78 ± 0.1 and 4.8 ± 0.1 log CFU/ml were used to inoculate the heads of Radicchio to be held for 1 h, 1 d and 5 d, respectively, before processing. After a 15 min immersion followed by 1 h of drying and 1 min of centrifugal drying to remove excess inoculum, Radicchio to be processed immediately, held for 1 d or 5 d, respectively, contained E. coli O157:H7 populations of 5.5 ± 0.3, 5.9 ± 0.3 and 4.3 ± 0.2 log CFU/g. After 1 d of storage at 4°C, numbers decreased to 5.6 ± 0.3 CFU/g, while after 5 d of storage at 22°C the population increased to 5.4 ± 0.4 CFU/g. No significant differences (P > 0.05) were observed between the numbers of E. coli O157:H7 adhering to the three products at the time of processing. 5.4.2 Attachment and biofilm formation. The SEM images of Radicchio leaves held 1 h before processing showed only a few attached cells with no evidence of biofilm formation (Figure 5.1A), whereas a 1 d hold at 4oC led to initial signs of an extracellular matrix, with the cells organized in small aggregates mainly within the stomates and leaf crevices (Figure 5.1B). After 5 days of storage at 22oC to mimic pre-harvest contamination, larger cell aggregates (~5 to 20 cells) were enveloped in a dense extracellular matrix, with this material forming a web covering the entire leaf surface (Figure 5.1C). 5.4.3 Radicchio. Populations of E. coli O157:H7 on the 1 h-, 1 d- and 5 d-held Radicchio decreased 1.7, 1.3, and 2.1 log CFU/g after processing, respectively, with the counts on 1 d-held product being significantly higher than those on the 1 h- and 5 d-held product (P < 0.05) (Figure 5.2). After processing the inoculated Radicchio, 18.8, 20.8, and 18.0 % (w/w) remained in the processing line for the 1 h, 1 d and 5 d held Radicchio, respectively. As expected, the largest percentage of remaining inoculated Radicchio, 3.1, 3.4, and 2.5 %, was recovered in the first 159 aliquot of the collected iceberg lettuce after processing the Radicchio inoculated 1 h, 1 d and 5 d prior to processing (Figure 5.3). After 45.4 kg of iceberg lettuce were processed, there was 17.3, 16.1, and 15.85 % Radicchio was distributed throughout the processed lettuce during the 1 h, 1 d and 5 d attachment-time experiments, respectively. 5.4.4 Iceberg lettuce. After processing 1 h-, 1 d- and 5 d-held Radicchio followed by the 45.4 kg of iceberg lettuce, E. coli O157:H7 was respectively found in all iceberg lettuce samples collected from ~ 8 baskets at mean levels of 0.3 ± 0.6, 0.0 ± 0.6 and -0.4 ± 0.6 log CFU/g with the counts on the 1 h-held product being significantly higher than the 1 d held product, followed by the 5 d held product (P < 0.05). Mean E. coli O157:H7 counts in the final basket of product collected at the end of processing were 0.2 ± 0.4, -0.6 ± 0.2 and -0.8 ± 0.4 log CFU/g for the 1 h, 1 d and 5 d held Radicchio, respectively (Figure 5.4). No statistical difference was observed (P > 0.05) in the mean numbers of E. coli transferred to the iceberg lettuce after the uninoculated products were processed during the 1 h- (0.3 log CFU/g), 1 d- (0.0 log CFU/g) and 5 d- (-0.4 log CFU/g) attachment-time experiments, respectively. Maximum E. coli O157:H7 transfer of 1.5, 1.0 and 0.4 log CFU/g was observed on the iceberg lettuce after processing 3.8, 4.2 and 2.0 kg of previously uninoculated product that followed the Radicchio that had been processed 1 h, 1 d, and 5 d post-inoculation, respectively. 5.4.5 Flume water. After refilling the water holding tank with ~890 L of tap water (~15oC), this water which was used to process the 45.4 kg of uninoculated lettuce, yielded E. coli O157:H7 populations of -1.2, 0.9 and -1.4 log CFU/ml at the end of processing for 1 h-, 1 d- and 5 d-held Radicchio, respectively (Figure 5.5) with no significant differences in terms of postinoculation hold time (P < 0.05). 160 5.4.6 Processing equipment surfaces. Generally no significant differences were seen in the numbers of E. coli O157:H7 recovered from the shredder, step conveyor, flume tank or shaker table (P < 0.05), which ranged from 0.7 to 1.8, -0.1 to 1.2 and -0.6 to 0.8 log CFU/g, for Radicchio heads held 1 h, 1 d and 5 d before processing, respectively (Table 5.1). However, statistically higher (P < 0.05) E. coli populations were seen on the surface of the conveyer belt after processing the 1 h-held product compared to that, which was held for 5 d at 22°C. The other equipment surfaces did not differ significantly based on the post-processing attachment time of the inoculated product (P < 0.05) (Table 5.1). In terms of the weight of Radicchio remaining on equipment after processing, the only significant difference (P < 0.05) was seen for the shredder when 1 d- (6.0 g) as opposed to 1 hheld product was processed with the remaining post-inoculation hold times not impacting the amount of Radicchio remaining on the equipment surfaces (Table 5.2). E. coli O157:H7 populations on the Radicchio did not differ significantly (P < 0.05) in terms of equipment location or holding time before processing (Table 5.2). 161 5.5 DISCUSSION The incidence, growth, attachment and survival of foodborne pathogens on fresh and processed produce is well documented (Beuchat 1996; Beuchat and Bracket 1990; Beuchat and others 2004; Gleeson and O'Beirne 2005; Harris and others 2003), however, the extent of bacterial transfer and cross-contamination that occurs during leafy green processing remains poorly understood. Taormina and others (2009) at the University of Georgia demonstrated the ability for E. coli O157:H7 to transfer from inoculated coring blades (2.72 and 1.67 log CFU/g) to 10 and 5 consecutive heads of iceberg lettuce during field coring, respectively. In our previous work, approximately 90% of the E. coli O157:H7 population on inoculated lettuce was shed in sanitizer-free water with this pathogen contaminating the product contact surfaces of the shredder, conveyor, flume tank, shaker table and dewatering centrifuge during processing (Buchholz and others 2012b) and then transferring to large quantities of previously uncontaminated product during subsequent processing (Buchholz and others 2012c and 2012d). This is the first study to look at the impact of microbial attachment and biofilm formation on cross-contamination from leafy greens contaminated in pre-harvest situations. In this study Radicchio inoculated at the 5 log CFU/g, representing a medium to high bacterial load, was used to visually track the spread of contaminated product during processing. This targeted inoculation level at the time of processing was based on previous work in Buchholz and others (2012b, 2012c, and 2012d), which showed that the data will have low variability between replicates and provide valuable information for future mathematical modeling and risk assessments. Higher inoculation levels can represent direct fecal contamination in the field from fecal matter, because cattle are a major source of this pathogen and can present a major contamination source of the environment (Williams and others 2008). While direct contact with produce is one source of 162 pre-harvest contamination, contaminated manure can also enter irrigation water to become a source for the spread of pathogens (Beuchat 2002; Ibekwe and others 2004). Uniform inoculation of the Radicchio by dipping was essential for determining the amount of crosscontamination and obtaining repeatable results. Lang and others (2004) previously reported greater recovery of E. coli O157:H7 from dip- as opposed to spray- or spot- inoculated lettuce when the same populations were initially applied. To evaluate the extent of transfer from contaminated to uncontaminated leafy greens with certainty, inoculated and uninoculated products having the same surface characteristics must be differentiated from each other based on color, dye or physical markings (Annous and others 2001; Nou and Luo 2010; Zhang and others 2009). In one such example, Golden Delicious and uninoculated Fuji apples were used to assess the efficacy of washing with a commercial flatbed brush washer (Annous and others 2001). Similarly, Nou and Luo (2010) used inoculated “Lollo Rossa” and uninoculated green-leaf lettuce to evaluate a chlorine wash prior to cutting the mixed leaves. In our study Radicchio (red colored leaf chicory) acted as a surrogate for contaminated iceberg lettuce which allowed visual tracking of the inoculated product through the processing line. Although belonging to different genera, both Radicchio and iceberg lettuce, are members of the “leafy vegetable” category. Using Radicchio made it possible to evaluate the impact of microbial attachment and biofilm production on transfer during leafy green processing 5 days post-inoculation because the heads did not wilt or decompose as quickly during storage unlike other leafy greens. Bacterial attachment and biofilm formation on the surface of leafy greens are two key factors that negatively impact the reduction of microorganisms during conventional washing (Carmichael and others 1999; Sapers 2006). When Lang and others (2004) assessed the impact 163 of different post-inoculation drying times on recovery of E. coli O157:H7 from iceberg lettuce and parsley, higher (or similar) populations were obtained from leafy greens that were dried for 2 h at 22°C as opposed to 2 h at 22°C followed by 22 h at 4°C. A similar trend was demonstrated in Buchholz and others (2012b), which showed that extending the time between the contamination event and processing from 1 h at 22°C to 1 h at 22°C followed by 1 d at 4°C decreased the removal and transfer of E. coli O157:H7 during iceberg lettuce processing. In support of these findings, our results showed that Radicchio leaves held for 1 d at 22°C retained more of the E. coli O157:H7 inoculum after processing than did the heads inoculated 1 h prior to processing and held at 22°C. Holding leafy greens at refrigeration temperatures replicates a scenario where products become contaminated during harvest or storage prior to processing. A second possible contamination scenario involves leafy greens becoming contaminated in the field during growth, several days prior to processing. Unexpectedly, Radicchio stored for 5 d retained fewer E. coli O157:H7 cells after processing. This was especially surprising because bacterial aggregates and an extensive extracellular matrix developed on the Radicchio leaves as seen by SEM. These findings may be attributed to the methods used for microbiological analysis resulting in E. coli O157 not being released sufficiently into the buffer solution during stomaching as expected. This explanation is further justified by the counts for E. coli O157:H7 transferred from Radicchio to the iceberg lettuce. In this case, as expected, the uninoculated leafy greens that had been processed following the Radicchio held for 5 days at room temperature had lower levels of E. coli O157:H7, followed by those held for 1 day at refrigeration temperature and 1 h at room temperature. While these findings imply that refrigerating the product for 1 day post-harvest or a contamination event in the field taking place a few days before harvest will decrease the potential spread of E. coli 164 O157:H7 during processing and lead to both fewer and smaller outbreaks, the oral infectious dose for E. coli O157:H7 of less than 100 cells (Meng and others 2007) makes this an unreliable method to prevent an outbreak. The high numbers of E. coli on iceberg lettuce collected after the inoculated batch of Radicchio was processed, indicates that cross-contamination of previously uninoculated product can occur through 1) residual water (approximately less than 2 L) from the flume tank, shaker table, centrifugal pump, hose or water recirculation tank, 2) direct contact with the inoculated Radicchio clinging to the shredder and conveyer belt surfaces, or 3) contact with the contaminated leafy greens/equipment surfaces throughout the processing line. In terms of transfer to the leafy greens, Buchholz and others (2012d) showed that E. coli O157:H7 was found throughout the entire batch of uninoculated iceberg (mean count of 0.3 CFU/g), and Romine lettuce (90.8 kg) (mean count of 0.7 log CFU/g) that was processed following 22.7 kg of leafy greens inoculated at ~4 log CFU/g. While the inoculation level in this current experiment was approximately 1 log higher and the ratio of inoculated to uninoculated products was larger, the same trend was apparent. Most of the transfer occurred to the first few baskets of initially uncontaminated product with a rapid decrease followed by a tailing effect. This study also mimicked the results described in Buchholz and others (2012c), which was a scaled up version of the current experiment using Radicchio held for 1 h at 22°C. Here we showed E. coli O157:H7 counts on the iceberg lettuce ranging from -0.1 to 3.0 log CFU/g (an overall average of 1.2 log CFU/g) after processing Radicchio inoculated at 5.9 log CFU/g. In our current experiment, E. coli O157:H7 transfer to iceberg lettuce from Radicchio held for 1 h before processing resulted in a slightly lower mean count of 0.3 log CFU/g and a lower maximum transfer level of log 1.5 165 CFU/g. These differences may be attributed to both a slightly lower initial inoculation level and a difference due to the scale of the experiments. Commercial production of fresh-cut leafy greens entails a washing step to remove dirt and debris and reduce the microbial load with commercial sanitizer washing systems only capable of reducing microbial populations on the product by 90 to 99% (Burnett and Beuchat and 2001; Keskinen and others 2009; Sapers 2001; Weissinger and others 2000; Zhang and others 2009). Wash water is an ideal vehicle for the transfer of pathogens during processing as demonstrated by Buchanan and others (1999) in a study looking at cross-contamination during washing of apples. The goal of our work was to quantify the numbers E. coli O157:H7 transferred during leafy green processing; therefore chemical sanitizers could not be used in the flume tank. If sanitizers had been used, the numbers of E. coli O157:H7 would have most likely been reduced to non-detectable levels in the flume water, preventing quantification of crosscontamination to the uncontaminated lettuce. Therefore, the processing water used to wash inoculated product was drained with the water holding tank then refilled before processing uninoculated product. The data collected in our study demonstrates that even with low bacterial counts in the flume water, any product that is washed in that water will become immediately inoculated with E. coli O157:H7 via direct contact with the contaminated water and that an extended storage time after contamination will not reduce the likelihood that a pathogen will be spread through an entire batch of product. This aligns very well with the efficacy of normal produce washing practices for the removal of microorganisms and shows that the ineffectiveness of current processing/washing methods is clearly a major concern as supported by the three recent leafy green recalls involving E. coli O157:H7, Salmonella and Listeria monocytogenes (FDA 2010a; FDA 2010b; FDA 2010c). 166 While hundreds of pieces of Radicchio were recovered from the surfaces of shredder, and step conveyor, few pieces were seen in the flume tank and on the shaker table after processing. Consequently, contaminated product can linger on processing equipment, eventually breaking free of the equipment surface and being carried downstream with uncontaminated product being processed, as evident by the numbers of small pieces of radicchio which were found even in the last baskets of iceberg lettuce collected. In a study examining the microbial changes of “Lollo Rosso” lettuce during processing, an increase of 1 log CFU/g was observed after shredding (Allende and others 2004), indicating that the shredder may be an important in-plant vehicle for spreading contamination to leafy greens during processing (Garg and others 1990; Stafford and others 2002). This type of transfer was the likely cause of an Australian Salmonella Bovismorbificans outbreak, in which 41 individuals became ill, when lettuce particulates were found behind the cutting wheel rim of the shredder used in the processing line (Stafford and others 2002). Although E. coli counts on the Radicchio pieces were relatively high, very few E. coli O157:H7 cells transferred to the product contact surfaces. While these numbers may be low, only a few bacterial cells are needed to cross-contaminate large quantities of product. Our results indicated that microbial attachment and biofilm formation may play an important role in bacterial transfer from inoculated leaves to processing equipment surfaces. As expected more bacteria were recovered from the surface of the conveyer belt after processing the 1 h-held product compared to that, which was held for 5 d at 22 °C. Hence, 1 h may be insufficient for E. coli O157:H7 to properly attach to Radicchio leaves before coming in direct contact with the polyurethane conveyer belt. In previous work (Buchholz and others 2012b and 1012c) the shredder and conveyer belt generally yielded higher E. coli O157:H7 counts than the flume tank 167 or shaker table after the initial inoculated products were processed with the bacteria being retained even after processing uninoculated product, but in Buchholz and others (2012c) we did not observe any differences between bacterial counts remaining on the equipment surfaces. Following again a similar trend to the data seen in Buchholz and others (2012c), no significant differences were observed in E. coli populations recovered from the shredder, step conveyor, flume tank or shaker table within each post-inoculation holding time experiment. The different result in this experiment be attributed to the fact that rather than looking at 50 individual locations on the equipment, like in Buchholz and others (2012b and 2012d), we created a composite of the equipment swab samples and looked at mean transfer to the equipment. This is the first report to assess the impact of microbial attachment and biofilm formation on E. coli O157:H7 cross-contamination during production of fresh-cut leafy greens. These results reveal that extending the time between a contamination event and post-harvest leafy green processing will reduce the ability to remove the attached bacteria from the surfaces of the leafy greens, but will decrease the potential spread of E. coli O157:H7 during those processing steps. Careful monitoring and cleaning is necessary to prevent spread of the contamination throughout a production lot. 168 5.6 FIGURES Figure 5.1: Scanning electron micrograph images showing attachment and biofilm formation E. coli and/or native microflora on Radicchio leaves after inoculation and holding for (A) 1 hour at 22°C, (B) 1 day at 4°C, and C) 5 days at 22°C. A 169 Figure 5.1 (cont’d) B C 170 Figure 5.2: E. coli O157:H7 populations (mean ± SD) on Radicchio after processing of iceberg lettuce 171 Figure 5.3: Percentage by weight (± SD) of Radicchio recovered from the iceberg lettuce after processing 4 1 hour 1 day 5 day % Radicchio 3 2 1 0 0 10 20 30 Cumulative Mass (kg) 172 40 50 60 Figure 5.4: E. coli O157:H7 populations (mean ± SD), on iceberg lettuce after processing. 2 1 hour E. coli O157:H7 (log CFU/g) 1 day 5 days 1 0 0 10 20 30 -1 -2 Cumulative Mass (kg) 173 40 50 60 Figure 5.5: E. coli O157:H7 populations (mean ± SD) in 890 L of recirculation water during processing of uninoculated iceberg lettuce 2 1 hour E. coli O157:H7 (logCFU/ml) 1 day 5 days 1 0 0 10 20 30 40 -1 -2 -3 Average Cumulative Mass (kg) 174 50 60 5.7 TABLES Table 5.1: E. coli O157:H7 populations (mean ± SD) recovered from the processing equipment surfaces after leafy green processing Product Processing Equipment E. coli O157:H7 recovered (log CFU/100 cm2) Radicchio held at 22°C for 1 hour Shredder 1.8 ± 1.2 Conveyer Belt 1.7 ± 0.8 Flume Tank 0.7 ± 0.2 Shaker Table 0.8 ± 0.2 Shredder 1.2 ± 0.4 Conveyer Belt 1.2 ± 0.3 Flume Tank -0.1 ± 0.9 Aa Aa Aa Aa Radicchio held at 4°C for 1 day 175 Aa Aa Aa a Table 5.1 (cont’d) Shaker Table 0.2 ± 0.7 Shredder 0.8 ± 0.5 Conveyer Belt 0.8 ± 0.5 Flume Tank -0.6 ± 1.3 Shaker Table -0.1 ± 0.8 Aa Radicchio held at 22°C for 5 days a Aa Aa Aa Aa Means not labeled with the same capital letter are E. coli O157:H7 counts that differ significantly in terms of processing equipment location while means not labeled with the same lowercase letter differ significantly in terms of post-inoculation holding time (P ≤ 0.05). 176 Table 5.2: E. coli O157:H7 populations (mean ± SD), Weight ± SD and Percentage ± SD of Radicchio recovered from equipment surfaces after processing a Product Processing Equipment E. coli O157:H7 recovered Weight (g) of the (log CFU/100 cm2) Radicchio Percent of the Radicchio Radicchio held at 22°C for 1 hour Shredder 4.6 ± 0.5 Conveyer Belt 5.1 ± 0.5 Flume Tank 3.6 ± 0.1 Shaker Table 4.1 ± 0.6 Shredder 4.6 ± 0.2 Conveyer Belt 5.2 ± 0.3 Flume Tank 4.3 ± 0.5 Aa Aa Aa Aa 0.5 ± 0.7 0.8 ± 0.6 0.0 ± 0.0 1.1 ± 0.8 Ab Aa Aa Aa 0.4 ± 0.4 1.0 ± 0.9 0.0 ± 0.0 1.09 ± 0.3 Radicchio held at 4°C for 1 day 177 Aa Aa Aa 6.0 ± 3.0 1.4 ± 1.7 0.6 ± 0.7 Aa Aa Aa 6.0 ± 2.9 1.4 ± 2.0 0.4 ± 0.7 Table 5.2 (cont’d) Shaker Table 4.6 ± 0.7 Shredder 4.6 ± 0.5 Conveyer Belt 5.1 ± 0.3 Flume Tank 4.2 ± 0.5 Shaker Table 4.1 ± 0.8 Aa 1.1 ± 0.4 Aa 1.06 ± 0.4 Radicchio held at 22°C for 5 days a Aa Aa Aa Aa 3.6 ± 2.0 0.9 ± 0.5 1.0 ± 0.8 0.5 ± 0.2 Aab Aa Aa Aa 3.8 ± 2.1 1.0 ± 0.5 0.7 ± 0.9 0.6 ± 0.2 Means not labeled with the same capital letter are E. coli O157:H7 counts that differ significantly in terms of processing equipment location while means not labeled with the same lowercase letter differ significantly in terms of post-inoculation holding time (P ≤ 0.05). 178 CHAPTER 6: Quantitative Transfer of Escherichia coli O157:H7 from Inoculated to Uninoculated Leafy Greens during Shaker Table Dewatering 179 6.1 ABSTRACT Leafy greens are prone to cross-contamination during commercial shredding, conveying and flume washing with the potential for further spread of bacterial pathogens including E. coli O157:H7 via direct leaf-to-leaf contact during shaker table dewatering. Consequently, the goal of this study was to determine the rate of E. coli O157:H7 transfer between products during partial dewatering on a commercial leafy green shaker table. In this study, one (0.03 g) or eight (0.38 g each) Radicchio pieces were manually cut from a single leaf, dip-inoculated to contain 4 (~10 CFU/g) of a 4-strain avirulent, GFP-labeled, ampicillin-resistant E. coli O157:H7 cocktail and dried for 1 h at 22°C. Thereafter, 300 or 297g of uninoculated iceberg lettuce was uniformly cut, wetted, briefly drained, and placed into each of seven square metal baskets (12’’ x 7’’ x 7’’) mounted to the side of a commercial-scale shaker table, followed by the addition of one (0.03 g) or eight (0.38 g each) pieces of inoculated Radicchio to the baskets while shaking to obtain inoculated to uninoculated (w/w) ratios of 1:100 and 1:10,000, respectively The baskets were removed after 0 sec, 10 sec, 1 min, 5 min, 10 min, 30 min or 60 min. All shreds of Radicchio and two 25-g samples of only iceberg lettuce were then retrieved from the baskets. Sample homogenates were quantitatively examined for E. coli O157:H7 by direct plating w/o prior membrane filtration using trypticase soy agar containing 0.6% yeast extract and 100 ppm ampicillin as the growth medium. At the 1:10,000 ratio, E. coli O157:H7 transfer plateaued after ~22 min with average populations of 0.11 log CFU/g recovered from previously uninoculated iceberg lettuce. As expected, transfer leveled off faster for the 1:100 ratio (~0.5 min) with E. coli O157:H7 counts on the lettuce averaging ~2.52 log CFU/g. These findings confirm that crosscontamination can occur via leaf-to-leaf contact with E. coli O157:H7-contaminated product within a very short time frame after a contamination event. 180 6.2 INTRODUCTION Between 1990 and 2008, 19 outbreaks have been specifically associated with the consumption of leafy greens (CSPI 2008b), many of which have been traced to produce grown in the Salinas Valley of California. The problems associated with contamination of leafy greens became national news in September of 2006, when consumption of pre-bagged baby spinach caused an E. coli O157:H7 outbreak in 26 states and Canada resulting in five deaths and 205 illnesses (Jay and others 2007). Three months later, two other E. coli O157:H7 outbreaks were traced to California-grown iceberg lettuce that was shredded and then purchased by two Mexican fast-food restaurant chains in the Midwest and Northeast (FDA 2006 and FDA 2007). These two outbreaks sickened a total of over 150 individuals (CSPI 2009a). The high number of foodborne outbreaks traced to fresh fruits and vegetables is partially attributed to current large-scale, centralized production and processing practices. Contamination can occur at any point in the farm to table continuum. Many of the previous outbreaks have pointed to contamination on the farm from wild and domestic animal droppings, water runoff from livestock operations, green or improperly composted manure, dust, insects, contaminated irrigation water, and poor handling practices. While pathogen contamination likely originates on the farm, current harvesting and post-harvest processing practices can increase the impact and spread pathogens to previously uncontaminated product during field coring (Taormina and others 2009), cooling, shredding, washing, conveying drying and subsequent handling. (Beuchat 1996; Beuchat and Ryu 1997; Gorny 2006), as evidenced by the recent multi-state outbreaks (Beuchat and Ryu 1997; Burnett and Beuchat 2001). . Once present on the surface of leafy greens E. coli O157:H7 can readily transfer to food contact surfaces on the processing equipment and to the wash water as demonstrated by 181 Buchholz and others (2012b). In addition, both E. coli O157:H7 and contaminated leafy greens can persist as transient residents on equipment surfaces throughout the processing line and slough off over time to contaminate large quantities of previously uncontaminated product during processing as shown by Buchholz and others (2012a, 2012c and 2012d). Our previous work investigated the different routes of bacterial transfer between contaminated products, equipment and water. This current study expands on previous work to determine if the extent of E. coli O157:H7 cross-contamination between inoculated and uninoculated leafy greens increases during partial dewatering following longer periods of time on the shaker table and a greater inoculated to uninoculated ratio. 182 6.3 MATERIALS AND METHODS 4 6.3.1 Experimental Design. In this study, Radicchio inoculated to contain 10 CFU/g E. coli O157:H7 was used to quantify leaf-to-leaf transfer. All experiments were conducted in triplicate. Transfer coefficients were determined for two different ratios of inoculated to uninoculated leafy greens - 1:100 and 1:10,000, with the red color of Radicchio allowing for easy differentiation between inoculated and uninoculated product. 6.3.2 Produce. Individually wrapped heads of iceberg lettuce (Lactuca sativa L.) (24 heads per case) and Radicchio (Cichorium intybus), (12 heads per case) were obtained from a local wholesaler (Stan Setas Produce Co., Lansing, MI). These products, which originated from California or Arizona depending on the growing season, were stored in a 4°C walk-in-cooler for a maximum of 5 days and then hand-cored immediately before use. 6.3.3 Bacterial strains. Four avirulent, GFP-labeled, ampicillin-resistant strains of E. coli O157:H7 (ATCC 43888, CV2b7, 6980-2, and 6982-2) were obtained from Dr. Michael Doyle at the Center for Food Safety, University of Georgia, Griffin, GA. Upon arrival, stock cultures of each strain were prepared in trypticase soy broth (Difco, Becton Dickinson, Sparks, MD) containing 0.6% yeast extract (Difco, Becton Dickinson) supplemented with 100 ppm ampicillin (ampicillin sodium salt, Sigma Life Science, St. Louis, MO) (TSBYE + amp) and 10% (v/v) glycerol (Sigma Chemical Company, St. Louis, MO) and stored at -80°C until needed. Working cultures were prepared by streaking each stock culture on trypticase soy agar (Difco, Becton Dickinson) containing 0.6% yeast extract and 100 ppm ampicillin (TSAYE + amp). After 18 - 24 h of incubation at 37°C, a single colony was transferred to 9 ml of TSBYE + amp and similarly incubated. Thereafter, a 0.2 ml aliquot of each culture was transferred to 200 ml of 183 TSBYE + amp, incubated for 18 – 20 h at 37°C and then combined in equal volumes to obtain the 4-strain cocktail. 5 6.3.4 Inoculation of Radicchio. An E. coli O157:H7 suspension containing 10 CFU/mL was obtained by adding 300 µL (quantities were determined based on the growth rates from Buchholz and others 2012) of the 4-strain cocktail to 3 L of chlorinated tap water (~ 1 ppm chlorine, ~15PoPC) in a 121 L plastic container (Rubbermaid, Wooster, Ohio). One (0.03 g) or eight (0.38 g each) pieces were manually cut from a single Radicchio leaf, immersed in the E. coli suspension for 15 min., drained, and air-dried for 1 h at 22°C before processing. Two 25-g Radicchio samples were also inoculated and used to determine the average inoculation level at the time of processing. 6.3.5 Processing equipment. The stainless steel mechanical shaker table for partial leafy green dewatering was equipped with a 1 HP Baldor washdown duty motor (Baldor Electric Co., Ft. Smith, AR) that operated at 1760 RPM. Water removed from the leafy greens during mechanical shaking passed through a screen on the bottom of the shaker table and was directed to the water holding tank via a water recirculation spout on the shaker table. This shaker table is part of a pilot-scale size leafy green processing line that consists of a lettuce shredder, conveyer belt, flume tank, shaker table and centrifugal dryer. Six rectangular 30 cm x 18 cm x 18 cm metal boxes with perforated (1 cm x 1 cm opening) bottoms were positioned side by side on the shaker table above the metal screen and secured in place with 8” c-clamps (Husky, Home Depot, Atlanta, GA). 6.3.6 Radicchio and iceberg lettuce processing and sample collection. Cored iceberg lettuce heads were manually cut into 1” x 1” pieces totaling 300 or 297 g to yield inoculated to uninoculated (w/w) ratios of 1:100 and 1:10,000, respectively. The iceberg lettuce pieces were 184 wetted with tap water (~15PoPC), briefly drained and placed into the six metal boxes mounted to the shaker table. Either one (0.03 g) or eight (0.38 g each) pieces of inoculated Radicchio were added to the boxes while shaking to obtain inoculated to uninoculated (w/w) ratios of 1:100 and 1:10,000, respectively. The six boxes were sequentially removed from the shaker table after 10 sec, 1 min, 5 min, 10 min, 30 min or 60 min. All Radicchio pieces were carefully separated from the iceberg lettuce to prevent any additional inoculum from transferring to the iceberg lettuce. Two 25-g samples of the iceberg lettuce were then collected from each box for microbial analysis. 6.3.7 Microbiological analyses. Iceberg lettuce samples (25 g) were added to 100 ml of sterile 1% (w/v) phosphate buffer (8.5 g/L NaCl, 1.44 g/L NaR2RHPOR4R, and 0.24 g/L KHR2RPOR4R, J.T. Baker, Mallinckrodt Baker Inc., Phillipsburg, NJ) in a Whirl-Pak™ filter bag (Nasco, Fort Atkinson, WI) and processed in a pulsifier (Pulsifier, Filtaflex Ltd., Almonte, Ontario, Canada) for 1 min. The Radicchio pieces were composited, diluted 1:5 in 0.1% phosphate buffer in Whirl-Pak™ bag and similarly pulsified. These iceberg lettuce and Radicchio samples were plated on TSAYE + amp or processed using 0.45 μm membrane filters (Millipore, Millipore Corporation, Billerica, MA), which were placed on 60-mm dia. Petri plates containing TSAYE + amp to quantify E. coli O157:H7. Following 18 - 20 h of incubation at 37°C, all green fluorescing colonies as seen under ultraviolet light (365 nm, Blak-Ray, Ultraviolet Product Inc. San Gabriel, CA) were counted as E. coli O157:H7. 6.3.9 Modeling of non-linear curves. JMP 8.0 software was used to conduct a nonlinear regression analysis and to calculate the 95% confidence intervals for the curves fitted to the E. coli O157:H7 counts on the iceberg lettuce. The cumulative Weibull model used was as follows: 185 n log N/N0 = -bt , where b and n are the scale and shape parameters, respectively. (Chen 2007, Gonzalez and others 2009, Marfart and others 2002, Peleg and Cole 1998). 186 6.4 RESULTS 6.4.1 Radicchio. The four-strain E. coli O157:H7 suspension used to inoculate the Radicchio pieces for inoculated to uninoculated (w/w) ratios of 1:100 and 1:10,000 contained 5.6 ± 0.5 and 4.7± 0.4 log CFU/ml, respectively. After inoculation by immersion, the Radicchio pieces contained E. coli O157:H7 populations of 4.9 ± 0.2 (1:100 (w/w) ratio) and 4.2 ± 0.3 (1:10,000 (w/w) ratio) log CFU/g. Numbers of E. coli O157:H7 decreased to 4.5 ± 0.5 log CFU/g on Radicchio used in the 1:100 ratio study, while no decrease was observed on Radicchio after shaking in the 1:10,000 ratio study (4.2 ± 1.0 log CFU/g). 6.4.2 Iceberg lettuce. At the 1:10,000 ratio, E. coli O157:H7 transfer plateaued after ~22 min with average populations of 0.11 log CFU/g recovered from previously uninoculated iceberg lettuce (Figure 6.1). After shaking for 60 min., 32.4 ± 6.4 % of the original inoculum transferred to uninoculated iceberg lettuce for the 1:10,000 ratio. As expected, transfer leveled off faster for the 1:100 ratio (~0.5 min) with E. coli O157:H7 counts on the lettuce averaging ~2.5 log CFU/g, with 52.4 ± 30.8 % of the original inoculum transferred after 60 min of shaking (Figure 6.1). 187 6.5 DISCUSSION Previous studies have demonstrated the potential for bacterial cross-contamination during field coring (Taormina and others 2009) and post-harvest processing of leafy greens (Buchholz and others 2012a, 2012b, 2012c and 2012d) and washing of produce (Buchanan and others 1999; Lopez-Galvez and others 2009; Lopez-Galvez and others 2010; Nou and Lou 2010). Bacterial transfer has also been demonstrated in various other food preparation scenarios, including between lettuce, chicken, and cutting boards (Chai and others 2008; Chen and others 2001; Fravalo and others 2009; Jiménez and others 2009; Ravishankar and others 2010), and cloths, hands, gloves, utensils and various food products (Chen and others 2001; Gill and Jones 2002; Montville and others 2001; Scott and Bloomfield 1990). Work in our own laboratory has shown that approximately 90% of the E. coli O157:H7 population on inoculated lettuce was shed in sanitizer-free water during washing of leafy greens in a flume tank with this pathogen contaminating product contact surfaces of the shredder, conveyor, flume tank, shaker table and dewatering centrifuge during processing (Buchholz and others 2012b) and then transferring to large quantities of previously uncontaminated product during subsequent processing (Buchholz and others 2012a, 2012b and 2012d). During post-harvest processing, inoculated and uninoculated leafy greens will inevitably become commingled, thus complicating the assessment of transfer from contaminated product to previously uncontaminated product (Buchholz and others 2012a, 2012c, and 2012d). In this study E. coli O157:H7-inoculated Radicchio was used as a colored surrogate for iceberg lettuce to track the spread of contaminated product during shaker table dewatering. Various methods for differentiating product are available based on colored dyes, physical markings (Zhang and others 2009), and similar products of different colors with Annous and others (2001) having used both 188 inoculated Golden Delicious and uninoculated Fuji apples to assess the efficacy of a commercial flatbed brush washer (Annous and others 2001). Similarly, Nou and Luo (2010) used inoculated “Lollo Rossa” and uninoculated green-leaf lettuce leaves to evaluate the efficacy of a chlorine wash in the production of mixed salad greens. In our study, Radicchio - a red colored leaf chicory, was used as a surrogate for iceberg lettuce to allow us to differentiate the inoculated and uninoculated lettuce pieces. Although not a member of the same genus as lettuce, Radicchio is in the same family with both considered to be “leafy vegetables”. In this study Radicchio inoculated at ~5 log CFU/g, representing a medium to high bacterial load, was used as an indicator to visualize inoculated from uninoculated product. This targeted inoculation level at the time of processing was used to minimize variability between replicates for future mathematical modeling. While less common in a real world scenario than lower pathogen loads, these high inoculation levels represent direct contamination from fecal matter in a field from cattle (the main carriers of E. coli O157:H7) that are a well-documented source of contamination in the field (Williams and others 2008). Cross-contamination between leafy greens has been previously demonstrated with similar findings. A study by Wachtel and Charkowski (2002) showed that mixing of one dry lettuce 5 piece, inoculated to contain ~10 CFU of EHEC per piece, with a large volume of dry lettuce, followed by storage at 4°C or 25°C for 20 h led to the contamination of all uninoculated leaves tested. When fitted to our data, the nonlinear Weibull model described the initial quick increase in E. coli O157:H7 numbers transferred to the uninoculated leaves followed by a leveling off as partial dewatering on the shaker table continued. A confidence interval of 95% was calculated for the E. coli O157:H7 counts based on non-linear regression analysis. As expected, faster cross-contamination was seen using an inoculated to uninoculated ratio of 1:100 than 1:10,000 189 (w/w) due to the higher number of E. coli O157:H7 cells available for transfer. In summary, these findings confirm that cross-contamination can occur via leaf-to-leaf contact with E. coli O157:H7-contaminated product within a very short time frame after a contamination event. 190 6.6 FIGURES Figure 6.1: E. coli O157:H7 populations (mean ± SD), on the iceberg lettuce after leafy green processing. 191 6.7 TABLES Table 6.1: Weibull model parameters fitted to E. coli O157:H7 populations on iceberg lettuce. Ratio Model Parameter Value ± SE Low CI 95% High CI 95% 1:100 b 2.4 ± 0.1 2.3 2.6 n 0.0 ± 0.0 -0.0 0.0 1:10,000 b -0.3 ± 0.1 -0.4 -0.1 n -0.4 ± 0.1 -0.7 -0.2 192 CONCLUSIONS AND FUTURE RECOMMENDATIONS Cross-contamination/multi-directional transfer of Escherichia coli O157:H7 during commercial shredding, conveying, fluming and drying of fresh-cut leafy greens has become a major public health concern, particularly because post-processing kill steps are currently not implemented in the industry. The first objective of the research was to quantify E. coli O157:H7 transfer to equipment during commercial production of fresh-cut leafy greens. These results indicated that during processing of iceberg lettuce, Romaine lettuce and baby spinach, approximately 90% of the inoculum was transferred to the sanitizer-free wash water, which generally agrees with the efficacy of normal produce washing practices for the removal of microorganisms. Very few E. coli O157:H7 cells transferred to the product contact surfaces compared to the water. However, in terms of the equipment surfaces, greatest transfer was seen to the shredder and conveyer belt, followed by the centrifugal dryer, flume tank and shaker table. Shred size does not appear to affect the numbers of E. coli O157:H7 transferred from contaminated leafy greens during processing; however, extending the time between contamination and processing will decrease the removal and transfer of this pathogen during production of fresh-cut leafy greens. Overall, ~29% of the remaining product inoculum was lost during centrifugal drying. Future research needs to be conducted to determine if routine microbial monitoring of the centrifugation water may be a useful strategy for minimizing and preventing the continued spread of bacterial pathogens to large quantities of product. The second objective of this research was to quantify the transfer of E. coli O157:H7 from equipment surfaces to iceberg and Romaine lettuce during simulated commercial processing. These findings indicated that E. coli O157:H7 is capable of persisting on equipment surfaces even after 90.8 kg of clean uncontaminated product are processed. Initially, greatest E. 193 coli O157:H7 transfer was seen from inoculated lettuce to the shredder and conveyer belt, with all equipment surface populations decreasing only 90 - 99% after processing 90.8 kg of uncontaminated product. Such cross-contamination can occur even when E. coli O157:H7 is initially present in very low levels. This study supports the importance of sanitizers in wash waters, or other interventions, to adequately reduce the microbial load on the product and in flume water during commercial processing to prevent future product recalls and outbreaks. Future research may focus on determining the efficacy of various commercially available sanitizers against E. coli O157:H7 in wash water and to prevent cross-contamination during pilot-scale leafy green processing. Based on the findings of the second objective, the third objective of this study was developed to track an E. coli O157:H7 inoculated batch of leafy greens through a processing line using Radicchio as visual marker. The results clearly indicate that E. coli O157:H7 persisted in the same processing line after 907.2 kg of clean uncontaminated iceberg lettuce had passed through. Hundreds of contaminated Radicchio shreds were recovered from the leafy green processing line with at least a few shreds remaining on each processing equipment piece. Radicchio build-up was most evident on the discharge chute and area surrounding the cutting wheel of the shredder, the front of the flume tank and the polyurethane conveyer belt, especially at the junction between the guard rails and the belt. These findings provide important insight into potential lapses in current cleaning/sanitizing practices and improved equipment designs. The results of objective 3 clearly showed that one small inoculated batch could contaminate subsequent large quantities of previously uncontaminated product during processing if an effective microbial intervention strategy is not implemented. This was demonstrated by the E. coli O157:H7 populations averaging 1 log CFU/g on the uninoculated product and the shreds of 194 Radicchio that were found in all but one bag of the ~38 bags of processed product. These data can later be used to validate a simulation model created to estimate the prevalence and percent of E. coli O157:H7 transferred during a cross-contamination event in fresh-cut bagged leafy greens under different scenarios. In the event of an outbreak, these findings could be used as a guide to help estimate the amount of product which may have become cross-contaminated during processing and would need to be recalled. Building on results from the first objective, the fourth objective of this study determined the effect that bacterial attachment time may have on cross-contamination during processing. After processing, E. coli O157:H7 was found in all previously uncontaminated iceberg lettuce samples with mean counts highest for 1 h- followed by 1 d- and 5 d-held product. Scanning electron microscopy images showed greater bacterial attachment for 5 d-held product, with larger cells aggregates enveloped in a dense extracellular matrix of material which formed a web covering the entire leaf surface. E. coli O157:H7 populations in the flume water or on the equipment surface samples did not differ significantly for the three post-inoculation holding times. Based on these findings, the point at which contamination occurs will impact the extent of E. coli O157:H7 that will detach from leafy greens during processing. Future work may explore whether the contaminant’s location on the leaf, the source of contaminant (e.g., animal feces vs. irrigation water), the region of the country the product was grown in, or the point in the growing season (early vs. late), may have an effect on bacterial transfer or the effectiveness of sanitizers frequently used during leafy green processing. The final objective of this study was to quantify the E. coli O157:H7 transfer from inoculated leafy greens to uninoculated leafy greens during partial dewatering. This project was developed when it was determined that objectives 2, 3 and 4 alone could not be used to 195 determine the extent of cross-contamination from direct leaf-to-leaf contact likely to occur on the conveyer belt, shaker table, etc. As expected, faster cross-contamination was seen using an inoculated to uninoculated ratio of 1:100 than 1:10,000 (w/w) until a maximum population was reached due to the higher number of E. coli O157:H7 cells available for transfer. In summary, these findings confirm that cross-contamination can occur during shaker table dewatering within a very short time frame after a contamination event. Leaf-to-leaf contamination can be minimized by ensuring that all product moves through the leafy green processing line without delay. Overall, this research shows that small numbers of E. coli O157:H7 in the water and on the equipment surfaces can lead to contamination of multiple batches of fresh-cut leafy greens in a processing facility with or without sanitizer. Flume tank wash water needs to contain a chemical sanitizer to minimize cross-contamination of leafy greens during post-harvest processing. A single shred of E. coli O157:H7-contaminated product can quickly spread E. coli O157:H7 to adjacent leaves via leaf-to-leaf contact. Monitoring the spent centrifugation water for microbial contaminants, including E. coli O157:H7, may be helpful in identifying potentially contaminated product. An extended the time between the contamination event and post-harvest leafy green processing will decrease the removal of E. coli O157:H7 from leaf surfaces during leafy green processing. Equipment designed to facilitate cleaning and sanitation and proper monitoring of the preventative controls are essential means to mitigate the extent of bacterial transfer and cross-contamination that can occur during commercial processing of leafy greens to prevent large-scale outbreaks. 196 APPENDICES 197 APPENDIX I Mean Escherichia coli O157:H7 populations on equipment surfaces after processing 198 Table AI.1: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 6 log CFU/g (n=3) 2 Mean ± SD E. coli O157:H7 (log CFU/100 cm ) Location on Processing Equipment Baby Spinach Romaine Lettuce Iceberg Lettuce 1) front left, stainless steel ND 4.69 ± 0.58 5.71 ± 0.39 1) front right, stainless steel ND 5.24 ± 0.14 5.62 ± 0.22 2) left side, feed belt, plastic ND 4.94 ± 1.14 5.34 ± 0.32 2) right side, feed belt, plastic ND 4.33 ± 0.54 5.28 ± 0.34 3) front left feed belt guide, stainless steel ND 5.13 ± 0.56 5.11 3) front right feed belt guide, stainless steel ND 5.71 ± 0.23 5.10 ± 0.34 3) end, left side, feed belt guide, stainless steel ND 1.63 ± 1.08 4.82 ± 0.81 3) end, right side, feed belt guide, stainless steel ND 1.63 ± 1.08 4.10 ± 1.07 4) below the cutting wheel, stainless steel ND 4.23 ± 0.92 5.04 ± 0.72 5) cutting wheel, stainless steel ND 4.58 ± 0.31 4.65 ± 0.31 6) discharge chute, stainless steel ND 5.28 ± 0.52 6.07 ± 0.58 Shredder 199 ± 0.92 Table AI.1 (cont’d) 6) discharge chute, stainless steel ND 5.45 ± 0.23 5.75 ± 0.44 6) discharge chute, stainless steel ND 5.89 ± 0.69 5.60 ± 0.33 6) discharge chute, stainless steel ND 5.70 ± 0.55 5.65 ± 0.56 1) conveyer belt, polyurethane ND 5.57 ± 0.15 5.88 ± 0.72 1) conveyer belt, polyurethane ND 5.23 ± 0.30 5.69 ± 0.42 2) bottom, left side, belt guide, plastic ND 5.38 ± 0.34 6.02 ± 0.52 2) bottom, right side, belt guide, plastic ND 4.75 ± 0.61 5.73 ± 0.40 2) center, left side, belt guide, plastic ND 4.87 ± 0.34 5.36 ± 0.13 2) center, right side, belt guide, plastic ND 4.06 ± 0.57 5.83 ± 0.59 1) top of the conveyer belt shelf, polyurethane ND 5.37 ± 0.47 5.88 ± 0.32 1) bottom of the conveyer belt shelf, polyurethane ND 4.21 ± 0.48 5.52 ± 0.13 1) slanted front, stainless steel 5.01 ± 0.51 5.31 ± 0.42 5.06 ± 0.19 2) left side, interior wall, 2’ from start, stainless steel 3.66 ± 0.58 2.13 ± 0.24 3.75 ± 0.62 Conveyer Belt Flume Tank 200 Table AI.1 (cont’d) 2) right side, interior wall, 2’ from start, stainless steel 3.23 ± 0.12 2.02 ± 0.95 2.68 ± 0.26 2) left side, interior wall, 4’ from start, stainless steel 2.89 ± 0.42 1.39 ± 0.68 2.61 ± 0.07 2) bottom, 4’ from front, stainless steel 3.02 ± 0.65 3.64 ± 0.17 3.94 ± 0.29 2) right side, interior wall, 4’ from start, stainless steel 2.27 ± 0.90 1.00 ± 0.00 2.44 ± 0.37 2) left side, interior wall, 6’ from start, stainless steel 2.25 ± 0.35 1.94 ± 0.82 2.14 ± 0.27 2) right side, interior wall, 6’ from start, stainless steel 1.97 ± 0.76 1.00 ± 0.00 2.00 ± 0.28 2) left interior wall, 8’ from start, stainless steel 2.11 ± 0.53 1.00 ± 0.00 1.65 ± 0.44 2) bottom, 8’ from start, stainless steel 2.41 ± 0.98 2.78 ± 0.56 3.08 ± 0.15 2) right interior wall, 8’ from start, stainless steel 1.70 ± 0.42 1.78 ± 0.68 1.48 ± 0.28 1) front, textured, stainless steel 3.08 ± 0.58 3.93 ± 0.26 3.67 ± 0.29 2) perforated dewatering screen, stainless steel 3.60 ± 0.68 4.42 ± 0.22 4.68 ± 0.01 3) 1’ from start, textured, stainless steel 3.92 ± 0.26 4.59 ± 0.16 4.71 ± 0.05 3) left side, end, textured, stainless steel 3.81 ± 0.45 4.25 ± 0.21 4.59 ± 0.16 4) end (vertical portion), stainless steel 3.04 ± 0.49 3.60 ± 0.20 3.80 ± 0.49 Shaker table 201 Table AI.1 (cont’d) 3) right side, end, textured, stainless steel 3.67 ± 0.11 4.59 ± 0.08 4.53 ± 0.14 3) end (slanted portion), stainless steel 2.50 ± 1.29 2.44 ± 0.24 3.54 ± 0.39 5) left side, recirculation spout, stainless steel 1.75 ± 0.63 1.68 ± 0.61 2.76 ± 0.15 6) right side, recirculation spout, stainless steel 1.28 ± 0.17 1.00 ± 0.00 2.51 ± 0.39 1) interior basket carrier, stainless steel 4.00 ± 0.31 4.95 ± 0.16 5.15 ± 0.63 1) interior basket carrier, stainless steel 4.06 ± 0.38 4.63 ± 0.49 5.14 ± 1.16 1) interior basket carrier, stainless steel 3.94 ± 0.23 4.95 ± 0.04 4.79 ± 0.24 1) interior basket carrier, stainless steel 4.30 ± 0.05 4.47 ± 0.18 4.69 ± 0.28 2) left side, interior wall, stainless steel 3.99 ± 0.22 4.05 ± 0.20 4.23 ± 0.55 2) right side, interior wall, stainless steel 3.99 ± 0.22 4.04 ± 0.28 4.21 ± 0.51 3) viewing window in the lid, plastic 2.62 ± 1.28 1.00 ± 0.00 2.10 ± 0.29 4) drain, stainless steel 5.22 ± 0.23 5.51 ± 0.10 5.95 ± 0.39 Dewatering Centrifuge 202 Table AI.2: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 4 log CFU/g (n=3) 2 Mean ± SD E. coli O157:H7 (log CFU/100 cm ) Location on Processing Equipment Baby Romaine Iceberg Iceberg Iceberg Lettuce Spinach Lettuce Lettuce Lettuce (held 24 h before (finely processing) shredded) Shredder 1) front left, stainless steel ND 4.04 ± 0.19 3.71 ± 0.21 2.59 ± 0.53 1.66 ± 0.77 1) front right, stainless steel ND 4.36 ± 0.07 3.93 ± 0.15 1.00 ± 0.75 2.35 ± 0.81 2) left side, feed belt, plastic ND 4.24 ± 0.07 3.71 ± 0.39 0.77 ± 0.52 1.63 ± 0.79 2) right side, feed belt, plastic ND 4.17 ± 0.17 3.62 ± 0.29 1.43 ± 0.72 1.67 ± 0.78 3) front left feed belt guide, ND 4.06 ± 0.44 3.89 ± 0.36 1.74 ± 0.65 1.48 ± 0.83 stainless steel 203 Table AI.2 (cont’d) 3) front right feed belt guide, ND 4.18 ± 0.28 4.06 ± 0.17 0.88 ± 0.11 2.59 ± 1.19 ND 4.10 ± 0.42 3.00 ± 0.80 1.13 ± 0.17 1.54 ± 0.57 ND 2.94 ± 0.54 2.54 ± 0.90 1.13 ± 0.21 1.52 ± 0.54 ND 2.95 ± 0.93 3.26 ± 0.40 0.93 ± 0.13 2.20 ± 1.67 5) cutting wheel, stainless steel ND 3.85 ± 0.14 2.27 ± 0.24 1.32 ± 1.09 2.33 ± 0.30 6) discharge chute, stainless steel ND 4.21 ± 0.23 4.22 ± 0.34 1.56 ± 1.56 1.32 ± 0.28 6) discharge chute, stainless steel ND 4.80 ± 0.14 3.80 ± 0.40 2.39 ± 0.83 3.08 ± 0.28 6) discharge chute, stainless steel ND 4.52 ± 0.12 3.13 ± 1.22 2.74 ± 0.45 1.28 ± 0.17 6) discharge chute, stainless steel ND 4.28 ± 0.63 3.57 ± 0.44 2.96 ± 0.61 2.81 ± 1.58 ND 4.80 ± 0.07 3.82 ± 0.86 2.99 ± 0.41 2.11 ± 0.85 stainless steel 3) end, left side, feed belt guide, stainless steel 3) end, right side, feed belt guide, stainless steel 4) below the cutting wheel, stainless steel Conveyer Belt 1) conveyer belt, polyurethane 204 Table AI.2 (cont’d) 1) conveyer belt, polyurethane ND 4.57 ± 0.11 4.22 ± 0.05 2.40 ± 0.77 2.80 ± 0.89 2) bottom, left side, belt guide, ND 4.40 ± 0.07 4.16 ± 0.32 3.02 ± 0.37 2.40 ± 0.53 ND 4.40 ± 0.31 4.10 ± 0.33 1.42 ± 1.49 2.35 ± 0.17 ND 4.40 ± 0.37 3.66 ± 0.26 1.59 ± 1.34 2.36 ± 0.90 ND 3.95 ± 0.12 3.73 ± 0.31 1.73 ± 1.33 3.03 ± 0.73 ND 3.94 ± 0.11 4.30 ± 0.11 2.92 ± 0.70 2.90 ± 0.66 ND 4.38 ± 0.23 3.73 ± 0.14 2.88 ± 0.34 2.74 ± 1.28 2.88 ± 0.87 4.46 ± 0.32 3.27 ± 0.08 2.46 ± 0.14 2.64 ± 0.73 plastic 2) bottom, right side, belt guide, plastic 2) center, left side, belt guide, plastic 2) center, right side, belt guide, plastic 1) top of the conveyer belt shelf, polyurethane 1) bottom of the conveyer belt shelf, polyurethane Flume Tank 1) slanted front, stainless steel 205 Table AI.2 (cont’d) 2) left side, interior wall, 2’ from 1.05 ± 0.40 4.00 ± 0.07 1.39 ± 0.13 1.33 ± 2.07 0.77 ± 1.76 2) right side, interior wall, 2’ from 1.12 ± 0.90 1.98 ± 0.64 1.03 ± 0.52 0.91 ± 1.70 0.20 ± 2.07 0.89 ± 0.26 1.28 ± 0.63 1.15 ± 0.23 0.77 ± 1.53 0.79 ± 1.07 1.46 ± 0.73 1.12 ± 0.81 1.86 ± 0.25 -0.35 ± 1.13 2.19 ± 1.53 2) right side, interior wall, 4’ from 0.96 ± 0.55 2.74 ± 0.17 0.66 ± 0.59 2.14 ± 1.70 1.34 ± 2.43 0.10 ± 0.17 0.56 ± 0.43 1.50 ± 0.14 1.45 ± 2.13 1.12 ± 2.51 2) right side, interior wall, 6’ from 0.17 ± 1.17 1.42 ± 0.43 1.04 ± 0.43 1.41 ± 2.10 0.14 ± 1.13 start, stainless steel start, stainless steel 2) left side, interior wall, 4’ from start, stainless steel 2) bottom, 4’ from front, stainless steel start, stainless steel 2) left side, interior wall, 6’ from start, stainless steel start, stainless steel 206 Table AI.2 (cont’d) 2) left interior wall, 8’ from start, 0.78 ± 0.88 1.16 ± 0.44 0.44 ± 0.42 1.70 ± 2.35 -1.00 ± 0.00 0.90 ± 0.79 0.42 ± 0.39 1.71 ± 0.47 2.19 ± 1.34 1.00 ± 0.39 ± 1.67 2.18 ± 0.11 0.65 ± 0.33 1.00 ± 1.74 -0.13 ± 1.51 1) front, textured, stainless steel 1.18 ± 1.03 1.15 ± 1.02 2.21 ± 0.26 2.18 ± 0.45 0.70 ± 1.47 2) perforated dewatering screen, 1.81 ± 0.79 2.86 ± 0.05 2.75 ± 0.05 0.81 ± 1.57 0.96 ± 1.71 3) 1’ from start, textured, stainless 1.80 ± 0.83 3.23 ± 0.15 2.70 ± 0.09 1.84 ± 0.27 1.17 ± 1.89 3.23 ± 0.12 3.05 ± 0.07 2.82 ± 0.32 1.55 ± 2.24 stainless steel 2) bottom, 8’ from start, stainless steel 2) right interior wall, 8’ from start, stainless steel 0.21 Shaker table stainless steel steel 3) left side, end, textured, stainless 1.81 ± 0.14 steel 207 Table AI.2 (cont’d) 4) end (vertical portion), stainless 1.43 ± 0.25 3.14 ± 0.05 2.30 ± 0.23 1.73 ± 0.06 1.39 ± 0.83 1.63 ± 0.14 2.81 ± 0.18 2.99 ± 0.17 3.44 ± 0.64 1.68 ± 2.33 0.93 ± 0.48 3.23 ± 0.21 1.97 ± 0.51 2.99 ± 1.57 1.33 ± 0.35 1.38 ± 1.07 1.84 ± 0.02 0.90 ± 0.63 2.12 ± 2.70 0.45 ± 1.61 0.67 ± 1.60 -0.22 ± 0.68 -0.03 ± 0.95 2.41 ± 2.95 0.07 ± 1.86 2.29 ± 0.02 0.06 ± 0.10 2.52 ± 0.34 3.28 ± 1.52 3.25 ± 0.27 2.29 ± 0.35 3.03 ± 0.78 2.80 ± 0.16 2.96 ± 1.19 3.01 ± 0.45 steel 3) right side, end, textured, stainless steel 3) end (slanted portion), stainless steel 5) left side, recirculation spout, stainless steel 6) right side, recirculation spout, stainless steel Dewatering Centrifuge 1) interior basket carrier, stainless steel 1) interior basket carrier, stainless steel 208 Table AI.2 (cont’d) 1) interior basket carrier, stainless 2.16 ± 0.14 3.06 ± 0.41 2.78 ± 0.19 3.14 ± 0.88 3.10 ± 0.10 2.35 ± 0.44 3.34 ± 0.61 2.85 ± 0.19 3.33 ± 0.66 2.73 ± 1.08 1.69 ± 0.28 2.97 ± 0.31 2.02 ± 0.42 3.38 ± 0.40 2.53 ± 0.09 2) right side, interior wall, stainless 1.73 ± 0.13 3.21 ± 0.16 2.11 ± 0.28 2.90 ± 1.07 2.23 ± 0.74 0.87 ± 0.58 3.07 ± 0.26 0.26 ± 1.58 3.12 ± 1.85 1.80 ± 0.77 3.08 ± 0.25 0.92 ± 0.82 3.87 ± 0.06 3.53 ± 0.68 3.48 ± 0.33 steel 1) interior basket carrier, stainless steel 2) left side, interior wall, stainless steel steel 3) viewing window in the lid, plastic 4) drain, stainless steel 209 Table AI.3: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 2 log CFU/g (n=5) 2 Mean ± SD E. coli O157:H7 (log CFU/100 cm ) Location on Processing Equipment Baby Spinach Romaine Lettuce Iceberg Lettuce 1) front left, stainless steel ND 1.31 ± 0.59 1.53 ± 0.87 1) front right, stainless steel ND 1.24 ± 0.58 1.69 ± 0.33 2) left side, feed belt, plastic ND 0.80 ± 1.48 1.40 ± 0.67 2) right side, feed belt, plastic ND 1.00 ± 0.70 1.16 ± 0.64 3) front left feed belt guide, stainless steel ND 1.18 ± 0.87 1.06 ± 1.29 3) front right feed belt guide, stainless steel ND 0.88 ± 0.55 1.95 ± 0.06 3) end, left side, feed belt guide, stainless steel ND -1.04 ± 0.08 -0.06 ± 1.22 3) end, right side, feed belt guide, stainless steel ND -1.04 ± 0.08 1.05 ± 0.76 4) below the cutting wheel, stainless steel ND 0.17 ± 0.82 1.27 ± 0.60 5) cutting wheel, stainless steel ND 0.86 ± 0.37 0.96 ± 0.55 6) discharge chute, stainless steel ND 1.33 ± 0.29 1.91 ± 0.32 Shredder 210 Table AI.3 (cont’d) 6) discharge chute, stainless steel ND 1.79 ± 0.68 1.84 ± 0.25 6) discharge chute, stainless steel ND 2.18 ± 0.65 1.58 ± 0.32 6) discharge chute, stainless steel ND 2.06 ± 0.40 0.85 ± 0.85 1) conveyer belt, polyurethane ND 1.71 ± 0.56 1.85 ± 0.40 1) conveyer belt, polyurethane ND 1.50 ± 0.70 1.57 ± 0.46 2) bottom, left side, belt guide, plastic ND 1.44 ± 1.04 1.92 ± 0.35 2) bottom, right side, belt guide, plastic ND 1.28 ± 1.37 1.94 ± 0.60 2) center, left side, belt guide, plastic ND 0.72 ± 0.44 1.30 ± 0.54 2) center, right side, belt guide, plastic ND 0.62 ± 0.48 1.50 ± 0.47 1) top of the conveyer belt shelf, polyurethane ND 0.61 ± 1.64 2.04 ± 0.38 1) bottom of the conveyer belt shelf, polyurethane ND 0.42 ± 1.00 1.71 ± 0.60 1) slanted front, stainless steel 1.92 ± 1.01 0.27 ± 1.21 1.52 ± 0.21 2) left side, interior wall, 2’ from start, stainless steel -0.79 ± 0.87 -1.14 ± 0.08 -0.06 ± 1.21 Conveyer Belt Flume Tank 211 Table AI.3 (cont’d) 2) right side, interior wall, 2’ from start, stainless steel -0.09 ± 1.09 -1.14 ± 0.08 -0.22 ± 1.27 2) left side, interior wall, 4’ from start, stainless steel -0.61 ± 0.79 -1.14 ± 0.08 -0.47 ± 0.64 2) bottom, 4’ from front, stainless steel -0.53 ± 0.95 -0.71 ± 0.64 0.13 2) right side, interior wall, 4’ from start, stainless steel -1.18 ± 0.00 -1.14 ± 0.08 -0.15 ± 1.41 2) left side, interior wall, 6’ from start, stainless steel -1.18 ± 0.00 -1.14 ± 0.08 -0.41 ± 1.18 2) right side, interior wall, 6’ from start, stainless steel -0.66 ± 1.16 -1.14 ± 0.08 -0.70 ± 1.06 2) left interior wall, 8’ from start, stainless steel -0.63 ± 1.21 -1.14 ± 0.08 -0.80 ± 0.84 2) bottom, 8’ from start, stainless steel -0.55 ± 0.86 -0.91 ± 0.51 -0.09 ± 1.26 2) right interior wall, 8’ from start, stainless steel -0.84 ± 0.75 -1.14 ± 0.08 -1.18 ± 0.00 1) front, textured, stainless steel -0.29 ± 1.28 -0.79 ± 0.78 -0.08 ± 0.65 2) perforated dewatering screen, stainless steel -0.36 ± 1.27 0.03 ± 0.80 0.52 ± 0.96 3) 1’ from start, textured, stainless steel -0.33 ± 1.33 0.22 ± 0.84 0.85 ± 0.49 3) left side, end, textured, stainless steel 0.01 ± 1.19 -0.27 ± 0.92 0.61 ± 1.01 4) end (vertical portion), stainless steel -0.51 ± 0.98 -0.81 ± 0.72 -0.31 ± 0.80 ± 0.92 Shaker table 212 Table AI.3 (cont’d) 3) right side, end, textured, stainless steel 0.46 ± 1.08 0.36 ± 0.97 0.35 ± 1.42 3) end (slanted portion), stainless steel 0.11 ± 1.32 -1.14 ± 0.08 0.15 ± 1.55 5) left side, recirculation spout, stainless steel -0.76 ± 0.93 -0.85 ± 0.74 -0.32 ± 1.20 6) right side, recirculation spout, stainless steel -1.18 ± 0.00 -1.14 ± 0.08 -0.14 ± 1.43 1) interior basket carrier, stainless steel 0.74 ± 0.34 0.75 ± 1.08 1.30 ± 0.61 1) interior basket carrier, stainless steel 0.43 ± 0.99 0.53 ± 1.44 0.93 ± 0.54 1) interior basket carrier, stainless steel 0.63 ± 0.11 0.56 ± 1.11 0.88 ± 0.57 1) interior basket carrier, stainless steel 0.26 ± 0.87 0.44 ± 1.04 0.70 ± 1.11 2) left side, interior wall, stainless steel -0.19 ± 0.94 -0.09 ± 0.92 0.30 ± 0.91 2) right side, interior wall, stainless steel 0.24 ± 0.89 -0.14 ± 0.80 -0.22 ± 0.88 3) viewing window in the lid, plastic -0.94 ± 0.53 -1.00 ± 0.00 -0.10 ± 1.06 4) drain, stainless steel 0.80 ± 1.18 1.79 ± 0.59 1.71 Dewatering Centrifuge 213 ± 0.47 Table AI.4: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 6 log CFU/g followed by 90.8 kg of uninoculated leafy greens (n=3) 2 Mean ± SD E. coli O157:H7 (log CFU/100 cm ) Location on Processing Equipment Romaine Lettuce Iceberg Lettuce 1) front left, stainless steel 3.30 ± 2.01 2.87 ± 0.39 1) front right, stainless steel 4.05 ± 0.89 3.74 ± 0.22 2) left side, feed belt, plastic 1.29 ± 0.51 2.60 ± 0.32 2) right side, feed belt, plastic 1.00 ± 0.00 3.93 ± 0.34 3) front left feed belt guide, stainless steel 1.45 ± 0.78 4.19 ± 0.92 3) front right feed belt guide, stainless steel 1.57 ± 0.99 4.84 ± 0.34 3) end, left side, feed belt guide, stainless steel 1.58 ± 0.51 3.28 ± 0.81 3) end, right side, feed belt guide, stainless steel 1.74 ± 0.69 3.31 ± 1.07 4) below the cutting wheel, stainless steel 3.51 ± 1.02 5.82 ± 0.72 5) cutting wheel, stainless steel 3.92 ± 0.59 5.20 ± 0.31 6) discharge chute, stainless steel 4.84 ± 0.81 4.53 ± 0.58 Shredder 214 Table AI.4 (cont’d) 6) discharge chute, stainless steel 3.05 ± 2.05 1.95 ± 0.44 6) discharge chute, stainless steel 4.35 ± 0.83 2.31 ± 0.33 6) discharge chute, stainless steel 3.81 ± 0.95 3.13 ± 0.56 1) conveyer belt, polyurethane 4.59 ± 0.77 2.71 ± 0.72 1) conveyer belt, polyurethane 4.80 ± 0.34 2.76 ± 0.42 2) bottom, left side, belt guide, plastic 4.60 ± 0.46 2.54 ± 0.52 2) bottom, right side, belt guide, plastic 4.43 ± 0.77 1.11 ± 0.40 2) center, left side, belt guide, plastic 3.23 ± 0.39 1.67 ± 0.13 2) center, right side, belt guide, plastic 3.66 ± 0.56 2.77 ± 0.59 1) top of the conveyer belt shelf, polyurethane 5.20 ± 0.18 1.61 ± 0.32 1) bottom of the conveyer belt shelf, polyurethane 4.25 ± 1.27 4.34 ± 0.13 1) slanted front, stainless steel 4.19 ± 0.79 3.80 ± 0.19 2) left side, interior wall, 2’ from start, stainless steel 1.69 ± 0.14 2.80 ± 0.62 Conveyer Belt Flume Tank 215 Table AI.4 (cont’d) 2) right side, interior wall, 2’ from start, stainless steel 0.85 ± 1.60 1.98 ± 0.26 2) left side, interior wall, 4’ from start, stainless steel 0.30 ± 1.17 4.50 ± 0.07 2) bottom, 4’ from front, stainless steel 1.61 ± 0.33 3.98 ± 0.29 2) right side, interior wall, 4’ from start, stainless steel -0.22 ± 0.68 3.96 ± 0.37 2) left side, interior wall, 6’ from start, stainless steel 0.99 ± 0.45 3.60 ± 0.27 2) right side, interior wall, 6’ from start, stainless steel 0.51 ± 0.35 3.26 ± 0.28 2) left interior wall, 8’ from start, stainless steel 0.28 ± 0.17 2.53 ± 0.44 2) bottom, 8’ from start, stainless steel 1.47 ± 0.37 4.79 ± 0.15 2) right interior wall, 8’ from start, stainless steel 0.61 ± 0.51 3.21 ± 0.28 1) front, textured, stainless steel 2.23 ± 0.20 3.58 ± 0.29 2) perforated dewatering screen, stainless steel 2.71 ± 0.22 1.56 ± 0.01 3) 1’ from start, textured, stainless steel 2.68 ± 0.24 1.08 ± 0.05 3) left side, end, textured, stainless steel 2.39 ± 0.41 0.59 ± 0.16 4) end (vertical portion), stainless steel 2.60 ± 0.46 1.22 ± 0.49 Shaker table 216 Table AI.4 (cont’d) 3) right side, end, textured, stainless steel 2.93 ± 0.16 -0.63 ± 0.14 3) end (slanted portion), stainless steel 1.88 ± 0.50 1.44 ± 0.39 5) left side, recirculation spout, stainless steel 0.68 ± 0.46 -0.68 ± 0.15 6) right side, recirculation spout, stainless steel 0.24 ± 1.15 -0.39 ± 0.39 1) interior basket carrier, stainless steel 3.30 ± 2.01 1.01 ± 0.63 1) interior basket carrier, stainless steel 4.05 ± 0.89 0.82 ± 1.16 1) interior basket carrier, stainless steel 1.29 ± 0.51 1.43 ± 0.24 1) interior basket carrier, stainless steel 1.00 ± 0.00 2.47 ± 0.28 2) left side, interior wall, stainless steel 1.45 ± 0.78 2.24 ± 0.55 2) right side, interior wall, stainless steel 1.57 ± 0.99 2.08 ± 0.51 3) viewing window in the lid, plastic 1.58 ± 0.51 2.03 ± 0.29 4) drain, stainless steel 1.74 ± 0.69 2.26 ± 0.39 Dewatering Centrifuge 217 Table AI.5: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 4 log CFU/g followed by 90.8 kg of uninoculated leafy greens (n=3) 2 Mean ± SD E. coli O157:H7 (log CFU/100 cm ) Location on Processing Equipment Romaine Lettuce Iceberg Lettuce 1) front left, stainless steel 0.77 ± 0.59 0.77 ± 0.59 1) front right, stainless steel -0.02 ± 0.90 -0.02 ± 0.90 2) left side, feed belt, plastic 0.34 ± 1.19 0.34 ± 1.19 2) right side, feed belt, plastic 1.60 ± 0.80 1.60 ± 0.80 3) front left feed belt guide, stainless steel 0.30 ± 1.22 0.30 ± 1.22 3) front right feed belt guide, stainless steel -0.17 ± 1.43 -0.17 ± 1.43 3) end, left side, feed belt guide, stainless steel -0.67 ± 0.58 -0.67 ± 0.58 3) end, right side, feed belt guide, stainless steel -1.00 ± 0.00 -1.00 ± 0.00 4) below the cutting wheel, stainless steel 1.00 ± 0.63 1.00 ± 0.63 5) cutting wheel, stainless steel 1.00 ± 0.50 1.00 ± 0.50 6) discharge chute, stainless steel 2.08 ± 0.59 2.08 ± 0.59 Shredder 218 Table AI.5 (cont’d) 6) discharge chute, stainless steel 1.42 ± 0.37 1.42 ± 0.37 6) discharge chute, stainless steel -0.04 ± 0.94 -0.04 ± 0.94 6) discharge chute, stainless steel -0.67 ± 0.58 -0.67 ± 0.58 1) conveyer belt, polyurethane 2.07 ± 0.95 2.07 ± 0.95 1) conveyer belt, polyurethane 1.61 ± 0.85 1.61 ± 0.85 2) bottom, left side, belt guide, plastic 1.88 ± 0.33 1.88 ± 0.33 2) bottom, right side, belt guide, plastic 1.52 ± 0.23 1.52 ± 0.23 2) center, left side, belt guide, plastic 0.81 ± 0.53 0.81 ± 0.53 2) center, right side, belt guide, plastic -0.41 ± 1.03 -0.41 ± 1.03 1) top of the conveyer belt shelf, polyurethane 1.85 ± 0.52 1.85 ± 0.52 1) bottom of the conveyer belt shelf, polyurethane 1.88 ± 0.97 1.88 ± 0.97 1) slanted front, stainless steel 2.23 ± 0.31 2.23 ± 0.31 2) left side, interior wall, 2’ from start, stainless steel -0.29 ± 0.78 -0.29 ± 0.78 Conveyer Belt Flume Tank 219 Table AI.5 (cont’d) 2) right side, interior wall, 2’ from start, stainless steel -0.78 ± 0.68 -0.78 ± 0.68 2) left side, interior wall, 4’ from start, stainless steel -0.68 ± 0.85 -0.68 ± 0.85 2) bottom, 4’ from front, stainless steel -0.11 ± 1.02 -0.11 ± 1.02 2) right side, interior wall, 4’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) left side, interior wall, 6’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) right side, interior wall, 6’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) left interior wall, 8’ from start, stainless steel -0.68 ± 0.85 -0.68 ± 0.85 2) bottom, 8’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) right interior wall, 8’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 1) front, textured, stainless steel 0.10 ± 0.17 0.10 ± 0.17 2) perforated dewatering screen, stainless steel 0.64 ± 0.19 0.64 ± 0.19 3) 1’ from start, textured, stainless steel -0.09 ± 0.95 -0.09 ± 0.95 3) left side, end, textured, stainless steel 0.10 ± 1.11 0.10 ± 1.11 4) end (vertical portion), stainless steel 0.26 ± 0.24 0.26 ± 0.24 Shaker table 220 Table AI.5 (cont’d) 3) right side, end, textured, stainless steel 0.44 ± 0.42 0.44 ± 0.42 3) end (slanted portion), stainless steel -0.68 ± 0.85 -0.68 ± 0.85 5) left side, recirculation spout, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 6) right side, recirculation spout, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 1) interior basket carrier, stainless steel 0.77 ± 0.59 0.77 ± 0.59 1) interior basket carrier, stainless steel -0.02 ± 0.90 -0.02 ± 0.90 1) interior basket carrier, stainless steel 0.34 ± 1.19 0.34 ± 1.19 1) interior basket carrier, stainless steel 1.60 ± 0.80 1.60 ± 0.80 2) left side, interior wall, stainless steel 0.30 ± 1.22 0.30 ± 1.22 2) right side, interior wall, stainless steel -0.17 ± 1.43 -0.17 ± 1.43 3) viewing window in the lid, plastic -0.67 ± 0.58 -0.67 ± 0.58 4) drain, stainless steel -1.00 ± 0.00 -1.00 ± 0.00 Dewatering Centrifuge 221 Table AI.6: E. coli O157:H7 populations (mean ± SD) on equipment surfaces after processing 22.7 kg of leafy greens inoculated at 2 log CFU/g followed by 90.8 kg of uninoculated leafy greens (n=5) 2 Mean ± SD E. coli O157:H7 (log CFU/100 cm ) Location on Processing Equipment Romaine Lettuce Iceberg Lettuce 1) front left, stainless steel -0.19 ± 0.99 -1.18 ± 0.00 1) front right, stainless steel 0.02 ± 1.13 -0.80 ± 0.84 2) left side, feed belt, plastic -0.57 ± 0.87 -0.71 ± 0.64 2) right side, feed belt, plastic -0.82 ± 0.80 -1.18 ± 0.00 3) front left feed belt guide, stainless steel -0.17 ± 1.51 -1.18 ± 0.00 3) front right feed belt guide, stainless steel -0.53 ± 0.90 -1.18 ± 0.00 3) end, left side, feed belt guide, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 3) end, right side, feed belt guide, stainless steel -0.71 ± 0.64 -1.18 ± 0.00 4) below the cutting wheel, stainless steel 0.01 ± 1.24 -1.18 ± 0.00 5) cutting wheel, stainless steel -0.85 ± 0.74 -1.18 ± 0.00 6) discharge chute, stainless steel 0.99 ± 1.28 -0.94 ± 0.53 Shredder 222 Table AI.6 (cont’d) 6) discharge chute, stainless steel 1.01 ± 0.98 -0.41 ± 1.06 6) discharge chute, stainless steel 0.60 ± 1.25 -0.94 ± 0.53 6) discharge chute, stainless steel -0.72 ± 1.02 -1.18 ± 0.00 1) conveyer belt, polyurethane 0.85 ± 1.26 0.02 ± 0.74 1) conveyer belt, polyurethane 1.04 ± 0.16 -0.55 ± 0.86 2) bottom, left side, belt guide, plastic 1.09 ± 0.60 -0.76 ± 0.93 2) bottom, right side, belt guide, plastic 0.96 ± 0.36 -1.18 ± 0.00 2) center, left side, belt guide, plastic -0.88 ± 0.66 -1.18 ± 0.00 2) center, right side, belt guide, plastic -0.40 ± 1.07 -1.18 ± 0.00 1) top of the conveyer belt shelf, polyurethane 0.18 ± 0.92 -0.57 ± 0.87 1) bottom of the conveyer belt shelf, polyurethane -0.51 ± 0.98 -1.18 ± 0.00 1) slanted front, stainless steel 0.37 ± 0.96 -0.71 ± 0.64 2) left side, interior wall, 2’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 Conveyer Belt Flume Tank 223 Table AI.6 (cont’d) 2) right side, interior wall, 2’ from start, stainless steel -0.88 ± 0.66 -1.18 ± 0.00 2) left side, interior wall, 4’ from start, stainless steel -0.94 ± 0.53 -1.18 ± 0.00 2) bottom, 4’ from front, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) right side, interior wall, 4’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) left side, interior wall, 6’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) right side, interior wall, 6’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 2) left interior wall, 8’ from start, stainless steel -0.88 ± 0.66 -1.18 ± 0.00 2) bottom, 8’ from start, stainless steel -0.88 ± 0.66 -1.18 ± 0.00 2) right interior wall, 8’ from start, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 1) front, textured, stainless steel -0.53 ± 0.90 -1.18 ± 0.00 2) perforated dewatering screen, stainless steel -0.53 ± 0.95 -1.18 ± 0.00 3) 1’ from start, textured, stainless steel -0.42 ± 1.05 -1.18 ± 0.00 3) left side, end, textured, stainless steel -0.76 ± 0.93 -0.94 ± 0.53 4) end (vertical portion), stainless steel -0.40 ± 1.08 -0.94 ± 0.53 Shaker table 224 Table AI.6 (cont’d) 3) right side, end, textured, stainless steel -0.55 ± 1.39 -1.18 ± 0.00 3) end (slanted portion), stainless steel -1.18 ± 0.00 -1.18 ± 0.00 5) left side, recirculation spout, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 6) right side, recirculation spout, stainless steel -1.18 ± 0.00 -1.18 ± 0.00 1) interior basket carrier, stainless steel -0.19 ± 0.99 -1.18 ± 0.00 1) interior basket carrier, stainless steel 0.02 ± 1.13 -0.80 ± 0.84 1) interior basket carrier, stainless steel -0.57 ± 0.87 -0.71 ± 0.64 1) interior basket carrier, stainless steel -0.82 ± 0.80 -1.18 ± 0.00 2) left side, interior wall, stainless steel -0.17 ± 1.51 -1.18 ± 0.00 2) right side, interior wall, stainless steel -0.53 ± 0.90 -1.18 ± 0.00 3) viewing window in the lid, plastic -1.18 ± 0.00 -1.18 ± 0.00 4) drain, stainless steel -0.71 ± 0.64 -1.18 ± 0.00 Dewatering Centrifuge 225 APPENDIX II Isolation and Identification of Gas-Producing Yeasts from Maraschino Cherries 226 Isolation and Identification of Gas-Producing Yeasts from Maraschino Cherries By Yinfa Zhang, Lei Zhang, Annemarie L. Buchholz, and Elliot T. Ryser AII.1 ABSTRACT Maraschino cherry production is a time-consuming and complex process with gas identified as an occasional quality defect. The objective of this study was to identify the specific yeasts responsible for gas production in jars of maraschino cherries. Thirty-eight samples of commercial maraschino cherry syrup were collected late in the production process and examined for yeasts by enriching in carbohydrate purple broth containing 1% high fructose corn syrup or corn syrup (pH 6.18 - 6.52), 50% aqueous cherry syrup (pH 3.66-3.81), and 50% cherry syrup diluted 1:1 in trypticase soy broth containing 0.6% yeast extract (Cherry:TSB-YE; pH 6.19 - 6.66). Yeasts were isolated on trypticase soy agar containing 0.6% yeast extract (TSA-YE), deMan-Rogosa-Sharpe agar (MRS), and potato dextrose agar (PDA) (48 h/28°C). Gas production from selected isolates was confirmed in the previous enrichment media using Durham tubes with each yeast isolate biochemically identified using API 20 C AUX test strips and through PCR. Eighteen of the 38 samples yielded gas-producing yeasts. Seventeen isolates were identified through PCR as Zygosaccharomyces bailii with one isolate identified as Clavispora lusitaniae. Mean D10-values at 60 °C of these three yeasts ranged from 4.3 to 6.9 seconds, indicating adequacy of the current pasteurization treatment 227 AIII.2 INTRODUCTION By definition, maraschino cherries are cherries that have been “dyed red, impregnated with sugar and packed in a sugar syrup flavored with oil of bitter almonds or a similar flavor” (FDA 1980). Maraschino cherry production begins with the bleaching of fresh sweet cherries in a brine solution containing 0.75 to 1.5% sulfur dioxide and 3000 to 5000 ppm calcium salts for 6 to 18 months. These bleached yellow cherries are then pitted, soaked in water to leach out the brine and impregnated with 40° Brix sugar syrup containing artificial colors and flavors (Nordlee and others 1985). After jarring cherries with or without pasteurization, the finished product is typically shelf-stable for a minimum of 3 years. While the microbiological safety of maraschino cherries has remained a non-issue, gas production of unknown origin is an occasional quality defect observed in jars of finished product. 228 AII.3 MATERIALS AND METHODS AII.3.1 Sample collection. Twenty-four samples of commercial maraschino cherry syrup and three yellow cherry samples were collected in Whirl-Pak TM bags at different points during maraschino cherry production in a large-scale processing plant. These samples, along with eleven maraschino cherry jar samples (2 prior to pasteurization and 9 post-pasteurization) were transported to Michigan State University during a period from October, 2008 to May, 2009. AII.3.2 Enrichment. Using Durham tubes for gas detection, 1 ml original sample aliquots were enriched in 12 ml of carbohydrate purple broth (Difco, Becton Dickinson, Sparks, MD) (CPB) + 1% high fructose corn syrup, CPB + 1% corn syrup (pH 6.18 6.52), 50% aqueous cherry syrup (provided by the maraschino cherry processor) diluted 1:1 in distilled water (pH 3.66 - 3.81), and 50% cherry syrup diluted 1:1 in trypticase soy broth (Difco, Becton Dickinson) + 0.6% yeast extract (Difco, Becton Dickinson) (Cherry:TSB-YE; pH 6.19 - 6.66). All tubes were incubated at 28 °C and visually examined for gas production during 30 days of incubation at 28 °C. AII.3.4 Isolation. A 10 µl aliquot from enrichments producing gas was streaked to plates of trypticase soy agar (Difco, Becton Dickinson) + 0.6% yeast extract (TSA-YE), deMan-Rogosa-Sharpe agar (Oxoid, Oxoid Ltd, Cambridge, UK) (MRS), and potato dextrose agar (Difco, Becton Dickinson) (PDA) and incubated for 48 h at 28 °C. Three colonies per plate were streaked on the initial culture medium (TSA-YE, MSR or PDA) in duplicate to obtain pure isolates. These isolates were then inoculated into tubes of the same media to identify gas-producing isolates. 229 AII3.5 Biochemical identification of yeast isolates. All gas-producing isolates initially classified as yeasts based on morphology and Gram-staining were biochemically speciated in duplicate using API 20 C AUX test strips following the detailed usage instructions provided by bioMérieux, Inc. Presence of hyphae was confirmed through the method recommended by bioMérieux, Inc. Isolates were streaked on one quadrant of a corn meal agar (CMA) w/ 1% Tween 80 plate, in 2 - 3 parallel lines a few mm apart, an 'S' was then streaked over these lines in a dollar sign configuration. A glass cover slip was placed over the streaked lines and the plates were incubated at room temperature (22 ± 2 °C) for 3 - 7 days. Using the 40X objective placed directly on the glass cover slip, hyphae presence or absence was observed first at the crossroads of the streaks in the center and moving from the center toward the outer edge. One isolate from each species identified with the API 20 C AUX test strips was selected for confirmation by sequencing and further analysis. AII3.6 Identification of yeast isolates through sequencing. Yeast colonies were grown for 1 day in YEPD (1% yeast extract, 2% peptone, 2% glucose) at 30 °C. The overnight liquid cultures (1.5 ml) were centrifuged at 14,000 rpm for 4 min and the yeast DNA was then extracted using DNA-Pure yeast genomic kit, from CPG Inc. Pure Biotech Product (Lincoln Park, NJ). DNA was amplified by PCR using the primers; NL1= 5¹-ATA TCA ATA AGC GGA GGA AAA G-3¹ NL4=5¹-GGT CCG TGT TTC AAG ACG G-3¹ The PCR cycling parameters were as follows: 1 cycle at 94 °C for 5 minutes, 36 cycles of 1 minute at 94 °C, 1 minute at 52°C, and 2 minutes at 72 °C., 1 cycle for final elongation of 10 minutes at 72°C. The PCR product was loaded onto a 0.7% agarose gel 230 and run at 75 volts and after 30 minutes, the DNA band was cut from the gel and extracted using Qiaquick gel extraction kit (Valencia, CA). The samples were then sequenced using the NL1 and NL4 primers at the Biotechnology Resource Center (Cornell University) and the nucleotide sequence was searched for homology using the NCBI Blast data bank for identity. AII.3.7 Gas confirmation during small-scale maraschino cherry production. Two lots of commercially brined and leached frozen yellow cherries were obtained from a large-scale maraschino processor. The small-scale (Figure AII.1) maraschino cherry production system comprised of 2 containers connected by the tube with a tap for flow control. Container 1 was placed ~24 inches lower than Container 2. The frozen yellow cherries were placed in Container 1 fill it with water to the top leaving no air space. The cherries defrosted with tap water ~3.5ml/min for a few hours and the temperature was monitored to determine when the products had thawed completely. The flow was halted and add the commercial sugar syrup (40 °Brix) was added to Container 2. The syrup was inoculated to have the same species of gas-producing yeasts identified with the API 20 C 5 AUX test strips at a level of ~10 CFU/ml. Syrup was allowed to flow from Container 2 into Container 1 until the air in the system was completely removed from the outlet tube connected with the top of the Container 1. The tap was adjusted so the liquid in Container 1 would flow out at ~ 20 ml/hr. or 0.33 ml/min from the outlet tube. The outlet tube was used to determine if gas was being produced during the conversion. The °Brix in the outflow was monitored to determine when the maraschino cherries were finished – once the sugar concentration of the outlet syrup increased from 0 °Brix to 40 °Brix over 3 - 5 231 days at ambient temperature (22 ± 1 °C). The system can run continuously. The lab-scale experiment was done in duplicate. AII.3.8 Survival curve for yeast isolates. Maraschino cherry syrup was diluted to achieve ratios of 1:7, 1:3, 1:1 and 3:1 commercial cherry syrup (OCCS) to distilled water 7 (DW). 10 µl aliquots of the three yeast isolates (10 CFU/ml) were transferred into individual autoclaved test tubes with Durham tubes containing 12 ml of diluted or undiluted cherry syrup (Table AII.1). The tubes were observed twice daily for gas production. Using the same OCCS to DW ratios, 50 ml sterile centrifuge tubes were filled with 12 ml sterile cherry syrup spiked individually with 10 µl of the three yeast isolates to determine generation time at ambient temperature (22 ± 1 °C). Daily or every two days, a 100 µl aliquot was removed from each tube to determine yeast generation time. Samples were appropriately diluted in sterile 1% (w/v) phosphate buffer (8.5 g/L NaCl, 1.44 g/L Na2HPO4, and 0.24 g/L KH2PO4, J.T. Baker, Mallinckrodt Baker Inc., Phillipsburg, NJ) (PBS) and plated on PDA. Following 48 h of incubation at 28°C, all colonies resembling yeasts were counted to obtain the CFU/ml at that specific time point. Each experiment was replicated in duplicate. AII.3.9 Thermal inactivation. Thermal resistance of three gas-producing yeasts was assessed in triplicate using a series of screw-capped 50 ml sterile centrifuge tubes containing 10 ml of autoclaved 40 °Brix syrup tempered to 55, 58 or 60 °C in a water 6 bath. These submerged tubes were individually inoculated to contain ~10 CFU/ml of each gas-producing yeast, removed at five predetermined intervals and rapidly cooled to < 36 °C in a ice water bath. A 1 ml aliquot from each tube was appropriately diluted and 232 plated on PDA to quantify survivors. D10-values and Z-values were calculated from the semi-log survivor curve. AII3.10 Laboratory pasteurization. Maraschino cherry jars (225 ml) with 24 months of shelf life at ambient temperature, obtained from a large-scale cherry processor, were tempered in a water bath in duplicate until the core temperature of each jar reached either 60 or 71.1 °C. The lids of the jar samples were removed and each jar was individually inoculated with a 1 ml aliquot of one of three yeast isolate suspensions 8 containing ~10 /ml. New sterile lids replaced the old lids. The jars were removed from the water bath and placed immediately into ice water, dropping the internal temperature to < 36 °C within 40 seconds. Jars were stored at ambient temperature (22±1 °C), and checked daily for a popped lid, indicating gas in the sample. AII.3.11 Statistical analysis. The generation time (hours) of yeast = {the incubation period (minutes)/[(log CFU/ml at end of incubation – log CFU/ml at beginning of incubation)/0.301]}/60. In order to correlate the syrup’s sugar concentration or pH-value with the minimum generation time of the yeasts, the minimum generation time was converted to log (minimum generation time) and submitted to Spearman’s correlation using SAS statistical software (Statistical Analysis System; SAS Institute Incorporation, Campus Drive Cary, NC, USA). The Student’s t-test was applied to determine if there were significant differences between mean D-values of three yeasts. For all tests, α = 0.05. 233 AII.4 RESULTS AII.4.1 Isolation of gas-producing yeasts. Eighteen (Table AII.2) of the 38 samples yielded gas-producing yeasts based on cellular morphology. Yeasts that fit the selection criteria were round, white colonies with fringed edges, appearing on the streaked plates after 48h incubation at 28 °C, Gram-positive and were positive for Durham tube gas production. These same yeast isolates produced gas following reinoculation into enrichment media with faster gas production seen in Cherry:TSB-YE compared to the other broths. Production of gas was visible one day after inoculation in Cherry:TSB-YE, two days after inoculation in CPB + 1% high fructose corn syrup or CPB + 1% corn syrup, and three days after inoculation in 50% aqueous cherry syrup diluted 1:1 in distilled water. AII.4.2 Identification of yeast isolates. Gas-producing yeast isolates from sixteen samples were identified as Candida krusei/inconspicua (33.9% ID) with one isolate each of Candida magnoliae (62.4% ID) and Candida lusitaniae/famata (C. lusitaniae 74.6 % ID, C. famata 59.1% ID) by the API 20 C AUX test strips. One representative, Y1, was selected from the sixteen isolates identified as Candida krusei/inconspicua, and used in further testing to confirm gas during lab-scale maraschino cherry production, survivor curve development and thermal inactivation in cherry syrup. C. magnoliae and C. lusitaniae/famata labeled Y2 and Y3, respectively, also underwent further analysis. Hyphae appeared on the outer edge of Y1 and Y2 colonies on PDA plates after ten days stored at ambient temperature (22 ± 1 °C). These findings indicate identifications may be different when using other media or longer incubation times for detection of hyphae/pseudo-hyphae than those recommended by bioMérieux, Inc. If the results of the 234 Y1 and Y2 hyphae test were positive, Y1 would have been identified as Candida norvegensis (ID 43.8 %) and Y2 as Geotrichum capitatum (ID 64.4 %). Percent confidence in the identification for the API 20 C AUX test strip results was small, implying that other test methods would be needed for further identification. After comparing the yeast isolates nucleotide sequences to the NCBI Blast Database, Y1 and Y2 were both identified as Zygosaccharomyces bailii. Isolate Y3 was identified as Clavispora lusitaniae. Y1 and Y2 only grew on PDA and MRS, with similar colony and cellular morphology, Gram reaction and the time needed for gas-production in maraschino cherry syrup. Y3 grew on all three plating media and produced larger colonies than the other yeast isolates. Gas production was slowest for Y3. Of thirty-eight samples, Y1 was isolated from fourteen of twenty-three maraschino cherry syrup samples and two maraschino jar samples before pasteurization. Y2 was found in just one maraschino cherry syrup sample, and Y3 was isolated from one of three yellow cherry samples. Yeast was not found in any of the nine pasteurized maraschino cherry jar samples. The results indicated Y1 could survive in products during many stages of cherry production prior to pasteurization and was the most widely distributed yeast causing gas defects (Table AII.2). AII.4.3 Gas confirmation during small-scale maraschino cherry production. Three yeast isolates were individually inoculated into 40 °Brix commercial cherry syrup (Figure AII.1A) to obtain 5.69 ± 0.34 log CFU/ml Y1, 6.41 ± 0.09 log CFU/ml Y2, and 6.93 ± 0.25 log CFU/ml Y3 in the syrup. During maraschino cherry production, gas was observed in the outlet tube of the container holding the cherries three days after the syrup 235 had been inoculated. These visual observations confirm that the yeasts originally isolated from the maraschino cherries and syrup are responsible for the observed gas defects. AII.4.4 Survival curve for yeast isolates. Gas was visible in the Durham tubes six days after inoculation with yeasts (Table AII.3). The time until gas was observed for all three yeasts was the shortest in 21.8 °Brix syrup while the amount of gas production within 20 days was the longest in 31.5 °Brix syrup. Y1 could survive in 50% aqueous cherry syrup for at least 279 days during the trial in the lab. A positive correlation between the syrup’s sugar concentration and log minimum generation time was determined through line regression and correlation analysis (Table AII.3). Figure AII.2 shows that the minimum generation time increases with the increase in sugar concentration indicating that the higher concentrations inhibit the growing of yeasts. The negative correlation between syrup pH and log minimum generation time determined through line regression and correlation analysis was not significant (P > 0.05, Table AII.3 and Figure AII.3). The survival curves of the yeast isolates in five different sugar concentrations of maraschino cherry syrup at ambient temperature (22 ± 1 °C) are displayed in Figures AII.4, AII.5, and AII.6. While the biochemical tests resulted in different identifications, the similarities observed in the growth kinetics of Y1 and Y2 agree with the identifications by sequencing and indicate that these two isolates may be the same species. AII.4.5 Thermal inactivation and laboratory pasteurization. Mean D10-values at 60 °C for the yeast isolates ranged from 4.3 to 6.9 seconds (Table AII.4), suggesting they are unable to survive the post-bottling pasteurization step (160 °F, 5 minutes) 236 adopted by the maraschino cherry making plant. Although no significant difference (P > 0.05) among D10-values of three yeasts at 60 °C was noted, the D10-value of Y3 at 55 °C was higher than D10-values of Y1 and Y2 (P < 0.05) indicating Y3 is more thermally resistance than the other two. The Z-value of these yeast isolates ranged from 3.7 to 4.7 °C (Table AII.4). Following lab scale pasteurization, the lids of the maraschino cherry jars remained depressed even seven months later after pasteurizing. These findings establish that reaching the critical central temperature needed to decrease the yeast populations present in cherries during manufacture is one important step in producing a final product of high quality. 237 AII.5 FIGURES Figure AII.1: Maraschino cherry production: (A) laboratory-scale set-up and (B) gas visible in the outlet tube A B 238 Figure AII.2: The correlation between syrup concentration (°Brix) and generation time of yeast isolates a a 2 The line was fit to Y1 data using the equation y = 0.8892 + 0.0145x where R = 2 0.8369. Y2 data was fit with the linear equation y = 0.7486 + 0.0205x where R = 2 0.8135. A line was fit to the Y3 data using the equation y = 0.7067 + 0.0353x where R = 0.9015. 239 Figure AII.3: The correlation between pH of syrup and generation time of yeast isolates a 2 The line was fit to Y1 data using the equation y = 5.4869 - 1.1859x where R = 0.7122. 2 Y2 data was fit with the linear equation y = 7.1695 + 1.6591x where R = 0.7684. A line 2 was fit to the Y3 data using the equation y = 13.595 - 3.3673x where R = 0.9638. 240 Figure AII.4: The survival curve of isolate Y1 in different °Brix syrup at ambient temperature 241 Figure AII.5: The survival curve of isolate Y2 in different °Brix syrup at ambient temperature 242 Figure AII.6: The survival curve of isolate Y3 in different °Brix syrup at ambient temperature 243 AII.6 TABLES Table AII.1: Sugar concentration (°Brix) and pH-value of cherry syrup Sugar concentration (°Brix) b pH Before Sample OCCS:DW After Before After autoclaving autoclaving autoclaving autoclaving a 1 1:7 5.5 5.6 3.68 3.82 2 1:3 11.2 11.6 3.62 3.70 3 1:1 21.8 22.6 3.50 3.57 4 3:1 31.5 32.7 3.28 3.48 5 No dilution 40.8 41.7 3.32 3.42 a During the study, ambient temperature (mean ± SD) was 22.9 ± 0.8 °C and cherry syrup temperature (mean ± SD) was 21.5 ± 0.3 °C b Sugar concentration and pH-value were measured by SPER Digital Refractometer 300017 (SPER Scientific, Scottsdale, AZ) (0 - 45 °Brix) and HANNA Microprocessor pH Meter (Hanna Instruments, Woonsocket, RI), respectively. 244 Table AII.2: The number gas-producing yeast positive cherry samples Samples positive Product Samples collected for gas-producing yeasts Yellow cherries 3 1 Maraschino cherry syrup 24 15 Cherry jars before pasteurization 2 2 Cherry jars after pasteurization 9 0 Total 38 18 245 Table AII.3: Gas production and minimum generation time in sugar syrup at 22 ± 1 °C within 30 days (n = 2) °Brix pH isolate Y1 Days until gas Gas Yeast production a Minimum generation production time (hours) 9.06 3.70 ++ 8 - 10 13.52 3.57 +++ 6-8 14.60 31.5 3.48 ++++ 7-8 16.72 40.8 3.42 + 14 - 15 20.71 5.5 3.82 - NA 8.76 11.2 3.70 ++ 8 - 10 9.11 21.8 3.57 +++ 6-7 15.14 31.5 3.48 ++++ 7-8 15.33 40.8 3.42 + 14 - 15 25.22 5.5 3.82 - NA 6.70 11.2 3.70 +- 9 - 10 10.40 21.8 3.57 +- 8-9 44.45 31.5 3.48 - NA 100.00 40.8 a NA 21.8 Y3 3.82 - 11.2 Y2 5.5 3.42 - NA 89.50 Gas production was rated on a scale where “++++” represented the greatest gas production followed by “+++”, “++”, “+” and “+-“, and where “-“ represented no visible gas production. 246 Table AII.4: The D10-values and Z-values of yeasts isolated from maraschino cherry syrup (n = 3) Yeast isolate Temperature (°C) D10-value (sec.) Y1 8.6 4.9 B 55 64.7 A 20.7 C 60 4.3 B 55 156.5 D 58 20.8 C 60 6.9 2 Z-value (°C) R B 58 a A 60 Y3 55.1 58 Y2 55 a 4.7 0.9296 4.3 0.9110 3.7 0.9949 B D10-values not labeled with by the same letter are significantly different (P ≤ 0.05). 247 APPENDIX III Pulsed-Field Gel Electrophoresis as a Predictor of Listeria monocytogenes Biofilm Formation 248 Pulsed-Field Gel Electrophoresis as a Predictor of Listeria monocytogenes Biofilm Formation By Gordon R. Davidson, Annemarie L. Buchholz, Zhinong Yan, and Elliot T. Ryser AIII.1 ABSTRACT Persistence and spread of Listeria monocytogenes in food manufacturing and retail environments is greatly aided by the ability of this pathogen to form biofilms. While numerous Listeria isolates from environmental surveys have been serotyped and subjected to various strain-specific molecular typing methods, relatively few studies have characterized such strains in terms of biofilm formation, which could lead to improved Listeria intervention strategies. In this study, the relationship between biofilm formation, serotype and pulsed-field gel electrophoresis (PFGE) type was determined in a set of 30 L. monocytogenes environmental isolates from one delicatessen. All 30 L. monocytogenes isolates were serotyped using a PCR-based method with five different primer sets and then subjected to PFGE typing using the PulseNet protocol with restriction enzymes AscI and ApaI. Biofilm formation was quantified using a standard microtiter plate assay with Modified Welshimer’s Broth as the growth medium. After averaging the 9 optical density readings from triplicate wells for each of three replicates, each isolate was classified as a weak, medium or strong biofilm former. Overall, 16, 7, 6 and 1 L. monocytogenes isolate belonged to serotypes 1/2b(3b), 1/2a(3a), 4b(d,e) and 4a/c, respectively. A total of 12 PFGE types grouped into three distinct clusters. All 7 strong biofilm-forming strains belonged to serotype 1/2a(3a) and to two closely related 249 PFGE types containing a unique high kbp doublet using ApaI that were only 60.9% similar compared to the remaining 23 isolates. These 7 strong biofilm formers were isolated from multiple locations over 9 months. In contrast, 4 of 15 weak and 2 of 8 medium biofilm-forming strains persisted for 10 months, thereby reinforcing previous correlations between biofilm formation and persistence. These findings demonstrate that PFGE typing can be a predictor of L. monocytogenes biofilm formation. However, other factors in addition to biofilm formation are also clearly important in persistence. 250 AIII.2 INTRODUCTION In 2002, consumption of Listeria-contaminated delicatessen meat was identified as the leading cause of listeriosis in the United States with a recent unpublished survey indicating a contamination rate 7 to 8-times higher for delicatessen- as opposed to manufacture-sliced products. Persistence and spread of L. monocytogenes in both retail and foodservice environments is greatly aided by the ability of this pathogen to form biofilms. Hence, 30 L. monocytogenes strains previously isolated from a commercial delicatessen and characterized by pulsed-field gel electrophoresis (PFGE) typing were now serotyped and assessed for biofilm formation (Yan and others 2008), which could lead to improved Listeria intervention strategies. 251 AIII.3 MATERIALS AND METHODS AIII.3.1 Study design. Thirty delicatessen isolates previously identified as L. monocytogens were characterized according to serotype, PFGE type and biofilm-forming ability. AIII.3.2 Serotyping. Five primer sets (D1, D2, GLT, FlaA and MAMA-C) were used to classify the strains into five serotype groups (1/2a(3a), 1/2b(3b), 1/2c(3c), 4b(d,e) and 4a/c) as described by Borucki and Call (2003). AIII.3.3 DNA isolation and PFGE typing. DNA isolation and PFGE analysis followed the PulseNet protocol developed by CDC (Graves and Swaminathan 2000). Two rare restriction enzymes, ApaI and AscI, were used. The PFGE types were obtained by combining both restriction enzyme patterns into one profile. The combined AscI and ApaI restriction profiles were analyzed by automated cluster analysis using the DICE coefficient (tolerance of 1.5%) and unweighted pair group method with arithmetic averages (UPGMA). A similarity score value of 95% served as the cutoff. AIII.3.4 Biofilm Formation. Biofilm formation was quantified using a standard microtiter plate assay with Modified Welshimer’s Broth inoculated at 102 CFU/ml serving as the growth medium (Djordjevic and others 2002). Assays were carried out at 22 ± 2°C for 4 d. Biofilms were stained with crystal violet which was resolubilized in glacial acetic acid. Optical densities were read at 570 nm using a spectrophotometer. After averaging the 9 optical density readings from triplicate wells for each of three replicates, each isolate was arbitrarily classified as a weak (OD < 0.50), medium (OD 0.51 – 0.70) or strong (OD > 0.70) biofilm former. 252 AIII.4 RESULTS Overall, 16, 7, 6 and 1 L. monocytogenes isolate belonged to serotypes 1/2b(3b), 1/2a(3a), 4b(d,e) and 4a/c, respectively (Table AIII.1). A total of 12 PFGE types grouped into three distinct clusters (Figure AIII.1). Fifteen weak, 8 medium and 7 strong biofilm formers were identified (Table AIII.2). The 7 strong biofilm-forming strains belonged to serotype 1/2a(3a) and to two closely related PFGE types containing a unique high kbp doublet using ApaI. These strains were only 60.9% similar compared to the remaining 23 isolates. All 7 strong biofilm formers were isolated from multiple locations over 9 months. In contrast, 4 of 15 weak and 2 of 8 medium biofilm-forming strains persisted for 10 months, thereby reinforcing previous correlations between biofilm formation and persistence. 253 AIII.5 FIGURES Figure AIII.1: L. monocytogenes PFGE patterns obtained using ApaI and AscI 254 AIII.6 TABLES Table AIII.1: L. monocytogenes serotypes Isolate D1 D2 GLT FlaA MAMA-C Division Serotype Control 1 + - - NA - I 4b Control 2 - + NA + NA II 1/2a Control 3 + - + NA NA I 1/2b Control 4 NA NA NA NA + III 4a Control 5 NA NA NA NA + III 4a 2-S25 + - - NA - I 4b(d/e) 10-S25 + - - NA - I 4b(d/e) 12-S25 + - - NA - I 4b(d/e) 13-S25 + - - NA - I 4b(d/e) 5-S32 + - - NA - I 4b(d/e) 5-S39 + - - NA - I 4b(d/e) 12-S3 + - + NA NA I 1/2b(3b) 12-S6 + - + NA NA I 1/2b(3b) 14-S8 + - + NA NA I 1/2b(3b) 12-S2 + - + NA NA I 1/2b(3b) 14-S4 + - + NA NA I 1/2b(3b) 3-S9 + - + NA NA I 1/2b(3b) 14-S9 + - + NA NA I 1/2b(3b) 12-S8 + - + NA NA I 1/2b(3b) 255 Table AIII.1 (cont’d) 12-S9 + - + NA NA I 1/2b(3b) 11-S8 + - + NA NA I 1/2b(3b) 11-S9 + - + NA NA I 1/2b(3b) 11-S19 + - + NA NA I 1/2b(3b) 11-S25 + - + NA NA I 1/2b(3b) 12-S1 + - + NA NA I 1/2b(3b) 12-S11 + - + NA NA I 1/2b(3b) 13-S9 + - + NA NA I 1/2b(3b) 10-S19 + - - NA + III 4a/c 4-S4 - + NA - NA II 1/2c(3c) 5-S16 - + NA + NA II 1/2a(3a) 10-S3 - + NA + NA II 1/2a(3a) 10-S6 - + NA + NA II 1/2a(3a) 10-S11 - + NA + NA II 1/2a(3a) 3-S8 - + NA + NA II 1/2a(3a) 13-S8 - + NA + NA II 1/2a(3a) 256 Table AIII.2: L. monocytogenes biofilm OD’s, serotypes, and PFGE types Isolate Sampling Area Sampling Time Mean OD Serotype PFGE 2-S25 Cheese case: Floor 8/2/2006 0.347 A 4b(d,e) A 10-S25 Cheese case: Floor 3/27/2007 0.325 A 4b(d,e) A 12-S25 Cheese case: Floor 6/5/2007 0.449 A 4b(d,e) A 13-S25 Cheese case: Floor 6/26/2007 0.422 A 4b(d,e) A 4b(d,e) B B Sandwich line: Slicer (Hobart) back 5-S32 plate A 11/1/2006 0.267 Deli: Slicer (Bizerba) collection A 5-S39 area 11/1/2006 0.386 4b(d,e) 12-S3 Basement: Produce walk-in floor 6/5/2007 0.574 B 1/2b(3b) C 12-S6 Basement: Drain under sink 6/5/2007 0.591 B 1/2b(3b) C 14-S8 Kitchen: Cooler floor 7/24/2007 0.580 B 1/2b(3b) C 12-S2 Basement: Catering walk-in entry 6/5/2007 0.383 A 1/2b(3b) D 14-S4 Basement: Produce walk-in entry 7/24/2007 0.513 B 1/2b(3b) E 3-S9 9/6/2006 0.605 B 1/2b(3b) F 14-S9 Kitchen: Cooler entry 7/24/2007 0.666 B 1/2b(3b) F 12-S8 Kitchen: Cooler floor 6/5/2007 0.547 B 1/2b(3b) G 12-S9 Kitchen: Cooler entry 6/5/2007 0.528 B 1/2b(3b) G 11-S8 Kitchen: Cooler floor 5/1/2007 0.350 A 1/2b(3b) H 11-S9 Kitchen: Cooler entry 5/1/2007 0.359 A 1/2b(3b) H 11-S19 Sandwich line: Floor 5/1/2007 0.365 A 1/2b(3b) H Kitchen: Cooler entry 257 Table AIII.2 (cont’d) 11-S25 Cheese case: Floor 5/1/2007 0.309 A 1/2b(3b) H 12-S1 Basement: Catering walk-in floor 6/5/2007 0.350 A 1/2b(3b) H 12-S11 Kitchen: Floor drain near cooler 6/5/2007 0.351 A 1/2b(3b) H 13-S9 Kitchen: Cooler entry 6/26/2007 0.398 A 1/2b(3b) I 10-S19 Sandwich line: Floor 3/27/2007 0.329 A 4a/c 4-S4 10/5/2006 0.883 C 1/2a(3a) K 5-S16 Sandwich line: Counter top (board) 11/1/2006 0.938 C 1/2a(3a) K 10-S3 Basement: Produce walk-in floor 3/27/2007 0.902 C 1/2a(3a) K 10-S6 Basement: Drain under sink 3/27/2007 1.015 C 1/2a(3a) K 10-S11 Kitchen: Floor drain near cooler 3/27/2007 0.991 C 1/2a(3a) K 3-S8 9/6/2006 0.793 C 1/2a(3a) L 6/26/2007 0.850 C 1/2a(3a) L Basement: Produce walk-in entry Kitchen: Cooler floor 13-S8 Kitchen: Cooler floor a J The letters A, B, and C in the Mean OD column indicate the biofilm forming ability of the isolates where A is weak, B is medium and C is strong. 258 REFERENCES 259 REFERENCES Abadias, M., J. Usall, M. Anguera, C. Solsona, and I. Viñas. 2008. Microbiological quality of fresh, minimally-processed fruit and vegetables, and sprouts from retail establishments. Int. J.. Food Microbiol. 123:121-129. Ailes, E. C., J. S. Leon, L. A. Jaykus, L. M. Johnston, H. A. Clayton, S. Blanding, D. G. Kleinbaum, L. C. Backer, and C. L. Moe. 2008. Microbial concentrations on fresh produce are affected by postharvest processing, importation, and season. J. Food Prot. 71:2389-2397. Allende, A., E. Aguayo, and F. Artes. 2004. Microbial and sensory quality of commercial fresh processed red lettuce throughout the production chain and shelf life. Int. J. Food Microbiol. 91:109-117. Anderson, M., L. Jaykus, S. Beaulieu, and S. Dennis. 2011. Pathogen-produce pair attribution risk ranking tool to prioritize fresh produce commodity and pathogen combinations for further evaluation (P3ARRT). Food Control 22:1865-1872. Annous, B. A., G. M. Sapers, A. M. Mattrazzo, and D. C. Riordan. 2001. Efficacy of washing with a commercial flatbed brush washer, using conventional and experimental washing agents, in reducing populations of Escherichia coli on artificially inoculated apples. J. Food Prot. 64:159-163. Annous, B.A., Fratamico, P.M., Smith, J.L. 2009. Quorum sensing in biofilms: Why bacteria behave the way they do. J. Food Sci. 74:R24-R37. Bell, B. P., M. Goldoft, P. M. Griffin, M. A. Davis, D. C. Gordon, P. I. Tarr, C. A. Bartleson, J. H. Lewis, T. J. Barrett, and J. G. Wells. 1994. A multistate outbreak of Escherichia coli O157:H7-associated bloody diarrhea and hemolytic uremic syndrome from hamburgers. The Washington experience. JAMA 272:1349-1353 Beuchat, L. R. 1996. Pathogenic microorganisms associated with fresh produce. J. Food Prot. 59:204-216. Beuchat, L. R. 2006. Vectors and conditions for preharvest contamination of fruits and vegetables with pathogens capable of causing enteric disease. Br. Food J. 108:38-53. Beuchat, L. R., 2002. Ecological factors influencing survival and growth of human pathogens on raw fruits and vegetables. Microb. Infect. 4:413-423. Beuchat, L. R., and J.H. Ryu. 1997. Produce handling and processing practices. Emerg. Infect. Dis. 3:459-65. 260 Beuchat, L. R., and R. E. Brackett. 1990. Survival and growth of Listeria monocytogenes on lettuce as influenced by shredding, chlorine treatment, modified atmosphere packaging and temperature. J. Food Sci. 55:755-758. Beuchat, L. R., B. B. Adler, and M. M. Lang. 2004. Efficacy of chlorine and a peroxyacetic acid sanitizer in killing Listeria monocytogenes on iceberg and Romaine lettuce using simulated commercial processing conditions. J. Food Prot. 67:1238-1242. Beuchat, L. R., B. V. Nail, B. B. Adler, and M. R. S. Clavero. 1998. Efficacy of spray application of chlorine in killing pathogenic bacteria on raw apples, tomatoes, and lettuce. J. Food Prot. 61:1305-1311. Borucki, M.K., and D.R. Call. 2003. Listeria monocytogenes serotype identification by PCR. J. Clin. Microbiol. 41: 5537-5540. Boyhan, G. E., W. C. Hurst, W. T. Kelly, G. W. Krewer, and K. C.Taylor. 2004. Postharvest handling and transportation of fruits and vegetables. Available at: http://www.fruit.cornell.edu/berry/postharvest/postharvestpdfs/FS100.pdf. Accessed 01 April 2012. Brandl, M. T. Plant Lesions Promote the Rapid Multiplication of Escherichia coli O157:H7 on Postharvest Lettuce. 2008. Appl. Environ. Microbiol. 74:5285–5289. Buchanan, R.L., S. G. Edelson, R. L. Miller, and G. M. Sapers. 1999. Contamination of intact apples after immersion in an aqueous environment containing Escherichia coli O157:H7. J. Food Prot. 62:444-450. Buchholz, A. L., G. R. Davidson, B. P. Marks, E. C. D. Todd, and E. T. Ryser. 2012a. Impact of Post-inoculation Hold Time on Escherichia coli O157:H7 Transfer during Commercial Production of Fresh-cut Leafy Greens. J. Food Prot. (In preparation). Buchholz, A. L., G. R. Davidson, B. P. Marks, E. C. D. Todd, and E. T. Ryser. 2012b. Quantitative transfer of Escherichia coli O157:H7 to equipment during small-scale production of fresh-cut leafy greens. J. Food Prot. 75: (in press). Buchholz, A. L., G. R. Davidson, B. P. Marks, E. C. D. Todd, and E. T. Ryser. 2012c. Tracking an Escherichia coli O157:H7 Contaminated Batch of Leafy Greens through a Commercial Processing Line. J. Food Prot. (In preparation). Buchholz, A. L., G. R. Davidson, B. P. Marks, E. C. D. Todd, and E. T. Ryser. 2012d. Transfer of Escherichia coli O157:H7 from Equipment Surfaces to Iceberg and Romaine Lettuce during Simulated Commercial Processing. J. Food Prot. (Submitted). Burnett, A. B., M. H. Iturriaga, E. F. Escartin, C. A. Pettigrew, and L. R. Beuchat. 2004. Influence of variations in methodology on populations of Listeria monocytogenes recovered from lettuce treated with sanitizers. J. Food Prot. 67:742-750. 261 Burnett, S. L., and L. R. Beuchat. 2001. Human pathogens associated with raw produce and unpasteurized juices, and difficulties in decontamination. J. Ind. Microbiol. Biotechnol. 27:104-110. Calvin, L. 2007. Outbreak Linked to Spinach Forces Reassessment of Food Safety Practices. Available at: http://www.ers.usda.gov/AmberWaves/June07/PDF/Spinach.pdf. Accessed 01 April 2012. Capriolo, A., S. Morabito, H. Brugere and E. Oswald. 2005. Enterohaemorrhagic Escherichia coli: emerging issues on virulence and modes of transmission. Vet. Res. 36:289–311 Carmichael, I., I. S. Harper, M. J. Coventry, P. W. J. Taylor, J. Wan, and M.W. Hickey. 1999. Bacterial colonization and biofilm development on minimally processed vegetables. J. Appl. Microbiol. 85:45S-51S. Center for Science in the Public Interest (CSPI). 2008a. CSPI Outbreak Alert Data: Info on Produce Outbreaks. Available at: http://www.cspinet.org/new/pdf/cspi_outbreak_alert.pdf. Accessed 01 April 2012. Center for Science in the Public Interest (CSPI). 2008b. Outbreak Alert! 2008. Available at: http://www.cspinet.org/new/pdf/outbreak_alert_2008_report_final.pdf. Accessed 01 April 2012. Center for Science in the Public Interest (CSPI). 2009a. Outbreak Alert! Analyzing Foodborne Outbreaks 1998 to 2007. Available at: http://www.cspinet.org/new/pdf/outbreakalertreport09.pdf. Accessed 01 April 2012. Center for Science in the Public Interest (CSPI). 2009b. The Ten Riskiest Foods Regulated by the US Food and Drug Administration. Available at: http://cspinet.org/new/pdf/cspi_top_10_fda.pdf. Accessed 01 April 2012. Centers for Disease Control and Prevention (CDC). 1996. Outbreak of Escherichia coli O157:H7 infections associated with drinking unpasteurized commercial apple juice -British Columbia, California, Colorado, and Washington, October 1996. Morb. Mortal. Wkly. Rep. 45:975. Centers for Disease Control and Prevention (CDC). 2006. Ongoing multistate outbreak of Escherichia coli serotype O157:H7 infections associated with consumption of fresh spinach—United States, September 2006. Morb Mortal. Wkly. Rep. 55:1045–1046. Centers for Disease Control and Prevention (CDC). 2011. Investigation Announcement: Multistate Outbreak of E. coli O157:H7 Infections Linked to Romaine Lettuce. Available at: http://www.cdc.gov/ecoli/2011/ecoliO157/romainelettuce/120711/index.html. Accessed 01 April 2012. 262 Chai, L. C., H. Y. Lee, F. M. Ghazali, F. Abu Bakar, P. K. Malakar, M. Nishibuchi, Y. Nakaguchi, and S. Radu. 2008. Simulation of cross-contamination and decontamination of Campylobacter jejuni during handling of contaminated raw vegetables in a domestic kitchen. J. Food Prot. 71:2448-2452. Chase-Topping, M. E., I. J. McKendrick, M. C. Pearce, P. MacDonald, L. Matthews, J. Halliday, L. Allison , D. Fenlon, J. C. Low, G. Gunn, and M. E. Woolhouse. 2007. Risk factors for the presence of high-level shedders of Escherichia coli O157 on Scottish farms. J. Clin. Microbiol. 45:1594-1603. Chen, H. 2007. Use of linear, Weibull, and log-logistic functions to model pressure inactivation of seven foodborne pathogens in milk. Food Microbiol. 24:197–204. Chen, Y, K. M. Jackson, F. P. Chea, and D. W. Schaffner. 2001. Quantification and variability analysis of bacterial cross-contamination rates in common food service tasks. J. Food Prot. 64:72-80. Cody, S. H., M. K. Glynn, J. A. Farrar, K. L. Cairns, P. M. Griffin, J. Kobayashi, M. Fyfe, R. Hoffman, A. S. King, J. H. Lewis, B. Swaminathan, R. G. Bryant, and D. J. Vugia. 1999. An outbreak of Escherichia coli O157:H7 infection from unpasteurized commercial apple juice. Ann. Intern. Med. 130:202-209. da Cruz, A. G., S. A. Cenci, and M. C. A. Maia. 2008. Microbiological hazards involved in fresh-cut lettuce processing. J. Sci. Food Agric. 88:1455–1463. Davis, H., J. P. Taylor, J. N. Perdue, G. N. Stelma Jr, J. M. Humphreys Jr, R. Rowntree 3rd, and K. D. Greene. 1988. A shigellosis outbreak traced to commercially distributed, shredded lettuce. Am. J. Epidemiol. 128:1312-1321. Djordjevic, D., M. Wiedmann, and L.A. McLandsborough. 2002. Microtiter plate assay for assessment of Listeria monocytogenes biofilm formation. Appl. Environ. Microbiol. 68: 2950- 2958. Doering, H. J., M. A. Harrison, R. A. Morrow, W. C. Hurst, and W. L. Kerr. 2009. Use of the systems approach to determine the fate of Escherichia coli O157:H7 on fresh lettuce and spinach. J. Food Prot. 72:1560-1568. Doyle, M. P., and M. C. Erickson. 2006. Closing the Door on the Fecal Coliform Assay. Microbe 1:162-163. Doyle, M. P., and M. C. Erickson. 2008. Summer meeting 2007 – the problems with fresh produce: an overview. J. Appl. Microbiol. 105:317–330. Erickson, M. C., C. C. Webb, J. C. Diaz-Perez, S. C. Phatak, J. J. Silvoy, L. Davey, A. S. Payton, J. Liao, L. Ma, and M. P. Doyle. 2010. Surface and internalized Escherichia coli 263 O157:H7 on field-grown spinach and lettuce treated with spray-contaminated irrigation water. J. Food Prot. 73:1023-1029. Erickson, M. C., C. C. Webb, J. C. Diaz-Perez, S. C. Phatak, J. J. Silvoy, L. Davey, A. S. Payton, J. Lia, L. Ma, and M. P. Doyle. 2010. Surface and internalized Escherichia coli O157:H7 on field-grown spinach and lettuce treated with spray-contaminated irrigation water. J Food Prot. 73:1023-1029. Ethelberg, S., M. Lisby, B. Böttiger, A. C. Schultz, A. Villif, T. Jensen, K. E. Olsen, F. Scheutz, C. Kjelsø, L. Müller. 2010. Outbreaks of gastroenteritis linked to lettuce, Denmark, January 2010. Euro. Surveill. 15:1-3. FAO-WHO (Food and Agriculture Organization of the United Nations and World Health Organization). 2008. Microbiological hazards in fresh fruits and vegetables. Microbiological Risk Assessment Series. Meeting Report (prepublication version). Available at: www.who.int/entity/foodsafety/publications/micro/MRA_FruitVeges.pdf. Accessed 01 April 2012. Farrell, B. L., A. B. Ronner, and A. C. Wong. 1998. Attachment of Escherichia coli O157:H7 in ground beef to meat grinders and survival after sanitation with chlorine and peroxyacetic acid. J. Food Prot. 61:817-822. FDA. 1980. Compliance Policy Guidance Manual Sec. 550.550 Maraschino Cherries (CPG 7110.16). Fett, W. F., 2000. Naturally occurring biofilms on alfalfa and other types of sprouts. J. Food Prot. 63:625-632. Flores, R. A. and M. L. Tamplin. 2002. Distribution patterns of Escherichia coli O157:H7 in ground beef produced by a laboratory-scale grinder. J. Food Prot. 65:18941902. Flores, R. A. and T. E. Stewart. 2004. Empirical distribution models for Escherichia coli O157:H7 in ground beef produced by a mid-size commercial grinder. J. Food Sci. 69:121-126. Food and Drug Administration (FDA). 2006. UPDATE: FDA Narrows Investigation of E. coli O157:H7 Outbreak at Taco Bell Restaurants. Available at: http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/2006/ucm108803.htm. Accessed 01 April 2012. Food and Drug Administration (FDA). 2007. FDA and states closer to identifying source of E. coli contamination associated with illness at Taco John’s restaurants. Available at: http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/2007/ucm108827.htm. Accessed 01 April 2012. 264 Food and Drug Administration (FDA). 2009. Guidance for Industry: Guide to Minimize Microbial Food Safety Hazards of Leafy Greens; Draft Guidance. Available at: http://www.fda.gov/Food/GuidanceComplianceRegulatoryInformation/GuidanceDocume nts/ProduceandPlanProducts/ucm174200.htm#post. Accessed 01 April 2012. Food and Drug Administration (FDA). 2010a. Fresh Express Announces Precautionary Recall of a Limited Number of Cases of Veggie Lovers Salad with an Expiration Date of August 10 Due to a Possible Health Risk. Available at: http://www.fda.gov/Safety/Recalls/ucm221943.htm. Accessed 01 April 2012. Food and Drug Administration (FDA). 2010b. Fresh Express Announces Recall of Expired Romaine Lettuce Products with Use-By Dates of July 8 to 12 and "S" in the Product Code Due to Possible Health Risk. Available at: http://www.fda.gov/Safety/Recalls/ucm219057.htm. Accessed 01 April 2012. Food and Drug Administration (FDA). 2010c. Fresh Express Recalls Romaine-based Salads with Use-by Dates of May 13-16th Due to Possible Health Risk. Available at: http://www.fda.gov/Safety/Recalls/ucm213247.htm. Accessed 01 April 2012. Food and Drug Administration (FDA). 2011. FDA Food Safety Modernization Act. Public Law 111-353, Jan. 4, 2011. Available at: http://www.gpo.gov/fdsys/pkg/PLAW111publ353/pdf/PLAW-111publ353.pdf. Accessed 01 April 2012. Francis, G. A., C. Thomas, and D. O’Beirne. 1999. Review paper: the microbiological safety of minimally processed vegetables. Int. J. Food Sci. Technol. 34: 1-22. Fravalo, P., M. J. Laisney, M. O. Gillard, G. Salvat, and M. Chemaly. 2009. Campylobacter transfer from naturally contaminated chicken thighs to cutting boards is inversely related to initial load. J. Food Prot. 72:1836-1840. Friesema, I., G. Sigmundsdottir, K. van der Zwaluw, A. Heuvelink, B. Schimmer, C. de Jager, B. Rump, H. Briem, H. Hardardottir, A. Atladottir, E. Gudmundsdottir, W. van Pelt. 2008. An international outbreak of shiga toxin-producing Escherichia coli O157 infection due to lettuce, September-October 2007. Euro. Surveill. 13:19065. Garcia-Villanova Ruiz, B. G., R. Galvez Vargas, and R. Garcia-Villanova. 1987. Contamination on fresh vegetables during cultivation and marketing. Int. J. Food Microbiol. 4:285–291. Garg, N., J. J. Churey, and D. F. Splittstoesser. 1990. Effect of processing conditions on the microflora of fresh-cut vegetables. J. Food Prot. 53:701-703. Gill, C.O., and T. Jones. 2002. Effects of wearing knitted or rubber gloves on the transfer of Escherichia coli between hands and meat. J. Food Prot. 65:1045-1048. 265 Gleeson, E., and D. O'Beirne. 2005. Effects of process severity on survival and growth of Escherichia coli and Listeria innocua on minimally processed vegetables. Food Control 16:677-685. Gobin, M., N. Launders, C. Lane, G. Kafatos, and B. Adak. 2011. National outbreak of Salmonella Java phage type 3b variant 9 infection using parallel case–control and case– case study designs, United Kingdom, July to October 2010. Euro. Surveill . 16:1-7. Gorny, J. R. 2006. Microbial contamination of fruits and vegetables, p. 3-32. In G. M. Sapers, J. R. Gorny, and A. E. Yousef (ed.), Microbiology of fruits and vegetables. CRC Press, Boca Raton, FL. Graves, L.M., and B. Swaminathan. 2000. PulseNet standardized protocol for subtyping Listeria monocytogenes by macrorestriction and pulsed-field gel electrophoresis. Int. J. Food Microbiol. 65: 55-62. Griffin P. M., and A. V. Tauxe. 1991. The epidemiology of infections caused by Escherichia coli O157:H7, other enterohemorrhagic E. coli and the associated hemolytic uremic syndrome. Epidemiol. Rev. 13:60–98. Harris, L. J., J. N. Farber, L. R. Beuchat, M. E. Parish, T. V. Suslow, E. H. Garrett, and F. F. Busta. 2003. Outbreaks associated with fresh produce: Incidence, growth, and survival of pathogens in fresh and fresh-cut produce. Compr. Rev. Food Sci. Food Safety. 2:78– 141. Herman, K. M., T. L. Ayers, and M. Lynch. 2008. Foodborne Disease Outbreaks Associated with Leafy Greens, 1973-2006. International Conference on Emerging Infectious Diseases 2008 Slide Sessions and Poster Abstracts. Available at: http://www.cdc.gov/eid/content/14/3/ICEID2008.pdf. Accessed 01 April 2012. Hill, V. R., A. L. Polaczyk, D. Hahn, J. Narayanan, T. L. Cromeans, J. M. Roberts and J. E. Amburgey. 2005. Development of a rapid method for simultaneous recovery of diverse microbes in drinking water by ultrafiltration with sodium polyphosphate and surfactants. Appl. Environ. Microbiol. 71:6878-6884. Holvoet, K., L. Jacxsens, I. Sampers, and M. Uyttendaele. 2012. Insight in prevalence and distribution of microbial contamination to evaluate water management in fresh produce processing industry. J. Food Prot. 75:671-681. Ibekwe, A. M., P. M. Watt, P. J. Shouse, and C. M. Grieve. 2004. Fate of Escherichia coli O157:H7 in irrigation water on soils and plants as validated by culture method and real-time PCR. Can. J. Microbiol. 50:1007-1014. Illinois Department of Public Health (IDPH). 2010. Update - Salmonella Illnesses in Illinois. Available at: 266 http://www.idph.state.il.us/public/press10/6.18.10Salmonella_Update.htm. Accessed 01 April 2012. Jackson, D. W., K. Suzuki, L. Oakford, J. W. Simecka, M. E. Hart, and T. Romeo. 2002. Biofilm formation and dispersal under the influence of the global regulator CsrA of Escherichia coli. J. Bacteriol. 184:290-301. Jay, M. T., M. Cooley, D. Carychao, G. W. Wiscomb, R. A. Sweitzer, L. CrawfordMiksza, J. A. Farrar, D. K. Lau, J. O'Connell, A. Millington, R. V. Asmundson, E. R. Atwill, and R.E. Mandrell. 2007. Escherichia coli O157:H7 in feral swine near spinach fields and cattle, central California coast. Emerg. Infect. Dis. 13:1908-1911. Jiménez, S. M., M. C. Tiburzi, M. S. Salsi, M. A. Moguilevsky, and M. E. Pirovani. 2009. Survival of Salmonella on refrigerated chicken carcasses and subsequent transfer to cutting board. Lett. Appl. Microbiol. 48:687-691. Johnston, L. M., L-A. Jaykus, D. Moll, M. C. Martinez, J. Anciso, B. Mora, and C. L. Moe. 2005. A Field Study of the Microbiological Quality of Fresh Produce. J. Food Prot. 68:1840–1847 Kaneko, K., H. Hayashidani, K. Takahashi, Y. Shiraki, S. Limawongpranee, and M. Ogawa. 1999. Bacterial contamination in the environment of food factories processing ready-to-eat fresh vegetables. J. Food Prot. 62:800-804. Kaper, J. B., J. P. Nataro, and H. L. Mobley. 2004. Pathogenic Escherichia coli. Nat. Rev. Microbiol. 2:123-140. Karmali, M. A., M. Petric, B. T. Steele, and C. Lim. 1983. Sporadic cases of haemolyticuraemic syndrome associated with faecal cytotoxin and cytotoxin producing Escherichia coli in stools. Lancet. 321:619-620. Kays, J. S. 1997. Heat, heat transfer, and cooling, p. 457–507. In Postharvest physiology of perishable plant products. Exon Press, Athens, Ga. Keene, W. E., K. Hedberg, D. E. Herriott, D. D. Hancock, R. W. McKay, T. J. Barrett, and D W. Fleming. 1997. A prolonged outbreak of Escherichia coli O157:H7 infections caused by commercially distributed raw milk. J. Infect. Dis. 176:815-818. Keskinen, L. A., A. Burke and B. A. Annous. 2009. Efficacy of chlorine, acidic electrolyzed water and aqueous chlorine dioxide solutions to decontaminate Escherichia coli O157:H7 from lettuce leaves. Int. J. Food Microbiol. 132:134-140. Keskinen, L. A., A. M. Burke, and B. A. Annous. 2009. Efficacy of chlorine, acidic electrolyzed water and aqueous chlorine dioxide solutions to decontaminate Escherichia coli O157:H7 from lettuce leaves. Int. J. Food Microbiol. 132:134-140. 267 Keskinen, L. A., E. C. Todd, and E. T. Ryser. 2008a. Impact of bacterial stress and biofilm-forming ability on transfer of surface-dried Listeria monocytogenes during slicing of delicatessen meats. Int.J. Food Microbiol. 127:298-304. Keskinen, L. A., E. C. Todd, and E. T. Ryser. 2008b. Transfer of surface-dried Listeria monocytogenes from stainless steel knife blades to roast turkey breast. J. Food Prot. 71:176-181. Kim, J. K. and M. A. Harrison. 2008. Transfer of Escherichia coli O157:H7 to romaine lettuce due to contact water from melting ice. J. Food Prot. 71:252-256. King, A. D., J. A. Magnusson, T. Torok, and N. Goodman. 1991. Microbial flora and storage quality of partially processed lettuce. J. Food Sci. 56:459–461. Koike, S. T. and M. Cahn. 2011. Spinach Production in California. Available at: http://anrcatalog.ucdavis.edu/pdf/7212.pdf. Accessed 01 April 2012. Kusumaningrum, H. D., G. Riboldi, W. C. Hazeleger, and R.R. Beumer. 2003. Survival of foodborne pathogens on stainless steel surfaces and cross-contamination to foods. Int. J. Food Microbiol. 85:227-236. Kyle, J. L., C. T. Parker, D. Goudeau, and M. T. Brandl. 2010. Transcriptome Analysis of Escherichia coli O157:H7 Exposed to Lysates of Lettuce Leaves. Appl. Environ. Microbiol. 76: 1375–1387. Lang, M. M., L. J. Harris, and L. R. Beuchat. 2004. Survival and recovery of Escherichia coli O157:H7, Salmonella, and Listeria monocytogenes on lettuce and parsley as affected by method of inoculation, time between inoculation and analysis, and treatment with chlorinated water. J. Food Prot. 67:1092-1103. Li, H., M. Tajkarimi, and B. I. Osburn. 2008. Impact of Vacuum Cooling on Escherichia coli O157:H7 Infiltration into Lettuce Tissue. Appl. Environ. Microbiol. 74:3138–3142. Lin, C. M., K. Takeuchi, L. Zhang, C. B. Dohm, J. D. Meyer, P.A. Hall, and M.P. Doyle. 2006, Cross-contamination between processing equipment and deli meats by Listeria monocytogenes. J. Food Prot. 69:71-79. López-Gálvez, F., A. Allende, M. V. Selma, and M. I. Gil. 2009. Prevention of Escherichia coli cross-contamination by different commercial sanitizers during washing of fresh-cut lettuce. Int. J Food Microbiol. 133:167–171. López-Gálvez, F., M. I. Gil, P. Truchado, M. V. Selma, and A. Allende. 2010. Crosscontamination of fresh-cut lettuce after a short-term exposure during pre-washing cannot be controlled after subsequent washing with chlorine dioxide or sodium hypochlorite. Food Microbiol. 27:199–204. 268 Low, J. C., I. J. McKendrick, C. McKechnie, D. Fenlon, S. W. Naylor, C. Currie, D. G. Smith, L. Allison, and D. L. Gally. 2005. Rectal carriage of enterohemorrhagic Escherichia coli O157 in slaughtered cattle. Appl. Environ. Microbiol. 71:93-97. Ma, L., G.D. Zhang, and M.P. Doyle. 2011. Green fluorescent protein labelling of Listeria, Salmonella, and Escherichia coli O157:H7 for safety-related studies. PLoS One. 4:e18083. Mafart, P., O. Couvert, S. Gaillard, and I. Leguerinel. 2002. On calculating sterility in thermal preservation methods: application of the Weibull frequency distribution model. Int. J. Food Microbiol. 72:107–113. Mandrell, R. E. 2009. Chapter 1. Enteric Human Pathogens Associated with Fresh Produce: Sources, Transport, and Ecology, p. 5-42. In X. Fan, B. A. Niemira, C. J. Doona, F. E. Feeherry and R. B. Gravani (ed.), Microbial Safety of Fresh Produce. Wiley-Blackwell, Oxford, UK. Manuel González, M., P. N. Skandamis, and M. Hänninen. 2009. A modified Weibull model for describing the survival of Campylobacter jejuni in minced chicken meat. Int. J. Food Microbiol. 136:52–58. Manvell, P. M. and M. R. Ackland. 1986. Rapid detection of microbial growth in vegetable salads at chill and abuse temperatures. Food Microbiol. 3:59 65. Meng, J., M. P. Doyle, T. Zhao and S. Zhao. 2007. Enterohemorrhagic Escherichia coli. p. 260. In M. P. Doyle and L. R. Beuchat (ed.), Food Microbiology: Fundamentals and Frontiers, 3rd Ed. ASM Press. Washington, D.C. Montville, R, D. W. Schaffner. 2003. Inoculum size influences bacterial cross contamination between surfaces. Appl. Environ. Microbiol. 69:7188-7193. Montville, R., Y. Chen, and D. W. Schaffner. 2001. Glove barriers to bacterial crosscontamination between hands to food. J. Food Prot. 64:845-849. Moore, C. M., B. W. Sheldon, and L. A. Jaykus. 2003. Transfer of Salmonella and Campylobacter from stainless steel to Romaine lettuce. J. Food Prot. 66:2231-2236. Morales-Morales, H. A., G. Vidal, J. Olszewski, C. M. Rock, D. Dasgupta, K. H. Oshima and G. B. Smith. 2003. Optimization of a reusable hollow-fiber ultrafilter for simultaneous concentration of enteric bacteria, protozoa, and viruses from water. Appl. Environ. Microbiol. 69:4098-4102. Morris, C. E., J. Monier and M. Jacques. 1997. Methods for observing microbial biofilms directly on leaf surfaces and recovering them for isolation of culturable microorganisms. Appl. Environ. Microbiol. 63:1570-1576. 269 Mukherjee. A., D. Speh, E. Dyck, and F. Diez-Gonzalez. 2004. Preharvest evaluation of coliforms, Escherichia coli, Salmonella, and Escherichia coli O157:H7 in organic and conventional produce grown by Minnesota farmers. J. Food Prot. 67:894-900. Niemira, B. A. and P. H. Cooke. 2010. Escherichia coli O157:H7 Biofilm formation on Romaine lettuce and spinach leaf surfaces reduces efficacy of irradiation and sodium hypochlorite Washes. J. Food Sci. 75: 270-277. Nordlee, J.A., L.B. Martin, and S.L. Taylor. 1985. Sulfite residues in maraschino cherries. J. Food Sci. 50:256-257. Nou, X. and Y. Luo. 2010. Whole-leaf wash improves chlorine efficacy for microbial reduction and prevents pathogen cross-contamination during fresh-cut lettuce processing. J. Food Sci. 75:283-290. Paton, J. C. and A. W. Paton. Pathogenesis and Diagnosis of Shiga Toxin-Producing Escherichia coli Infections. 1998. Clin. Microbiol. Rev. 11:450-479. Peleg, M., and M.B. Cole. 1998. Reinterpretation of microbial survival curves. Crit. Rev. Food Sci., 38:353–380. Pérez Rodríguez, F., D. Campos, E.T. Ryser, A.L. Buchholz, B.P. Marks, Z, Gonzalo, and E.C.D. Todd. 2011. A mathematical risk model for Escherichia coli O157:H7 crosscontamination of lettuce during processing. Food Microbiol. 28:694-701. Rangaranjan, A., M. Pritts, S. Reiners, and L. Pedersen. 2000. Reduce Microbial Contamination with Good Agricultural Practices. Available at: http://www.gaps.cornell.edu/Educationalmaterials/Samples/PamphletEng.pdf. Accessed 01 April 2012. Ravishankar, S., L. Zhu, and D. Jaroni. 2010. Assessing the cross contamination and transfer rates of Salmonella enterica from chicken to lettuce under different foodhandling scenarios. Food Microbiol. 7:791-794. Riley, L. W., R. S. Remis, S. D. Helgerson, H. B. McGee, J. G. Wells, B. R. Davis, R. J. Hebert, E. S. Olcott, L. M. Johnson, N. T. Hargrett, P. A. Blake, and M. L. Cohen. 1983. Hemorrhagic colitis associated with a rare Escherichia coli serotype. N. Engl. J. Med. 308:681-685. Ryu, J., and L. R. Beuchat. 2005. Biofilm formation by Escherichia coli O157:H7 on stainless steel: Effect of exopolysaccharide and curli production on its resistance to chlorine. Appl. Environ. Microbiol. 71:247–254. Ryu, J. H., H. Kim, and L. R. Beuchat. 2004. Attachment and biofilm formation by Escherichia coli O157:H7 on stainless steel as influenced by exopolysaccharide production, nutrient availability, and temperature. J. Food Prot. 67:2123-2131. 270 Sapers, G.M. 2001. Efficacy of washing and sanitizing methods for disinfection of fresh fruit and vegetable products. Food Technol. Biotechnol. 39:305-311. Sapers, G.M. 2006. Washing and sanitizing treatments for fruits and vegetables, p. 375400. In G. M. Sapers, J. R. Gorny, and A. E. Yousef (ed.), Microbiology of fruits and vegetables. CRC Press, Boca Raton, FL. Scott, E., and S. F. Bloomfield. 1990. The survival and transfer of microbial contamination via cloths, hands and utensils. J. Appl. Bacteriol. 68:271-278. Seo, K. H, and J. F. Frank. 1999. Attachment of Escherichia coli O157:H7 to lettuce leaf surface and bacterial viability in response to chlorine treatment as demonstrated by using confocal scanning laser microscopy. J. Food Prot. 62:3-9. Shaw, R. K., C. N. Berger, B. Feys, S. Knutton, M. J. Pallen, and G. Frankel. 2008. Enterohemorrhagic Escherichia coli exploits EspA filaments for attachment to salad leaves. Appl. Environ. Microbiol. 74:2908-2914 Sheen, S., and C. A. Hwang. 2008. Modeling transfer of Listeria monocytogenes from slicer to deli meat during mechanical slicing. Foodborne Pathog. Dis. 5:135-146. Shepherd, M. W., J. Kim, X. Jiang, M. P. Doyle, and M. C. Erickson. Evaluation of Physical Coverings Used To Control Escherichia coli O157:H7 at the Compost Heap Surface. Appl. Environ. Microbiol. 77:5044-5049 Shepherd, M.W., J. Kim, X.P. Jiang, M.P. Doyle, and M.C. Erickson. 2011. Evaluation of physical coverings used to control Escherichia coli O157:H7 at the compost heap surface. Appl. Environ. Microbiol. 77:5044-5049. Solomon, E. B., C. J. Potenski, and K. R. Matthews. 2002. Effect of irrigation method on transmission to and persistence of Escherichia coli O157:H7 on lettuce. J. Food Prot. 65:673-676. Stafford, R. J, B. J. McCall, A. S. Neill, D. S. Leon, G. J. Dorricott, C. D. Towner, and G. R. Micalizzi. 2002. A statewide outbreak of Salmonella Bovismorbificans phage type 32 infection in Queensland. Comm. Dis. Intell. 26:568-573. Sun-Times Media LLC. 2011. Salad suspected in salmonella outbreak. Available at: http://couriernews.suntimes.com/news/5549282-418/salad-suspected-in-salmonellaoutbreak.html. Accessed 01 April 2012. Swerdlow, D. L., B. A. Woodruff, R. C. Brady, P. M. Griffin, S. Tippen, H. D. Donnell, E. Geldreich, B. J. Payne, A. Meyer, J. G. Wells, K. D. Greene, M. Bright, N. H. Bean, and P. A. Blake. 1992. A waterborne outbreak in Missouri of Escherichia coli O157:H7 associated with bloody diarrhea and death. Ann. Intern. Med. 117:812-819. 271 Takeuchi, K., and J. F. Frank. 2000. Penetration of Escherichia coli O157:H7 into lettuce tissues as affected by inoculum size and temperature and the effect of chlorine treatment on cell viability. J. Food Prot. 63:434-440. Taormina, P. J., L. R. Beuchat, M. C. Erickson, L. Ma, G. Zhang, and M. P. Doyle. 2009. Transfer of Escherichia coli O157:H7 to iceberg lettuce via simulated field coring. J. Food Prot. 72:465-472. Tarr, P. I., C. A. Gordon, W. L. Chandler. 2005. Shiga-toxin-producing Escherichia coli and haemolytic uraemic syndrome. Lancet. 365:1073-1086. Tarver, T. 2009. Biofilms: A threat to food safety. Food Technol. 63:46-52. Thunberg, R. L., T. T. Tran, R. W. Bennett, R. N. Matthews, and N. Belay. 2002. Microbial evaluation of selected fresh produce obtained at retail markets. J. Food Prot. 65:677-682. Tomás-Callejas, A., G. López-Velasco, A. B. Camacho, Artés, F. Artés-Hernández, and T. V. Suslow. 2011. Survival and distribution of Escherichia coli on diverse fresh-cut baby leafy greens under preharvest through postharvest conditions. Int.J Food Microbiol. 151:2216–2224. Turini, T. 2011. Iceberg Lettuce Production in California. Available at: http://anrcatalog.ucdavis.edu/pdf/7215.pdf. Accessed 01 April 2012. U.S. Department of Health and Human Services and U.S. Department of Agriculture (HHS/USDA). 2005. Dietary Guidelines for Americans. 6th Ed. U.S. Government Printing Office. Washington, DC. U.S. Environmental Protection Agency (EPA). 2009. Drinking Water Contaminants. Available at: http://www.epa.gov/safewater/contaminants/index.html. Accessed 01 April 2012. Uhlich, G. A, D. P. Rogers, and D. A. Mosier. 2010. Escherichia coli serotype O157:H7 retention on solid surfaces and peroxide resistance is enhanced by dual-strain biofilm formation. Foodborne Pathog Dis. 7:935-943. Vorst, K. L, E. C. D Todd, and E. T. Ryser. 2006. Transfer of Listeria monocytogenes during mechanical slicing of turkey breast, bologna, and salami. J. Food Prot. 69:619626. Wachtel, M. R., and A. O. Charkowski. 2002. Cross-contamination of lettuce with Escherichia coli O157:H7. J. Food Prot. 65:465-470 272 Wachtel, M. R., L. C. Whitehand, and R. E. Mandrell. 2002. Association of Escherichia coli O157:H7 with preharvest leaf lettuce on exposure to contaminated irrigation water. J. Food Prot. 65:18–25. Weissinger, W. R., W. Chantarapanont, and L. R. Beuchat. 2000. Survival and growth of Salmonella baildon in shredded lettuce and diced tomatoes, and effectiveness of chlorinated water as a sanitizer. Int. J. Food Microbiol. 62:123-131. Williams, A.P., H. Gordon, D.L. Jones, N.J. C. Strachan, L.M. Avery, and K. Killham. 2008. Leaching of bioluminescent Escherichia coli O157:H7 from sheep and cattle faeces during simulated rainstorm events. J. Appl. Microbiol. 105:1452-1460. Xicohtencatl-Cortes, J., E. Sánchez Chacón, Z. Saldaña, E. Freer, and J. A. Girón. 2009. Interaction of Escherichia coli O157:H7 with leafy green produce. J. Food Prot. 72:1531-1537. Yan, Z., A.L. Buchholz, L. Zhang, and E.T. Ryser. 2008. Prevalence, persistence, and spread of Listeria spp. in a commercial delicatessen. Abst. P5-72, Ann. Mtg. Intern. Assoc. Food Prot., Columbus, OH. Zhang, G, L. Ma, V. H. Phelan and M. P. Doyle. 2009. Efficacy of antimicrobial agents in lettuce leaf processing water for control of Escherichia coli O157:H7. J. Food Prot. 72:1392-1397. Zhang, S. and J. M. Farber. 1996. The effects of various disinfectants against Listeria monocytogenes on fresh-cut vegetables. Food Microbiol. 13:311–321. 273